13.9.1 Diabetes 2464

13.9.1 Diabetes 2464

CONTENTS 13.9.1 Diabetes  2464 Colin Dayan and Julia Platts 13.9.2 Hypoglycaemia  2531 Mark Evans and Ben Challis 13.9.1  Diabetes Colin Dayan and Julia Platts ESSENTIALS Diabetes mellitus can be defined as a state of chronic hyperglycaemia sufficient to cause long-​term damage to specific tissues, notably the retina, kidney, nerves, and arteries. It is due to inadequate produc- tion of insulin and/​or ‘resistance’ to the glucose lowering and other actions of insulin, and is a significant and growing threat to global health, affecting more than 400 million people worldwide. Definitions—​normal fasting blood glucose concentration is in the range 3.5 to 5.5 mmol/​litre, and even large carbohydrate loads do not raise the concentration above 8 mmol/​litre. Widely accepted diag- nostic criteria for diabetes and other hyperglycaemic states are (1) dia- betes mellitus—​fasting glucose more than 7.0 mmol/​litre (126 mg/​dl) and/​or a value exceeding 11.1 mmol/​litre (199 mg/​dl), either at 2 h during a 75-​g oral glucose tolerance test or in a random sample; or an HbA1c value in a standardized assay of more than 48 mmol/​mol; (2) impaired glucose tolerance—​2-​h oral glucose tolerance test value between 7.8 and 11.1 mmol/​litre (140–​199 mg/​dl); (3) impaired fasting glucose—​fasting glucose 5.6 to 6.9 mmol/​litre (100–​125 mg/​dl). Impaired glucose tolerance is a not a stable state: within 5 years, about 25% of subjects deteriorate into type 2 diabetes, while a further 25% revert to normoglycaemia. Type 1 diabetes This condition, previously referred to as ‘juvenile-​onset’ or ‘insulin-​ dependent’ diabetes, most commonly develops in childhood, with highest incidence in northern European countries, and accounts for 5 to 15% of all cases of diabetes. Aetiology—​type 1 diabetes is caused by an autoimmune, pre- dominantly T-​cell-​mediated process that selectively destroys the pancreatic β cells. Genetic factors explain 30 to 40% of total sus- ceptibility: at least 59 loci are involved, with the HLA class II locus IDDM1 having by far the greatest effect. Environmental factors that have been implicated but not confirmed include viral infection (par- ticularly coxsackie B), bovine serum albumin from cow’s milk (by immunological cross-​reactivity), and other toxins. Notable β-​cell selective autoantibodies that are commonly found are those that recognize GAD65 (a heat shock protein), IA-​2 (a protein tyrosine phosphatase-​like molecule), ZnT8 (a zinc transporter molecule) and insulin itself, but these are clearly not the immediate cause of the disease. Several years of progressive autoimmune damage usually precede the clinical onset of diabetes. Pathogenesis—​in untreated type 1 diabetes at diagnosis, insulin concentrations are generally 10 to 50% of non​diabetic levels in the face of hyperglycaemia which would normally greatly increase insulin secretion. Such severe deficiency cannot sustain the normal anabolic effects of insulin and leads to runaway catabolism in carbohydrate, fat, and protein metabolism. A similar clinical picture of insulin de- pendence can be caused by other forms of severe pancreatic damage. Clinical features—​classical presentation of untreated or poorly controlled type 1 diabetes is with onset over days or a few weeks of polyuria (caused by osmotic diuresis due to hyperglycaemia), thirst, weight loss, and general tiredness/​malaise. Other features can include blurred vision (due to hyperglycaemia-​related refractive changes in the lens), infection (particularly genital candidiasis), and diabetic keto- acidosis. Chronic diabetic complications are not seen at presentation. Type 2 diabetes Type 2 diabetes (previously referred to as ‘non-​insulin-​dependent’ or ‘maturity-​onset’) is a heterogeneous condition, diagnosed em- pirically by the absence of features suggesting type 1 diabetes. It is most commonly diagnosed in those over 40 years of age, with peak incidence between 45 and 64 years, but it is increasingly being diag- nosed in children. Type 2 diabetes accounts for 85 to 90% of diabetes worldwide, but with striking geographical variation (prevalence <1% in rural China, 50% in Pima Indians of New Mexico). Currently more than three-​quarters of people with diabetes live in low-​ or middle-​ income countries. 13.9 Diabetes and hypoglycaemia

13.9.1  Diabetes 2465 Aetiology—​type 2 diabetes is due to the combination of insulin re- sistance and β-​cell failure. Genetic factors explain 60 to 90% of total susceptibility, with a polygenic pattern reflecting the inheritance of a critical mass of minor diabetogenic polymorphisms in genes that in- fluence insulin secretion, insulin resistance, pancreatic development, and obesity. An important specific risk factor for type 2 diabetes, which aggravates insulin resistance, is obesity—​particularly if this develops after the early twenties, and especially around the waist. Increasing rates of obesity and sedentary activity has led to a very marked (2–​3-​fold) increase in incidence of type 2 diabetes world- wide, especially in developing countries. The mechanism of β-​cell failure in human type 2 diabetes is not known. Clinical features—​in type 2 diabetes significant hyperglycaemia may have been present for several years at the time of diagnosis, hence cases are often discovered by screening or at routine health checks. Many cases present with classical symptoms of osmotic diuresis, blurred vision, and genital candidiasis. The hyperosmolar non​ketotic state can present with confusion or coma, but diabetic ketoacidosis is rare. Chronic diabetic complications may be a presenting feature. Monogenic and other types of diabetes Maturity-​onset diabetes of the young (MODY, now referred to as monogenic diabetes)—​is most often caused by mutations in the genes for glucokinase (MODY2) and HNF-​1α (MODY3). This diag- nosis should be considered if there is a family history of young-​ onset diabetes in more than one generation, with at least one family member diagnosed under the age of 25; affected members are not markedly obese; there is no evidence of insulin resistance; fasting C-​peptide is detectable and within the normal range; islet cell or anti-​GAD autoantibodies are absent; other associated features may be present (e.g. renal cysts in HNF-​1β, deafness in mitochondrial genetic diabetes). The likelihood of monogenic diabetes can be pre- dicted from an online risk calculator (https://​www.diabetesgenes. org). Individuals are often mistakenly treated with insulin when they would benefit from other treatments such as sulfonylureas (HNF-​1α) or do not require treatment (glucokinase). Other types of diabetes include those related to pancreatic disease (chronic pancreatitis, cystic fibrosis, haemochromatosis) and gesta- tional diabetes. Management of diabetes General aspects—​management requires tackling cardiovascular risk factors and obesity in addition to hyperglycaemia. Important issues include (1)  dietary modification—​reducing total energy in- take in patients who are overweight (body mass index >28 kg/​m2), improving dietary composition (fat <30% total energy intake, with saturated animal fat <10%; carbohydrates—​preferably pulses, root/​ leaf vegetables and fruit—​>55% total energy intake; sodium <6 g/​day); (2) increasing physical activity; (3) smoking cessation; and in some patients (4) antiobesity drugs and/​or bariatric surgery. Glucose-​lowering drugs—​these include (1)  insulin—​soluble (regular, or short-​acting) insulin injected subcutaneously begins to lower glucose within 30 min, has a peak effect between 1 and 2 h and lasts 3 to 5 h; long-​acting preparations (e.g. isophane and lente insulins) are used to cover basal insulin requirements; (2) in- sulin analogues—​have improved physicochemical characteristics for subcutaneous absorption and can be fast acting (e.g. insulin lispro and insulin aspart), or long acting (e.g. insulin glargine (Lantus), insulin detemir (Levemir) insulin degludec (Tresiba)); (3) oral hypo- glycaemic agents—​(a) sulphonylureas and meglitinides—​insulin secretagogues; (b) metformin—​a biguanide that acts primarily by inhibiting gluconeogenesis in the liver; (c) thiazolidinediones—​act to improve insulin sensitivity; (d) α-​glucosidase inhibitors—​partly block digestion of complex carbohydrates and so damp postprandial gly- caemic rises, but are of low efficacy and poorly tolerated; (e) incretin mimetics—​augment insulin secretion—​GLP-​1 analogues or inhibi- tors of the enzyme that breakdown GLP-​1, dipeptidyl peptidase (DPP IV)—​gliptins (f) renal glucose transporter (SGLT2) inhibitors—​ gliflozins—​that result in renal glucose wasting. Type 1 diabetes—​patients must be given insulin immediately and for life. Standard treatment involves giving a short-​acting insulin 20 to 30 min before eating or a fast-​acting insulin immediately before eating, and a twice (sometimes once) daily dose of a long-​acting insulin. Common practice is to commence with low dosages of long-​acting insulin (e.g. 8–​12 U in the morning and 4–​6 U at night), with short/​fast-​acting insulin then added to cover excessive pran- dial hyperglycaemia. Premixed insulins (e.g. 30% short-​acting with 70% long-​acting) can be given twice daily and are more convenient than giving short-​ and long-​acting insulins separately, but they lack flexibility. Administration is usually by conventional syringes or pen injection devices, but pumps can be used to administer continuous subcutaneous infusions of insulin with greatly enhanced flexibility. Type 2 diabetes—​remission of type 2 diabetes can be achieved by significant weight loss (e.g. >15 kg) and first line management is by appropriate diet and increased exercise. The first-​line oral hypogly- caemic agent for so-​called ‘dietary failure’ is metformin, with a (usually) sulphonylurea or (sometimes) thiazolidinedione added as second-​line treatment. A once-​daily dose of a long-​acting insulin can be combined effectively with metformin. Insulin therapy can range from once-​ or twice-​daily long-​acting insulin in subjects with residual insulin, to the more intensified basal and prandial regimens used in type 1 diabetes (>200 U/​day may be required in very obese, insulin-​resistant patients). However, DPPIV inhibitors (glitazones), GLP-1 analogues and SGLT2 inhibitors are increasingly being used second-​line as an alternative to insulin therapy to avoid weight gain and hypoglycaemia. Treatment targets for blood glucose—​these have been selected to reduce the risk of chronic diabetic complications. Avoiding acute episodes of hyper-​ and hypoglycaemia is also important. In most patients the treatment goal should be to reduce HbA1c to <53 mmol/mol (<7.0%), but a target of 42–48 mmol/litre (6.0–6.5%) can be employed in selected patients (e.g. short disease duration, long life expectany) if achievable without significant hypoglycaemia, and a target of <64 mmol/mol (<8.0%) may be appropriate for some (e.g. limited life expectancy). Multidisciplinary care—​diabetes is best managed by the combined efforts of a well-​trained primary care team and a team of special- ists with complementary and overlapping skills: physician, specialist diabetes nurse, dietitian, and chiropodist. Patients require education about diabetes, with key elements including (1) causes of hypergly- caemia and diabetic symptoms; (2) own treatment—​diet and lifestyle; drawing up and injecting insulin; oral agents; recognizing and treating hypoglycaemia ‘hypos’; (3) self-​monitoring technique—​targets and danger levels; how to respond to poor control; (4) ‘sick-​day’ rules—​ monitoring during intercurrent illness; how to adjust own treatment; when and how to call for help (never stop taking your insulin; check

section 13  Endocrine disorders 2466 your blood glucose every 4 h; test your urine for ketones; call for help if you start vomiting, have glucose over 15 mmol/​litre that does not come down after insulin, get hypos, get ketones in the urine, are wor- ried, and do not know what to do). Acute metabolic complications of diabetes Diabetic ketoacidosis—​uncontrolled hyperglycaemia with hyper­ ketonaemia severe enough to cause metabolic acidosis. Precipitating factors include new presentation of type 1 diabetes, omission or underdosing of insulin by patients known to have type 1 diabetes, the use of SGLT2 inhibitors and intercurrent illness (compounded by failure to monitor blood glucose and take appropriate action). Usual presentation is with classical hyperglycaemic symptoms together with acidotic (Kussmaul) breathing and ketotic foetor, vomiting, evidence of dehydration and hypovolaemia, and signs of any precipitating con- dition. Drowsiness and coma are late features. Diagnosis is confirmed with a finger-​prick blood glucose measurement and urine or blood analysis for ketones: other investigations should include a biochem- ical screen, full sepsis screen, arterial blood gas analysis, and ECG. Management requires (1) fluid replacement—​usually with 0.9% saline
(typically 1–​2 litres in 2 h, then 1 litre in 4 h, then 4 litres in next 24 h); (2) potassium replacement—​typically 40 mmol of KCl to each litre of intravenous fluid if K+ is normal (3.5–​5.0 mmol/​litre), but adjusted in response to frequent monitoring; (3) intravenous insulin—​initially
at a rate of 6 U/​h or 0.1 U/​kg, and continued (if necessary in combin- ation with 10% dextrose infusion) until ketosis has resolved (blood ketones <0.6  mmol/​litre); (4)  treatment, when possible, of any precipitating condition. Intravenous fluids and insulin can be discon- tinued when the patient can eat and drink, and they can be restarted on their usual insulin regimen (or a typical maintenance regimen can be introduced). Hyperosmolar Hyperglycaemic (formerly known as non​ketotic) state—​is distinguished from diabetic ketoacidosis by the absence (be- cause circulating insulin levels are high enough to suppress lipolysis and ketogenesis) of marked hyperketonaemia and metabolic acid- osis. Presentation is typically with classical hyperglycaemic symp- toms; confusion, drowsiness, and coma are more common than in diabetic ketoacidosis. Typical biochemical features include severe hyperglycaemia (>30 mmol/​litre) and hypernatraemia (sodium often

155 mmol/​litre). Management is largely as for diabetic ketoacidosis, excepting that (1) 0.45% saline is often given if plasma sodium is over 150 mmol/​litre or osmolality over 350 mosmol/​kg; (2) intravenous insulin infusion at low doses (0.05 IU/​kg/​hr) rapidly controls hyper- glycaemia in most cases—​indeed recent guidelines suggest initially treating with aggressive fluid replacement alone (3–​6 litres in the first 12 hours), and only introducing insulin if the glucose if falling at less than 5 mmol/​litre per hour; (3) the risk of thrombotic events is particularly high, hence prophylactic doses of low molecular weight heparin should be given. Hypoglycaemia—​an inevitable side effect of antidiabetic drugs that raise circulating insulin levels. Typical features include (1) auto- nomic symptoms—​pallor, sweating, tremor, and tachycardia, and (2) symptoms of neuroglycopenia—​commonly drowsiness, confu- sion, incoordination, and dysarthria, but also automatic or disinhib- ited behaviour and focal neurological deficits. Diagnosis is confirmed with a finger-​prick blood glucose measurement below 3.5 mmol/​litre in an appropriate clinical context. Treatment is with (1) oral glucose or sucrose or other carbohydrate—​if the patient can swallow safely; or (2) intravenous glucose (15–​20 g as 10% or 50% solution) or intra- muscular glucagon (1 mg)—​if the patient is not able to swallow safely. Chronic complications of diabetes Long-​term tissue damage is the major burden of diabetes, the greatest source of fear for people living with diabetes, and the most expensive item in the diabetes healthcare budget. Pathogenesis—​possible mechanisms for diabetic complications include glycation of proteins and macromolecules, overactivity of the polyol pathway, activation of protein kinase C and abnormal microvascular blood flow. Microvascular complications—​retinopathy, neuropathy, and nephropathy—​are specific to diabetes and reflect damage inflicted on the microcirculation throughout the body. Macrovascular disease is atherosclerosis, which behaves more ag- gressively than in non​diabetic people, and causes typical coronary heart disease, stroke, and peripheral arterial disease. Diabetic eye disease—​is the commonest cause of blindness in people of working age in most Westernized countries, although blind- ness rates are falling following the introduction of retinal screening programmes. Stages of diabetic retinopathy are (1)  background—​ microaneurysms, hard exudates, haemorrhages (flame, dot, blot), cotton wool spots (<5); (2)  preproliferative—​rapid increase in microaneurysms, intraretinal microvascular abnormalities, multiple deep haemorrhages, cotton wool spots (>5), venous beading/​loops/​ duplication; (3) proliferative—​new vessels on the disc or elsewhere, fibrous proliferation on the disc or elsewhere, preretinal or vitreous haemorrhages; (4) advanced eye disease—​retinal detachment, retinal tears, rubeosis iridis, neovascular glaucoma. Disease of the macula (maculopathy), serious enough to affect central vision, can accom- pany any stage of diabetic retinopathy including background, and may be present in newly diagnosed type 2 patients. Management requires (1) general preventive measures—​tight glycaemic control, control of hypertension, stopping smoking, regular (annual) eye screening; and (2) specific treatments—​laser photocoagulation can preserve useful vision in many cases of proliferative retinopathy and maculopathy. Intraocular (intravitreal) injections of vascular endo- thelial growth factor (VEGF) blocking agents have been shown to improve vision in diabetic maculopathy. Diabetic neuropathies—​recognized clinically distinct syndromes include (1) diffuse symmetrical polyneuropathy—​classically a distal ‘glove and stocking’ peripheral polyneuropathy that affects all sizes of sensory and motor fibres; (2) autonomic neuropathy—​manifest as sexual difficulties (erectile failure, ejaculatory failure), postural hypotension, disturbed gastrointestinal motility, abnormal sweating, neuropathic bladder, abnormal blood flow, sudden unexplained death; (3) acute mononeuropathy; (4) diabetic amyotrophy; (5) cra- nial and other nerve palsies. Management is difficult: specific treat- ments have so far been disappointing. Numb feet are at greatly increased risk of ulceration and require sensible shoes and good foot care. Poor glycaemic control should be corrected. Pain may be diffi- cult to treat: simple analgesics are generally ineffective but drugs for neuropathic pain including specific antidepressants and anticonvul- sants are of more value. Autonomic neuropathic symptoms may be treated as follows: (1) erectile failure with oral phosphodiesterase type 5 inhibitors (e.g. sildenafil); (2) postural hypotension with compres- sion stockings, fludrocortisone, and/​or midodrine; (3) gastroparesis with erythromycin, metoclopramide or domperidone; (4)  exces- sive sweating with oral clonidine or topical glycopyrrolate cream;

13.9.1  Diabetes 2467 (5) neuropathic bladder with regular bladder training, but intermit- tent self-​catheterization may be needed. Diabetic nephropathy—​see Chapter 21.10.1. Macrovascular disease—​(1) dyslipidaemia—​first-​line treatment is with statins, ideally aiming to reduce low-​density lipoprotein chol- esterol (LDL-C) to below 1.8 mmol/​litre (approximately equivalent to 2.5  mmol/​litre non-​high-​density lipoprotein-​c cholesterol), especially in those with overt cardiovascular disease or neph- ropathy (2)  hypertension—​clinic blood pressure should be re- duced to a target of less than 140/85–90 mmHg in most patients (130/80 mmHg may be appropriate in some), with angiotensin-​ converting enzyme (ACE) inhibitors often recommended as first line; (3) coronary heart disease—​there should be a low threshold for referring patients with diabetes presenting with typical or atypical chest pain suggestive of angina for further evaluation (as coronary ischaemia is often asymptomatic of with minimal chest pain in dia- betes); (4) stroke—​investigation and management are conventional; (5) peripheral vascular disease—​investigation and management are conventional. Diabetic foot disease—​ulceration and severe ischaemia leading to gangrene of the toes or forefoot are the commonest problems. Many problems can be avoided by teaching the patients basic foot care, by regularly checking their feet (especially the soles of the feet) and shoes, and by providing prophylactic podiatry and special foot- wear as appropriate. Typical manifestations include (1) neuropathic ulcers—​occur at high-​pressure sites (heel, metatarsal heads) and ap- pear cleanly punched out of the surrounding callus; (2) ischaemic ulcers—​tend to affect the edges of the foot and toes; (3) traumatic damage (e.g. symmetrical damage across the toes and margins of the feet from tight shoes); with (4) all lesions prone to be complicated by infection (including osteomyelitis). Management requires the prevention of further trauma, treatment of infection, and optimiza- tion of the circulation. Charcot’s arthropathy most commonly affects the ankle and joints in the mid-​ and forefoot, which in advanced cases degenerate (usually painlessly) into a ‘bag of bones’: treatment is often unsatisfactory—​offloading pressure with a plaster-​cast boot may temporarily halt bone destruction; bisphosphonate infusions may slow the disease process by inhibiting osteoclast activity but the evidence base is weak. Introduction Diabetes mellitus can be defined as a state of chronic hypergly- caemia sufficient to cause long-​term damage to specific tissues, notably the retina, kidney, nerves, and arteries, but this functional label gives little insight into the long and colourful history of this disease, its clinical and scientific importance, or its immense per- sonal and socioeconomic impact. Diabetes was recognized in an- tiquity, and its clinical features (with empirical treatment guidelines) were recorded over 3500 years ago in the Egyptian Ebers papyrus. Our understanding of the disease has advanced greatly, especially during the last two decades, but many aspects of its management remain imperfect. The American Diabetes Association has pro- posed a generally accepted classification of diabetes mellitus into four types: type 1 is associated with β-​cell destruction leading to ab- solute deficiency of insulin, is immune-​mediated and of unknown root cause; type 2 is associated with a relative insulin deficiency and insulin resistance—​a range of abnormalities occur, and in some pa- tients a secretory defect predominates; gestational diabetes; and the fourth type is ‘other specific types of diabetes’ is used to encompass diabetes caused by specific defects, other endocrine abnormalities, and drug-​induced diabetes, and accounts for about 5% of patients. The incidence of all types of diabetes is rising. Extensive studies have shown a strong relationship between glycaemic control, the fraction of glycated haemoglobin (HbA1c), and disease outcomes. Treatment of raised blood pressure and lowering blood lipids also contributes to improved outcomes for nephropathy, retinopathy, myocardial infarction, and stroke. In both type 1 and type 2 dia- betes, there are compelling data to show that outcomes are improved by intensive therapy: better glycaemic control appears to have bene- fits on macrovascular complications in very long-​term studies, and lowering blood pressure has significant and more immediate bene- fits on both small-​vessel (microvascular) and macrovascular disease. The use of SGLT2 inhibitor drugs reduces the risk of death from heart failure in those at high risk. Diabetes is a significant and growing threat to global health. Worldwide, diabetes affects more than 400  million people. This number was more than 10-​fold less in 1985 (30 million) and the world prevalence is predicted to reach 640 million by 2040, with 10% of all adults affected. Diagnosis of diabetes Blood glucose concentrations are normally tightly regulated: fasting values lie between 3.5 and 5.5 mmol/​litre and even large carbohy- drate loads do not raise the concentration more than 8 mmol/​litre. It is logical to define diabetes by the blood glucose concentrations which cause the chronic complications of the disease, but the choice of the diagnostic glucose levels has been contentious (and has stirred up much passion among epidemiologists). One difficulty is that some diabetic complications show a ‘threshold’ effect with the risk rising above a cut-​off level (e.g. fasting plasma glucose of >7 mmol/​ litre for retinopathy), whereas macrovascular disease (atheroma) and the complications of pregnancy do not (see later). Another problem is that even the current criteria are not self-​consistent (e.g. up to 30% of patients with a diagnostic raised fasting glucose will have a 2 h value in the glucose tolerance test that is below the diag- nostic cut-​off). The current diagnostic criteria for diabetes and other hypergly- caemic states (see Fig. 13.9.1.1) based on blood glucose levels have been approved by the World Health Organization (WHO) and most national diabetes associations. All values refer to venous plasma glu- cose concentrations (see next). Since 2011, the American Diabetes Association and the WHO have additionally proposed that HbA1c measurements can be used, which are more convenient as they can be measured at any time of day, and that measurement of fasting glucose is not required. • Diabetes mellitus:  fasting glucose greater than 7.0 mmol/​litre (126 mg/​dl) and/​or a value exceeding 11.1 mmol/​litre, either at 2 h during a 75 g oral glucose tolerance test or in a random sample. The corresponding levels in non-​SI units are 126 and 200 mg/​dl, respectively. The diagnostic fasting glucose level was lowered from the previous value of 7.8 mmol/​litre to reflect more accurately the risk of developing diabetic retinopathy. The HbA1c range used

section 13  Endocrine disorders 2468 to diagnose diabetes is HbA1c in excess of 48 mmol/​mol (6.5%). Note that values less than 48 mmol/​mol do not exclude diabetes if the glucose values indicate diabetes. • Impaired glucose tolerance (WHO): 2 h oral glucose tolerance test value between 7.8 and 11.1 mmol/​litre (140–​199 mg/​dl). • Impaired fasting glucose:  fasting glucose 5.6 to 6.9 mmol/​litre (100–​125 mg/​dl). The lower value for this range was reduced from 6.0 mmol/​litre to 5.6 mmol/​litre by the American Diabetes Association in 2003. The HbA1c range 39–​46  mmol/​mol (5.7—​6.4%) has been proposed to correspond to impaired fasting glucose. Impaired glucose tolerance (IGT) and the more recently distin- guished impaired fasting glucose (IFG) are intermediate categories of hyperglycaemia (sometimes referred to as ‘prediabetes’) that carry definite risks and so require follow-​up and risk-​factor management (see next). They are often transient stages and overlap to some ex- tent: about one-​third of subjects with impaired fasting glucose also have impaired glucose tolerance, while one-​quarter of those with impaired glucose tolerance also show impaired fasting glucose. Recent criteria put much emphasis on the fasting plasma glucose concentration. However, the time-​consuming oral glucose tolerance test is still required in some cases with borderline fasting hypergly- caemia, because the 2-​h oral glucose tolerance test value in such patients may be high enough to put them at risk of microvascular complications. Moreover, the oral glucose tolerance test remains the only way to define impaired glucose tolerance. However, the use of HbA1c measurements is being increasingly accepted as a con- venient method of screening for diabetes. At an HbA1c cut-​off level of 48 mmol/​mol (6.5%), up to 30% of cases of diabetes positive on fasting glucose would be missed, but it is argued that the increased practicality of the tests means that overall more people will be tested, and hence more patients will be correctly diagnosed. Practical screening and diagnostic procedures Figure 13.9.1.2 shows an algorithmic approach to screening for and diagnosis of diabetes and its associated hyperglycaemic states. Certain high-​risk groups need to be actively screened for type 2 dia- betes, which may be present (and causing complications) for several years before it is noticed. Risk factors include having a first degree Venous plasma glucose (mmol/litre) Diabetes 11.1 IGT Random 12 10 8 6 4 2 0 Fasting 7.8 7.0 IFG 2 h after OGTT Diabetes Diabetes 11.1 5.6 Fig. 13.9.1.1  Diagnostic thresholds for diabetes, impaired glucose tolerance (IGT), and impaired fasting glucose (IFG). For conversion to mg/​ dl, multiply values in mmol/​litre by 18. Source data from Genuth S, et al. (2003). Follow-​up report on the diagnosis of diabetes mellitus. Diabetes Care, 26, 3160–​67. Fig. 13.9.1.2  Screening algorithm for diagnosing diabetes, impaired glucose tolerance, and impaired fasting glucose. All glucose values relate to venous plasma (mmol/​litre). Adapted from data in Shaw JE, Zimmet P (2000). Do we know how to diagnose diabetes and do we need to screen for the disease? In: Gill GV, Pickup JC, Williams G, (eds) Difficult diabetes, pp. 3–​21. Blackwell Science, Oxford.

13.9.1  Diabetes 2469 relative with type 2 diabetes, body mass index (BMI) more than 25, racial origin (including south Asian and Afro-​Caribbean descent), high-​density lipoprotein (HDL) less than 0.9 mmol/​litre, triglyceride more than 2.8 mmol/​litre, and giving birth to a baby weighing more than 4 kg. The American Diabetes Association (ADA) recommend three-​yearly screening for everyone age more than 45  years and for people with one or more risk factors aged more than 25 years. However, this would be a huge challenge for healthcare systems and would still miss many young-​onset cases. At present no screening system has been universally adopted, and a ‘low threshold for oppor- tunistic screening’ approach is widely practised. Diabetes is not a trivial diagnosis, and certain practical points must be carefully observed: • Glucose should be measured in venous plasma using a quality-​ controlled laboratory method. Capillary (finger-​prick) samples contain higher glucose levels than venous blood, from which glu- cose has been extracted by the tissue bed; whole-​blood glucose levels are lower than in plasma, because red cells actively metab- olize glucose and so contain only low concentrations. These dif- ferences may reach 0.5 to 1.0 mmol/​litre. Portable glucose meters correlate well with laboratory glucose methods, but because of po- tential technical errors they should not be used to make or refute the diagnosis. • An oral glucose tolerance test is indicated for borderline hypergly- caemia (Fig. 13.9.1.2). After an overnight fast, the subject drinks 75 g of anhydrous glucose dissolved in 250 ml water (or 419 ml of a glucose drink such as Lucozade Energy Original—​73 kcal/​ 100 ml); venous blood is sampled at baseline and 2 h later. Food intake should be normal during the preceding few days: poor nu- trition can cause delayed hyperglycaemia with a raised 2 h value (the lag curve). • Abnormal values need confirmation. Postchallenge glucose levels in particular can vary considerably. Because of this and possible laboratory error, the diagnosis of diabetes should be verified using a further sample on another day unless there is a clear history of symptoms of hyperglycaemia confirming that this value is not a one-​off result. • HbA1c is now widely accepted as an alternative test for type 2 dia- betes diagnosis that does not require fasting or repeated samples. Values greater than 48  mmol/​mol (6.5%) are considered diag- nostic of diabetes and the range 38–​47  mmol/​mol (5.7–​6.5%) indicates increased risk of diabetes in the future (equivalent to impaired glucose tolerance). The limitations of this test should be borne in mind; the levels are affected by haemoglobinopathies and anaemia and in circumstances where the hyperglycaemia may be acute, such as type 1 diabetes and gestational diabetes, HbA1c should not be used as a test of exclusion of the diagnosis. Glycosuria depends on the renal threshold for glucose reabsorption and its presence does not necessarily indicate hyperglycaemia; con- versely, glucose may be absent from the urine in diabetic subjects who also have a high renal threshold. However, abnormal results with any of these tests suggest diabetes and indicate the need for formal blood glucose screening. Impaired glucose tolerance Impaired glucose tolerance is a not a stable state: within 5 years, about 25% of subjects with impaired glucose tolerance deteriorate into type 2 diabetes, while a further 25% revert to normoglycaemia. The degree of hyperglycaemia in impaired glucose tolerance falls, by definition, below the threshold for microvascular complications but is enough to predispose to cardiovascular disease (see later). Subjects found to have impaired glucose tolerance must be fol- lowed up because of the hazards of both diabetes and macrovascular disease. An oral glucose tolerance test or HbA1c should be repeated at least annually, and dietary and lifestyle advice given to decrease metabolic and cardiovascular risks; increased physical activity, a low-​fat diet and weight loss convincingly reduce both the pro- gression to type 2 diabetes (by 58%) and cardiovascular risk. Risk factors such as smoking, hypertension, dyslipidaemia, and obesity should be managed actively. Specific antihyperglycaemic treatments also reduce progression to type 2 diabetes—​metformin (24%), pioglitazone (effective in preventing type 2 diabetes following an episode of gestational diabetes)—​in addition to pharmacological (orlistat) or physical (bariatric surgery) weight-​loss interventions (see Chapter 11.6). These measures should be used in combination with lifestyle intervention, which is recommended for all subjects with impaired glucose tolerance. Impaired fasting glucose As with impaired glucose tolerance, the 5-​year risk of progressing to type 2 diabetes appears to be about 25%, and IFG predisposes to cardiovascular disease. Long-​term monitoring and management should therefore be as for impaired glucose tolerance. Metabolic basis of diabetes Diabetes is due to inadequate production of insulin and/​or ‘resistance’ to the glucose lowering and other actions of insulin. To put this in context, key aspects of normal metabolism will be briefly reviewed. The islets of Langerhans There are about 1  million islets of Langerhans in the normal adult: insulin is produced by the β cells, which make up the bulky core of each islet; β cells also synthesize the peptide known as amylin or islet-​associated polypeptide. The other islet cell types, mostly sur- rounding the β-​cell core, are the α cells that produce glucagon, the δ cells that produce somatostatin, and the pancreatic polypeptide (PP) cells that synthesize pancreatic polypeptide. All islet cells are derived embryologically from the buds of gut endoderm which also give rise to the exocrine pancreatic tissue. The various islet cell types communicate with each other through the hormones they secrete into the islet’s rich capillary plexus and probably by paracrine effects on adjacent cells; these inter- actions presumably regulate hormone secretion. Insulin inhibits release of glucagon, while glucagon powerfully stimulates insulin secretion—​an action exploited in the testing of β-​cell reserve (see next). Somatostatin suppresses the secretion of insulin and glu- cagon. Amylin can inhibit insulin and glucagon secretion as well as reduce appetite and gastric emptying. Its physiological role is uncer- tain but amylin analogues when used as pharmacotherapy have been shown to reduce weight as well as blood glucose levels. Amylin also polymerizes outside the β cell to produce fibrils of amyloid material, which have been implicated in the progressive β-​cell damage of type 2 diabetes.

section 13  Endocrine disorders 2470 Insulin Insulin is a 5800 Da protein made up of an A chain (21 amino acid residues) and a B chain (30 residues), joined covalently by two disul- phide bridges. The precursor molecule, proinsulin, consists of the A and B chains linked end-​to-​end through a connecting (C) peptide which is cleaved off during insulin processing. In the circulation, insulin is monomeric but in crystals and more concentrated solu- tions (e.g. in the insulin vial and the subcutaneous injection site), six insulin molecules self-​associate around a central Zn2+ ion. Self-​ association influences the pharmacokinetic properties of subcuta- neously injected insulin: the rate-​limiting dissociation of hexamers into monomers slows the absorption of even fast-​acting insulin. Insulin regulates metabolism in birds, fish, and reptiles as well as mammals, and its structure is remarkably well conserved across the phyla. Three species of insulin are used therapeutically; the human sequence differs from porcine at a single residue (B30) and from bovine at two others. These differences affect the pharmacokinetic and immunogenic characteristics of the insulins (see next). The physicochemical behaviour of insulin has been successfully ma- nipulated in synthetic ‘designer’ insulins that have improved absorp- tion profiles: modification of the C terminus of the B chain, a region crucial for self-​association, produces analogues that remain in the monomeric state and are therefore absorbed faster than the native soluble insulin (see later). Insulin biosynthesis and processing Insulin is a product of the INS gene, located on the short arm of chromosome 11, whose coding region contains three exons. Translation of INS mRNA in the rough endoplasmic reticulum pro- duces preproinsulin, which is successively cleaved during its passage through the Golgi vesicles and secretory vesicles to yield first pro- insulin and finally insulin and C-​peptide. Proinsulin is converted into insulin by the proteolytic excision of the C-​peptide chain; the two intermediate cleavage products (with either end of the C-​ peptide remaining attached to insulin) are called split products of proinsulin. Normally, almost all proinsulin is processed through this regulated pathway to yield equimolar amounts of insulin and C-​peptide. However, a constitutive pathway may predominate in dysfunctional β cells (e.g. in type 2 diabetes and insulinoma), when processing is not complete and large quantities of proinsulin and split products may be released into the circulation. C-​peptide is generally regarded as an inert by-​product of insulin production. However, its structure is also conserved across species and it may have vasoactive and other properties. Insulinopathies are point mutations in the INS gene which either produce a mutant insulin (e.g. insulin Chicago:  a phenylalanine for leucine substitution at residue B25) or interfere with one of the cleavage sites of proinsulin so that the mutant split product cannot be further processed (e.g. proinsulin Tokyo). These conditions are inherited as autosomal dominant traits; circulating insulin-​like or proinsulin-​like immunoreactivities may be extremely high, but glu- cose intolerance is often surprisingly mild. Insulin secretion Glucose is the main insulin secretagogue; this action of glucose is modulated by other ingested nutrients, by hormones released by the islets and the gut, and by the autonomic innervation of the islet. The process gives insight into the mode of action of the sulphonylureas and related drugs, and some of the causes of maturity-​onset diabetes of the young and neonatal diabetes (see next). Glucose-​stimulated insulin secretion The amount of insulin released by the normal β cell is tightly coupled to blood glucose levels and begins to increase immediately when blood glucose rises. The ability of the β cell to sense ambient glucose levels accurately and rapidly depends on the glucose transporter isoform GLUT-​2 and the glucose metabolizing enzyme glucokinase, while insulin release hinges on depolarization of the β-​cell mem- brane which is controlled by a specific ion channel, the ATP-​sensitive K+ channel. The characteristics of GLUT-​2 allow glucose at physio- logical concentrations to freely enter the β cell, where it is immedi- ately converted by glucokinase into glucose 6-​phosphate—​the point of entry into the glycolytic pathway which ultimately yields ATP; ATP production within the β cell is therefore proportionate to extra- cellular glucose. ATP binds to and closes the ATP-​dependent K+ channel; when open, this channel allows K+ ions to leave the β cell along their con- centration gradient and thus helps to maintain the negative charge inside the β-​cell membrane. ATP-​induced closure of the channel therefore causes K+ ions to accumulate within the cell and the mem- brane to depolarize, which triggers the opening of specific (voltage-​ gated) Ca2+ channels in the membrane. Ca2+ ions then flood into the β cell from the outside and activate the contractile proteins which drag the secretory vesicles containing insulin and C-​peptide to the cell surface. Here, the vesicles fuse with the cell membrane and re- lease their contents into the extracellular space (exocytosis), from where insulin and C-​peptide enter the islet capillaries. Other factors affecting insulin secretion Sulphonylureas induce insulin secretion by closing the same ATP-​ sensitive K+ channel as glucose: they bind to a specific sulphonylurea receptor (SUR1) linked to the K+ channel protein (called Kir 6.2). Repaglinide also closes this K+ channel, but binds to a different site from the sulphonylureas. By contrast, diazoxide locks the channel open, hyperpolarizing the β-​cell membrane and inhibiting insulin secretion—​hence its use in treating insulinoma. Glucagon and glucagon-​like peptide 1 7–​36 amide (GLP-​1; a gut peptide with insulin secretagogue (incretin) actions) both stimulate insulin secretion by raising cytosolic Ca2+ concentrations; binding to their receptors increases generation of cAMP which blocks re- moval of Ca2+ into intracellular organelles. Conversely, somatostatin and possibly amylin act to decrease production of cAMP and inhibit insulin secretion. Arginine stimulates insulin secretion, possibly by depolarizing the β-​cell membrane as it enters the cell (it is cationic). The autonomic nervous system is an important modulator of in- sulin secretion; it is stimulated by the parasympathetic (vagal) out- flow and inhibited by the sympathetic. Vagal stimulation is mediated by acetylcholine acting via muscarinic receptors, while the inhibi- tory sympathetic neurotransmitter is noradrenaline, interacting with α2-​adrenoceptors. Defects in insulin secretion due to mutations affecting glucokinase are responsible for 20% of cases of maturity-​onset dia- betes of the young (MODY), that is, glucokinase-​dependent MODY (MODY 2). This impairs ATP production from glucose, blunting

13.9.1  Diabetes 2471 the insulin response of the β cell to rising glucose and resulting in variable hyperglycaemia (see next). By contrast, familial neonatal hyperinsulinism is caused by inactivating mutations in ABCC8 (SUR1) or KCNJ11 (Kir6.2) that result in closure of the ATP-​sensitive K+ channel, leading to sustained insulin secretion and severe hypo- glycaemia soon after birth. Activating mutations of KCNJ11 (Kir6.2) cause impaired ATP-​sensitive K+ channel closure and have recently been shown to be a cause of persistent neonatal diabetes that can be treated with high-​dose sulphonylureas. Normal pattern of insulin secretion Insulin concentrations in peripheral blood show basal levels of about 10 mU/​litre (1 mU/​litre is approximately equivalent to 6.5 pmol/​ litre) that tend to fall overnight, on which are superimposed pran- dial peaks reaching 80 to 100 mU/​litre, roughly proportionate to the amount eaten. The prandial peaks are elicited by the insulin secreta- gogue effects of glucose and other nutrients, augmented by incretin gut peptides (such as GLP-​1) and the vagal outflow (the early ceph- alic phase of insulin release). Very frequent sampling (every minute) shows that ‘basal’ insulin secretion is in fact pulsatile, with clear but low-​amplitude peaks every 9 to 13 min. This may help to keep the target tissues sensitive to insulin; loss of this pulsatility is an early sign of β-​cell dysfunction in type 2 diabetes. An acute insulin secretagogue challenge (e.g. an intravenous glucose bolus) induces a sharp ‘first-​phase’ insulin peak, loss of which is another early abnormality in type 2 diabetes. The insulin response elicited by eating is larger than when an equivalent nutrient load is given intravenously. This is because glu- cose entering the gut stimulates neuroendocrine cells in the gut wall to release ‘incretin’ hormones which act on the β cell to enhance insulin secretion (the enteroinsular axis: see Chapter 15.9.1). An im- portant incretin appears to be GLP-​1, a product of alternative pro- cessing of the preproglucagon gene (glucagon itself is not produced, in contrast to in the islet α cell). GLP-​1 released from the small intes- tine augments insulin release in the presence of glucose, slows gas- tric emptying, and acts on the central nervous system to generate a feeling of satiety, effects currently being exploited in the treatment of type 2 diabetes by use of long-​acting GLP-​1 analogues or inhibitors of GLP-​1 breakdown (see next). Peripheral insulin levels are lower than those in the portal vein, into which the islets drain, because up to 30% of insulin is removed on its first pass through the liver—​one of the main targets for insulin action. The kidney also actively clears and degrades insulin; the cir- culating half-​life is only a few minutes. C-​peptide provides a robust measure of residual β-​cell function, because it is cleared more slowly than insulin and its plasma concen- trations are therefore more stable. C-​peptide is generally measured after intense β-​cell stimulation with the powerful insulin secreta- gogue glucagon; alternatives are a heavy oral load of carbohydrate, mixed meal stimulation including amino acids (such as Boost or Sustacal) or simply the measurement of 24-​h secretion of C-​peptide in urine (it is cleared largely intact through the kidneys). In normal subjects and most with type 2 diabetes, peak C-​peptide concentra- tions at 6 min after 1 mg of intravenous glucagon are 1 to 4 nmol/​litre, whereas type 1 diabetic individuals are typically C-​peptide negative, with peak levels less than 0.2 nmol/​litre after 5 years. At diagnosis of type 1 diabetes there may be overlap with levels in patients with type 2 diabetes and accordingly the test is not used diagnostically (see Fig. 13.9.1.6). However, measurement of C-​peptide concentrations, and spe- cifically urine C-​peptide:creatinine ratio after a mixed meal, has recently been proposed as a valuable method of distinguishing MODY from type 1 diabetes. The insulin receptor and signal transduction The insulin receptor belongs to the family that also includes the insulin-​like growth factor 1 (IGF-​1) receptor. Insulin receptors are found in the obvious insulin target tissues (fat, liver, and skeletal muscle) but also in unexpected sites, such as the brain and gonads, in which glucose uptake does not depend on insulin. The insulin receptor is a 400-​kDa heterotetramer composed of two α and two β glycoprotein subunits, interconnected by disulphide bridges (Fig. 13.9.1.3). Both α and β subunits are encoded within a complex gene (22 exons) on chromosome 19q. The α subunit (135 kDa) lies entirely extracellularly, while the β subunit (95 kDa) spans the cell membrane and extends into the cytoplasm. Part of the intracytoplasmic tail functions as a tyrosine kinase, attaching phosphate groups from ATP to tyrosine residues elsewhere on the receptor (autophosphorylation) and on other intracellular proteins. This tyrosine kinase activity is essential for insulin signalling and for insulin to exert its many effects on its target tissues. Insulin binds to a site on the extracellular α subunits, and binding triggers a conform- ational change in the receptor which activates the tyrosine kinase domain of the β subunits. P P P P P P P IRS-1 IRS-2 IRS-3 IRS-4 PI 3-kinase Akt mTOR GSK3 FOXO-1 Lipid synthesis Glucose/protein metabolism Cell growth differentiation Raf MEK MAP kinase SREBP-1c PKCλ Insulin receptor P85 P110 Ras Grb2 P62 GAB-1 Sos Shc Fig. 13.9.1.3  The insulin receptor and signal transduction pathways within insulin’s target cells. Binding of insulin to the extracellular α subunits of the receptor activates the tyrosine (Tyr) kinase domain of the intracellular β subunit. This results in phosphorylation of more than
10 substrate proteins including four structurally related proteins of the insulin receptor substrate (IRS) family which vary in their tissue distribution, as well as Shc, Cbl, p62dok, and Gab-​1. These activated proteins then trigger other reactions that result in the biological actions of insulin, including enhanced glucose uptake, anabolic effects, and cell growth. MAP kinase, mitogen-​activated protein kinase; PI3 kinase, phosphatidylinositol-​3-​kinase. Akt is also known as protein kinase B (PKB). Source data from Biddinger SN, Kahn CR (2006). From mice to men: insights into the insulin resistance syndromes. Annu Rev Physiol, 68, 123–​58.

section 13  Endocrine disorders 2472 Postreceptor mechanisms The activated receptor phosphorylates tyrosine residues on spe- cific intracellular proteins which initiate the signal transduction pathway within the target cell. One group of proteins is the insulin receptor substrate family (IRS 1–​4) that vary in their tissue distri- bution and subcellular localization. Additional substrates include Shc, Cbl, p62dok, and Gab-​1. All these substrates carry docking sites for proteins possessing specific Src homology region SH2 do- mains. Docking of these proteins by the IRS molecules and other substrates begins a cascade of intracellular reactions that lead ultim- ately to the effects of insulin on glucose, lipid, and protein metab- olism and its many other actions (see Fig. 13.9.1.3). A key element is the phosphatidylinositol 3-​kinase pathway which appears to me- diate almost all of insulin’s effects on glucose transport, lipogenesis, and glycogenesis. The mitogen-​activated protein kinase pathway, by contrast, is particularly relevant to insulin’s actions on cell growth, with less relevance to its metabolic effects (see Fig. 13.9.1.3). Receptor turnover Receptors that bind insulin are internalized (i.e. taken up into the target cell by an invagination of the cell membrane that is coated with the protein clathrin). Bound insulin is degraded in the lyso- somes, while most of the insulin receptors are carried back to the cell surface and reinserted into the membrane. The density of receptors on the cell surface is therefore a dynamic quantity, regulated partly by new receptor synthesis and partly by receptor recycling, which in turn is determined by insulin binding. Prolonged exposure to high insulin concentrations increases the proportion of internalized re- ceptors and so decreases the density of receptors available on the cell surface. This downregulation of receptors reduces the sensitivity of the target tissue to insulin. Disorders due to insulin receptor defects Many mutations have now been described in the insulin receptor, including point mutations that cause single-​residue substitutions or truncation of the α or β subunits. The most severe mutations af- fect the insulin-​binding extracellular domain and result in so-​called leprechaunism, while less severe mutations affect the tyrosine kinase domain and interfere with insulin signalling (Rabson–​Mendenhall syndrome). Both syndromes are associated with severe insulin re- sistance (type A) as well as serious mental and physical abnormal- ities, confirming the importance of insulin in fetal development. Antibodies may develop against the insulin receptor and usually cause insulin resistance with variable hyperglycaemia (the type B insulin resistance syndrome); rarely, hypoglycaemia results from antibodies that activate the receptor (analogous to thyrotoxicosis in- duced by antibodies to the thyroid-​stimulating hormone receptor in Graves’ disease). Metabolic actions of insulin Insulin functions as an anabolic hormone, favouring the uptake, utilization, and storage of glucose, the storage of lipids as trigly- ceride, and preventing the breakdown of protein. Effects on carbohydrate metabolism Insulin lowers blood glucose in two main ways (Fig. 13.9.1.4). At low basal concentrations (overnight and between meals) it shuts off the production of glucose by the liver, which is the main determinant of fasting glycaemia. Hepatic glucose output is fuelled by both glycogen breakdown (glycogenolysis) and gluconeogenesis (i.e. glucose synthesis from substrates including lactate, glycerol, and alanine and other amino acids); the rate-​limiting enzymes for these processes are powerfully inhibited by insulin. Conversely, insulin stimulates glycogen synthesis. At higher concentrations, such as after meals, insulin also stimu- lates glucose transport into skeletal muscle (where it is utilized to provide energy via glycolysis or stored as glycogen) and into fat (where it is used to synthesize triglycerides). In both these tissues, insulin enhances glucose uptake through a specific glucose trans- porter protein, GLUT-​4 (Fig. 13.9.1.4). Insulin causes GLUT-​4 units to be translocated rapidly to the cell surface and inserted into the membrane: there, GLUT-​4 units act as hydrophilic pores through which glucose can cross the otherwise impermeable membrane into the cell, following its concentration gradient. Insulin also stimulates GLUT-​4 synthesis. Overall, insulin acting via GLUT-​4 can increase glucose uptake into muscle and fat by up to 40-​fold over the basal, non-​insulin-​mediated, glucose uptake. Non-​insulin-​mediated glu- cose uptake occurs through other glucose transporter isoforms that operate in the absence of insulin, notably GLUT-​1 in peripheral tis- sues and erythrocytes and GLUT-​3 in brain. Effects on lipid metabolism Insulin inhibits triglyceride breakdown (lipolysis), while promoting its synthesis (lipogenesis). Lipolytic enzymes that split triglyceride into glycerol and free fatty acids are powerfully inhibited by insulin, even at low basal insulin concentrations. Profound insulin defi- ciency, such as in untreated type 1 diabetes, is therefore required before uncontrolled lipolysis occurs and generates enough free fatty acids to cause ketoacidosis (see next pargraphs). Effects on protein metabolism Insulin inhibits protein catabolism and thus reduces the gener- ation of amino acids which can act as gluconeogenic precursors to Peripheral glucose uptake + + + + + Other GLUTS CNS and other tissues Gluconeogenic precursors Lactate Glycerol Amino Fat Triglyceride Glucose Glucose Glucose Glycogen Muscle Glycolysis Glucose Glycerol Other GLUTS Other GLUTS GLUT4 Ins GLUT4 Ins Ins Ins Ins – – Ins acids Gluconeogenesis Liver Glucose Glycogenolysis Glycogen Hepatic glucose output Fig. 13.9.1.4  Effects of insulin on glucose homeostasis. Insulin inhibits gluconeogenesis and glycogen breakdown in the liver, thus decreasing hepatic glucose output. Blood glucose is also lowered by increased glucose uptake into fat and skeletal muscle, mediated by the insulin-​ stimulated glucose transporter, GLUT-​4. (Non-​insulin-​mediated glucose uptake is effected by other GLUT proteins.)

13.9.1  Diabetes 2473 enhance glucose production by the liver and kidney. Insulin also promotes protein synthesis and cellular and tissue growth. Other actions of insulin These include vasodilatation, mediated by endothelial production of nitric oxide; growth and differentiation of the fetal nervous system; and enhanced tubular reabsorption of Na+ ions by the kidneys. Measurements of insulin action Glucose lowering is the most easily tested biological action of insulin, and forms the basis for most measurements of insulin resistance. Several methods are used in the research setting; the- oretically, the simplest could be used in clinical diabetes care, to identify patients with marked insulin resistance who might benefit particularly from insulin-​sensitizing drugs such as the thiazolidinediones: • Homeostatic model assessment (HOMA) is an index derived by mathematical modelling of the relationship between the fasting glucose and insulin concentrations: with decreasing insulin sen- sitivity, insulin secretion increases in an attempt to maintain euglycaemia, resulting in compensatory hyperinsulinaemia. Homeostatic model assessment yields measures of both insulin resistance and β-​cell function; the test can be performed on a single fasting blood sample and the results compare well with the insulin–​glucose clamp. • Insulin–​glucose (hyperinsulinaemic–​euglycaemic) clamp. Insulin is infused intravenously to achieve constant high concentrations and a separate infusion of glucose is adjusted to maintain blood glucose ‘clamped’ at a normal value. The more glucose required, the greater is the insulin sensitivity. The clamp is generally re- garded as the gold standard method but demands blood glu- cose measurements every few minutes and takes some hours to perform. • Intravenous glucose tolerance test. An intravenous glucose bolus stimulates insulin release, and mathematical modelling of the re- lationship between the insulin peak and the decay in blood glu- cose levels can yield indices of both insulin secretion and insulin sensitivity. Insulin resistance Insulin resistance (or insensitivity) is a poorly defined term signi- fying decreased biological activity of insulin, and which is usually equated with impaired glucose lowering. There is no universal normal range for insulin sensitivity, be- cause the ability of insulin to lower glucose varies considerably between and within individuals—​it is influenced (e.g. by levels of physical activity and fitness). Subjects with ‘insulin-​resistant’ con- ditions such as type 2 diabetes or essential hypertension commonly show reductions of 40 to 60% in glucose disposal (measured by the clamp technique), as compared with matched healthy controls, yet many apparently normal subjects also have comparable decreases in insulin sensitivity. There is no argument about extreme examples of insulin resistance:  in some patients with leprechaunism, over 20 000 U/​day of insulin have failed to control hyperglycaemia and ketosis. A working definition of clinically relevant insulin resistance in insulin-​treated diabetic patients is a daily requirement of more than 1.5 U/​kg. Causes of insulin resistance Inherited causes Inherited causes include the very rare mutations affecting the insulin receptor or postreceptor signalling pathways which can lead to extreme insulin resistance (type A  insulin resistance syndrome); milder polygenic defects contribute to the insulin resistance of type 2 diabetes (see next). Insulin receptor muta- tions cause clinically distinct syndromes, often with acanthosis nigricans and, in women, features of polycystic ovary disease and masculinization; hyperglycaemia is variable. Specific syndromes include the speculatively named leprechaunism and various in- herited lipodystrophies in which fat is lost from subcutaneous and other depots in defined but unexplained anatomical patterns. Recently, mutations affecting the PPARG gene (the target for the thiazolidinedione drugs; see next) have been shown to modify insulin sensitivity. Several mutations in loci that predispose to obesity have recently been reported, but interestingly, only a subset of these also predispose to type 2 diabetes (e.g. LEP, FTO) suggesting that genetic influences in addition to obesity are re- quired for the generation of diabetes. Obesity Obesity induces insulin resistance, especially in skeletal muscle, and weight loss can improve insulin sensitivity in the obese. Insulin resistance is particularly associated with truncal (central) obesity, where fat is deposited in and around the abdomen; both the sub- cutaneous and intra-​abdominal (visceral) fat depots have been implicated to various degrees that may reflect ethnic and other dif- ferences. Visceral fat (and especially intrahepatic fat) appears to be particularly associated with insulin resistance and diabetes risk because it represents ‘ectopic’ fat deposition occurring when the larger, safer, and more metabolically inert subcutaneous fat stores are saturated. Consistent with this, individuals with lipodystrophy syndromes that are unable to store fat subcutaneously develop high levels of visceral fat and are frequently insulin resistant without have a particularly high BMI. It is still not clear at the molecular level how an increased fat mass can decrease whole-​body insulin sensitivity, but cir- culating fat-​derived products are presumed to be responsible. Intra-​abdominal fat depots can secrete potentially diabetogenic mediators into the portal circulation—​where they would be de- livered directly to the liver—​and this may explain how visceral adiposity causes insulin resistance. Possible candidates include free fatty acids and the cytokine tumour necrosis factor-​α (TNFα); both are secreted by adipocytes and, under experimental condi- tions at least, interfere with aspects of insulin action. Levels of free fatty acids are raised in obese subjects, apparently because lip- olysis is enhanced, and free fatty acids may cause hyperglycaemia by competing with glucose metabolism in liver and muscle. In liver, free fatty acids enhance gluconeogenesis by stimulating the rate-​limiting enzyme pyruvate carboxylase and so increase hep- atic glucose production. In muscle, free fatty acids inhibit gly- colysis at the level of phosphofructokinase and glucose oxidation via pyruvate dehydrogenase, causing a decrease in glucose utiliza- tion and a secondary reduction in glucose uptake (the glucose–​ fatty acid or Randle cycle). In vitro, TNFα inhibits the tyrosine kinase activity of the insulin receptor that is crucial for insulin

section 13  Endocrine disorders 2474 signalling. Production of TNFα by adipose tissue is increased in obesity but its role as a mediator of insulin resistance in human obesity is uncertain. Recently, an adipocyte product, adiponectin, has been shown to enhance insulin sensitivity in rodents; intri- guingly, circulating adiponectin concentrations are decreased in human obesity. Several other recently identified hormones re- leased by adipose tissue (adipokines) also have effects on insulin action (e.g. resistin, visfatin, interleukin-​6) but their role in type 2 diabetes is less well defined. Obesity is also accompanied by the ectopic deposition of triglyceride in liver and skeletal muscle, and the accumulation of triglyceride is correlated with impairment of insulin action in these tissues. Physical inactivity strongly predisposes to obesity and also pro- motes insulin resistance which can be reversed by regular exer- cise. The mechanism is unknown but physical training is known to stimulate translocation of GLUT-​4 glucose transporters to the sur- face of muscle cells independently of insulin. In addition, muscle contraction enhances expression of the enzyme AMP kinase which mediates improved glucose transport and fatty acid metabolism. AMP kinase has recently been shown to be a molecular target of two drugs known to reduce insulin resistance (metformin and thiazolidinediones). Other acquired causes There are several other acquired causes of insulin resistance. Intrauterine growth retardation may contribute (see the Barker–​ Hales hypothesis later). Physiological states of insulin resistance, due to the appropriate oversecretion of the counterregulatory hormones whose metabolic actions oppose those of insulin, are puberty and pregnancy (see gestational diabetes, Chapter 14.10). Endocrine diseases that induce insulin resistance and can cause glucose intolerance and overt diabetes through excessive pro- duction of anti-​insulin hormones include acromegaly (preva- lence of diabetes and impaired glucose tolerance each c.25%), Cushing’s disease (diabetes c.30%), thyrotoxicosis, and the very rare glucagonoma (diabetes in >90% of cases). In these disorders, diabetes is mostly non​ketotic, although insulin may be needed to control hyperglycaemia. Intercurrent illnesses (e.g. myocardial infarction, stroke, or severe infections) induce the secretion of counterregulatory stress hor- mones that can cause marked insulin resistance—​insulin-​treated diabetic patients may need twice their usual insulin dosages during such episodes. Many drugs decrease insulin sensitivity, including glucocorticoids, β2 adrenoceptor agonists (ritodrine, salbutamol), and certain oral contraceptive pills containing high-​dose oestrogen or levonorgestrel; glucocorticoid-​induced hyperglycaemia com- monly requires insulin treatment. Acquired lipodystrophies, most notably that induced by drugs used to treat HIV, especially protease inhibitors and nucleoside analogue inhibitors of viral reverse tran- scriptase (NRTIs), are also associated with insulin resistance and the development of diabetes. The type B insulin resistance syndrome is due to the development of autoantibodies against the insulin receptor which interfere with insulin binding and/​or signalling. Most patients are young women, usually with pre-​existing autoimmune diseases such as lupus erythematosus, and masculinization often occurs. ‘Immune insulin resistance’ describes insulin-​treated patients with very high insulin requirements (sometimes several thousand U/​day) because of high titres of insulin-​binding antibodies that bind and inactivate admin- istered insulin. This has become very rare since the introduction of highly purified human-​sequence insulin preparations with low im- munogenicity (see next). Metabolic and clinical features of insulin resistance The metabolic disturbance due to insulin-​resistant syndromes ranges from subclinical glucose intolerance to severely symptomatic hyperglycaemia, sometimes with ketosis. A crucial determinant is the capacity of the individual’s β cells to secrete insulin in response to the rises in blood glucose that are due to impaired insulin action. The resulting hyperinsulinaemia is extremely variable, with plasma insulin levels ranging from twice normal in many obese subjects to 500 times normal in patients with defects of insulin receptors. Near normoglycaemia can be maintained as long as hyperinsulinaemia can compensate for the underlying defect in insulin signalling; diabetes occurs when β-​cell failure supervenes and insulin secre- tion falls below a critical level. In the total absence of functional insulin receptors (e.g. in leprechaunism), massive endogenous hyperinsulinaemia or administration of industrial insulin dosages cannot prevent severe diabetes, although very high insulin concen- trations may exert some metabolic actions through ‘cross-​talk’ with the IGF-​1 receptor. Acanthosis nigricans, a characteristic skin manifestation of se- vere insulin resistance, may be due to high insulin concentrations activating growth factor receptors (perhaps the IGF1 receptor) that drive the proliferation of keratinocytes and melanocytes. Hyperplasia of these cells leads to a velvety thickening and variable darkening of the skin, especially in the axillae (often with prolifer- ation of skin tags), groin, and nape of the neck (see Chapter 23.8). Widespread acanthosis nigricans can also accompany gut tumours, which may also secrete dermal growth factors. Increased androgen concentrations may lead to hirsutism and oc- casionally virilization in women with severe insulin resistance; high insulin concentrations may stimulate androgen production by the ovaries, which often show a polycystic appearance. Insulin resist- ance is a feature of polycystic ovary syndrome, especially in obese patients (Chapter  13.6.1). Enhancing insulin sensitivity through weight loss or treatment with metformin or the thiazolidinediones can decrease androgen levels and improve hirsutism and menstrual dysfunction. The metabolic syndrome The metabolic syndrome (syndrome X) denotes the co-​occurrence of insulin resistance and glucose intolerance (ranging from mild to overt type 2 diabetes), with truncal obesity, dyslipidaemia (raised triglycerides and a high low-​density lipoprotein:high-​density lipo- protein (HDL/​LDL) ratio), and hypertension (see Fig. 13.9.1.5). These abnormalities are all common in most Westernized populations, and it is still not clear whether or not this constella- tion of cardiovascular risk factors represents a genuine syndrome with a common underlying cause. Reaven and others have argued that insulin resistance is the central abnormality, and that the key features can be explained either by loss of specific actions of in- sulin or by the effects of the compensatory hyperinsulinaemia on organs that remain relatively insulin sensitive. For example, raised

13.9.1  Diabetes 2475 insulin levels could contribute to hypertension by enhancing re- tention of Na+ by the kidney; conversely, blood pressure could also be raised through loss of the direct vasodilator action of in- sulin. The pattern of abnormalities would therefore require in- sulin resistance to affect certain tissues and specific actions of insulin but not others. Other proatherogenic defects identified in subjects with various features of syndrome X include increased coagulability of the blood (e.g. increased levels of plasminogen activator inhibitor-​1) and impaired endothelial-​mediated vaso- dilatation. The relationship of these abnormalities to insulin re- sistance is uncertain. Obesity, dyslipidaemia, hypertension, and glucose intolerance are all independent cardiovascular risk fac- tors; any possible proatherogenic role of hyperinsulinaemia per se remains controversial. The aetiology of syndrome X is unresolved; indeed, it has been argued that it is not a distinct entity, but simply represents the vari- able association of several abnormalities that are relatively common in all populations, and especially those that generally overeat and are too sedentary. Adiposity, insulin sensitivity, and blood pres- sure show variable strengths of familial transmission that differ be- tween populations and generally suggest polygenic inheritance of multiple minor genes. On the other hand, Barker and Hales have suggested that fetal malnutrition programmes insulin resistance, hypertension, and dyslipidaemia in middle to late adult life. The underlying mechanisms remain elusive. Because obesity leads to insulin resistance and glucose intolerance, dyslipidaemia, hyper- tension, and atheroma, weight gain in middle age may be particu- larly hazardous in subjects who were underweight at birth. Clustering of these metabolic and cardiovascular risk factors is important clinically because it predisposes to atheroma for- mation and substantially increases the risk of dying prematurely from myocardial infarction or stroke. Treatment is currently based on correcting any factors (e.g. type 2 diabetes, hypertension, and dyslipidaemia) present in the individual patient. Lifestyle and dietary modification that achieves weight loss can improve most aspects of the syndrome. Several drugs have been shown to slow progression from impaired glucose tolerance to type 2 diabetes (e.g. metformin, pioglitazone, and weight-​loss drugs such as orlistat), although their role in treatment of the metabolic syn- drome remains controversial. Types and classification of diabetes mellitus The current WHO classification is based on aetiology (see Table 13.9.1.1). Type 1 and type 2 diabetes together account for 90 to 95% of cases and will be described in detail. Type 1 diabetes Type 1 diabetes is due to autoimmune destruction of the β cells (the type 1 process). A  similar clinical picture of insulin dependence can be caused by other forms of severe pancreatic damage such as pancreatitis. Epidemiology and demographic features Type 1 diabetes is considerably rarer than type 2, accounting for be- tween 5 and 15% of all diabetes and 30 to 50% of insulin-​treated cases in various populations. It appears predominantly in childhood, with a peak age at presentation of about 11 years in girls and 14 years in boys—​hence the old description of juvenile onset. However, it can develop at any age; up to 50% of all cases are diagnosed over the age of 18 and about 5% of newly diagnosed white diabetic patients over 65 years are considered to have type 1 diabetes. The prevalence of type 1 diabetes varies considerably throughout the world. Incidence is highest in northern European countries (about 30 to 35 cases per 100 000 children per year in Finland and Scotland) and declines progressively towards the equator; there are some isolated hot spots such as Sardinia, where the incidence is as high as in Finland. High susceptibility is found in European populations throughout the world, while African and East Asian populations are relatively spared (incidences of less than 1 per 100 000 per year). Superimposed on this geographical variation are time-​related changes in incidence that hint at the import- ance of the environment in causing the disease. Type 1 diabetes presents more frequently during the winter months, particularly in children aged 10 to 14 years. In many countries (e.g. Norway, Poland, Sweden, and the United Kingdom), there have been sharp 30 to 50% increases in incidence over 10-​ to 20-​year periods, al- though the explanation and significance of these secular trends are not clear. In particular, there has been a shift to diagnosis at a younger age with a particularly marked rise in cases being diag- nosed under the age of 5. Susceptibility to type 1 diabetes shows no gender bias in cases diagnosed before puberty, but in older individuals there is a male preponderance in new cases of 1.5:1 to 1.8:1 Aetiology Type 1 diabetes is an autoimmune, predominantly T-​cell-​mediated process that selectively destroys the β cells. Susceptibility is multi- factorial, resulting from the impact of environmental agents in a genetically disadvantaged subject. Of these two components, the en- vironment appears more important; genetic factors explain only 30 to 40% of total susceptibility. Genetic factors Over 50 genetic loci have been robustly associated with type 1 diabetes, most of which are believed to be causal. The genetic re- gions are tagged by around 100 single nucleotide polymorphisms, around half of which are shared with other autoimmune diseases Hypertension Atheroma Dyslipidaemia ( TG, HDL) Procoagulant tendency Glucose intolerance Truncal obesity Insulin resistance + Hyperinsulinaemia Fig. 13.9.1.5  Metabolic syndrome, a constellation of atherogenic risk factors which may each be related to insulin resistance and/​or the hyperinsulinaemia that accompanies insulin-​resistant states. ↓HDL, reduced high-​density lipoprotein cholesterol; ↑TG, hypertriglyceridaemia.

section 13  Endocrine disorders 2476 Table 13.9.1.1  Classification of diabetes mellitus according to aetiology Type 1 diabetes β-​Cell destruction, usually leading to absolute insulin deficiency (10–​15% of cases in Europe and United States): 1A-​Immune mediated 1B-​Idiopathic (e.g. fulminant type 1 diabetes) Type 2 diabetes May range from predominantly insulin resistance with relative insulin deficiency to a predominantly secretory defect with insulin resistance (80–​85% of cases in Europe and United States) Other specific types Other types with specific causes (5% of cases in Europe and United States)   A Genetic defects of β-​cell function MODY: see Table 13.9.1.4 Others, including mitochondrial DNA defects (MELAS syndrome) Neonatal diabetes: mutations in KCNJ11; imprinting abnormality in ZAC and HYMAI—​may be transient   B Genetic defects of insulin action Type A insulin resistance syndrome Leprechaunism Rabson–​Mendenhall syndrome Congenital lipodystrophies   C Diseases of the exocrine pancreas Pancreatitis, chronic and acute Carcinoma of the pancreas Haemochromatosis Cystic fibrosis Pancreatectomy, trauma Fibrocalculous pancreatopathy   D Endocrinopathies Acromegaly Cushing’s disease and syndrome Phaeochromocytoma Glucagonoma Hyperthyroidism Somatostatinoma Aldosteronoma   E Drug-​ or chemically-​induced Glucocorticoids β-​Blockers Thiazides Diazoxide Others—​phenytoin, pentamidine, nicotinic acid, interferon-​α   F Infections Congenital rubella Cytomegalovirus   G Uncommon forms of
  immune-​mediated diabetes Type B insulin resistance (insulin receptor antibodies) ‘Stiff man’ syndrome   H Other genetic syndromes Prader–​Willi syndrome Wolfram’s syndrome (DIDMOAD) Down’s syndrome Turner’s syndrome Klinefelter’s syndrome Others (e.g. Laurence–​Moon–​Biedl syndrome) Gestational diabetes Gestational diabetes or glucose intolerance DIDMOAD, diabetes insipidus, diabetes mellitus, optic atrophy, and deafness; MELAS, myopathy, encephalopathy, lactic acidosis, and stroke-​like episodes (associated with type 1 or type 2 diabetes); MODY, maturity-​onset diabetes of the young. Adapted from American Diabetes Association (2007). Diagnosis and classification of diabetes. Diabetes Care, 30, S42–​S47.

13.9.1  Diabetes 2477 (e.g. coeliac disease, RA, psoriasis, inflammatory bowel disease, or autoimmune thyroid disease). The best characterized are the HLA class II locus (HLA-​DQB1, IDDM1) and the insulin gene promoter region (IDDM2), of which the HLA class II locus has by far the greatest effect (odds ratio for diabetes is 7–​13:1 for sus- ceptible alleles). The HLA class II locus lies within the major histocompatibility complex region on chromosome 6, that encodes several proteins in- timately involved in immune responses. Of particular importance is HLA-​DQB1; this encodes the DQB1 peptide chain, which forms part of the cleft in the surface of the HLA class II molecule that is crucial in presenting peptide fragments of antigen to the T-​helper lymphocyte. Changes in the structure of the DQB1 molecule could therefore influence the coupling between the class  II molecule–​ peptide complex and the T-​lymphocyte receptor, and thus modulate the immune response against the (auto)antigenic peptide. Specific DQB1 polymorphisms have been shown to predispose to type 1 diabetes (e.g. DQB10302), whereas others (e.g. DQB10602) are protective—​at least in certain racial groups. The relationships of these polymorphisms to the long-​recognized influences of the DR3 and DR4 class II antigens (which increase several-​fold the risk of type 1 diabetes) and of the protective DR2 are discussed further in Chapter 13.12.2. The IDDM2 locus corresponds to the insulin gene (INS) whose uniqueness as a β-​cell product makes it an obvious candidate gene. The insulin coding sequence is unchanged in type 1 diabetes. However, variation is observed in a region upstream of the insulin gene in which there is a variable number of repeats of the consensus sequence, 5′-​ACAGGGGTGTGGGG-​3′ one after another, known as the variable number of tandem repeats (VNTR) minisatellite. The short class I VNTR alleles (26–​63 repeats) predispose to diabetes, while class III alleles (140–​210 repeats) have a dominant protective effect (odds ratio for type 1 diabetes with class I vs. class III alleles is 2.2:1). This protective effect appears to be mediated by a two-​ to threefold increased expression of insulin in the thymus. Insulin, like many other self proteins, is normally expressed at low level in the thymus as part of the process which promotes central tolerance to self antigens among T cells. The further increased levels associated with class III alleles thereby results in a relative reduction in the risk of autoimmunity. Additional loci confirmed to predispose to type 1 diabetes show a strong predominance of genes affecting the immune system, including PTPN22 and CTLA4, both negative regulatory molecules of the immune system, IL2RA (CD25), the high-​affinity interleukin-​ 2 receptor, interleukin-​27, and IFIH1, a cytoplasmic helicase that mediates induction of interferon in response to viral RNA. Work is in progress to define more precisely how disease predisposition is increased by the high-​risk alleles and should ultimately shed light on the pathogenesis of the disease. Environmental factors Viruses have long been popular candidates as an environmental trigger for diabetes. Some (e.g. mumps, Coxsackie, cytomegalo- virus, and rubella) infect the pancreas but normally damage the entire gland, particularly the exocrine tissue, rather than causing se- lective β-​cell injury. Certain viruses target the β cell in animals (e.g. the Kilham rat virus) and can cause insulin-​dependent diabetes, either through their direct cytolytic effects or by provoking a type 1-​like autoimmune process. Important contenders in humans are coxsackieviruses (especially B4), rubella, and rotaviruses. Serological studies indicate that recent coxsackie B infections and possibly other enteroviruses are relatively common among newly diagnosed patients with type 1 diabetes; these could represent the final insult in the disease’s long natural history, since the autoimmune process can be detected many years prior to this. Coxsackieviruses capable of damaging rodent β cells have also been isolated post-​ mortem from the islets of some type 1 diabetic subjects. About 20% of children who survive intrauterine rubella infection develop type 1 diabetes, with typical autoimmune markers. Endogenous retro- viruses were previously implicated as aetiological agents, but this has not been confirmed in further studies. For other viruses, the epi- demiological data are conflicting; for example, the eradication of ru- bella by vaccination has not reduced the incidence of type 1 diabetes in Finland while the prevalence of coxsackie infections is lower in Finland than in the adjacent Russian Karelian population which is genetically related but has a substantially lower type 1 diabetes risk. Viruses could trigger or maintain autoimmune β-​cell damage in various ways. Acute or persistent viral infection of β cells could re- lease β-​cell antigens that are normally sequestered beyond the reach of the immune cells. Certain viral proteins may elicit an immune re- sponse which cross-​reacts with specific β-​cell antigens that happen to be similar (molecular mimicry): e.g. peptide sequences of the P2-​ C capsid protein of coxsackie B viruses may cross-​react with glu- tamate decarboxylase-​65 (GAD65) in the β-​cell membrane. Other environmental factors are suggested to include bovine serum albumin from cow’s milk and various toxins. Bovine serum albumin contains a peptide sequence that may cross-​react with a β-​ cell surface protein (see next); this was suggested as an explanation for an apparent excess risk of type 1 diabetes among children fed with cow’s milk in the neonatal period, although a protective effect for breastfeeding remains controversial. Various toxins selectively damage β cells, including streptozotocin, a nitrosourea used to in- duce experimental diabetes in rodents. Related nitrosamine com- pounds have been blamed for the higher risk of type 1 diabetes in the children of women who eat fermented smoked mutton (a traditional delicacy in Iceland). To try to resolve the controversies in this complex area, an international consortium—​The Environmental Determinants of Diabetes in the Young (TEDDY; https://​teddy.epi.usf.edu/​)—​was established. This followed several thousand children with high-​risk HLA genotypes from birth until adolescence to identify infectious agents and dietary or other environmental factors that trigger β-​cell autoimmunity in genetically susceptible people. Autoimmune features Type 1 diabetes has strong associations with endocrine and other autoimmune diseases, including Schmidt’s syndrome (with hypo­ thyroidism and adrenocortical failure) and the autoimmune polyendocrinopathy–​candidiasis–​ectodermal dystrophy (APECED) syndrome caused by mutations in the AIRE gene which controls self-​ tolerance by influencing thymic expression of autoantigens. Type 1 dia- betes is also a feature of the IPEX syndrome (immunodysregulation, polyendocrinopathy, and enteropathy, X-​linked syndrome) caused by mutations in the key T-​cell regulatory gene, FOXP3. Most β-​cell damage is probably inflicted by T lymphocytes. Insulitis—​infiltration of the islets with immune cells, mostly

section 13  Endocrine disorders 2478 cytotoxic/​suppressor (CD8+) T lymphocytes—​is a pathognomonic feature of the disease, and circulating T-​helper lymphocytes can be identified that react against β-​cell antigens including proinsulin, GAD65, IA-​2, and IGRP (islet-​specific glucose-​6-​phosphatase catalytic subunit-​related protein). Various circulating autoanti- bodies also occur. Some target antigens are unique to the β cell, while other autoantigens are shared by other islet cell types. Notable β-​cell selective autoantibodies are those that recognize GAD65, a heat shock protein (hsp60), and insulin itself. GAD catalyses the conversion of glutamic acid to γ-​aminobutyric acid, whose role in the β cell is uncertain. Studies in rodents with type 1 diabetes suggest that the level of GAD65 expression influences the inten- sity of the autoimmune attack on the β cells. The GAD67 isoform of the enzyme is also expressed in the central nervous system, and autoimmune damage of GABAergic neurons is presumed to explain the association of type 1 diabetes with the rare ‘stiff man’ syndrome (Chapter 24.19.4). High frequencies of autoantibodies to the protein tyrosine phosphatase-​like molecule IA-​2 are also seen. Most recently, autoantibodies against the cation efflux transporter, zinc transporter 8 (ZnT8) and the transmembrane glycoprotein, tetraspanin-​7 have been identified. GAD65 antibodies are present in 70 to 90% of newly diagnosed type 1 patients, insulin antibodies in 40 to 70%, IA-​2 autoantibodies in around 50 to 60%, and ZnT8 antibodies in around 70%. Islet cell antibodies detected by immunofluorescence on tissue sections are present in 80 to 90% of newly diagnosed patients but are technic- ally difficult to measure. Recent studies suggest that automated combined testing for GAD and IA-​2 has equivalent sensitivity and specificity to islet cell antibodies (ICA) testing. This is increasingly replacing ICA assays, although undoubtedly ICA reactivity encom- passes more (as yet undetermined) antigens than insulin, GAD 65, IA-​2, and ZnT8 alone. These antibodies cannot explain the selective destruction of β cells; although some islet cell surface antibodies are complement-​fixing, most of the islet cell destruction is believed to be caused by T cells. High titres of each of these classes of antibodies have some value in predicting diabetes in high-​risk individuals—​the combination of high titres of three autoantibodies (GAD, IA-​2, and insulin or ICA) among family members of subjects with type 1 diabetes is 90% pre- dictive of disease, although hyperglycaemia may not develop for 20 years or more. However, they are clearly not the immediate cause of the disease: single autoantibody-​positive individuals rarely pro- gress to disease, suggesting that these autoantibodies are general markers of autoimmunity against the β cell, rather than evidence of β-​cell destruction, which is primarily cell mediated. Titres of all these antibodies tend to be high at presentation and (according to prospective studies of high-​risk subjects) during the months leading up to this. Thereafter, antibody levels decline progressively (espe- cially to insulin and ZnT8) and may even become undetectable, possibly through dwindling of the antigen load that perpetuates autoimmunity as any remaining β cells disappear. Natural history of type 1 diabetes The damage to β cells might be initiated by direct viral attack, en- vironmental toxins, and/​or a primary immune attack against specific β-​cell antigens such as GAD65, perhaps via molecular mimicry. T-​helper lymphocytes (CD4+) are activated by β-​cell antigens presented together with diabetogenic class II antigens by antigen-​presenting cells (dendritic cells). Activated T-​helper cells produce cytokines that attract T and B lymphocytes and encourage them to proliferate in the islet, leading to insulitis. B lymphocytes might then damage β cells by producing antibodies against released β-​cell antigens, while cytotoxic (CD8+) T lymphocytes directly at- tack β cells carrying the target autoantigens. Insulitis is a patchy and unpredictable process that might flare up after encounters with new environmental triggers such as viral infections, but which can also fade and abort for unknown reasons. Several years of progressive autoimmune damage usually precede the clinical onset of diabetes. This long prediabetic phase is asymp- tomatic, although careful testing (e.g. with the intravenous glucose tolerance test) reveals loss of the first phase, then increasingly obvious disturbances of insulin and C-​peptide secretion, and eventually glu- cose intolerance. Finally, when the β-​cell mass has been eroded to a critical level (probably 5 to 10% of normal), falling insulin secretion can no longer restrain hyperglycaemia and clinical diabetes develops. Residual β-​cell mass is variable at presentation of type 1 dia- betes: some newly diagnosed type 1 patients are C-​peptide positive, and β-​cell secretion may improve temporarily during the ‘honeymoon period’ that can follow the lowering of blood glucose when insulin treatment is started (see following section). As a result, it is not possible to absolutely distinguish type 1 and type 2 diabetes by measurement of C-​peptide at diagnosis although levels tend to be very much lower in type 1 diabetes (see Fig. 13.9.1.6, panel (a). With continuing β-​cell destruction, endogenous insulin production declines progressively, and more than 90% of type 1 patients become permanently C-​peptide negative within 5 years of presentation. The loss of C-​peptide is more rapid in individuals diagnosed in childhood than in new onset dis- ease in adults (see Fig. 13.9.1.6, panels (b) and (c). Ultimately, insulitis burns itself out and the immune cells retreat, leaving islet remnants that are devoid of β cells, but which still contain intact α, δ, and PP cells. Interestingly, there is a concomitant 50% reduction in the size of the exocrine pancreas in patients with long-​standing type 1 dia- betes: this appears not to result in clinically significant malabsorption and the mechanism by which it occurs is unknown. The protracted prediabetic phase provides an opportunity to prevent subjects with active insulitis from developing clinical disease. Progression is more rapid in subjects with circulating islet autoantibodies directed at more than one islet antigen (see Fig. 13.9.1.7); indeed, the risk of diabetes in subjects with three anti- bodies appears to be more than 90% although it may take 15 years or more to present. A combination of autoantibody titres and gen- etic markers (HLA haplotypes) can be used to predict the chances of the disease developing in high-​risk subjects, such as the siblings of children with type 1 diabetes; various immunosuppressive and immunomodulatory treatments are currently undergoing clinical trials as forms of early intervention or prevention. Preservation of residual insulin at diagnosis has potential to substantially improve metabolic control, including a 50% reduction in hypoglycaemia, 2–​3-​fold more individuals reaching glycaemic targets and reduced long-​term complications. Metabolic disturbances of type 1 diabetes In untreated type 1 diabetes, insulin concentrations are generally 10 to 50% of non​diabetic levels in the face of hyperglycaemia which would normally greatly increase insulin secretion. Such severe de- ficiency cannot sustain the normal anabolic effects of insulin and

13.9.1  Diabetes 2479 10.0 C-peptide (ng/ml) 5.0 4.0 3.0 2.0 T1DM T2DM *µ = 2.66 *µ = 0.38 Adults (N = 2432) N = 1694 1.0 0.9 0.8 0.7 0.6 0.5 Stimulated C-peptide (mmol/litre) Stimulated C-peptide (mmol/litre) 0.4 0.3 0.2 0.1 0.0 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 1 2 3 4 5 6 7 8 Duration of T1DM (years) Duration of T1DM (years) Adoleseents (N = 1304) N = 838

N = 466 92% 6% 2% 15% 11% 0.4% 2.6% 97% 22% 67% 33% 52% 9 10 11 12 13 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 14 15 N = 738 *P<0.001 1.0 0.0 (a) (b) (c) Fig. 13.9.1.6  Loss of insulin production in type 1 diabetes. Panel (a) compares fasting insulin C-​peptide levels at diagnosis in children with type 1 vs. type 2 diabetes showing that although the levels are very different, there is still some degree of overlap; the lower panels show the peak C-​peptide during a mixed meal tolerance test in patients more than 18 years of age (panel (b), adults) or less than 18 years of age (panel (c)) at onset of diabetes and when tested after 1 to 15 years duration of diabetes, as part of screening for entry into the DCCT study. Note that C-​peptide levels fall faster in younger subjects so that after 5 years 8% of adults but only 3% of children have a peak C-​peptide level > 200 pmol/​litre. Panel (a) adapted from Katz LE, et al. (2007). Fasting C-​peptide and insulin-​like growth factor-​binding protein-​1 levels help to distinguish childhood type 1 and type 2 diabetes at diagnosis. Pediatr Diabetes, 8, 53–​9; panels (b) and (c) source data from Palmer JP, et al. (2004). C-​peptide is the appropriate outcome measure for type 1 diabetes clinical trials to preserve β-​cell function: report of an ADA workshop, 21–​22 October 2001. Diabetes, 53, 250–​64.

section 13  Endocrine disorders 2480 leads to runaway catabolism in carbohydrate, fat, and protein me- tabolism. Each of these processes accelerates hyperglycaemia, while the oxidation of excess free fatty acids generated by triglyceride breakdown can result in diabetic ketoacidosis. Carbohydrate metabolism Basal hyperglycaemia is due mainly to unrestrained production of glucose by the liver and is accentuated after eating by the failure of glucose to be cleared peripherally (see Fig. 13.9.1.4). Hepatic glu- cose output is boosted, especially by increased gluconeogenesis: the normal inhibition of the process by insulin is lost, while the supply of gluconeogenic precursors (glycerol from lipolysis, amino acids such as alanine from protein breakdown) is increased. Enhanced gluconeogenesis in the kidney may also contribute. Postprandial glu- cose uptake into muscle and fat, mediated by insulin and GLUT-​4, is greatly decreased; this is partly offset by increased non-​insulin-​ dependent glucose uptake into peripheral tissues, via glucose transporters that do not require insulin. The overall result is hyper- glycaemia, commonly in the range of 15 to 25 mmol/​litre and higher after meals. Glucose concentrations of over 40 mmol/​litre are not uncommon during intercurrent illness and especially when insulin treatment is omitted or not increased sufficiently. Fat metabolism Lipolysis is stimulated by severe insulin deficiency, generating gly- cerol (a gluconeogenic precursor) and free fatty acids, the substrate for ketone formation. Ketogenesis is particularly enhanced by con- comitant glucagon excess (see next). Mobilization of body fat con- tributes to the marked weight loss in untreated type 1 diabetes. Protein metabolism Loss of the net anabolic effect of insulin encourages catabolism of proteins (primarily through the proteasome-​mediated pathway), thus generating amino acids including gluconeogenic precursors such as alanine and glutamine. Muscle wasting may be prominent. Role of counterregulatory hormones The effects of hypoinsulinaemia are compounded by the counter­ regulatory hormones which are secreted in excess in response to stress (e.g. infections, myocardial infarction, trauma, surgery) and when circulating volume falls (e.g. in hyperglycaemic dehydrated patients). Insulin deficiency also leads to increased glucagon secre- tion, because insulin normally inhibits the α cells. Glucagon increases hepatic glucose production, both by driving glycogen breakdown and by increasing uptake of glucogenic amino acids by the liver and enhancing gluconeogenesis. It also stimu- lates ketogenesis by increasing entry of free fatty acids (as their fatty acyl-​CoA derivatives) into liver mitochondria (see Fig. 13.9.1.12). Glucagon excess is an important factor that promotes diabetic keto- acidosis, acting synergistically with insulin deficiency (see next). Cortisol and catecholamines enhance gluconeogenesis. Cortisol, catecholamines, and growth hormone oppose the lipogenic action of insulin and favour lipolysis, in the presence of hypoinsulinaemia. Cortisol is a powerful inducer of proteolysis, whereas growth hor- mone cooperates with insulin to stimulate protein synthesis. Clinical features of type 1 diabetes The classical presentation of untreated or poorly controlled type 1 diabetes reflects the consequences of catabolism and hypergly- caemia (see Table 13.9.1.2). These features usually develop progres- sively and quite rapidly over a period of a few days to a few weeks. Diuresis is due mainly to the osmotic effect of glucose remaining in the renal tubule, when its concentration exceeds the reabsorption threshold for glucose (corresponding generally to plasma glucose levels of about 10 mmol/​litre). The osmotic loads of urinary ketones and of electrolytes that are obligatorily lost with glucose also con- tribute. Urine output may reach several litres per day, causing poly- uria, nocturia, and in children, enuresis. Thirst generally parallels urine output and can be very intense; it is characteristically made worse by sugar-​rich drinks. Taking water to bed at night is a useful sign of pathological thirst. A high fluid in- take is an important homeostatic response to diuresis, and patients unable to drink (e.g. through nausea in ketoacidosis) can rapidly be- come dehydrated and hypovolaemic. Weight loss, due to loss of fat and muscle and later to dehydration, can be dramatic and reach several kilograms over a few weeks. The energy deficit caused by catabolism and urinary losses of glucose can amount to several hundred calories per day. Appetite is often increased; the mechanism in humans is not known; falls in circu- lating leptin and insulin, both of which act on the central nervous system to inhibit feeding, are probably responsible for hyperphagia in diabetic rodents. Systemic symptoms include tiredness, malaise, lack of energy, and muscular weakness. Blurred vision is commonly due to (reversible) changes in the shape of the lens due to osmotic shifts, typically causing long-​ sightedness. Rarely, acute ‘snowflake’ cataracts develop because of reversible refractile changes, rather than the permanent denatur- ation of lens proteins in senile cataract. Infections are often present because hyperglycaemia predisposes to infections and also because infections stimulate the secretion 100 80 60 40 20 0 0 5 None None Number of islet autoantibodies 1 islet 1 islet 2 islet 2 islet 3 islet 3 islet Number of events Age (Years) Proportion of patients without type 1 diabetes (%) 358 227 474 430 168 250 112 82 272 5253 20 19 118 1161 .. 1 9 44 12318 8875 10 15 20 Fig. 13.9.1.7  Proportion of patients without type 1 diabetes in relation to the number of islet autoantibodies after being followed up from birth. Source data from Ziegler AG, et al. (2013). Seroconversion to multiple islet autoantibodies and risk of progression to diabetes in children. JAMA, 309, 2473–​79.

13.9.1  Diabetes 2481 of stress hormones. Genital candida infections, causing recurrent pruritus vulvae in women and balanitis in men, are frequent and should always prompt testing for diabetes. Pyogenic skin infections and urinary tract infections, sometimes complicated by severe renal damage, are also common, and certain rare infections have a par- ticular predilection for diabetic people (see next). Diabetic ketoacidosis presents with hyperglycaemic symptoms, which are usually severe, together with nausea and vomiting, acidotic (Kussmaul) breathing, the smell of acetone on the breath, and, espe- cially in children, altered mood and clouding of consciousness that may progress to coma. Diabetic ketoacidosis is described in detail later. Unlike type 2 diabetes, which is often present for several years be- fore diagnosis, hyperglycaemia in newly presenting type 1 patients develops too acutely for chronic diabetic complications to appear. Because obvious symptoms appear quickly, very few cases are picked up fortuitously, although doctors who have forgotten to think of dia- betes in their differential diagnosis of weight loss or hyperventila- tion may be surprised when hyperglycaemia is detected by routine screening. With the rising incidence and awareness of diabetes in the general population (due to rising rates of type 2 diabetes), an increasing number of cases of type 1 diabetes are detected before ketosis develops—​giving rise, especially in adults, to confusion over whether the diagnosis is type 1 or type 2 diabetes. Prognosis of type 1 diabetes Before the introduction of insulin during the early 1920s, type 1 diabetes was invariably fatal, usually within months. With various semistarvation diets, hyperglycaemic symptoms could be improved somewhat and life extended by a few miserable months. Over the last 20 years the mortality rate from diabetic ketoacidosis has fallen from 8% to 0.67%, although one-​third of deaths in diabetic children and young adults are still due to metabolic emergencies, not- ably ketoacidosis. The main threat to survival with type 1 diabetes is now chronic tissue damage, particularly renal failure from nephropathy, and vascular disease, notably myocardial infarction and stroke. Throughout adult life, the overall risk of dying within 10 years is about fourfold higher for patients with type 1 diabetes than for their non​diabetic peers. Life expectancy There is encouraging evidence from Europe and the United States of America that the outlook for type 1 diabetes has improved over the last 10 to 20 years, with definite declines in the incidence of microvascular complications and extended survival—​at least in countries able to af- ford effective diabetes care. This is partly attributable to tighter control of hyperglycaemia, which can reduce by 30 to 40% the risks of nephrop- athy and retinopathy developing or progressing to a clinically signifi- cant degree (see next). Other measures have undoubtedly contributed, including better treatment of raised blood pressure and blood lipids. Tragically, however, in many parts of the world patients with type 1 diabetes still die today as they did a century ago, simply because insulin is not available or is not affordable. Type 2 diabetes Type 2 diabetes is a heterogeneous condition, diagnosed empirically by the absence of features suggesting type 1 diabetes (see Table 13.9.1.2) and of the many other conditions that cause hyperglycaemia (see Table 13.9.1.1). Diagnostic accuracy may depend on the thoroughness of investigation; for example, up to 10% of subjects with presumed type 2 diabetes show evidence of autoimmune β-​cell damage and thus prob- ably have slowly evolving type 1 diabetes (so-​called latent autoimmune diabetes in adults, LADA) and a further 1% will have monogenic diabetes. The term ‘type 2’ replaces ‘non-​insulin-​dependent’ and ‘maturity-​ onset’ which were both clumsy and misleading: many type 2 patients require insulin to control hyperglycaemia and increasingly type 2 diabetes is being diagnosed in (overweight) children. Epidemiology and demographic features Type 2 diabetes accounts for 85 to 90% of diabetes worldwide. It is very common, affecting at least 3 to 4% of the white populations in Table 13.9.1.2  Typical features of type 1 and type 2 diabetes, with some distinguishing characteristics Type 1 diabetes Type 2 diabetes Osmotic and glycosuric symptoms: polyuria, nocturia, enuresis; thirst, polydipsia; blurred vision;
genital candidiasis (pruritus vulvae, balanitis)

• → ++ ± → ++ Systemic symptoms: malaise, tiredness, lack of energy
• → ++ 0 → ++ Catabolic features: recent weight loss; muscle wasting and weakness
• → ++ 0 → + Ketoacidosis Spontaneous Rare; mostly precipitated by intercurrent illness Diabetic microvascular complications at presentation –​ ± Age at presentation Young > old Old > young Obesity Unusual ++ (Almost invariable in white people) Family history ±

Clinical insulin dependence (weight loss and hyperglycaemia without insulin replacement) + –​ Special investigations: C-​peptide Low, especially 5 years after diagnosis Normal or raised HLA DR3 or DR4 ++ –​ Islet cell antibodies: ICA, GAD, IA-​2, ZNT8 ++ –​ GAD, glutamic acid decarboxylase; HLA, human leukocyte antigen; ICA, islet cell antibodies; IA-​2, insulinoma associated antigen-​2, ZnT8 –​zinc transporter-​8

section 13  Endocrine disorders 2482 most countries, with rates rising to between 8 and 11% in Eastern Europe and North America. The prevalence rises with age to well over 10% of those over 70 years. It is substantially more common in certain immigrant populations living in more affluent countries (e.g. 10–​15% of adults in some Asian or Afro-​Caribbean groups in the United Kingdom are affected, compared with a prevalence of 4% in the white population). Type 2 diabetes is most commonly diagnosed in those over 40 years of age and the incidence rises to a peak at 45 to 64 years. However, much younger people are now presenting with type 2 dia- betes, following the rapid rise in childhood obesity. Up to one-​third of North Americans diagnosed as diabetic under 20 years of age have type 2 diabetes, with Afro-​Caribbean and Hispanic populations being at particular risk. Monogenic diabetes or maturity-​onset dia- betes of the young (MODY) due to single-​gene defects, commonly presents before 25 years of age in more than one generation, and is now classified separately (read on for more details). The prevalence of type 2 diabetes shows striking geographical variation—​entirely different from that of type 1—​and ranges from less than 1% in rural China to 50% in the Pima Indians of New Mexico. Prevalence is also rising rapidly, especially in developing countries and, worldwide, will increase by at least 50% within 10 to 15 years. This pandemic can be largely explained by Westernization, and is following in the wake of the obesity that is spreading throughout the world. The Pima Indians illustrate this process especially viv- idly; most developed and developing countries are now showing the same phenomenon, albeit more slowly. Diabetes was rare while the Pima tribes led a frugal existence in desert conditions and were lean and physically active. Following urban resettlement and exposure to overnutrition and inactivity, there were rapid increases in the preva- lence of obesity (currently 80% of adult Pima Indians have a BMI of over 30 kg/​m2) and later of type 2 diabetes. There is a 3:2 male preponderance among subjects with type 2 diabetes in Western countries although worldwide there is a 10% excess of females. Aetiology Type 2 diabetes is due to the combination of insulin resistance and β-​ cell failure, the latter preventing sufficient insulin secretion to over- come insulin resistance. These two components vary in importance between different individuals, who may be clinically quite similar, and each has numerous possible causes. Susceptibility is determined by the interactions between genes and environment. The steeply rising prevalence of type 2 diabetes suggests that diabetogenic genes are common and are now enjoying an unparalleled opportunity to express themselves through the global spread of Westernized life- style and obesity. Genetic factors Overall genetic susceptibility to type 2 diabetes is probably 60 to 90%, rather less than was previously deduced from twin studies. Generally, transmission does not follow simple mendelian rules, and this polygenic pattern reflects the inheritance of a critical mass of minor diabetogenic polymorphisms which interfere with insulin action and/​or insulin secretion. Having a first-​order relative with the disease increases an individual’s chances of developing it fivefold, representing a lifetime risk in white people of about 40%. However, this figure will be strongly influenced by modifiable factors, notably the BMI and physical activity of the at-​risk individual. Much progress has been made recently in identifying the gene loci predisposing to type 2 diabetes by using genome-​wide scanning in large population databases. Importantly, these findings have been verified by repeat analyses in other data sets to exclude spurious statistical findings arising from the very large number of statistical comparisons performed. At least 30 loci have been confirmed, with predicted effects on insulin resistance (PPARG) and obesity (FTO), but interestingly, a greater number of confirmed loci seem to relate to pancreas development and/​or insulin secretion (TCF7L2, KCNJ11, HHEX–​IDE, CDKAL1, CDKN2, IGF2BP2, and SLC30A8)—​see Fig. 13.9.1.8. Although confirmed, the influence of each locus is relatively weak: the strongest association is with TCF7L2 (odds ratio for dia- betes of high-​risk polymorphism is 1.5) with the remaining loci con- ferring odds ratios of 1.1 to 1.25. Taken together, the known loci still only explain a small proportion of the inheritance of type 2 diabetes, indicating that there are many more minor loci to be identified. Interestingly, only three of the defined loci for common polygenic type 2 diabetes are the same as those identified to cause the much rarer monogenic diabetes syndromes of maturity-​onset diabetes of the young (MODY, see next—​glucokinase, HNF-α, HNF-​1 β). Environmental factors These clearly play a critical part, because obesity and type 2 dia- betes are spreading too rapidly to be explicable by changes in the genome; environmental factors are also important in practice be- cause they may be modified to treat and prevent the disease. Known environmental diabetogenic factors mostly induce insulin resist- ance (e.g. obesity, pregnancy, intercurrent illness, certain drugs). Hyperglycaemia per se can both impair insulin sensitivity and in- hibit insulin secretion (glucotoxicity). Specific risk factors for type 2 diabetes Obesity, itself determined by both genes and environment, is one of the most important risk factors, apparently due to aggravation of insulin resistance (see earlier). The diabetogenic properties of excess fat depend not only on its bulk but also on its anatomical distribution and the time of life at which it is laid down. The risks of developing type 2 diabetes begin to increase steeply once the BMI exceeds 28 kg/​ m2; some studies estimate the risk at a BMI over 35 kg/​m2 to be 80-​ fold higher than for individuals with a BMI of less than 22 kg/​m2—​a lifetime risk of about 50%. Fat in the truncal (central) distribution is more diabetogenic than that deposited around the hips and thighs, and the visceral (intra-​abdominal) depot is strongly associated with insulin resistance. Increasing adiposity after the early twenties, espe- cially around the waist, aggravates the risk of a high BMI. Physical inactivity, especially from the twenties onwards, is an in- dependent predictor of diabetes in middle age, the risk increasing by about threefold for sedentary people as compared with regular athletes. This is due to worsening insulin resistance, which can be improved by physical training and may in part be due to changes in activity of the enzyme AMP kinase in skeletal muscle. The Barker–​Hales hypothesis suggests that poor fetal growth can programme enduring metabolic and vascular abnormalities that are manifested in adult life, especially in people who were underweight at birth but then become obese. These abnormalities include key features of the metabolic syndrome (hyperglycaemia, hypertension, dyslipidaemia), resulting in atheroma formation, myocardial infarction, and stroke (see earlier). Evidence, mainly

13.9.1  Diabetes 2483 from animals, suggests that maternal and therefore fetal malnutri- tion during a critical early phase of fetal development can reduce β-​cell mass and permanently impair insulin secretory reserve; de- ficiencies of sulphur-​containing amino acids may be responsible in experimental animals but the relevance to humans is unknown. Other studies suggest that insulin sensitivity may also be reduced into adult life. β-​Cell failure in type 2 diabetes β-​Cell failure is an obligatory defect in the pathogenesis of type 2 diabetes: near normoglycaemia can be maintained even in severe insulin resistance (e.g. due to mutations in the insulin receptor), as long as the β cell can respond to the challenge and secrete enough insulin to overcome the resistance. Subtle abnormalities of insulin secretion, including loss of the physiological pulses and of the first-​phase response to intravenous glucose injection, are seen in normoglycaemic subjects who later de- velop the disease. These defects presumably indicate that the β cell is already stressed in trying to produce enough insulin to overcome in- sulin resistance. Normoglycaemic first-​order relatives of type 2 dia- betic subjects also show loss of pulsatility of insulin secretion which might indicate an inherited tendency to β-​cell failure. The key role of β-​cell failure in predisposing to type 2 diabetes has recently been underlined by the finding that most of the confirmed genetic suscep- tibility loci for type 2 diabetes relate to islet cell function or devel- opment rather than insulin resistance (see earlier and Fig. 13.9.1.8). The mechanism of β-​cell failure in human type 2 diabetes is not known. Histologically, the islets in type 2 diabetes show no features of type 1 autoimmune insulitis, and β-​cell mass is not so dramatic- ally reduced. Animal models of the disease suggest various causes, including synchronized β-​cell apoptosis (possibly mediated by ni- tric oxide) in the Zucker diabetic fatty rat, and the deposition of amyloid fibrils (see earlier) in the rhesus monkey. Amyloid deposits are also prominent in the islets of some type 2 diabetic patients but may merely be due to dysfunctional β-​cell hypersecretion rather than the cause of β-​cell damage. Once hyperglycaemia is estab- lished, glucotoxicity per se may further worsen both insulin secre- tion and insulin resistance. Elevated free fatty acid levels resulting from insulin resistance have also been proposed to impair β-​cell function—​so-​called lipotoxicity—​but this remains controversial. In established type 2 diabetes, insulin secretion is unequivocally subnormal and tends to decline progressively with time, as illus- trated by the long-​term follow-​up data from the United Kingdom Prospective Diabetes Study. Initially, plasma insulin levels may be higher than in non​diabetic subjects but are still inappropriately low, as the normal pancreas would produce much higher insulin concentrations in response to diabetic levels of blood glucose. Conventional radioimmunoassays may overestimate insulin levels in type 2 diabetic patients because of cross-​reaction with incom- pletely processed insulin precursors (proinsulin and its split prod- ucts) released by the constitutive pathway which operates in the malfunctioning β cell (see earlier). Many type 2 patients ultimately need insulin replacement; this indicates relatively severe insulin deficiency, although still not as profound as in type 1 diabetes. Some type 2 patients who require insulin early have autoimmune markers characteristic of type 1 diabetes, suggesting that they in CDKALI, CDKN2A CDKN2B Reduced β-cell mass β-cell dysfunction Obesity Insulin resistance not due to obesity Reduced insulin secretion Insulin resistance Predisposition to type 2 diabetes MTNR1B, TCF7L2, KCNJ11 FTO IRS1, PPARG Fig. 13.9.1.8  Pathways to type 2 diabetes implicated by identified common variant associations. Type 2 diabetes results when pancreatic β cells are unable to secrete sufficient insulin to maintain normoglycemia, typically in the context of increasing peripheral insulin resistance. The β-​cell abnormalities fundamental to type 2 diabetes are thought to include both reduced β-​cell mass and disruptions of β-​cell function. Insulin resistance can be the consequence of obesity or of obesity-​ independent abnormalities in the responses of muscle, fat, or liver to insulin. Examples of susceptibility variants that, given current evidence, are likely to influence predisposition to type 2 diabetes by means of each of these mechanisms are shown. Reproduced from McCarthy MI (2010). Genomics, type 2 diabetes, and obesity. N Engl J Med, 363, 2339–​50. Copyright © 2010 Massachusetts Medical Society.

section 13  Endocrine disorders 2484 fact have an indolent variant of the disease. Although patients with type 1 diabetes have significantly lower insulin C-​peptide levels at diagnosis than in type 2, there remains overlap in the ranges (see Fig. 13.9.1.6) such that C-​peptide alone only has a sensitivity of 83% in diagnosing type 2 diabetes even in children. Natural history Longitudinal and cross-​sectional studies indicate that insulin re- sistance develops first and that compensatory increases in insulin secretion can initially maintain near normoglycaemia. Worsening insulin resistance is thought to drive the β cells towards maximal insulin output, a metastable stage that probably corresponds to impaired glucose tolerance (see earlier). Rescue is still possible if insulin resistance is decreased (e.g. through weight loss or insulin-​ sensitizing drugs): about 25% of subjects with impaired glucose tolerance return to normoglycaemia within 5 years. However, if insulin resistance persists or worsens, the β cells fail and insulin production falls. At this point, the brake-​limiting hyperglycaemia is released and blood glucose rises into the diabetic range. The bell-​ shaped response of insulin secretion, initially increasing to com- pensate but ultimately failing, has been termed the ‘Starling curve’ of the β cells because it recalls the classical plot of cardiac output against preload in heart failure. In common type 2 diabetes, these events usually take many years, and significant hyperglycaemia may have been present for several years at the time of diagnosis. The whole process can be greatly ac- celerated by acute increases in insulin resistance as those induced by steroid treatment or pregnancy, to give just two examples. Metabolic disturbances in type 2 diabetes Hyperglycaemia is the most obvious abnormality, the extreme case being the hyperosmolar non​ketotic state. Lipid metabolism is also disturbed but true ketoacidosis occurs only exceptionally and is usu- ally provoked by intercurrent events such as infections or myocar- dial infarction. Blood glucose concentrations are raised both in the basal (fasting) state and after eating. This reflects the impairment of insulin action in both liver and skeletal muscle, where insulin respectively shuts off hepatic glu- cose production and stimulates glucose uptake after meals. Hepatic glu- cose output is increased, due mainly to unsuppressed gluconeogenesis, and this is largely responsible for hyperglycaemia overnight and before meals. In muscle, GLUT-​4 activity and glycogen synthesis are especially decreased; this reduces insulin-​stimulated glucose uptake into muscle after meals, although basal glucose uptake (non-​insulin-​mediated glu- cose uptake; see earlier) is higher than in normal subjects because of the mass action effect of hyperglycaemia. The degree of hyperglycaemia varies widely: many patients have fasting plasma glucose levels of 8 to 13 mmol/​litre with postprandial peaks of up to 20 mmol/​litre, while values exceeding 60 mmol/​litre are not uncommon in the hyperosmolar non​ ketotic state. Insulin deficiency is less profound than in type 1 diabetes, so mo- bilization of triglyceride (loss of body fat, ketoacidosis) and catab- olism of protein (muscle breakdown) are not usually pronounced. Diabetic ketoacidosis may develop in patients with apparently typical type 2 diabetes who can subsequently be controlled by oral hypoglycaemic agents rather than insulin (see ‘Flatbush or ketone-​ prone diabetes’ next). Diabetic ketoacidosis is usually precipitated by severe intercurrent illness (e.g. myocardial infarction, stroke, or pneumonia) in which excessive secretion of counterregulatory stress hormones exacerbates the metabolic disturbance caused by relative insulin deficiency. Clinical features Many cases present with classical symptoms of osmotic diuresis, blurred vision due to hyperglycaemia-​related refractive changes in the lens, and genital candidiasis (see Table 13.9.1.2). Weight loss may occur but is generally less dramatic than with newly presenting type 1 diabetes and may not be obvious because many type 2 patients—​over two-​thirds in the United Kingdom—​are obese. Rapid or severe weight loss in patients who otherwise appear to have type 2 diabetes should be regarded with suspicion as it may point to an early need for insulin replacement (and possibly type 1 diabetes itself) or to coexisting illness: a well-​recognized but unex- plained association with recent onset type 2 diabetes is carcinoma of the pancreas. The hyperosmolar non​ketotic state can present with confusion or coma (see next); as mentioned earlier, diabetic ketoacidosis is rare. Chronic diabetic complications may be a presenting feature, be- cause hyperglycaemia severe enough to cause tissue damage may already have been present for several years. Extrapolating the num- bers of microaneurysms (which only develop at diabetic glucose concentrations) in type 2 patients at various intervals after diagnosis suggests that significant hyperglycaemia is present for an average of 5 to 7 years before diagnosis. Common problems are arterial disease (myocardial infarction, stroke, and peripheral vascular disease), cataracts—​which are especially common in the older population—​ and retinopathy, especially maculopathy, which can damage central vision, and foot ulceration. Increasing numbers of people with diabetes are detected by screening, either in high-​risk groups such as the obese and those with cardiovascular disease, or at routine health checks. Many of these are nominally asymptomatic but will admit to symptoms such as nocturia or perineal irritation if asked directly. Prognosis of type 2 diabetes A long-​held and prevalent misconception is that type 2 diabetes is mild. Some patients do have relatively unexciting or asymptomatic hyperglycaemia, but this can still be enough to cause complications which wreck the patient’s life just as much as in type 1 diabetes. Moreover, hyperglycaemia can be as hard to control (even with in- sulin) as in type 1 patients. Overall, life expectancy is shortened by up to a quarter in pa- tients with type 2 diabetes presenting in their forties, with vas- cular disease (myocardial infarction and stroke) being the main cause of premature death. Renal failure from diabetic nephrop- athy is becoming more common in type 2 patients as their survival from vascular complications improves, and the disease is now the most frequent pathology among people waiting for renal replace- ment therapy in the United States of America and some European countries. Type 2 diabetes is therefore an important threat to the patient’s health and survival, and must be taken seriously by patients and their medical attendants, even if the blood glucose concentrations are not dramatically raised. Accordingly, treatment guidelines for the disease are rigorous (see Table 13.9.1.3 and discussion later in this chapter).

13.9.1  Diabetes 2485 Monogenic diabetes: Maturity-​onset diabetes of the young (MODY) and neonatal diabetes Maturity-​onset diabetes of the young (MODY) While the vast majority of diabetes is polygenic in origin, there is now an expanding list of single-​gene loci that are associated with diabetes either in the neonatal period, in childhood, or in early adulthood. In 1974, Tattersall described a rare familial form of non-​insulin-​dependent diabetes that he distinguished from the generality of cases by its early age of onset, autosomal dominant inheritance, and an apparently low risk of microvascular compli- cations. The term ‘maturity-​onset diabetes of the young’ (MODY) came to be applied to individuals in which (1) a diagnosis of type 2 diabetes had been made under the age of 25; (2) there is evidence of autosomal dominant inheritance (diagnosis under the age of 25 in more than one generation); and (3) subjects can be managed without insulin. In 1992, the first conclusive evidence for the existence of mono- genic diabetes was provided when a subset of MODY (now known as MODY 2) was linked to the glucokinase gene locus. There are now at least eleven forms of MODY, accounting for around 1% of all cases of Table 13.9.1.3  Treatment targets for patients with diabetes Patient sub-group Source of recommendation Specific goal
of therapy Notes HbA1c Most EASD and ADA <53 mmol/mol (<7.0%) Regarded as ‘generally accepted’ (EASD) a ‘reasonable goal for many nonpregnant adults’ (ADA) Some EASD and ADA 42-48 mmol/mol (6.0-6.5%) Selected patients (e.g. short disease duration, long life expectancy, no ASCVD) if can be achieved without significant hypoglycaemia or other adverse effects Some ADA <64 mmol/mol (<8.0%) May be appropriate for patients with a history of severe hypogly­ caemia, also in circumstances including limited life expectancy, advanced complications, extensive comorbid conditions Blood pressure Most EASD <140/85 mm Hg Nephropathy with ‘overt proteinuria’ EASD <130/80 mm Hg If tolerated 10-year ASCVD risk <15% ADA <140/90 mm Hg Existing ASCVD or 10-year
ASCVD risk >15% ADA <130/80 mm Hg If can be attained safely Lipids T1DM and T2DM at ‘very high
risk’ (ASCVD, severe CKD, or
one or more CV risk factors
and/or target organ damage) EASD LDL-C <1.8 mmol/litre
(<70 mg/dL) Statin treatment recommended T1DM at ‘high risk’ (no other CV risk factors, no target organ damage) EASD Statin treatment may be considered irrespective of basal LDL-C T2DM at ‘high risk’ (no other CV risk factors, no target organ damage) EASD LDL-C <2.5 mmol/litre (<100 mg/dL) Statin treatment recommended Age <40 and no ASCVD or
10-year ASCVD risk <20% ADA Statin treatment not generally recommended, but moderate intensity statin may be considered based on presence of ASCVD risk factors and risk/benefit analysis Age <40 with ASCVD or
10-year ASCVD risk >20% ADA High intensity statin treatment recommended Age >40 and no ASCVD or
10-year ASCVD risk <20% ADA Moderate intensity statin treatment recommended Age >40 with ASCVD or
10-year ASCVD risk >20% ADA High intensity statin treatment recommended BMI T2DM - 25.0-26.9 kg/m2 ADA Advise diet, physical activity and behavioural therapy T2DM - 27.0-29.9 kg/m2 ADA

• consider pharmacotherapy T2DM - 30.0-34.9 kg/m2 ADA
• consider metabolic surgery in appropriate patients who do not achieve durable weight loss and improvement in comorbidities with nonsurgical methods T2DM - 35.0-39.9 kg/m2 ADA
• recommend metabolic surgery to appropriate patients who do not achieve durable weight loss and improvement in comorbidities with nonsurgical methods T2DM - >40.0 kg/m2 ADA
• recommend metabolic surgery to appropriate patients ASCVD, atherosclerotic cardiovascular disease; CV, cardiovascular; LDL-C, LDL cholesterol; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus. (1) lifestyle measures (e.g. smoking cessation, diet, exercise) recommended for all patients; (2) lipids—in patients with ASCVD, ADA recommends additional LDL-lowering therapy, e.g. ezetimibe PCSK9 inhibitor, if LDL-C >1.8 mmol/litre (>70 mg/dL) on maximally tolerated statin dose. Collated from various sources including the American Diabetes Association (ADA) website (https://care.diabetesjournals.org/content/42/Supplement_1) and European Association for the Study of Diabetes (EASD) website (https://www.easd.org/statements.html), accessed October 2019. Cardiovascular risk calculated as from http://tools.acc.org/ ascvd-risk-estimator-plus/#!/calculate/estimate/.

section 13  Endocrine disorders 2486 diabetes, in which the gene has been identified (see Table 13.9.1.4). Two forms predominate and have a distinctive clinical picture. MODY 2 (glucokinase mutations) is similar to the initial cases described by Tattersall with mild, non​progressive fasting hyperglycaemia and a very low risk of long-​term complications even without treatment. By contrast, MODY 3 (HNF1A mutations) is associated with progressive decline in glycaemic control. In addition, the renal threshold for glu- cose is low and there is a high risk of long-​term complications. Of par- ticular importance, MODY 3 is exquisitely sensitive to sulphonylureas and most patients wrongly diagnosed as having type 1 diabetes have been successfully transferred from insulin to sulphonylureas with im- provement in glycaemic control. Doses required may be as low as one-​ quarter of the normal adult starting dose. A diagnosis of MODY should be considered if: • There is a family history of young-​onset diabetes in more than one generation with at least one family member diagnosed under the age of 25. • Affected members are not markedly obese or of normal weight. • There is no evidence of insulin resistance—​no acanthosis nigricans, low insulin doses if insulin treated, high-​density lipoprotein greater than 1.2 mol/​litre. • Fasting serum C-​peptide (or urine C-​peptide:creatinine ratio) is detectable and within the normal range (not elevated). • Islet cell or anti-​GAD autoantibodies are absent. • Other associated features are present (see Table 13.9.1.4). None of these criteria are absolute and where doubt exists, advice from an expert centre should be sought before requesting genetic screening. Detection of MODY 3 is of particular importance be- cause of the excellent response to treatment with sulphonylureas. Neonatal diabetes Diabetes diagnosed under the age of 6 months is very unlikely to be type 1 (autoimmune) diabetes and alternative causes should be sought. Neonatal diabetes is insulin-​requiring diabetes, usually diagnosed with the first 3 months of life, and two subgroups have now been identified. Transient neonatal diabetes mellitus resolves around 3 months after birth although it can return in later life in up to 50% of cases. The most common cause is an imprinting ab- normality in the ZACN (ZAC) and HYMAI genes on chromosome 6 at the 6q24 locus. Macroglossia occurs in 23% of cases and is the only non​pancreatic feature. Presenting blood glucose levels are high (from 12 to >50 mmol/​litre) and insulin is required: if relapse oc- curs, this is normally not insulin requiring, at least in the initial stages. MODY 5 and KCNJ11 (Kir6.2) mutations (see next) occa- sionally also present as transient neonatal diabetes mellitus. Permanent neonatal diabetes mellitus requires continual insulin treatment from diagnosis. The most common cause is a mutation in the KCNJ11 gene, encoding the Kir6.2 subunit of the β-​cell KATP channel. Ninety per cent (90%) of cases are due to spontaneous (new) mutations so there is no family history. Affected individuals may have a range of neurological abnormalities that in the most se- vere form are referred to as DEND syndrome (developmental delay, epilepsy, and neonatal diabetes). Patients with Kir6.2 mutations be- have as insulin deficient, with a 30% risk of ketoacidosis and low or undetectable C-​peptide levels. However, most patients respond well to high doses of sulphonylureas, given at up to four times the normal Table 13.9.1.4  Maturity-​onset diabetes of the young (MODY) Type Genetic defect OMIM Frequency (% of MODY) Clinical features Sensitive to sulphonylureas MODY 1 HNF-​4α 125850 1% Rare. Similar to MODY 3 but renal threshold normal. Consider if MODY 3 screen negative May be sensitive MODY 2 Glucokinase 125851 20% Mild, non​progressive fasting hyperglycaemia (5.5–​8.5 mmol/​litre, HbA1c < 6%). Glucose increment < 3.5 mmol/​litre on OGTT. Complications rarely develop. Frequently do not response well to drug treatment and do not require it No MODY 3 HNF-​1α 600496 60% Young-​onset diabetes. Not particularly overweight and not insulin requiring (no ketosis) or surprisingly good control for several years on little insulin. Detectable C-​peptide beyond 3 years postdiagnosis. Low renal threshold. Large glucose increment (>5 mol/​litre) on OGTT. Progressive deterioration in glycaemic control and high risk of complications Extremely sensitive MODY 4 IPF-​1 606392 1% Rare. Possibly later-​onset disease. Some affected family members may not be diabetic Not determined MODY 5 HNF-​1β(TCF2) 137920 1% Renal cysts and diabetes. Renal, uterine, and/​or genital developmental abnormalities are typical initial presentation especially renal cysts. Gout, abnormal LFTs. Subclinical pancreatic exocrine insufficiency No MODY 6 NEUROD1 606394 <1% Rare Not determined MODY 7 KLF-​11 610508 <1% Rare Not determined MODY 8 CEL 609812 <1% Rare Not determined MODY 9 PAX-​4 612225 <1% Rare Not determined MOFY 10 INS 613370 <1% Rare Not determined MODY 11 BLK 613375 <1% Rare Not determined MODY X Unknown 15% Not defined Not determined HNF, hepatocyte nuclear factor; IPF-​1, insulin promoter factor 1; LFT, liver function test; NEUROD1, neurogenic differentiation 1 transcription factor; KLF-​11, Kruppel-​like Factor 11; CEL, carboxyl ester lipase; PAX-​4 (transcription factor); INS, insulin gene; BLK, B lymphocyte tyrosine kinase; OGTT, oral glucose tolerance test.

13.9.1  Diabetes 2487 adult therapeutic dosage (e.g. glibenclamide 0.5–​1 mg/​kg per day), with the restoration of insulin secretion. Occasionally, MODY 2 and 4 may also present as permanent neonatal diabetes mellitus as can other rare genetic syndromes (see next). Other types of diabetes (See Table 13.9.1.1.) Diabetes in pancreatic disease Chronic pancreatitis, most commonly due to alcohol abuse, causes diabetes that needs insulin in about one-​third of cases. Widespread flecks of fine to medium calcification are often scattered through the pancreas, outlining it on a plain abdominal radiograph. Concomitant destruction of the islet α cells means that glucagon secretion is lost as well as insulin; diabetic ketoacidosis is therefore rare, while hypo- glycaemia can be profound and prolonged—​a particular hazard in those who continue to drink alcohol. Acute pancreatitis causes acute hyperglycaemia in 50% of cases but few develop permanent diabetes. Carcinoma of the pancreas is associated with newly presenting type 2 diabetes, and should be suspected in older patients with weight loss (especially when accompanied by abdominal or back pain and jaundice). The mechanism is unknown but appears to be due to tu- mour products that cause insulin resistance rather than to β-​cell loss. Genetic diseases that cause diabetes through pancreatic damage in- clude haemochromatosis and cystic fibrosis. In one-​half of cases of haemochromatosis, heavy deposition of haemosiderin in the islets causes diabetes, usually requiring insulin; associated features are slate-​ grey skin pigmentation due to deposition of iron in the dermis (‘bronze diabetes’), cirrhosis, secondary gonadal failure, and pyrophosphate arthropathy. MRI shows abnormal signals in the liver and pancreas, while serum ferritin concentrations are greatly elevated; diagnosis is usually possible by means of molecular analysis of the HFE gene but Perls’ stain for iron deposition in a liver biopsy may be necessary (see Chapter 12.7.1). Diabetes due to excessive iron deposition in the pan- creas is also seen in children surviving thalassaemia major. Cystic fibrosis causes pancreatic exocrine failure, with an increasing risk of diabetes (often requiring insulin) that approaches 25% in subjects who survive beyond 20 years of age. Gestational diabetes This includes all degrees of hyperglycaemia (impaired glucose tolerance as well as overt diabetes) diagnosed during pregnancy in previously normoglycaemic women. The American Diabetes Association has re- cently removed type 2 diabetes diagnosed in the first trimester from the definition of gestational diabetes. It is covered in Chapter 14.10. Malnutrition-​related diabetes This controversial diagnostic category was omitted from the most recent WHO classification. It included ‘fibrocalculous pancreatic diabetes’ and ‘protein-​deficient diabetes mellitus’. Fibrocalculous pancreatic diabetes was identified by dense pancreatic fibrosis, the formation of discrete and often spectacularly large stones in the dilated pancreatic ducts, and recurrent abdominal pain; protein-​ deficient diabetes mellitus was a vaguer entity that lacked the pan- creatic stones. Patients conforming to these ‘syndromes’ were rare even in the tropical zones where they were described (<5% of all diabetes), and the current consensus is that they represent type 2 diabetes or chronic pancreatitis superimposed on malnutrition. Flatbush or ketone-​prone diabetes The term ‘Flatbush diabetes’ has been used to refer to diabetes in young Afro-​Caribbeans who present with profound diabetic keto- acidosis but later prove to be non​insulin dependent. It appears that at diagnosis they have both marked insulin resistance and impaired insulin secretion but the latter later recovers, sometimes sufficiently for them to go into prolonged remission. In at least one report there was an excess of HLA DR3 and DR4 alleles, but anti-​GAD autoanti- bodies are negative. More recently, the term ketone-​prone diabetes has been used for patients who experience an episode of ketoacidosis but have clinical features of type 2 diabetes (e.g. obesity). A distinc- tion is made between those with and without islet autoantibodies. This area remains to be more accurately classified. Fulminant type 1 diabetes This form of diabetes was first described in 2000 in Japan and refers to presentation with severe diabetic ketoacidosis but low HbA1c (<8.5%) relative to their initial marked hyperglycaemia, thus indicating an abrupt onset. Additional typical features include a short history of symptoms (2–​10 days), raised pancreatic enzyme levels, and negative anti-​GAD (and other) autoantibodies. Prevalence in Japanese and Korean populations may approach 20–​30% of cases of rapid onset diabetes with ketosis, especially where the presentation is in adult- hood and/​or in pregnancy. It is rare in other races including white populations. Pancreatic biopsy reveals T-​cell infiltrates in the exo- crine pancreas, but without insulitis or features of acute pancreatitis. Mitochondrial diabetes Maternal transmission of mutations in mitochondrial DNA (mtDNA), especially the A3243G substitution in the leucine tRNA gene, can result in maternal inheritance of diabetes. Typical clinical presentation includes a presentation age of 20 to 50 with associated sensorineural deafness and short stature as in MIDD syndrome (maternally inherited diabetes and deafness). There is progressive non-​autoimmune β-​cell failure which may progress rapidly to in- sulin dependence (40% are insulin-​dependent within 4 years) The same mutation occurs in MELAS syndrome (mitochondrial my- opathy, encephalopathy, lactic acidosis, and stroke-​like episodes) and both MIDD and MELAS can occur in the same family. The ratio of mutant to wild-​type mtDNA in the blood (i.e. the degree of heteroplasmy) at diagnosis does not correlate with disease pheno- type or severity, presumably because it does not reflect the degree of heteroplasmy in other tissues such as the pancreas. Management of diabetes The treatment of diabetes has traditionally concentrated on correcting hyperglycaemia with the aim of abolishing the symptoms of hypergly- caemia, preventing acute complications such as diabetic ketoacidosis and preventing long-​term complications. Intensive control of hyper- glycaemia particularly in the early years of diagnosis has been shown to reduce both microvascular and macrovascular complications in type 1 and type 2 diabetes. The principal cause of morbidity and pre- mature death is cardiovascular disease. Intensive control of hypergly- caemia in those with long-​standing diabetes does not confer the same cardiovascular protection and some studies have shown an increased mortality in this group. Control of hyperlipidaemia and hypertension

section 13  Endocrine disorders 2488 are the key factors in reducing cardiovascular disease in this group of patients however reducing hyperglycaemia remains important to reduce the risk of microvascular disease. The current treatment tar- gets for both type 1 and type 2 diabetes (see Table 13.9.1.3 - low risk column) are therefore more holistic, tackling cardiovascular risk fac- tors and obesity in addition to hyperglycaemia. This section describes the roles of lifestyle modification and antidiabetic drugs, followed by specific treatment strategies for type 1 and type 2 diabetes. Diet and lifestyle modification and
management of obesity About 80% of patients with type 2 diabetes are obese, as are at least 30% of those with type 1 disease. Obesity is arguably one of the greatest obstacles to successful management of diabetes: it worsens insulin resistance, dyslipidaemia, and hypertension, and is now rec- ognized in its own right as a risk factor for coronary heart disease. Proven benefits of 10% weight loss in type 2 patients with a BMI of 30 to 40 kg/​m2 include falls in fasting glucose of 2 to 4 mmol/​litre and a 1% decrease in HbA1c—​comparable with sulphonylureas or metformin—​and reduced dosages of antidiabetic drugs, including insulin. There may also be variable improvements in blood pressure and dyslipidaemia (decreased triglycerides and low-​density lipopro- tein cholesterol, increased high-​density lipoprotein). The traditional focus on obesity has been on type 2 diabetes, but there is no reason to assume that the cardiovascular hazards of obesity do not also apply to type 1 diabetes. Weight reduction is important in the management of people living with diabetes who are obese. The aetiology of obesity is often com- plex with psychological factors as important as physiological fac- tors and this may result in weight loss being surprisingly difficult to achieve. Modern approaches include peer support programmes, be- haviour modification techniques such as motivational interviewing and incorporating food labelling and shopping advice into structured education programmes and such measures have all shown benefit. It is important that all members of the diabetes team recognize the importance of lifestyle management so that consistent advice and ap- propriate emphasis may be placed on this. The notion of the ‘diabetic diet’ must now finally be laid to rest. Traditionally, carbohydrate intake was restricted because of the sim- plistic assumptions that sugar alone raised blood glucose and might even be diabetogenic; this strategy favoured a high fat intake that undoubtedly helped to sustain obesity and probably predisposed to atheroma. Current advice is close to the healthy eating recommenda- tions for the whole population and can therefore be suggested for the patient’s entire family, which will greatly increase the chances of com- pliance. It may also have benefit for the rest of the family as having a spouse with type 2 diabetes is a risk factor for type 2 diabetes. The following diet and activity recommendations apply to both type 1 and type 2 diabetes. The aims are to: • correct obesity, which worsens insulin resistance, reduces the ef- ficacy of glucose-​lowering, antihypertensive, and lipid-​modifying drugs, and is an independent risk factor for macrovascular disease (management of obesity is discussed in detail in Chapter 11.6); • reduce cardiovascular risk, by limiting fat, cholesterol, sodium, and alcohol intakes; • avoid hypoglycaemia in patients receiving insulin or sulphonylureas by optimizing the timing and content of meals. The steps in designing dietary advice for the individual patient are shown in Fig. 13.9.1.9. Reducing total energy intake This should be reduced by 500 to 600 kcal/​day (2100–​2520 kJ/​day) in patients who are overweight (BMI >28 kg/​m2). This energy deficit mobilizes fat preferentially, whereas protein, glycogen, and water are also lost with more aggressive energy restriction; initially, the rate of weight loss will be 0.5 to 1.0 kg/​week (adipose tissue contains c.7000 cal/​kg or 29 400 J/​kg). (unchanged) The desired energy intake should be calculated from standard for- mulae that employ the subject’s age, sex, weight, and level of physical activity to estimate energy expenditure, which must equal energy intake under steady state conditions. The standard dietary history is not useful for trying to assess energy intake, because overweight subjects consistently under-​report how much they eat. Specific ad- vice about how to cut energy intake is best left to the dietitian, but hinges on reducing fat intake—​a simple message that can be re- inforced by the entire diabetes care team. Fat-​rich foods not only have the highest energy density (9 cal/​g or 38 J/​g, compared with 4 cal/​g (17 J/​g) for carbohydrate and protein), but also have poor sa- tiating effects and so tend to encourage overeating. The initial target should be a 10% loss of starting weight, not the ‘ideal’ body weight or BMI, which is only rarely attained by obese diabetic patients. When energy intake is cut acutely, type 2 patients often show an immediate fall in blood glucose, due to a drop in hep- atic glucose output, even before weight loss begins. Weight loss during an energy deficit of 500 to 600 cal/​day (2100–​2520 J/​day) is a slow process: for a 100 kg patient, a 10% weight loss may take several months. Frequent contact and encouragement are the best predictors of success, and the patient should be re- assured that weight loss by a small but tolerable change in lifestyle is much more likely to be maintained than weight lost by a crash diet. As weight falls, resting energy expenditure also declines: it is proportional to lean body mass, which also decreases, although at a slower rate than fat. This means that greater reductions in energy intake (>600 cal/​day or 2520 J/​day) will be needed to maintain the same rate of weight loss. If the 10% target is met, further loss to- wards an ‘ideal’ BMI of around 23 kg/​m2 may be feasible. Total energy intake Calculate current Reduce by 500–600 cal/d if BMI>28 Protein (10–15%) Other recommendations Salt < 6g/d Alcohol < 3 U/d for men, < 2 U/d for women Avoid ‘diabetic’ foods Percentage of total energy intake Carbohydrate (> 55%) Complex, fibre-rich Sucrose < 50g/d Fat (< 30%) Saturated < 10% Cholesterol < 250 mg/d Fig. 13.9.1.9  Dietary recommendations for people with diabetes. These guidelines now reflect healthy eating for the general population, rather than a diabetic diet.

13.9.1  Diabetes 2489 Weight loss is harder to achieve in diabetic patients than in their non​diabetic counterparts; possible reasons include fears about sugar rather than fat, and the adipogenic effects of insulin, sulphonylureas, and thiazolidinediones. In practice, weight loss of even 10% is not commonly achieved by diet and lifestyle modifi- cation alone; only 15 to 30% of newly diagnosed type 2 diabetic patients can normalize glycaemia initially by this means, and fewer than 10% can sustain this for 5 years or more. The progressive β-​cell dysfunction in type 2 diabetes (see earlier) makes it inevitable that the proportion of ‘dietary failures’ will increase steadily. However, a recent trial suggested that 46% of patients on a meal-replacement diet could achieve remission of diabetes within 6 years of diag- nosis, especially if weight loss >15 kg was achieved. Improving dietary composition Intakes of fat, salt, and refined sugar are generally too high in Westernized populations. Current recommendations for healthy eating are based on evidence of beneficial effects on body weight, glycaemic control, lipids, and blood pressure (see Fig. 13.9.1.9). Fat should provide less than 30% of total energy intake (in most industrialized countries, it accounts for 40%). Polyunsaturated or monounsaturated fats (e.g. sunflower or olive oils, respectively) are preferred to saturated animal fats, which should comprise less than 10% of total energy intake. Patients may need to be reminded that ‘good’ unsaturated fats still contain 9 cal/​g (38 J/​g) and therefore sustain obesity just as effectively as the others. Cholesterol should be limited to less than 250 mg/​day (less if dyslipidaemia is present). Carbohydrates should account for more than 55% of total en- ergy intake, preferably in the form of foods rich in soluble fibre (e.g. pulses, root and leaf vegetables, and fruit); the current WHO recommendation for the general population is for the consump- tion of at least five portions of fruit or vegetables per day. Sugary drinks (especially fizzy glucose solutions that are supposed to give energy) should be avoided, except to treat hypoglycaemia. The pre- sent recommendation, which seems reasonable but is not based on evidence, is to limit added sucrose to less than 25 g/​day and total sucrose intake to less than 50 g/​day. Protein should contribute 10 to 15% of total energy—​close to current levels in the general population. Sodium intake should be less than 6 g/​day, and less in patients with hypertension. Alcohol contains 7 cal/​g (29 J/​g), and beers and wines in par- ticular can be fattening. Intake should not exceed three units (30 g) per day in men and two units (20 g) per day in women, and should be further limited or avoided in those with hyperten- sion or obesity. Alcohol can delay recovery from hypoglycaemia (see next); ‘diabetic’ beers (low in sugar, but strong in alcohol) and spirits with sugar-​free mixers are especially likely to provoke hypoglycaemia. Moderate amounts of sucrose are acceptable (see earlier), and non​caloric sweeteners (such as aspartame) have no adverse metabolic effects. Diabetic sweets and foods contain sorbitol or fructose instead of glucose, and are an expensive way to get diar- rhoea; they should be avoided by patients, and withdrawn by the manufacturers. Optimizing meal patterns Judging the size and content of meals so as to limit glycaemic excur- sions remains an art rather than a science, and a skill which some patients develop with experience. Dosages of glucose-​lowering drugs that act acutely to cover meals (short-​acting insulin and sulphonylureas) can be tailored reasonably accurately to meals of similar composition but may not be matched to other meals, even when the total weights of carbohydrate, fat, and protein are similar. There has been much interest in the ability of various foods to raise blood glucose, usually measured as the ‘glycaemic index’; that is, the area under the curve of the rise in plasma glucose after eating a standardized load (50 g) of the food, expressed as a per- centage of the area under the glucose curve after ingesting 50 g of glucose. Foods with a low glycaemic index include pulses and cereals, probably because of their high fibre and complex carbo- hydrate contents, while bread has a surprisingly high index. The glycaemic index of many foods such as potatoes and pasta varies widely according to the method of cooking (and even the shape of the pasta), and mixing different foods in a real-​life meal has unpre- dictable effects on the overall postprandial glucose rise. It may be sensible to base meals around components with a low glycaemic index, but it is clearly not feasible to use the index to adjust dosages of antidiabetic medication. Appropriate portion size in meals is also important in limiting overall calorie intake. Portion size has crept up inexorably in res- taurants in many countries and probably contributes to the observed association between excessive weight gain and eating outside the family home. Increasing physical activity Short-​term exercise and improved physical fitness both in- crease insulin sensitivity, partly through increased translocation of GLUT-​4 units to the surface of skeletal muscle cells resulting in increased glucose uptake; this effect is independent of insulin and can enhance glucose uptake (under clamp conditions) better than metformin or the thiazolidinediones. Physical training also improves muscle blood flow. Several studies, notably the Finnish and American Diabetes prevention trials, have demonstrated that regular physical exercise reduces by over 50% the risk of im- paired glucose tolerance progressing to type 2 diabetes. There is also evidence that it significantly decreases cardiovascular events. Exercise must therefore be encouraged in all diabetic patients, but the advice must be realistic, achievable, and safe. Brisk walking for 30 to 40 min every day is better physiologically than a hectic workout in the gym once or twice a week, and is within almost everyone’s reach. Potential hazards of exercise include hypoglycaemia in patients on sulphonylureas or insulin, which may be delayed by several hours (see next), and cardiac disease. Patients at risk should have an ECG, with consideration for an exercise tolerance test and echo- cardiography, and appropriate treatment for ischaemic heart disease or heart failure. Exercise remains beneficial and important in these cases but should be built up gradually. Specific advice on exercise may be needed for those with neuropathy or active foot disease to avoid precipitating or exacerbating a foot lesion. Antiobesity drugs and bariatric surgery in diabetes Antiobesity drugs may be indicated in selected obese diabetic pa- tients with a BMI over 28 kg/​m2 and who have demonstrated, by losing weight beforehand through diet and exercise alone, that they are prepared to make long-​term changes in their lifestyle. Without this commitment, clinically useful weight loss is unlikely to be achieved or maintained beyond the period of drug prescription; the

section 13  Endocrine disorders 2490 medical and pharmacoeconomic benefits of modest weight loss for a couple of years in the obese patient’s middle age are not known but are probably not dramatic. Therefore, these medications should al- ways be regarded as an adjunct to dietary modification and exercise rather than an alternative to these interventions. Until recently the only drug available in many countries was orlistat, a gastrointestinal lipase inhibitor. With this, up to 30% of obese type 2 patients lose 10% or more of body weight within 6 to 12 months, HbA1c can fall by 1% or more, and dosages of glucose-​ lowering drugs, including insulin, may be decreased. Previously, the selective type 1 cannabinoid (CB1) receptor antagonist rimonabant was introduced; this reduces insulin resistance and may have the additional benefit of promoting smoking cessation. However, the exacerbation of pre-​existing depression or anxiety has resulted in the drug being withdrawn by the manufacturer. Sibutramine, a com- bined serotonin/​noradrenaline reuptake inhibitor, has also been withdrawn in Europe because of evidence of increased cardiovas- cular events. Recently liraglutide and semaglutide, incretins which stimulate insulin secretion and are used to treat diabetes (see later in chapter) have received regulatory body approval for the treat- ment of obesity. This is an expensive treatment option and is likely to limit use. Surgical treatment (bariatric surgery) with gastric banding, sleeve gastrectomy, gastric bypass and duodenal switch operations is indi- cated in selected patients with a BMI over 40 kg/​m2 or a BMI over 35 kg/​m2 if the diabetes is of recent onset (Table 13.9.1.3, and see Chapter 11.5). The number of operations performed in the United Kingdom has risen from 470 in 2003/​4 to 6500 in 2009/​2010. These operations are generally safe when performed by an experienced team and can achieve dramatic weight loss (up to 70% of excess fat, maintained for several years). Remission of diabetes occurs in ap- proximately 60% of patients with type 2 diabetes undergoing gastric bypass, with significant improvement in glycaemic control in others. The improvement in glucose metabolism precedes the weight loss and this may be due to the increase in GLP-​1 secretion and the re- duced glucagon levels which occurs in gastric bypass procedures. The reduction in morbidity and mortality as a result of bariatric sur- gery is likely to result in an increase in this treatment modality. Smoking Smoking is at least as common among diabetic patients as in the general population. Smoking greatly amplifies macrovascular risk in diabetic subjects: 10-​year mortality (mainly from myocardial in- farction) is about 50% higher than in diabetic non​smokers and twice as high as in non​diabetic non​smokers. Smoking may also accelerate the progression of nephropathy and possibly retinopathy. Many people living with diabetes, especially young women, con- tinue to smoke as a means of keeping thin, and because they fear gaining weight if they stop. Nicotine reduces fondness for sweet, energy-​dense, foods and may also be mildly thermogenic. Weight gain after stopping smoking averages 3 kg but about 20% of cases gain more than 6 kg; much of this weight is often lost within the fol- lowing 1 to 2 years, and it can be limited or prevented by careful dietetic support beforehand and in the months after cessation. Moreover, the risks of continuing to smoke are much greater than this degree of weight gain, especially in people living with dia- betes. Pharmacological support to overcome nicotine dependence including the use of nicotine replacement, antidepressants (e.g. bupropion, nortriptyline), and the nicotine receptor partial agonist varenicline each increases the chance of quitting by around two-​ to threefold. Electronic cigarettes (e-​cigarettes) are battery-​powered devices with cartridges that contain nicotine, flavours, and other chemicals which are inhaled as a vapour. These are a popular substi- tute for smoking as they are cheaper, widely available, and probably less harmful than cigarette smoking. However, there are concerns about their long-​term safety as clinical trials to ascertain safety have not been performed; there are also concerns that the sweet flavours and bright colours may encourage children to use the products and develop a nicotine addiction. Glucose-​lowering drugs Insulin Insulin is the cornerstone of treatment for type 1 diabetes and many with type 2 diabetes will eventually require insulin. Unfortunately, subcutaneously injected insulin cannot match the physiological pro- file of normal insulin secretion (see Fig. 13.9.1.10) and is a poor substitute for the finely tuned β cell with its nearly instantaneous capacity for ‘in-​flight’ adjustment. Moreover, insulin given subcuta- neously is absorbed into the systemic circulation rather than se- creted into the portal system where an immediate effect on the liver, and first pass clearance by that organ, are important in regulating the metabolic actions of insulin. History of insulin Insulin was traditionally extracted from pork and beef pancreases in acid ethanol and purified by precipitation and recrystallization. Plasma insulin concentration Nondiabetic insulin profile 24 Insulin lispro Human soluble Premixed (30:70, soluble:NPH) NPH Insulin glargine Clock time (h) 06 24 17 12 B L D 08 06 17 12 08 06 06 Fig. 13.9.1.10  A time course of insulin preparations, compared
with the normal diurnal profile of plasma insulin concentrations in
non​diabetic subjects (top). Breakfast (B), lunch (L), and dinner (D) were given as shown. Fast-​acting analogues (such as lispro) act more rapidly than conventional soluble ones but are still sluggish compared with normal prandial insulin release. Premixed insulins injected in the early evening cover the evening meal adequately, but the long-​ acting component can cause hyperinsulinaemia and troublesome hypoglycaemia in the small hours. None of the conventional long-​acting insulins reliably lasts 24 h; new long-​acting analogues such as insulin glargine or insulin detemir may provide adequate background insulin levels with once-​daily injections.

13.9.1  Diabetes 2491 Soluble (or ‘crystalline’) insulin prepared in this way was contam- inated with other islet proteins, including glucagon and pancreatic polypeptide, which had an adjuvant-​like effect and enhanced the im- munogenicity of the injected insulin; immune reactions were rela- tively common with the ‘dirty’ animal insulins in use until the 1970s (see next). More sophisticated purification techniques including gel filtration yield ‘highly purified’ or ‘monocomponent’ insulins which only rarely provoke immune reactions. Biosynthetic human-​sequence insulin, produced by recombinant DNA technology, entered clinical practice in the early 1980s and was the first genetically engineered protein to be used therapeutic- ally. The current approach is to introduce a synthetic gene for recom- binant proinsulin or a novel insulin precursor into yeast; the secreted product is then cleaved enzymatically to yield insulin and C-​peptide. There are some clinically relevant differences between the three species used therapeutically, although the shortcomings of insulin therapy relate mainly to the general pharmacokinetic misbehaviour of injected insulin. Human insulin is more lipophilic than porcine and bovine insulins and is slightly more rapidly absorbed: human soluble insulin especially may lower glucose faster and patients being transferred from other species should be warned of this and prandial doses reduced initially by one-​third. Human ultralente has a shorter and steeper action profile than its animal counterparts, particularly the bovine preparation; in real life, human ultralente behaves similarly to lente or isophane insulins and does not provide adequate basal levels for a full 24 h. Human insulin has been sug- gested to interfere with awareness of hypoglycaemia, but the balance of evidence does not support this view (see next). In many countries, animal insulins are no longer available and in most countries, it is very unusual to initiate animal insulin in a patient naïve to insulin. Typical users of animal insulin were started on it many years ago and have either experienced adverse effects of switching to human insulin or fear these effects. It important that a supply of animal in- sulin is available for such patients. Insulin absorption Absorption of insulin injected subcutaneously is slow and unpredict- able. Individual day-​to-​day variability in the amount absorbed within a few hours can exceed 50%. This means that small changes (<10%) in insulin dosage are unlikely to influence glycaemic control, and that insulin treatment should generally not be adjusted on a daily basis. Insulin absorption is influenced by the physical state of the insulin (soluble or delayed action), its speed of dissociation into monomers, the lipophilicity of the insulin species, and by blood flow and other characteristics of the injection site. Absorption is accelerated and may lead to noticeably faster falls in blood glucose, by stimulating general or local blood flow through exercise, hot climate, saunas, and/​or massaging the injection site. Conversely, absorption is slowed when subcutaneous blood flow is reduced (e.g. in cold conditions or hypovolaemic states). Lipohypertrophy, which may develop at frequently used injection sites, can significantly delay absorption—​ another reason for avoiding such areas. The anatomical site of injection also influences the rate of sub- cutaneous absorption. It is fastest in the abdomen (also a good site to limit any effects of exercise) and arm, and slower in the leg. These differences are often eclipsed by the overall variability in absorption. Absorption from muscle is faster, presumably because of its higher blood flow, and this route is preferred for the emergency treatment of hyperglycaemia or ketoacidosis if the best option, controlled intravenous infusion, is not practicable. Insulin preparations Soluble (regular or short-​acting) insulin injected subcutaneously begins to lower glucose within 30 min, has a peak effect between 1 and 2 h and lasts 3 to 5 h (see Fig. 13.9.1.9). This action profile is suit- able for covering meals or hyperglycaemic emergencies and for use in insulin pumps or infusions. However, it would have to be injected several times per day to control hyperglycaemia around the clock, at the cost of frequent hypoglycaemia. Long-​acting preparations are therefore used to cover basal insulin requirements. Various approaches have been used to slow and prolong insulin absorption, especially the chemical combination of insulin into complexes that release it slowly. More recently, synthetic analogues have been designed whose structure promotes precipitation when injected subcutaneously (see Fig. 13.9.1.9). Isophane insulins are also known as NPH (neutral protamine Hagedorn, from the director of the Danish laboratory where they were developed). They consist of a microcrystalline complex of insulin and the highly basic protein protamine (intriguingly isolated from fish sperm), together with trace amounts of Zn2+. Isophanes were derived from protamine–​zinc insulin which has a longer but highly unpredictable action profile. Isophanes produce peak plasma insulin levels at variable intervals between 4 and 8 h after injection, and their glucose-​lowering action wears off rapidly after 10 to 12 h. Insulin–​zinc suspensions (lente insulins) employ higher Zn2+ concentrations which encourage insulin to form crystalline lattices. Varying the reaction pH can produce either larger crystals which are particularly slow to dissolve (ultralente) or the amorphous semilente which releases insulin faster; the familiar lente is a 70:30 mixture of ultralente and semilente. Ultralente made with bovine insulin has a long, relatively flat action profile that can last 24 h or more, while human ultralente and the lente insulins of all three species have glucose-​lowering profiles similar to that of isophane. These long-​ acting insulins have a cloudy appearance and need to be shaken be- fore use to bring the insulin into suspension; visibly large particles or discoloration indicates that the insulin has become denatured and will have lost activity. Both lente and isophane insulins can be in- jected alone or mixed with soluble insulin. Premixed insulins contain a short-​acting soluble component to- gether with a longer-​acting lente or isophane. The aim is to provide prandial cover and then basal levels for several hours thereafter. A  variety of proportions of short-​acting insulin were previously available however a mixture with a 30:70 ratio is the only insulin generally available as most insulin manufacturers have reduced the available range in favour of analogue premixed insulins. All these insulin types have been produced with porcine-​, bovine-​ and human-​sequence insulins, and are available in cartridges for pen injection devices and some in disposable pen devices. Insulin analogues The pharmacokinetic properties of native insulins of any species are poorly suited to subcutaneous injection: soluble insulins (des- pite their high-​speed trade names) are too slow and prolonged in duration, while long-​acting insulins do not provide reliable enough 24-​h basal levels to be given once daily. Various synthetic

section 13  Endocrine disorders 2492 insulin analogues, designed by molecular modelling, have improved physicochemical characteristics. Fast-​acting analogues are modified at the C-​terminal end of the B chain, an area crucial in the self-​association of insulin molecules, so as to resist dimerization and hexamerization. Insulin hexamers formed in the subcutaneous injection site, dissociate slowly into absorbable monomers, and this is a rate-​limiting step in insulin absorption. Faster-​acting analogues include insulin lispro (interchanging the B28 lysine and B29 proline residues of the normal human sequence), in- sulin aspart, which carries aspartic acid at position B28 instead of the usual proline and insulin glulisine, which substitutes two amino acids. They have an appreciably faster and shorter action profile (see Fig. 13.9.1.9), and day-​to-​day variability in absorption and glycaemic responses may also be decreased. They can therefore reduce both prandial hyperglycaemia and the risk of postprandial hypoglycaemia. A newer formulation with vitamin B3 and L-arginine has even faster absorption. Despite theoretical advantages, meta-​analyses show only very modest reductions in HbA1c (c.0.1%) and reductions in hypo- glycaemic episodes when a fast-​acting analogue is substituted for soluble insulin, but there are significant improvements in quality of life generally attributable to the convenience of injecting immediately before or after meals rather than 30 min beforehand. Long-​acting insulin analogues have also been developed and have become the most commonly prescribed insulin in many countries. These are designed to give a smoother 24-​h profile than isophane (‘peakless’ insulin). At present three forms are available. Insulin glargine (A21 glycine, with two extra arginine residues extending the C-​terminal of the B chain) has an altered isoelectric point such that it is soluble in the vial or cartridge at pH 4 but precipitates under the skin at pH 7. Insulin detemir has a delayed action due to the addition of a fatty acyl chain that binds to plasma proteins such as albumin. It has a slightly shorter half-​life than glargine and can be given once or twice daily. Insulin degludec has a duration of action up to 42 hours (the addition of hexadecanedioic acid to lysine at B29 results in hexamer formation which creates a subcutaneous depot) which results in a basal insulin with less variation and greater flexibility of timing of injection. These analogues are clear in the vial or cartridge—​potentially a source of confusion with rapidly acting insulin. Claims have been made for improved HbA1c levels and less daytime hypoglycaemia, as well as weight loss or neutrality for detemir in type 2 diabetes, but the most robust finding appears to be a reduction in nocturnal hypoglycaemia. Side effects of insulin Hypoglycaemia is the most common complication of insulin treatment and can be unpleasant, debilitating, and occasionally life-​threatening. Mild hypoglycaemia is common—​many insulin-​treated patients have at least one episode most weeks—​but serious attacks causing unconsciousness or requiring the assistance of others are rare, about once every 3 patient-​years. Predictably, the frequency of both mild and severe attacks rises progressively when mean blood glu- cose levels are lowered by intensive insulin therapy; hypoglycaemia was three times more frequent in the tightly controlled group of the Diabetes Control and Complications Trial than in conventionally treated patients (see next). The manifestations and treatment of hypoglycaemia are covered in detail later. As discussed there, there is no convincing evidence that the use of human as opposed to animal insulins specifically interferes with awareness of hypoglycaemic symptoms. Weight gain is due to the anabolic effects of insulin, compounded by energy saved from glycosuria and sometimes by overeating after hypoglycaemia. Fear of weight gain discourages some patients, es- pecially young women, from taking their full insulin dosages; sur- prisingly often, deliberate omission or underdosing of insulin may be used by patients wishing to stay thin. Lipohypertrophy is the local thickening of subcutaneous tissue at frequently used injection sites, and is probably due to the lipogenic effects of high local insulin concentrations. Lipohypertrophy can be unsightly and can significantly delay insulin absorption. It can be prevented by rotating injections around several sites, and large le- sions can be removed by liposuction. Insulin allergy, now very rare with highly purified (especially human) insulins, can include local IgE-​mediated erythematous re- actions or even anaphylaxis. The commonest manifestation is re- peated pain at the site of injection. Lipoatrophy (localized pitting of the skin due to loss of subcutaneous fat) is apparently related to a chronic immune response generated around insulin crystals. Immune insulin resistance was seen with impure animal and espe- cially bovine insulins; high titres of insulin-​binding antibodies mop up free insulin from the circulation, resulting in very high insulin re- quirements (occasionally more than 10 000 U/​day), sometimes with unpredictable hypoglycaemia following the release of antibody-​ bound insulin. Insulin oedema is rare, and is usually seen in patients recovering from ketoacidosis who have been deprived of insulin for long periods. Fluid retention is probably due to the sodium-​conserving effects of insulin on the renal tubule, and may cause ankle or gener- alized oedema. It usually resolves within a few days, although treat- ment with diuretics or ephedrine may be required. Insulin neuritis refers to severe, persistent neuropathy following the use of insulin in individuals with very poor control glycaemic control; however, this is a consequence of the sudden improvement in metabolic state rather than a side effect of the insulin itself. Usually the pain resolves after some months. Insulin regimens Different individuals may need quite different insulin regimens, depending on their residual insulin reserve and severity of insulin resistance, as well as the desired tightness of control and the incon- venience that the patient will accept. Specific insulin schedules used in type 1 and type 2 diabetes are described later. Insulin dosage The healthy pancreas secretes about 40 to 60 U of insulin daily. Therapeutic insulin requirements range from less than this in thin type 1 patients (notably during the ‘honeymoon period’) to more than 200 U/​day in very obese, insulin-​resistant type 2 patients. High insulin requirements are often due to insulin resistance (see earlier), whereas low or falling dosages may be caused by weight loss (including anorexia nervosa), coeliac dis- ease, or loss of counterregulatory hormones in Addison’s dis- ease or hypothyroidism—​all these conditions being associated with type 1 diabetes. Changing dosages, especially in previously stable subjects, should prompt investigation of these possibilities. Some patients with ‘brittle’ diabetes or psychological maladapta- tion to life with diabetes may pretend to take very high or very low dosages (see later). Interestingly, insulin requirements via

13.9.1  Diabetes 2493 continuous subcutaneous infusion are typically 30% less than by intermittent injections. Types of insulin Formularies contain a bewildering assortment of insulins, many distinguished by imaginative claims about their action profile. Practically, prescribers should become familiar with regimens based on one or two preparations from the following broad classes: • Fast-​acting insulin:  either a soluble (regular) insulin such as Humulin S or Actrapid, injected 20 to 30 min before eating, or a faster-​acting analogue (e.g. lispro or aspart) which can be given immediately before or even shortly after eating. Fast-​acting in- sulin is either given as a fixed dose or as a ratio to the ingested carbohydrate. • Long-​acting insulin: either a lente insulin (e.g. Humulin Zn or Insulatard) or an isophane (e.g. Humulin I or Monotard). With either, circulating insulin falls to below useful levels after 10 to 14 h; they therefore need to be given twice daily in C-​peptide negative patients, although those with residual insulin secretion (or who are given three premeal injections of soluble insulin) may be able to maintain good glycaemic control with a single bedtime injection. Bovine (but not human) ultralente can last a full 24 h, but its absorption is erratic and it is rarely used. The long-​acting analogues currently available (such as insulin glargine, detemir, or degludec) have flat, steady action profiles that can provide basal insulin levels with a single daily injection. The timing of long-​acting insulin injections does not have to be yoked to meal- times as tightly as for soluble insulin. It is convenient to inject the dose at bedtime rather than together with the before-​supper sol- uble dose. This is because the action profile of long-​acting insulin clashes with the physiological changes in insulin sensitivity that occur overnight. Growth hormone is normally secreted in large spikes on entering deep sleep, typically between 24.00 and 02.00 h; this induces delayed insulin resistance which raises blood glucose during the hours leading up to breakfast. This ‘dawn phenomenon’ is accentuated if insulin levels are falling simultaneously—​as hap- pens if long-​acting insulin is injected in the early evening. Another hazard with this timing is potentially dangerous nocturnal hypo- glycaemia when insulin levels peak during the early morning (typ- ically 02.00–​04.00). Both problems can be reduced by delaying the long-​acting injection until bedtime (22.00–​23.00), when the risk of nocturnal hypoglycaemia is lower, and insulin levels gen- erally persist long enough to counteract the insulin resistance of the dawn phenomenon. If a second injection is required, this can be given with the before-​breakfast soluble insulin. Note that the long-​acting analogue insulins (glargine and detemir) should not be mixed in the same syringe as short-​acting insulin. • Premixed insulins (e.g. 30% short-​acting with 70% long-​acting) are obviously more convenient than giving short-​ and long-​acting insulins separately, but they lack flexibility. Premixed insulin in- jected 30 to 40 min before breakfast can achieve good glycaemic control through the morning and afternoon, but timing the evening dose is problematic: giving it before supper will tend to cause both early morning hypoglycaemia and fasting hypergly- caemia because of the time course of the long-​acting component, and simply increasing the evening dosage often makes nocturnal hypoglycaemia worse while failing to lower the before-​breakfast glucose. Premixed preparations including rapidly acting ana- logues such as insulin aspart or lispro and isophane are also avail- able and may be of some advantage. Premixed analogue insulin which has 50% short-​acting analogue and 50% isophane is some- times given three times a day with the main meals to those who are insulin resistant and this dosing regime has shown benefit. Insulin strengths Until recently all insulins in the United Kingdom were available at a U100 strength (100 units per ml). Recently insulins of higher strength have been introduced such as Tresiba U200 (insulin degludec U200) and Lantus U300 (insulin glargine U300). The rationale for the introduction of these insulins is to allow a smaller volume injection to be given. These insulins may have altered properties compared to the U100 formulation, for example Lantus U300 is a longer-​acting insulin compared to the U100 form and has shown a lower incidence of nocturnal hypoglycaemia during initiation in trials, although a higher dose may be required to achieve targets. These insulins come in pens which display the actual dose given however they should not be given via a standard U100 insulin syringe as this may result in double or triple the dose administered. Biosimilar insulins As the patents expire on some of the analogue insulins, less ex- pensive generic formulations have been created referred to as biosimilar insulins. These insulins contain the same molecule as the original insulin however differences in manufacturing pro- cesses may alter some of the properties so that the biosimilar in- sulin is not identical. The main advantage is that these insulins are less expensive. Insulin injections Most insulin formulations are now available for both conventional syringes or pen injection devices. Pen injectors are compact, con- venient, and easy to use: the required dose is ‘dialled up’ and injected by pressing the plunger; the ratchet mechanism of most pens gives an audible click that can help blind patients to count dosages. Syringes and pens carry very fine (28–​31 G) needles that allow in- sulin to be injected almost painlessly. The needle should be pushed in vertically and the insulin injected over a few seconds. A needle does not need to be longer than 4 mm in all patients to reach the subcutaneous space. If using a needle longer than 8 mm then it is advisable to inject into a pinched-​up fold of skin to avoid intramus- cular injection in places where there is limited subcutaneous tissue. Backtracking of insulin to the skin surface, which can occasionally cause loss of several units of insulin, may be reduced by leaving the needle in place for a short while. A spot of bleeding may occur; very rarely, sudden hypoglycaemia may be due to direct injection of in- sulin into a subcutaneous vein. Injections can be given into any site that is accessible and well-​ padded with adipose tissue, especially the abdomen, thighs, buttocks, and upper arms. The abdomen has the advantage (theor- etically at least) of relatively faster absorption that is less influenced by exercise, as compared with the limbs. Rotating injection sites (e.g. between the abdomen and leg, or around the quadrants of the ab- domen) helps to avoid local reactions, especially lipohypertrophy which can make insulin absorption slow and erratic.

section 13  Endocrine disorders 2494 Jet injectors fire a metered dose of insulin as a high-​pressure aerosol that penetrates the skin. These have obvious appeal to pa- tients with needle phobia, although there may be bruising and de- layed discomfort at the injection site. Jet injectors are bulky and expensive and do not offer any pharmacokinetic advantages over conventional injections. Inhaled insulin Several companies have developed an aerosol formulation of in- sulin that can be inhaled into the lower airways (insulin is not ab- sorbed from the nasal passages). Inhaled insulin has almost identical pharmokinetic characteristics to subcutaneously injected soluble in- sulin and so its use might be considered to be predominantly a matter of convenience to avoid injections, especially in those with injection site problems or needle phobia. Sophisticated pharmaceutical prep- aration and delivery devices are required to ensure accurate dosing. It cannot be used by current smokers (as absorption is variably en- hanced to an unpredictable degree) or subjects with chronic airways disease, including asthma and chronic obstructive pulmonary dis- ease. Transient cough may occur. Regular lung function testing is advised, as there is a progressive fall in lung function although in most people this is no more rapid than the reduction with age. An increase in insulin autoantibodies has been noted although the sig- nificance is uncertain. Inhaled insulin can be used in both type 1 and type 2 diabetes, although in type 1 diabetes a subcutaneous in- jection of intermediate acting insulin is still required. The long-​term risks of inhaling insulin over many years are not known and there is a theoretical concern of an increased risk of lung neoplasia. A pre- vious preparation of inhaled insulin was withdrawn after poor sales; however, a new product is now available with a smaller device for inhalation. Insulin pumps Portable insulin pumps that administer continuous subcutaneous insulin infusion were developed by Pickup and colleagues in the late 1970s. Modern pumps are compact and light and worn in a belt or holster. Soluble insulin in a special cartridge is delivered through a fine-​bore butterfly-​type cannula, which is inserted subcutaneously in the anterior abdominal wall or other suitable site and generally left in place for 2 to 4 days; the pump can be safely removed for up to 60 min for bathing or other activities. Different basal rates can be preprogrammed, and mealtime boluses are selected and given by pressing a button. Typical basal rates are 0.5 to 1.5 U/​h during the day and 0.5 to 1 U/​h overnight, with mealtime boluses (given imme- diately before meals or snacks) amounting to about 50% of the total daily dose. Most centres use rapid acting analogues in pumps and there is trial evidence to support this. Continuous subcutaneous insulin infusion (CSII) CSII can achieve relatively steady insulin levels under laboratory conditions and can partly overcome the variability of subcuta- neous insulin absorption seen with intermittent injections of larger doses. When used carefully by highly motivated patients who are supported by an experienced diabetes care team, continuous sub- cutaneous insulin infusion can achieve glycaemic control which is better than that achieved with multiple injections; the two were used side by side in the Diabetes Control and Complications Trial. Insulin pumps are expensive (£2600–​£3500 or (US) $5000–​6000) as are consumables (another £1800 per year); medical backup can also be costly to provide. Continuous subcutaneous insulin infusion is indicated for well-​informed patients with type 1 diabetes who are prepared to monitor their blood glucose frequently, learn carbo- hydrate counting, and take responsibility for adjusting the pump. It provides more flexibility for varied lifestyles than multiple daily doses. Randomized trials suggest modest reductions in HbA1c and reduced hypoglycaemia. Although not all randomized trials confirm this, with careful patient selection these benefits are frequently seen in clinical practice and many patients refuse to return to convention- ally delivered insulin. CSII appears to be most beneficial in patients striving hard to improve glycaemic control who are limited by recur- rent hypoglycaemia. It is widely used in the United States of America and many European countries as well as other parts of the world. Infections at the infusion site with pyogenic skin commensals or unusual organisms (e.g. atypical mycobacteria) are uncommon but can be troublesome and cause rapid deterioration in glycaemic con- trol. An increased rate of diabetic ketoacidosis was reported with earlier and less reliable pumps. With CSII, the subcutaneous insulin depot is only a few units, and any interruption of insulin delivery (e.g. with pump failure or cannula blockage) can lead to rapid rises in blood glucose and especially ketone levels. However, modern pumps carry no excess risk of diabetic ketoacidosis as compared with inten- sified injection therapy. Similarly, the risk of hypoglycaemia due to the pump overrunning is now very low. Continuous glucose monitors measure interstitial fluid glucose; this has a good correlation with capillary glucose when the glucose level is stable however may lag behind capillary glucose when this is changing rapidly. Insulin pumps may be used simultaneously with continuous glucose monitoring systems, this form of therapy known as sensor augmented pump therapy. In some systems the glucose reading is continuously displayed on the pump, with indicators that show if the glucose is rising or falling and alarms set to alert the user to low or high glucose levels. This may be particularly useful for those who have no awareness of hypoglycaemia. The use of sensor aug- mented pump therapy may also improve overall HbA1c level if used regularly. Closed loop systems using insulin pumps and continuous glucose monitoring systems are now being introduced. A limiting factor has been accommodating food into the algorithms, however trials have demonstrated benefits in particular circumstances such as in pregnancy. The current systems of sensor augmented pump therapy include devices that may automatically suspend insulin de- livery when the glucose falls below a certain parameter and automat- ically restarts delivery as the glucose rises. Continuous intraperitoneal infusion The peritoneum is a good route for insulin administration: absorp- tion is very rapid across its large surface area and insulin enters the portal circulation. Continuous intraperitoneal insulin infusion has been used in some cases, mostly employing a pump and reservoir implanted subcutaneously in the abdomen and delivering insulin through a flexible cannula sewn into the peritoneal cavity. The reser- voir is filled with soluble insulin through an injection port lying just beneath the skin and is emptied by a liquid/​gas compression system at a rate that can be varied by an external electromagnetic control. Continuous intraperitoneal insulin infusion can provide basal in- sulin; meals need to be covered by additional insulin, either injected subcutaneously or triggered by an external control device.

13.9.1  Diabetes 2495 Intraperitoneal pumps are expensive, and convincing indications for their use are rare. They have been successful in some patients with apparently very high subcutaneous insulin dosages but surpris- ingly normal intravenous requirements. It is now clear that this situ- ation is not due to a mysterious syndrome of ‘subcutaneous insulin resistance’, and that most, if not all, of these patients are interfering with their own treatment (see next). In this setting, continuous intraperitoneal insulin infusion is probably effective because these pumps are difficult to sabotage. Pramlintide Amylin is a 37 amino acid peptide which is cosecreted from the β cell with insulin and is deficient in type 1 diabetes and relatively defi- cient in type 2 diabetes. Amylin slows gastric emptying and regulates postprandial glucagon release. Pramlintide is an amylin analogue which may be injected at mealtimes. This has demonstrated an im- provement in postprandial glucose excursions and HbA1c without an increase in weight or hypoglycaemia; however, it is not widely available outside the United States. Oral hypoglycaemic agents Sulphonylureas and meglitinides The sulphonylureas were the first orally active glucose-​lowering drugs to be used and were discovered in the 1930s when early sul- phonamide antibiotics were found to cause hypoglycaemia. The first generation (chlorpropamide, tolbutamide) have since been super- seded by the second generation (e.g. gliclazide and glibenclamide) and by newer agents such as glimepiride. Repaglinide, a meglitinide, acts in a similar way to the sulphonylureas. Mode of action  Sulphonylureas are insulin secretagogues but in- sulin synthesis is not stimulated. Insulin levels peak within 1 to 2 h and decline within 4 to 6 h for the short-​acting drugs (such as gliclazide) but may remain elevated for much longer with chlorpropamide and glibenclamide, which therefore carry a greater risk of hypoglycaemia. An extrapancreatic action has also been at- tributed to sulphonylureas (i.e. improving insulin sensitivity). This effect is small and is probably explained by the non​specific decrease in insulin resistance (glucotoxicity) when hyperglycaemia is cor- rected by any means. Repaglinide acts in a similar way to the sulphonylureas but is structurally different. It is derived from the non​sulphonylurea part of the glibenclamide molecule (called meglitinide), which was found fortuitously to have glucose-​lowering activity of its own. Nateglinide behaves in a similar fashion and both of these drugs are particularly effective at increasing insulin levels after meals. Meglitinides may be useful as a substitute for sulphonylureas if a patient experiences hypoglycaemia with sulphonylureas particularly with exertion. Efficacy and potency  The ability of these agents to lower glycaemia depends on how much insulin is available for release from the β
cells (which are already stimulated by hyperglycaemia) and by the severity of insulin resistance. In practice, all sulphonylureas lower basal and postprandial glucose levels by no more than 2 to 4 mmol/​litre and HbA1c by 1 to 2%; mild hyperglycaemia may therefore be corrected but patients with fasting glucose in excess of 13 mmol/​litre are very unlikely to achieve normoglycaemia (pri- mary failure). Moreover, as β-​cell function declines progressively in type 2 diabetes, many patients who initially respond well to sulphonylureas will subsequently need additional glucose-​ lowering drugs; this secondary failure overtakes 5 to 10% of pa- tients per year, in a cumulative fashion. These limitations apply to all sulphonylureas and repaglinide: the more potent drugs have lower therapeutic dosages than the earlier agents but cannot lower glycaemia any further. Pharmacokinetics  Most are taken twice daily with meals; glimepiride is taken once daily and repaglinide with each meal. Chlorpropamide has a very long action profile, while glibenclamide shows variable and sometimes prolonged hypoglycaemic activity. Sulphonylureas and repaglinide bind to circulating proteins and may be displaced by other strongly protein-​bound drugs, causing hypoglycaemia (see next). All these drugs are cleared through the kidneys and can accumulate in renal failure, causing frequent hypo- glycaemia and other side effects. Gliquidone and tolbutamide are metabolized mainly in the liver and may be slightly less hazardous in patients with renal impairment, although insulin is usually indi- cated in these cases. Side effects  Weight gain is due to the anabolic effects of hyper­ insulinaemia, compounded by reduced losses of energy through glycosuria. Weight gain is typically 2 to 3 kg greater than with diet alone or metformin. Hypoglycaemia is rarer than with insulin, but the risk is greater with longer-​acting sulphonylureas (glibenclamide, chlorpropamide), in renal failure, and especially in older people. Sulphonylurea-​ induced hypoglycaemia may be more protracted than that caused by insulin and is more likely to result in hospital admission. All patients taking sulphonylureas should be aware of this side effect and should have the means to check their capillary glucose, particularly if they are drivers. Sulphonylureas can cause allergic reactions including skin rashes (notably Stevens–​Johnson syndrome) and marrow dyscrasias, and can precipitate acute intermittent porphyria. Side effects exclusive to chlorpropamide include the syndrome of inappropriate secretion of antidiuretic hormone (see Chapter  21.2.1) and acetaldehyde-​ mediated facial flushing on drinking alcohol. The cardiovascular safety of sulphonylureas has remained under a cloud since tolbutamide was associated with an excess of cardio- vascular deaths during an essentially uninterpretable study (the University Group Diabetes Program or UGDP) conducted in the 1970s; the presence of the ABCC9 (SUR2) receptor on cardiomyocytes has recently reinforced suspicions that these drugs may trigger is- chaemia and arrhythmias (by preventing preconditioning). However, the long-​term United Kingdom Prospective Diabetes Study found no evidence that patients treated with sulphonylureas suffered cardiovas- cular events more often than those treated with insulin. Glimepiride is highly selective for ABCC8 (SUR1). Indications and contraindications  These drugs can be used as first-​line therapy for non​obese subjects with type 2 diabetes in whom lifestyle and dietetic measures have failed to control hyper- glycaemia. However, because of their tendency to increase weight, in the overweight majority of type 2 diabetes patients, sulphonylureas are used as second-​line agents, typically combined with metformin, which may partly offset the weight gain. Sulphonylureas also have a less durable effect than other agents for treating type 2 diabetes.

section 13  Endocrine disorders 2496 Insulin secretagogues are inappropriate for severely insulin-​ deficient patients or during intercurrent illness, when insulin is needed, and are unlikely to be effective if fasting glucose ex- ceeds 13 mmol/​litre. Sulphonylureas are contraindicated in renal failure: all should be stopped, and insulin started if serum creatinine exceeds 250 µmol/​litre. Most sulphonylureas cross the placenta and are contraindicated in pregnancy; however, glibenclamide does not cross the placenta. Studies show that it may be used to control hyperglycaemia in pregnancy in type 2 diabetes and gestational diabetes but not as effectively as insulin. Therefore, it is usually used as a substitute for insulin in pregnancy if insulin is not an acceptable therapy to the patient or if insulin is not available or affordable (see Chapter 14.10). Sulphonylureas are the therapy of first choice in patients with HNF1α MODY, since these subjects are exquisitely sensitive to these agents, and in patients with the Kir6.2 mutation, who may require very high doses (see earlier). Many drugs interact with sulphonylureas, the most common out- come being hypoglycaemia due to displacement and/​or decreased clearance of protein-​bound sulphonylureas (e.g. by sulphonamides, fibrates, salicylates, and probenecid). Potential interactions must al- ways be checked for any drug being contemplated in patients re- ceiving sulphonylureas. Choice of drug  There is little to choose between the newer agents; chlorpropamide is now obsolete. Glibenclamide should be avoided in older people because of its unpredictable tendency to cause hypoglycaemia. Metformin Metformin and phenformin are biguanides, the class of compounds responsible for the mild hypoglycaemic action of goat’s rue Galega officinalis (an otherwise undistinguished weed). Phenformin is no longer available in many countries because it carries a 10-​fold greater risk of lactic acidosis, and metformin has only fairly recently entered clinical use in the United States of America. Mode of action  Metformin acts primarily by inhibiting gluconeo­ genesis in the liver, thus reducing the raised hepatic glucose output which underpins basal and overnight hyperglycaemia; this effect- ively enhances the action of insulin on the liver. AMP kinase, a key enzyme that balances anabolic and catabolic processes in the liver and other tissues, is an important target for metformin action. Peripheral glucose uptake may also be increased, while gastrointes- tinal side effects may help to reduce fondness for food. Metformin does not stimulate insulin secretion. Overall, metformin lowers blood glucose (especially postpran- dial) by 2 to 4 mmol/​litre and HbA1c by 1 to 2%, which is compar- able to the effect of sulphonylureas. On its own, metformin does not cause hypoglycaemia, although this can obviously occur when it is combined with either a sulphonylurea or insulin. Weight does not usually increase with metformin, and may fall. Metformin may have beneficial cardiovascular effects, as the United Kingdom Prospective Diabetes Study found a reduction in vascular events in the metformin-​treated group only (see next). It is not clear whether this is related to the specific metabolic effects of metformin (improved insulin sensitivity), to its mild antiobesity properties, or to other actions such as reported reductions in blood pressure and coagulability. Reduced cancer risk is also reported with metformin. Pharmacokinetics  Metformin is given twice or three times daily with meals. It is cleared mainly through the kidneys, and the in- crease in plasma levels in renal failure is a major risk factor for lactic acidosis. A slow-​release preparation is also available, which is taken once daily and appears to produce fewer gastrointestinal side effects. Side effects  Gastrointestinal symptoms (30% of cases) include al- tered taste, loss of appetite, heartburn, abdominal discomfort and bloating, and diarrhoea (metformin is the most common cause of this in the diabetic clinic). These problems are mostly mild, but may discourage the patient from taking the drug; they can be reduced by starting with a low dosage and increasing it slowly. Lactic acidosis is very rare with metformin (about three cases per 100 000 patient-​years) if it is carefully prescribed. This stems from the mode of action of metformin, namely the inhibition of hepatic gluconeogenesis—​a process that constantly consumes the lactate produced by glycolysis. Blood lactate levels are modestly raised in patients receiving biguanides, and can escalate rapidly and cause life-​ threatening acidosis if lactate is overproduced (e.g. in respiratory or cardiac failure), or is not cleared by the liver (hepatic failure), or if metformin accumulates in renal failure. The risk is also increased in the presence of excessive amounts of alcohol. Lactic acidosis is described in detail later. Megaloblastic anaemia can occur due to impaired absorption of vitamin B12 and 5-​yearly vitamin B12 estima- tions have been recommended. Indications and contraindications  Metformin is now considered the first-​line treatment in type 2 diabetes patients whose hypergly- caemia does not respond adequately to modification of diet and life- style; as it does not tend to cause weight gain, and may even reduce weight, it is especially valuable in obese patients. Recent American Diabetes Association guidelines propose starting metformin con- currently with lifestyle interventions, but this is not universally accepted. The addition of metformin can also be helpful in obese patients who are poorly controlled by sulphonylureas or insulin. Metformin has also proved beneficial in other insulin-​resistant conditions such as polycystic ovary syndrome (resulting in im- proved fertility, reduced hirsutism, and oligomenorrhoea) and im- paired glucose tolerance where it reduces progression to diabetes by around 25%. Contraindications include all the major organ failures—​renal, hepatic, cardiac, and respiratory. It should not be used when serum creatinine concentration exceeds 150 µmol/​litre or the estimated glomerular filtration rate (GFR) is less than 30 ml/​min. Studies have reported beneficial outcomes when patients with stable cardiac failure are treated with metformin. It must also be discontinued for 2 days after receiving radiographic contrast media to reduce the risk of lactic acidosis if contrast mediated nephropathy occurs. Thiazolidinediones Thiazolidinediones are a class of glucose-​lowering drugs which improve insulin sensitivity. There are distinct differences between individual thiazolidinediones which influence their therapeutic spectrum and safety. Pioglitazone is currently available in many countries; troglitazone has been withdrawn because it caused rare

13.9.1  Diabetes 2497 but life-​threatening hepatic damage, and rosiglitazone because of concerns of cardiovascular effects. Mode of action and pharmacokinetics  Thiazolidinediones bind to specific receptors in the nucleus which have the cumbersome title of peroxisome proliferator activating receptor-​γ (PPAR​γ). PPAR​γ and the related PPAR​α (the target for the fibrate class of lipid-​lowering drugs) are ligand-​activated transcription factors whose natural lig- ands appear to be fatty acid derivatives. PPAR​γ that has bound a thiazolidinedione forms a heterodimeric complex with another nu- clear receptor, retinoid X receptor, bound to its own endogenous ligand, retinoic acid. The heterodimer then binds to specific recog- nition motifs found in the promoter sequences upstream of many genes, notably those involved in adipocyte and lipid metabolism. The affinity of individual thiazolidinediones at PPAR​γ par- allels their glucose-​lowering ability in animal models of type 2 diabetes, but their precise mode of action remains uncertain. Thiazolidinediones exert concerted effects that encourage the storage of triglyceride in mature adipocytes, including the differen- tiation of preadipocytes into adipocytes and enhanced expression of lipogenic enzymes; overall, circulating levels of free fatty acids fall and this may reduce hepatic glucose production and increase glucose uptake into muscle as described earlier. The net effect is to enhance the action of insulin—​hence their description as insulin sensitizers. Thiazolidinediones have negligible glucose-​lowering action unless insulin resistance and hyperglycaemia are present. As with metformin, they do not cause hypoglycaemia when used alone, but can exaggerate the hypoglycaemic effects of insulin or sulphonylureas. Efficacy and potency  Alone, all thiazolidinediones lower glu- cose by 2 to 3 mmol/​litre and HbA1c by 1%, somewhat less than the sulphonylureas. However, in some individuals they can result in marked falls in HbA1c, up to 4%. For unknown reasons, blood glucose declines slowly during thiazolidinedione treatment, and a maximal effect may not be reached for up to 6 months. Pharmacokinetics  All are metabolized in the liver and cleared chiefly through the kidney. They are highly protein bound. Side effects  Weight gain, averaging 1 to 4 kg, is due mainly to sub- cutaneous fat deposition. This appears to spare the visceral depot associated with insulin resistance and does not negate the glucose-​ lowering action. Fluid retention of unknown aetiology may cause a mild dilutional anaemia (haemoglobin typically falls by 1–​2 g/​dl) and ankle oedema (in 5–​10% of cases); heart failure may also be precipitated in patients with pre-​existing myocardial dysfunction, especially if they are also treated with insulin. Meta-​analyses have suggested that rosiglitazone is associated with an increased risk of myocardial ischaemic events, and this has resulted in its withdrawal in most countries. Hepatic damage, ranging from subclinical elevations of hepatic enzymes to fulminant and fatal hepatic necrosis (about one case per 1000 patient-​years), has been reported with troglitazone but does not appear to be a risk with rosiglitazone or pioglitazone. Indeed, early indications suggest that thiazolidinediones may be helpful in reducing and possibly reversing steatosis (fat deposition) in the liver that is associated with obesity and insulin resistance and can pro- gress to cirrhosis. An unexpected class side effect of the thiazolidinediones in clin- ical trials is an increase in fractures in the limbs rather than the axial skeleton. This is especially a concern in postmenopausal women. Mechanisms appear to include increased bone resorption and sup- pression of osteoblast formation from mesenchymal progenitors. There is concern about a small increased risk of bladder cancer with pioglitazone and this should be avoided in those with previous bladder cancer or haematuria. Indications and contraindications  Thiazolidinediones are gener- ally regarded as second-​ or third-​line drugs for treating type 2 dia- betes when sulphonylureas or metformin (or the combination of the two) are ineffective or unsuitable. They can be combined with either a sulphonylurea or metformin, when HbA1c may fall by more than 1%; if HbA1c has not fallen by more than 1% within 6 months of adding a thiazolidinedione, it should be discontinued especially in view of the recent concerns over heart failure and fractures. When used alone, they have a lower rate of failure than metformin or sulphonylureas alone, but cost and potential side effect concerns argue against using them as monotherapy. When pioglitazone is used with insulin, in- sulin dosage can be reduced but weight gain may be problematic; rarely, heart failure may be precipitated. Subjects with impaired glu- cose tolerance treated with a thiazolidinedione have a lower risk of progressing to overt type 2 diabetes, and the drugs can improve hir- sutism and menstrual dysfunction (sometimes inducing ovulation) in women with polycystic ovary syndrome. Contraindications include congestive heart failure. α-​Glucosidase inhibitors Acarbose (and the related miglitol and voglibose) are inhibitors of α-​glucosidase, an enzyme of the brush border of the small intestine essential for the breakdown of dietary starch to disaccharides, which are then hydrolysed to the absorbable monosaccharides. They partly block digestion of complex carbohydrates and so damp postpran- dial glycaemic rises, but the therapeutic effect is small: postprandial glucose may fall by 1 to 2 mmol/​litre, with predictably little impact on overnight glucose, and HbA1c by 0.5% or less. Side effects due to carbohydrate malabsorption (flatus, abdominal bloating, gassy diar- rhoea) are common and probably damage compliance. Incretin mimetics These drugs mimic or enhance the action of the incretin hormones that augment insulin secretion. GLP-​1 is an incretin that stimu- lates insulin secretion and may also induce satiety, particularly by delaying gastric emptying. Blood glucose can be lowered compar- ably to sulphonylureas with GLP-​1 infused intravenously. Exenatide (exendin-​4) is an analogue of GLP-​1, first identified in the saliva and concentrated in the tail of the American venomous lizard, the Gila monster, which by an interesting coincidence lives alongside the diabetes-​prone Pima Indians of Arizona. Exenatide shares 50% homology with GLP-​1 but has a considerably longer half-​life in vivo and is now available as a twice daily subcutaneous injection at a dose of 5 or 10 μg and can be used in combination with metformin, a sulphonylurea or insulin. When used alone or in combination with metformin the risk of hypoglycaemia is low. Mean falls in HbA1c of 0.8 to 1% are seen with the higher dose and direct comparison sug- gested that these were similar to the results of addition of insulin with less associated hypoglycaemia. In contrast to the weight gain

section 13  Endocrine disorders 2498 seen with insulin, exenatide is associated with a modest weight loss of around 4 kg, due in part to direct inhibition of appetite. The main side effect is nausea, which occurs in more than 50% of patients, and precludes continuing therapy in around 10% of patients. Pancreatitis has been reported rarely and the use of these drugs is contraindicated in people who have had a previous history of pancreatitis. Animal studies show that exenatide is trophic for β cells; confirmation of this very valuable effect in humans is awaited. Other once daily GLP-​1 analogues, liraglutide, and lixisenatide are available and appear to produce less nausea and equal if not greater glucose lowering. Once weekly exenatide is available; this takes some weeks to develop max- imum efficacy, tends to be better tolerated than the shorter acting pre- parations and can be given up to 3 days after the dose is due. However, the injection requires a more complicated procedure than the shorter acting preparation. Recently more convenient and effective once weekly preparations have been introduced including dulaglitide and semaglutide and an oral version of semaglutide is now available. The GLP-​1 agonists are appropriate for patients who are obese and for whom additional weight gain would have a deleterious effect on their health. They are expensive and should only be continued if cer- tain targets are achieved. In the United Kingdom it is suggested they should be discontinued if the HbA1c is not reduced by 1% or if the patient does not lose 3% of body weight. They should be used in caution with severe renal dysfunction. Increasingly GLP-​1 agonists are used in combination with a basal insulin. Combination products containing liraglutide and insulin degludec are available and trials have shown improved control, weight loss, and reduced hypogly- caemia when compared to multiple-​dose insulin regimes. The gliptin class of drugs (including sitagliptin, vildagliptin saxagliptin, linagliptin, and alogliptin) are oral selective inhibitors of dipeptidyl peptidase IV (DPP IV), the enzyme that causes the break- down of circulating GLP-​1. They therefore prolong the survival and enhance the action of endogenous GLP-​1. These drugs are better tolerated than GLP-​1 agonists but do not result in weight loss and have less impact on HbA1c levels. An unexpected side effect is an increase in infections, notably sinusitis which is linked to the expres- sion of DPPIV on the surface of lymphocytes (CD26). Generally, they are well tolerated and several are licensed at reduced doses to be used at low eGFR values. Linagliptin is excreted hepatically and therefore may be used without a change in dose at all levels of eGFR. Saxagliptin has demonstrated reassuring cardiovascular outcome data with the exception of increased hospital admissions with heart failure; however, recent sitagliptin data have shown this is not a class effect. As a result of the low incidence of hypoglycaemia and side effects, these agents have become a popular choice for treating older people although outcome data are limited in this group. Sodium-​glucose cotransporter 2 inhibitors (SGLT2 inhibi- tors)  Ninety per cent of reabsorption of filtered glucose occurs in the proximal tubule of the kidney and is mediated by SGLT2. The re- maining 10% is reabsorbed in the distal tubule mediated by SGLT1. SGLT2 inhibitors prevent the reabsorption of glucose and result in glycosuria and subsequently an osmotic diuresis. Several SGLT2 inhibitors are licensed (dapagliflozin, canagliflozin, and empagliflozin) to treat type 2 diabetes in combination with various oral agents (although the licensed combinations vary between agents) and insulin. These agents lower the HbA1c by approximately 0.8%. The loss of 70 g of glucose in the urine each day results in a gradual and sustained weight loss. The therapeutic glycosuria has a consequence of genital candida infections for around 5% of patients although stopping the SGLT2 inhibitor is not usually necessary. A small drop in blood pressure is often seen. The SGLT2 inhibitors should not be used together with loop diuretics as the resulting reduction of plasma volume may result in a decline in renal function. There have been re- ports of ketoacidosis with the SGLT2 inhibitors and patients should be informed of this. Recent data published on all SGLT2 inhibitors have demonstrated improved cardiovascular outcomes compared to other agents routinely used to treat type 2 diabetes, with especially dramatic reductions in death from heart failure. This has prompted their use in patients with heart failure without diabetes. There is also evidence that this class of drugs reduced progression from microalbuminuria to end stage renal failure, although currently their use is contra-indicated with an eGFR<45 ml/min. Bromocriptine  A quick release formulation of bromocriptine is licensed for the treatment of type 2 diabetes in some countries. It modulates central glucose and metabolism pathways which results in reduced hepatic glucose production and a reduction of post- prandial glucose. HbA1c typically is reduced by 0.5%. The doses of bromocriptine used to treat diabetes are very much lower than those used to treat Parkinson’s disease. Colesevelam  Bile acid sequestrants are primarily lipid-​lowering drugs however also have a glucose-​lowering effect and are licensed for the treatment of type 2 diabetes. The HbA1c is lowered by ap- proximately 0.5% although the exact mechanism is unclear. Practical management of hyperglycaemia Most newly diagnosed diabetic patients are relatively easily allocated to either type 1 or type 2 on clinical criteria (see Table 13.9.1.2) and treatment is started accordingly. However, initial impressions may be misleading: a thin young patient may not need insulin because he has monogenic diabetes, whereas a person who has classical features of type 2 diabetes may lose weight rapidly and develop ketoacidosis because he has type 1 diabetes. Continuing monitoring and vigi- lance are therefore essential. The diagnostic pitfalls of Flatbush and fulminant type 1 diabetes have been mentioned earlier. Type 1 diabetes These patients must be given insulin immediately and for life. The insulin regimen will depend particularly on any remaining en- dogenous insulin, the patient’s body weight, lifestyle, and motiv- ation. Patients with residual insulin secretion, especially newly presenting and particularly during the ‘honeymoon period’ (see next), can often fill in gaps in insulin replacement and enjoy good glycaemic control with few injections and low insulin dosages. However, C-​peptide negative patients will require exogenous insulin to cover both basal and prandial needs (see Fig. 13.9.1.9) to achieve good control. Regimens include: • Basal insulin given once a day (insulin glargine or degludec) or twice a day (insulin detemir or isophane) with short-​acting in- sulin (human soluble insulin) or short-​acting analogue (insulin aspart, lispro, or glulisine) given with meals. In type 1 diabetes the use of analogue insulins is associated with a reduction of hypogly- caemia and is the preferred option. An intensified insulin regime early in the course of type 1 diabetes has been shown to be asso- ciated with short-​term and long-​term reduction of microvascular and macrovascular complications and should be considered the

13.9.1  Diabetes 2499 first-​line treatment option unless insulin pump therapy is imme- diately available. The short-​acting analogue insulin may be given as per an insulin to carbohydrate ratio rather than as a fixed dose and this may improve overall control. • Premixed insulins injected before breakfast and before the evening meal may be preferred by some patients however the limitations of this therapy should be discussed. Intensive glucose control may be achieved with insulin pump therapy which has demonstrated reduced HbA1c without an in- crease in hypoglycaemia. It is necessary that staff involved should have particular expertise in insulin pump therapy and thorough pa- tient education should be available for this therapy to be safe and successful. Insulin dosages should be titrated according to blood glucose and HbA1c monitoring (see Table 13.9.1.3). Metformin and SGLT2 inhibitors in type 1 diabetes The addition of metformin to insulin is valuable in patients with type 1 diabetes who are overweight as this may help to reduce insulin resistance, lower insulin dose, and prevent weight gain. Recently, an SGLT2 inhibitor has been licensed for use in type 1 diabetes fol- lowing data showing reduced glycaemic variability, lower HbA1c without increased hypoglycaemia. However here is an increased risk of ketoacidosis, especially euglycaemic ketoacidosis and hence it should only be initiated by specialist teams in patients compliant with insulin therapy. Starting insulin therapy Patients at risk of ketoacidosis may need hospital admission, but most patients are clinically well and can start insulin as an outpatient, supervised by a specialist diabetes nurse. Basal insulin is usually started once or twice a day and short-​acting insulin is added to cover prandial hyperglycaemia. Wherever practicable, patients should be encouraged to give their own injections as soon as possible. Newly diagnosed patients starting insulin need to be warned about a possible ‘honeymoon period’ of good glycaemic control, when the fall in glucose levels allows partial though temporary recovery of the remaining β cells. Blood glucose can often be easily controlled with low insulin dosages (and exceptionally, without exogenous insulin) but the honeymoon ultimately ends usually within a few months: blood sugar levels and insulin requirements then escalate, because of the progressive loss of remaining β cells over the next 1–​5 years. Poor diabetic control and ‘brittle’ diabetes In real life, relatively few type 1 patients approach the high-​quality glycaemic control aspired to in Table 13.9.1.3. This largely reflects the pharmacokinetic shortcomings of current insulin preparations and the unpredictable nature of subcutaneous absorption. The patient’s compliance is a crucial determinant of overall diabetic con- trol; teenagers are notoriously resistant to advice about diabetes, as with other matters, and many have markedly elevated HbA1c con- centrations. This clearly increases the risk of future diabetic compli- cations. However, it should be noted that compliance with complex insulin regimes is very demanding and less than 20% of people with type 1 diabetes achieve levels of glycaemic control (HbA1c <53 mmol/​ml) that prevent long-​term complications. A few patients have such poor metabolic control that they cannot live a normal life. Most have chronically high blood glucose and suffer recurrent hospital admissions with ketoacidosis; some suffer frequent hypoglycaemia, while others have an unstable or ‘brittle’ blood glucose profile that can swing rapidly between hyper-​ and hypoglycaemia. Occasionally, endocrine or intercurrent illnesses are found to be responsible (see Table 13.9.1.5), but most cases remain idiopathic after even intensive investigation. It is now clear that poor compliance, often aggravated by deliberate inter- ference with treatment, is responsible in many of these patients. Most are young women who are generally hyperglycaemic despite apparently high insulin dosages; when tested under controlled conditions, however, their intravenous and subcutaneous insulin requirements are unremarkable. Many are probably omitting in- sulin or taking only small doses: common motives include escape from difficulties at school or home, or wanting to stay thin (dis- turbances of body image are common in this group). Coexistent eating disorders, such as anorexia and bulimia nervosa, are com- monly seen in these individuals. Initially, such patients may ap- pear to lead charmed lives despite frequent hospital admissions but many die prematurely (especially from ketoacidosis or hypo- glycaemia); significant diabetic complications frequently develop during their twenties or thirties. Management can be extremely difficult. Patients with sustained poor control should be admitted selectively for intensive educa- tion, observation, and exclusion of other possible causes (see Table 13.9.1.3). In some cases, it may be necessary to confirm that insulin is effective at conventional doses (for more information see the paper by Schade and Duckworth listed in ‘Further reading’). Even close super- vision in hospital does not exclude ingenious interference with insulin Table 13.9.1.5  Causes of poor glycaemic control in type 1 diabetic patients Characteristics Cause High insulin requirements, chronic hyperglycaemia ± recurrent ketoacidosis Obesity Puberty Endocrine diseases: Cushing’s syndrome, thyrotoxicosis Drugs: especially glucocorticoids Immune insulin resistance Low insulin requirements, recurrent hypoglycaemia Weight loss Loss of hypoglycaemia awareness Other endocrine diseases: adrenocortical failure, hypothyroidism, growth hormone deficiency, hypopituitarism Gastroparesis Coeliac disease Liver disease Erratic glycaemic profile, frequent hyper-​ and hypoglycaemia (‘brittle’ diabetes) Compliance issues Pancreatic damage Overtreating hypoglycaemia Gastroparesis Injection site problems (lipohypertrophy) Recurrent or chronic infections: tuberculosis, sinusitis For all three characteristics, always consider: unsuitable insulin regime; poor diabetes education; deliberate noncompliance; appetite disorders (anorexia nervosa, food bingeing).

section 13  Endocrine disorders 2500 treatment or glucose monitoring. Intensified insulin schedules or con- tinuous subcutaneous insulin infusion may help in some cases and increasingly whole pancreas transplantation is being considered as an option if patients are willing to take the associated risks (see next). Experimental and future treatments for type 1 diabetes Whole pancreatic transplantation, usually performed in conjunc- tion with renal transplantation for patients with diabetic nephrop- athy, can achieve good results including long-​term withdrawal of exogenous insulin (> 5 years) in up to 70% of cases. The whole gland or a segment is transplanted into the pelvis and anastomosed to the iliac vessels; to avoid damage from pancreatic exocrine secretions, the pancreatic duct is drained either into the gut or into the bladder (when urinary amylase excretion can indicate the health of the graft). Outcomes for both the pancreas and the kidney are better when sim- ultaneous transplantation is performed as the early treatment of re- jection, which is easier to identify in the kidney by serum creatinine and or biopsy, preserves both organs, and the improved glycaemic control from the pancreas is beneficial to the kidney. Problems in- clude the need for lifelong immunosuppression (required anyway for renal transplantation) and the global shortage of donor organs. An increasing number of pancreas transplants alone are being per- formed in type 1 diabetes but the balance of risks (especially of ma- lignancy and infection from the immunosuppression) and benefits (from improved glycaemic control) requires careful assessment in each individual patient. The optimal indications for this procedure, where available, remain poorly defined but it is generally performed for persistent poor metabolic control with or without recurrent ketoacidosis or recurrent, intractable hypoglycaemia. Relentless progression of complications despite good glycaemic control is a further indication as generally the progression of complications is halted but not reversed. Although there are attendant risks from the surgery and immunosuppression, recurrent ketoacidosis and poor metabolic control itself carries a not insignificant risk of death. Introduction of an improved immunosuppressive regimen (which omits glucocorticoids) by Shapiro and colleagues reported in 2000 has led to a resurgence in pancreatic islet transplantation. The most widely used method is by transcutaneous injection into the portal vein of islets isolated from a donor pancreas; these col- onize and function well in the liver, the first stop for insulin secreted physiologically. Even with less toxic (to the β cell) immunosuppres- sion, two or three donor pancreases are currently needed for each re- cipient, and only 10% of patients are insulin independent at 5 years, but newer immunosuppressive regimes may improve this further. Nevertheless, up to 90% of patients report significant reductions in the rate of hypoglycaemia; hence recurrent severe hypoglycaemia unresponsive to changes in insulin therapy or the use of CSII re- mains the main indication for this procedure. Healthcare professionals should always remain sensitive to the burden of diabetes for the individual and aware that glycaemic control may vary depending on the various physical, psychological, and emo- tional stresses of life. Often diabetes control will decline in the face of adversity. Healthcare professionals should avoid blaming a patient for such deterioration in control and remain supportive and helpful. The transition from paediatric services to adult services occurs at a particularly vulnerable time in life when a young person has competing physical and emotional issues, is developing a fragile in- dependence from their family, often participates in high-​risk behav- iour and has a need for peer group conformity. The removal of a familiar team who take a holistic, family approach and replacing this with an unfamiliar team that treats the young person as an autono- mous adult often results in disengagement from services and very poor glycaemic control. Transition services should be designed to prevent this and improve outcomes for young people. Prevention of type 1 diabetes by aborting insulitis during the long prediabetic phase by immunosuppression in high-​risk subjects, or preserving islet cell function in newly diagnosed patients, is a major goal of current research. Trials in the 1980s demonstrated that ciclosporin can achieve this, but the cost in terms of side effects of continuous therapy is too high. Newer immunomodulatory agents, such as anti-​CD20 (rituximab) or CTLA4-​Ig (abatacept), anti-​CD2 (alefacept) or treatments, that regulate rather than suppress immune responses such as non​depleting anti-​CD3, significantly improve the risk–​benefit ratio to the point of acceptability but are not currently available, and trials of further agents are underway. Observational studies suggest that preservation of even small amounts of en- dogenous insulin (stimulated C-​peptide >200 omol/​litre) is associ- ated with a 50% reduction or more in hypoglycaemia, 50% or more individuals achieving optimal glycaemic control and reduced long-​ term complications. Much effort is also being invested in promoting the regeneration of β cells either from pancreatic tissue or more generic stem cells. These studies remain at a preliminary stage, but potentially offer a renewable therapy. Although they may not require immunosup- pression for allograft rejection, it remains to be seen whether such new cells would be retargeted by the autoimmune process in subjects with type 1 diabetes. Management of type 2 diabetes Dietary and lifestyle measures form an essential foundation for the management of type 2 diabetes and must be maintained throughout, even though fewer than 10% of patients can be controlled satisfac- torily for more than a year by these means alone. Recent evidence from the Early ACTID study suggests that intensive dietary support may be as effective as promoting diet and exercise together in the early stages after diagnosis. Patients who fail to meet the glycaemic targets set out in Table 13.9.1.3 should generally follow the steps outlined next, although compromises may be more appropriate in older people or those at risk of hypoglycaemia. Progress should be reviewed every 3 months or less if blood glucose is unacceptably high. The mainstay of treating type 2 diabetes over the last 50 years has been to introduce agents to manage the inexorable deterioration in β cell function in type 2 diabetes for example using metformin followed by a sulphonylurea followed by insulin. Over the last 10 years there has been the introduction of agents that have a more durable effect and preserve the β cell function such as thiazolidinediones, GLP-​1 agonists, and SGLT2 inhibitors. It seems logical to use such agents early in the course of the disease; however, this optimism needs to be balanced by a caution that long-​term effects of these agents are not known, and they are generally more expensive. However, out- come data are becoming available due to increased vigilance from the regulatory bodies, for example the reduced mortality seen with empagliflozin therapy in recent trials. The first-​line oral hypoglycaemic agent for dietary failure is metformin. Metformin has some of the most robust outcome data for benefit, is weight neutral, and has an extremely low risk of hypoglycaemia.

13.9.1  Diabetes 2501 Several guidelines advocate beginning metformin immediately at diagnosis, but the effect this has in undermining commitment to lifestyle measures has not been defined. The addition to metformin of a sulphonylurea represents standard second-​line treatment in many guidelines. However, there is general acceptance that therapy should be individualized taking into account factors such as the risk of hypoglycaemia, cardiovascular/heart failure risk, the adverse effects of further weight gain, the need to monitor capillary blood glucose, and the loss of β-​cell function. In these cir- cumstances, the use of DPP IV inhibitors, GLP-1 agonists and SGLT2 inhibitors should be considered and treatment algorithms are evolving. Around 40% of people living with type 2 diabetes manage their diabetes with insulin, although this is decreasing with the increasing use of new non-insulin therapies. The standard method of commen- cing insulin is to start basal insulin in the form of isophane or a basal analogue such as insulin Lantus at a dose of 10 units or 0.1–​0.2 units per kg at night. Most will continue with oral therapy alongside this regime. The insulin is then titrated by 2 units every 3 to 5 days until the fasting glucose is persistently below 7 mmol/​litre. Self-​titration by patients using standard algorithms may be more effective than that done by healthcare professionals. Short-​acting insulin or short-​ acting analogue may be added at mealtimes if needed. At this point it is usual to stop the sulphonylurea, however, metformin is continued if possible and there may be benefits to continuing other agents such as gliptins, GLP-​1 agonists, SGLT 2 inhibitors, and thiazolidinediones. Premixed insulins may be used rather than a multiple-​dose injection regime; advantages to this may be convenience in those who have a regular lifestyle. Generally, the more effective the regime at HbA1c reduction, the greater the weight gain and the risk of hypoglycaemia. The evidence for hypoglycaemia reduction using analogues rather than human insulin is less convincing in type 2 diabetes compared to type 1 diabetes. Many health providers are suggesting human insulin as a first-​line insulin option in type 2 diabetes reserving the analogues for the small proportion who may develop problems with hypogly- caemia, as substantial cost saving may be made. Intensive insulin therapy at the time of diagnosis for 2 to 3 weeks has been associated with remission of diabetes, presumably sec- ondary to a reduction of glucotoxicity. The usual indications for early insulin initiation are marked hyperglycaemia or the features of weight loss or ketosis. Obesity (and therefore insulin resistance) may worsen when in- sulin treatment is started. The average weight gain is around 6 kg; possible reasons include reduced loss of energy through glycosuria, a tendency to relax dietary restriction when a more effective means of lowering glycaemia is introduced, and sometimes overeating during hypoglycaemic episodes. Increasing insulin resistance may lead to escalating insulin dosages. The addition of GLP-​1 agonists or SGLT2 inhibitors may help to achieve targets while preventing weight gain or allowing weight loss. Monitoring diabetic control Treatment targets for blood glucose in type 1 and type 2 diabetes (see Table 13.9.1.3) have been selected to reduce the risk of chronic diabetic complications. Avoiding acute episodes of hyper-​ and hypo- glycaemia is also important. Blood glucose monitoring Blood glucose concentration can be easily and quickly measured in small drops of blood (a few microlitres or less), using various test strips; the ability to perform such measurements is an essential skill for all professionals delivering diabetes care and for most diabetic patients. Test strips contain glucose oxidase (which catalyses the oxi- dation of glucose to gluconic acid) together with a detection system to measure specific reaction products, either electrochemically or colorimetrically (using dyes sensitive to hydrogen peroxide). The signal is read by a reflectance meter or electrically, and converted into the glucose concentration in the sample. Colour-​based test strips can also be read by eye against a printed standard scale, al- though this may be difficult for partially sighted or colour-​blind patients. A drop of blood is obtained by pricking the sides of the fingertip, avoiding the sensitive pads; various lancets and automatic finger-​ pricking devices are available. Blood must cover the reaction area completely and be left in contact for exactly the period stipulated; modern meters read out automatically at this point, whereas older strips must be wiped dry and left for the colour to develop. Failure to follow the manufacturer’s instructions is the main cause of in- accurate readings, which are disturbingly frequent. With attention to detail, readings correspond closely to laboratory measurements of glucose (which also employ the glucose oxidase reaction) but are not reliable enough to be used for diagnosing diabetes. Monitoring schedules Monitoring is not essential in patients treated with diet alone or a single oral agent. Generally, those treated with diet or medications which have a very low risk of hypoglycaemia (such as metformin, glitazones, gliptins,) do not need to routinely monitor their capillary blood glu- cose. The exception would be when significant hyperglycaemia is suspected and it would be inappropriate to wait to see the response of the HbA1c blood test. Those receiving sulphonylureas are suscep- tible to hypoglycaemia and therefore should have the means to test their blood glucose levels if they feel unwell or related to driving. Insulin-​treated patients need more frequent monitoring to ad- just insulin dosages. Bedtime and premeal testing (4-​point) as well as ideally 2-​hour postprandial (7-​point) testing is recommended. Fasting glucose is determined by the previous evening’s long-​acting insulin. Prandial short-​acting insulin dosages can be titrated from the glucose rise 90 to 120 min after eating. Readings can be scat- tered across these time points on different days; most patients can be persuaded to check their glucose levels once or twice per day but to achieve tight glycaemic control targets without hypoglycaemia, more frequent blood glucose testing is required. Written records help to bring out general patterns in glucose con- trol and many modern meters can be downloaded to display the pat- tern in different formats. Patients must also be encouraged to check their glucose if they feel unwell and, crucially, at frequent intervals during intercurrent illness. Occasional tests during the night (es- pecially between 02.00 and 04.00) are useful in patients at risk of nocturnal hypoglycaemia, including those injecting long-​acting or premixed insulins in the early evening. Checking the self-​monitoring technique and the patient’s action plan when glucose levels fall outside the target range is a core part of the patient’s diabetic education. The Freestyle Libre glucose monitor consists of a sensor inserted into the subcutaneous tissue of the upper arm that may stay in place for up to 2 weeks. The user can scan the sensor with the hand-​held device as frequently as they wish to obtain the current glucose result

section 13  Endocrine disorders 2502 and the last 8 hours of data. There is no need to do any calibration capillary blood tests. This is the first such system available for direct purchase from the manufacturer and it has proved very popular. Continuous glucose monitoring systems have a sensor which lasts up to a week, may store data for a greater period however they re- quire calibration with capillary blood tests four times a day and are more expensive. They have demonstrated an improvement in HbA1c when used regularly and in particular circumstances such as pregnancy. HbA1c and fructosamine These tests measure the non​enzymatic reaction of glucose with circulating proteins (see next), and therefore reflect longer-​term blood glucose levels. Glycated (glycosylated) haemoglobin (HbA1) results from the combination of glucose with the N-​terminal valine residue of the B chain of adult Hb (HbA), and can be separated from unaltered HbA by electrophoretic and other methods. HbA1 includes the stable HbA1c fraction, which is most closely related to average blood glucose levels over the preceding 6 to 8 weeks. The various assay methods for HbA1c were initially matched to a standard used in the Diabetes Control and Complications Trial (DCCT), giving a result in percentage terms which defined the long-​term risks of diabetic microvascular complications (see next). More recently, values have been matched to a standard from the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) and are now expressed as mmol HbA1c per mol of haemoglobin (mmol/​mol) to define the fact that they are refer- enced to a different standard. Table 13.9.1.6 gives the conversion of common values and the conversion formula. For assays con- forming to IFCC standards, non​diabetic HbA1c ranges from 14.8 to 36.6 mmol/​mol Hb (3.5–​5.5% in DCCT units), with good con- trol defined as values less than 53 mmol/​mol (7%) and poor control as more than 64 mmol/​mol (8%); some poorly compliant patients have HbA1c concentrations of 130–​150 mmol/​mol (14–​16%). HbA1c measurements are a useful index of medium-​term glycaemic con- trol, but may be invalidated by abnormal red cell turnover (values are spuriously low in haemolysis, bleeding, and pregnancy), in renal failure (carbamylated HbA coelutes with HbA1c, falsely raising levels), and with abnormal haemoglobins such as hetero-​ or homo- zygous sickle cell disease (HbF also comigrates with HbA1). Modern analytical methods for HbA1c detect the presence of abnormal haemoglobins and hence spurious results are usually highlighted by the laboratory. Conversion tips:

1. For whole numbers of %HbA1c:  ‘minus 2 minus 2 rule’, e.g. 6.0% = (6 –​ 2), (6 –​ 2 –​ 2) (i.e. 42).
2. Remember 6.0% = 42, then add 11 for each 1%.
3. Normal less than 42, target less than 53 (7.0%).
4. Formula:  IFCC–​HbA1c (mmol/​mol)  =  [DCCT–​HbA1c (%) –​2.15] × 10.929. Serum albumin also undergoes glycation, which is measured by the fructosamine reaction. As albumin turns over faster than haemoglobin, the fructosamine concentration reflects mean blood glucose over the previous 1 to 2 weeks. Assays are cheap but not standardized between laboratories, and are generally less reliable and reproducible than measurements of HbA1c. Measurements of urinary glucose and ketones Urinary glucose concentrations can be measured easily using glu- cose oxidase test strips, but are of limited use: urinary glucose con- centration depends on the renal threshold (which can lie between 7 and 13 mmol/​litre), urine output, and the time since the bladder was last emptied. Crucially, hypoglycaemia cannot be detected. Urinary glucose measurements are acceptable in type 2 diabetic patients with a normal renal threshold who are not receiving hypoglycaemic medication (insulin or sulphonylureas) and in patients who decline to prick their fingers. Urinary ketone measurements can be useful for predicting impending ketoacidosis, particularly during intercurrent illness when blood glucose is high. Moderate ketonuria can be caused by fasting or undereating, including during infections. Modern blood testing meters can measure blood ketones with appropriate testing strips. Values less than 0.3 mmol/​litre are normal; levels in excess of 3 mmol/​litre indicate ketoacidosis. Structures for diabetes care Diabetes is best managed by the combined efforts of a well-​trained primary care team and a team of specialists with complementary and overlapping skills: physician, specialist diabetes nurse, diet- itian, and podiatrist. The specialist diabetes nurse has a crucial role in educating patients about diabetes and its practical man- agement, and in starting and adjusting therapy. Many patients are more receptive and responsive to information given by pri- mary care teams and specialist nurses than by doctors. For com- plex cases, there must be frequent contact with and easy access to other specialists (ophthalmologist, vascular surgeon, renal phys- ician, obstetrician, and clinical psychologist), ideally in the setting of combined clinics. Each member of the team has a particular niche, but all must agree common strategies (such as dietary ad- vice for obesity) to avoid giving the patients conflicting or incon- sistent information. Diabetes care can be delivered effectively by well-​informed general practitioners or practice nurses, hospital-​based clinics, community mini-​clinics, or shared care schemes that bridge the primary and secondary sectors. Because of the unpredictable course and potential complications of diabetes, all patients must be thoroughly reviewed each year and be rapidly referred for Table 13.9.1.6  Conversion between DCCT aligned and IFFC values for HbA1c DCCT HbA1c (%) IFFC HbA1c (mmol/​mol) 6.0 42 6.5 48 7.0 53 7.5 59 8.0 64 9.0 75 10 86 DCCT, Diabetes Control and Complications Trial; IFCC, International Federation of Clinical Chemistry and Laboratory Medicine.

13.9.1  Diabetes 2503 specialist help if the need arises. A check list for the annual review is suggested in Table 13.9.1.7. Diabetes education Living and coping with diabetes is a considerable burden that is poorly appreciated by many doctors and nurses. Careful education about diabetes, its complications, and its practical management can provide great reassurance to patients and also reduce emergency hospital admissions and complications such as foot ulceration and amputation. Diabetes education is most effectively provided by a trained practice nurse or specialist diabetes nurse, but all members of the diabetes care team should understand the key messages, and check and reinforce these whenever possible. Evidence suggests Table 13.9.1.7  Routine annual review of a diabetic patient: key points include those specific to patients taking insulin History and discussion Examination Investigations Diabetic treatment Diet, physical activity Weight, height, BMI Weight and change Waist–​hip ratio Glucose-​lowering drugs Insulin injection sites Diabetic control Self-​monitoring results (± check technique) Hyperglycaemic symptoms HbA1c Hypoglycaemia frequency and awareness of symptoms Liver function tests in type 2 diabetes if known fatty liver disease Diabetes education and skills (± family or associates) General knowledge Treatment targets ‘Sick-​day’ rules, driving rules, pregnancy plans Hypoglycaemia treatment Insulin injection technique Diabetic complications: Macrovascular Ischaemic heart disease (angina, MI, failure, arrhythmias) Examine heart, including signs of failure. Blood pressure, lying and standing Fasting lipid screen (total, HDL, and LDL cholesterol; triglycerides). ECG (if other risk factors or age >40) Peripheral vascular disease (claudication, stroke, TIA) Peripheral pulses, strength, and bruits Smoking history Other risk factors (hypertension, dyslipidaemia, family history) Eyes (retinopathy and cataract) Altered acuity, loss of vision Visual acuity (± corrected). Retinal examination (digital imaging, fundoscopy through dilated pupils) Nephropathy Blood electrolytes, urea, creatinine. Microalbuminuria screen (e.g. albumin:creatinine ratio) or timed urinary albumin excretion Neuropathy Altered or reduced sensation Sensory testing screen as appropriate (feet: see next) Pain Postural blood pressure drop Weakness in limbs Autonomic symptoms (sweating, postural dizziness, gastrointestinal) Feet Pain, numbness General condition: posture, callus, footwear appropriate Ulceration: current and previous Pulses and perfusion Footwear (sensible?) Oedema Foot care Sensory deficits (vibration sense, pin prick, monofilament) Sexual function Erectile and ejaculatory problems (men) Other illnesses Other medication (possible effects on glycaemic control and interactions with antidiabetic drugs) BMI, body mass index; MI, myocardial infarction; HDL, high-​density lipoprotein; LDL, low-​density lipoprotein; TIA, transient ischaemic attack.

section 13  Endocrine disorders 2504 that education in a group setting is often more effective than on a one to one basis and may promote informal support networks. Key elements of the education programme include: • causes of hyperglycaemia and diabetic symptoms • own treatment: diet and lifestyle; drawing up and injecting in- sulin; oral agents; recognizing and treating hypoglycaemia • self-​monitoring technique; targets and danger levels; how to re- spond to poor control • ‘sick-​day’ rules: monitoring during intercurrent illness; how to ad- just own treatment; when and how to call for help (Box 13.9.1.1) Structured education This term refers to programmes to facilitate self-​management of diabetes. Examples of these programmes are the Dose Adjustment For Normal Eating (DAFNE) programme, which is a programme for people with type 1 diabetes which teaches management of in- sulin depending on the ingested carbohydrate, activity, and illness and X-​PERT, which is a self-​management programme for people living with type 2 diabetes. There are various requirements of a structured education programme; there should be defined goals and policies, input from stakeholders and experts to ensure the quality of the programme, the needs of the population to be educated should be considered, educated staff, a curriculum, individualization of the programme as needed, ongoing support and measurement of progress, as well as quality assurance. Structured education pro- grammes are generally considered to be the gold standard of patient education. Employment, driving, and insurance Until recently people on insulin therapy were barred from holding heavy goods vehicle and public service vehicle driving licences. Similarly, certain occupations were restricted for those on insulin such as the armed forces, police service, fire service, and working on aircrafts. Following legislation preventing discrimination against people with disabilities including those with diabetes, each case must be reviewed individually. This has had particular impact for those with type 2 diabetes who previously delayed insulin treatment with the subsequent threat to their health to prevent losing their em- ployment. Licensing for taxi drivers varies between local authorities. Specific diabetic complications, notably sight-​threatening retinop- athy, may preclude particular jobs or pastimes. Patients must inform the driving licence authorities and their driving insurer that they have diabetes, and those receiving in- sulin or with clinically significant retinopathy may require peri- odic medical confirmation of fitness to drive. Hypoglycaemia with loss of awareness of symptoms is a bar to driving. Currently the use of GLP-​1 agonists carries no specific restrictions in the United Kingdom except for heavy goods vehicle or public service vehicle drivers taking these agents in conjunction with sulphonylureas, in which case the driving authority will make an assessment on individual basis. Special life insurance policies are available from companies en- dorsed by patient-​centred organizations such as Diabetes UK and the American Diabetes Association. Many patients find it valuable to join these organizations. Intercurrent events in diabetes and their management Infections People with diabetes probably have increased susceptibility to pyo- genic bacterial infections, especially when diabetes is poorly con- trolled. Hyperglycaemia can impair the killing of microorganisms by neutrophils and macrophages and may also interfere with the function of T lymphocytes. Some infections particularly associated with poorly controlled diabetes include: • recurrent and sometimes invasive candidiasis • tuberculosis, often widespread and cavitating • necrotizing fasciitis, rapidly spreading necrosis of subcutaneous tissues down to muscle, usually due to β-​haemolytic streptococci with staphylococci and often anaerobes • gas-​forming infections with anaerobes and clostridia, including emphysematous pyelonephritis, cholecystitis, cystitis, and foot in- fections; plain radiography shows gas in the affected tissues • diabetic foot ulcers (see next), which are often infected, with the risk of osteomyelitis and deep soft tissue spread • recurrent oral and genital candida infections • urinary tract infections, which may be complicated by ascending infections with pyelonephritis and renal or perinephric abscess (sometimes with gas), and occasionally acute papillary necrosis; severe loin pain and systemic symptoms, with deteriorating renal function, should suggest these possibilities and the need for ur- gent imaging • ‘malignant’ or necrotizing otitis externa, due to pseudomonas in- fection, which can invade the skull and facial nerve • periodontal infections, sometimes causing tooth loss—​these are common • rhinocerebral mucormycosis, a highly invasive fungal infection that originates in the sinuses but often spreads into the orbit and cranial cavity; mortality is about 50%, even with debridement and high-​dose intravenous amphotericin B The bacterial infections often require aggressive intravenous anti- biotic treatment with cover against anaerobes. Fastidious and rare organisms should be considered when standard antibiotic regimens are ineffective. Box 13.9.1.1  ‘Sick-​day’ rules for patients with type 1 diabetes If you feel unwell, and even if you think it is a minor infection: • Never stop taking your insulin—​you often need more when you are unwell • Check your blood glucose every 4 h—​glucose levels can rise very fast during infections • Test your urine or blood ketones if your glucose is greater than 14 mmol/​litre or you are feeling unwell with vomiting • Contact your doctor at once if you: — start vomiting — get high glucose levels (>15) that do not come down after insulin — get hypos (glucose <3) — get ketones in the urine or blood ketones greater than 1.0 mmol/​litre — are worried and do not know what to do

13.9.1  Diabetes 2505 Diabetic control during infections Minor viral infections rarely disturb diabetic control, but increased secretion of counterregulatory stress hormones during severe infec- tions, especially with fever, can rapidly worsen insulin resistance in both type 1 and 2 diabetes. Type 1 patients may need twice as much insulin as usual, even if they are unable to eat. Failure to increase the insulin dosage will therefore allow glucose to rise, sometimes dramatically fast, and risk precipitating ketoacidosis. It is therefore essential to continue taking insulin, to monitor blood glucose frequently, and to increase insulin if sustained hyperglycaemia develops. An increase of 30 to 50% in long-​acting insulin is often enough, but requirements will be deter- mined by blood glucose levels and should be decided in consultation with the diabetes care team. Avoidable deaths still occur every year because poorly educated patients (sometimes advised by ignorant doctors) reduce or even stop taking insulin because they feel ill, are not eating, and are worried about becoming hypoglycaemic. Clear ‘sick-​day’ rules (see Box 13.9.1.1) are a crucial part of diabetes edu- cation, which must be regularly checked and reinforced. During severe infections, insulin requirements may fluctuate rap- idly and the safest way to give insulin is by continuous intravenous infusion, backed up by frequent (hourly) blood glucose measure- ments (see next). Type 2 patients may similarly lose glycaemic con- trol, and are often best transferred temporarily to subcutaneous or intravenous insulin. It seems best to maintain blood glucose be- tween 5 and 10 mmol/​litre during intercurrent infections, although this is not firmly evidence-​based. Myocardial infarction This is discussed in detail later in this chapter. Surgery Guidelines for the United Kingdom can be found at: http://​ www.diabetologists-​abcd.org.uk/​JBDS/​JBDS_​IP_​Surgery_​Adults_​ Full.pdf Surgery can be hazardous to patients with diabetes:  the counter­regulatory stress response to surgical trauma can rapidly lead to hyperglycaemia and ketoacidosis, especially in insulin-​ deficient patients, while poorly controlled diabetes accelerates catabolism and delays wound healing. Moreover, insulin and the sulphonylureas can cause severe hypoglycaemia in fasted or anor- exic patients, which can be particularly dangerous during general anaesthesia. Glycaemic control must therefore be meticulous throughout the perioperative period. A  routine management policy should be agreed between the diabetes care team, surgeons, anaesthetists, and ward staff, and this will greatly reduce the risks of operating on diabetic people. Fitness for surgery should be carefully assessed, in view of cardiovascular or other complications. Patients may need to be admitted some days before operation to optimize their treatment. For type 2 patients who are well controlled by diet or oral agents and undergoing minor surgery only (anticipated to miss only one meal), long-​acting sulphonylureas (glibenclamide) should be changed to short-​acting ones (e.g. gliclazide) some days before sur- gery to reduce the risk of hypoglycaemia. Oral agents and breakfast should be omitted on the morning of operation and blood glucose should be monitored closely. Persistent hyperglycaemia should be treated with the intravenous insulin and fluids described next. For all other patients with diabetes, insulin should be given as a variable-​rate intravenous infusion (VRII, previously referred to as sliding scale insulin) adjusted according to hourly blood glucose measurements, which provides greater flexibility. If the patient is in steady state, this will both maintain satisfactory glycaemic control and prevent hypokalaemia (insulin enhances potassium entry into skeletal muscle). This regimen should be started on the morning of surgery and continued until the patient is able to eat and drink normally, when the usual treatment can be resumed. Recent guide- lines advise that the patient’s usual long-​acting insulin should be continued during the period of VRII. Adjustments to the insulin infusion rate are made aiming to keep blood sugars between 6 and 10  mmol/​litre (acceptable levels 4–​12  mmol/​litre). When using a variable-​rate intravenous insulin infusion the capillary glucose should be checked hourly and the serum electrolytes measured daily. A typical infusion scale is shown in Table 13.9.1.8, but in- sulin infusion rates can be increased for a given glucose range if glucose levels are not falling. Acute metabolic complications of diabetes
and their treatment Diabetic ketoacidosis This is uncontrolled hyperglycaemia with hyperketonaemia severe enough to cause metabolic acidosis. It remains a major cause of death in patients with type 1 diabetes under 20 years of age, although the mortality has fallen from 8% to 0.67% with modern treatment (higher in older patients with diabetic ketoacidosis precipitated by infection or myocardial infarction). Prompt diagnosis and careful management can prevent many deaths. Causes Diabetic ketoacidosis only develops when severe insulin defi- ciency, compounded by an excess of glucagon, stimulates lipolysis and a massive increase in ketogenesis (see earlier). It therefore Table 13.9.1.8  Starting insulin infusion scale for perioperative patients on variable-​rate insulin infusion Bedside capillary blood
glucose (mmol/​litre) Initial rate of insulin infusion
(units per hour) <4.0 0.5 (0.0 if a long-​acting background insulin has been continued) 4.1–​7.0 1 7.1–​9.0 2 9.1–​11.0 3 11.1–​14.0 4 14.1–​17.0 5 17.1–​20 6

20 Seek diabetes team or medical advice Reproduced, with permission from NHS Diabetes, from K Dhatariya et al., Management of adults with diabetes undergoing surgery and elective procedures: improving standards (2011), http://​www.diabetes.nhs.uk/​document.php?o=2178

section 13  Endocrine disorders 2506 almost always occurs in untreated or poorly treated type 1 dia- betes and is generally regarded as the hallmark of that disease. However, diabetic ketoacidosis can occur in subjects with type 2 diabetes who are relatively insulin deficient, especially when the secretion of counterregulatory hormones (especially glucagon) is increased by severe intercurrent illness. Precipitating factors include: • newly presenting type 1 diabetes • omission or underdosing of insulin by established type 1 diabetic patients, which may be deliberate in patients with disturbances of body image • intercurrent illness, such as infections, myocardial infarction, stroke, trauma, surgery, and burns; many patients (and their doc- tors) fail to increase insulin dosages or monitor blood glucose during such events About 30 to 40% of episodes are unexplained; omitted or inad- equate insulin treatment should always be suspected if no obvious infective or other cause is found. Pathophysiology Diabetic ketoacidosis is due to the accumulation of ketones, that is, acetoacetate and its derivatives, 3-​hydroxybutyrate (or β-​hydroxybutyrate) and acetone (see Fig. 13.9.1.11). They are gen- erated by β-​oxidation of free fatty acids within the mitochondria of the liver. Free fatty acids enter the cytoplasm of hepatocytes and combine with coenzyme A (CoA) to form their fatty acyl-​CoA de- rivatives. These are then transported into the mitochondria by the carnitine shuttle, a complex of two linked enzymes, carnitine palmitoyltransferase I (CPT-​I) on the outer mitochondrial mem- brane and carnitine palmitoyltransferase II (CPT-​II) on the inner. CPT-​I, and the overall activity of the shuttle, is powerfully inhibited by insulin and stimulated by glucagon. Once inside the mitochon- dria, free fatty acids undergo β-​oxidation to yield ATP (the process of oxidative phosphorylation) and acetyl-​CoA. The latter is con- verted to acetoacetate, which may be oxidized to 3-​hydroxybutyrate or undergo condensation to produce acetone. Ketones are transported out of the liver and are used as metabolic fuels by various tissues including the brain; they supply a few% of total energy needs after an overnight fast, but the proportion rises to over one-​third during prolonged fasting. When produced in ex- cess, they can accumulate rapidly, especially if plasma levels exceed 5 mmol/​litre (about 10 times normal), when tissue uptake mechan- isms become saturated. Ketogenesis is greatly enhanced in uncon- trolled type 1 diabetes because of the combination of low insulin with increased glucagon concentrations: lipolysis is unrestrained, and the uptake into liver mitochondria of the increased amounts of fatty acyl-​CoA is stimulated by the synergistic effects on CPT-​I of high glucagon and low insulin. The main consequences of raised cir- culating ketone levels are shown in Fig. 13.9.1.11 and listed next: • Acidosis:  acetoacetate and 3-​hydroxybutyrate are both moder- ately strong organic acids and lower the extracellular pH when the buffering capacity of plasma proteins is exceeded. Ion exchange across cell membranes leads to intracellular acidosis which com- promises cellular metabolism because many crucial enzymes operate within a narrow pH range. Clinical measurements of acid-​ base status are confined to the extracellular fluid and may under- estimate the severity of intracellular acidosis. • Diuresis: ketones are filtered in the urine and are osmotically ac- tive. They therefore exacerbate the osmotic diuresis caused by glycosuria and the resulting polyuria, electrolyte losses, dehydra- tion, and hypovolaemia. • Nausea: through direct stimulation of the chemoreceptor trigger zone in the medulla. Clinical features Diabetic ketoacidosis usually presents with classical hyperglycaemic symptoms (see Table 13.9.1.2), together with features of acidosis and hyperketonaemia: • Acidotic (Kussmaul) breathing is deep, sighing hyperventilation which has been mistaken for panic attacks, pulmonary embolism, and left ventricular failure. • Nausea and vomiting are ominous signs, because dehydration de- velops quickly in polyuric patients unable to drink. • Drowsiness and coma occur late and may indicate early cerebral oedema. The patient generally looks ill and may show postural hypoten- sion and other signs of dehydration and hypovolaemia. Acetone is volatile and may be smelled on the breath (ketotic foetor; ‘nail var- nish remover’ odour). Some patients are hypothermic due to heat loss from peripheral vasodilation, and this may mask the pyrexia of infection. Children with diabetic ketoacidosis often complain of abdominal pain, sometimes mimicking acute appendicitis or other surgical emergencies. A full examination is essential to identify any intercurrent illness. Investigations and diagnosis Once suspected, the diagnosis can be confirmed on the spot with a finger-​prick blood glucose measurement and urine or blood analysis for ketones. Recent guidelines place emphasis on the value of bed- side finger-​prick testing for blood ketones both in diagnosis (plasma ketone level >3.0 mmol/​litre, glucose <11.0 mmol/​litre or previous diabetes diagnosis, bicarbonate <15  mmol/​litre, and/​or pH <7.3) and in subsequent management. Treatment with intravenous saline and insulin should begin immediately, and baseline investigations Insulin deficiency + glucagon excess Cellular dysfunction Cerebral oedema Shock ? ? Vomiting Osmotic diuresis Blood ketones Fluid and electrolyte depletion Acidosis Blood glucose Fig. 13.9.1.11  Pathophysiological changes in diabetic ketoacidosis. Cellular dysfunction induced by intracellular acidosis, as well as cerebral oedema and shock are potentially life-​threatening.

13.9.1  Diabetes 2507 carried out. Venous blood is taken for biochemical screening and can also be used to assess acid-​base status as the difference between arterial and venous pH is 0.02 to 0.15 units and for bicarbonate it is 1.88 mmol/​litre, differences that are small enough not to influence management, particularly in view of the much greater convenience of venous sampling. High ketone concentrations cause a large anion gap (i.e. plasma [Na+ + K+] exceeds [HCO3–​ + Cl–​] by more than 17 mmol/​litre). Additional tests to identify the cause of the episode should include a full blood count, urine and blood culture, chest radiograph, and, especially in older patients, ECG and cardiac enzymes or troponin levels. Typical values and some diagnostic pitfalls in diabetic ketoacid- osis are shown in Fig. 13.9.1.12. Management Diabetic ketoacidosis is a potentially life-​threatening medical emer- gency that requires urgent treatment with scrupulous clinical and biochemical monitoring: many avoidable and serious accidents still happen because the patient is abandoned once treatment has been started. Severe diabetic ketoacidosis is best managed initially on a high-​dependency or intensive care unit. The highest priority is to correct hypovolaemia and dehydration, which will often improve acidosis and hyperglycaemia. Insulin re- placement must also be started urgently. However, it now appears likely that the high mortality of diabetic ketoacidosis has been partly due to overenergetic replacement of intravenous fluids (especially bicarbonate) and perhaps insulin, which may predispose to the de- velopment of cerebral oedema. The treatment guidelines (see Fig. 13.9.1.12) are based on large studies that have reported very low mortality and morbidity. Fluid replacement Good intravenous access is crucial: a large peripheral vein may be used but a central venous cannula is safest for severely hypovol- aemic patients and for older people or those at risk of heart failure, in whom monitoring of central venous pressure is essential. Most patients recover rapidly with slower fluid replacement than was previously recommended. For those who are not shocked give: • 1 to 2 litres in 2 h, then • 1 litre over the next 4 h, then • 4 litres over the next 24 h Fluid losses in urine or vomit should be added to these volumes. Shocked or oliguric patients may require faster fluid repletion, pos- sibly with plasma expanders rather than saline, while slower re- placement is safer in those with signs of fluid overload, myocardial infarction, heart failure, or any suspicion of cerebral oedema. Urine output must be monitored closely, as must blood pressure, central venous filling, and signs of pulmonary or peripheral oedema. Saline containing potassium is the logical fluid to replace the losses of Na+, K+, and Cl–​ induced by the osmotic diuresis of diabetic ketoacidosis. The use of intravenous bicarbonate to try to correct acidosis is contentious, both in terms of biochemistry and clinical outcome (see next). Isotonic (0.9%) saline is used initially. Half isotonic (0.45%) sa- line has been suggested to empirically replace 1 or 2 litres of iso- tonic saline, if severe hyperosmolarity (>350 mosmol/​kg) and/​or hypernatraemia (>150 mmol/​litre) are present. However, the ra- tionale may be flawed: 0.9% (normal) saline is already hypotonic with respect to the patient’s hypertonic plasma, and the use of even more hypotonic solutions would seem likely to exacerbate the intra- cellular movement of water which may lead to cerebral oedema. It is now recommended that 10% dextrose is added (125 ml/​h) as an add- itional infusion alongside the rehydration with saline when plasma glucose has fallen to below 14 mmol/​litre to prevent hypoglycaemia (insulin is still required to prevent ketogenesis and promote glucose utilization in the tissues). Intravenous sodium bicarbonate was previously recommended for severe acidosis. However, the hope that adding alkali will cor- rect acidosis may be oversimplistic. HCO3–​ and H+ ions (from 3-​ hydroxybutyric and acetoacetic acids) combine extracellularly to produce H2CO3, which dissociates to produce water and CO2; this may reduce extracellular acidosis, but as cell membranes are im- permeable to HCO3–​ ions, the all-​important intracellular acidosis is not improved. Indeed, CO2 can enter cells where it can combine with water to produce H2CO3, itself a weak organic acid that can dissociate into H+ and HCO3–​ ions. Paradoxically, therefore, intra- venous bicarbonate administration could worsen intracellular acid- osis and there is evidence from animal models of acidosis that this occurs. Worryingly, a recent study identified bicarbonate admin- istration as the most important independent predictor of cerebral oedema in children with moderately severe diabetic ketoacidosis. Another problem with high-​strength (8.4%) sodium bicarbonate so- lution is the intense thrombophlebitis it causes when given intraven- ously, which can obliterate even large central veins. Extravasation can also cause severe tissue necrosis. The current consensus is that bicarbonate is unlikely to be benefi- cial but runs the risk of doing harm, and that it should not be used in the treatment of diabetic ketoacidosis Potassium replacement Hyperkalaemia and hypokalaemia are the main causes of death in diabetic ketoacidosis (DKA) in adults. Diabetic ketoacidosis always Fig. 13.9.1.12  Guidelines for the management of diabetic ketoacidosis.

section 13  Endocrine disorders 2508 depletes total body K+ stores to a variable degree because of elec- trolyte losses through osmotic diuresis, but H+/​K+ exchange across the plasma membrane encourages K+ to leak out of cells in acidosis. Plasma K+ levels can therefore be low, normal, or high, and dan- gerous hyperkalaemia can be present, especially if severe hypovol- aemia causes prerenal failure. During insulin replacement, K+ is carried intracellularly with glucose, and plasma K+ levels can fall rap- idly. Frequent monitoring of K+ (every 2 hours) is therefore essential in the safe management of diabetic ketoacidosis, and patients with marked K+ disturbances should have continuous ECG monitoring. Potassium replacement should be determined by current plasma K+ levels: • Add 40 mmol of KCl to each litre of intravenous fluid if K+ is normal (3.5–​5.0 mmol/​litre). • Add 40 mmol/​litre of KCl to each litre if plasma K+ is less than 3.5 mmol/​litre and review in 1 h. • Omit KCl if plasma K+ is more than 5.0 mmol/​litre, because of the risk of precipitating arrhythmias. Repeat K+ in 2 hours and re- place if fallen below 5.0 mmol/​litre. Insulin replacement Continuous fixed rate intravenous infusion is the best way to give insulin in diabetic ketoacidosis; subcutaneous and intramuscular absorption are too erratic to be safe and the rate of fall of glucose (one of the factors implicated in cerebral oedema) cannot be easily controlled. A dose of 50 U soluble insulin should be added to 50 ml isotonic saline (i.e. 1 U/​ml) and delivered by a syringe driver pump, either into a separate vein or piggy-​backed into the intravenous fluids line only if a one-​way valve is present. Because the half-​life of insulin in the circulation is only a few minutes, blood glucose and ketone levels will rise rapidly if insulin delivery is interrupted; hourly monitoring of blood glucose is there- fore mandatory during intravenous insulin. Failure of glucose to fall usually means that the pump has been turned off or that the infusion cannula is blocked. It is now recommended that a fixed rate infusion at 0.1 U/​kg/​hr is used (to take account of the fact that some over- weight individuals will be insulin resistant), rather than a variable rate or ‘sliding scale’. The rationale is to promote as rapid resolution of ketosis as possible. When blood glucose falls to below 14 mmol/​ litre an infusion of 10% dextrose 1 litre per 8 hours is then recom- mended, along with any saline rehydration, to prevent hypogly- caemia and to allow the same rate of insulin infusion to correct the ketosis. Adequate response to insulin is judged by a fall in blood ke- tones by more than 0.5 mmol/​litre per hour, or a fall in blood glucose by more than 3 mmol/​h. If these parameters are not being met the infusion system should be carefully checked and, if working prop- erly, increases in the fixed rate insulin infusion by 1 U/​h are made until an adequate response is achieved. Resolution of ketosis is said to have occurred when the following criteria are satisfied: plasma ketones less than 0.6 mmol/​litre and venous pH greater than 7.3. The urine may misleadingly still test positively for ketones at this stage due to excretion of the previous ketone load. Once resolution of ketosis is confirmed, the patient can be converted back to their usual subcutaneous insulin regimen (or begun on a regimen if this is a new diagnosis of diabetes), assuming they are able to eat and drink normally. If it is impossible to give a controlled intravenous infusion, then intramuscular soluble insulin can be injected every 4 h or so, starting with 20 U and attempting to titrate subsequent dosages (e.g. 5–​10 U hourly). Other complications Intercurrent illness must be treated energetically. Broad-​spectrum antibiotics are often given prospectively. Myocardial infarction (see next) has a poor prognosis if it causes diabetic ketoacidosis. Shock may lead to prerenal failure and sometimes acute tubular necrosis. Plasma expanders and inotropes may occasionally be re- quired for severe hypotension, although rehydration, as already mentioned, is usually adequate. Cerebral oedema still accounts for 50% of fatalities in diabetic ketoacidosis, especially in children, although modern management protocols with slower fluid replacement and low-​dose intravenous insulin infusion can markedly reduce its incidence. The cause is thought to be shifts of ions and water into the brain, particularly the movement of water into dehydrated, hypertonic cells when relatively hypotonic fluids reach the extracellular space. Such shifts would be predicted with the administration of isotonic and particularly with hypotonic fluids. Risk factors for cerebral oedema include over-​ rapid falls in blood glucose, excessive fluid replacement, and high insulin dosages. Insulin can affect various ion transport mechanisms in the brain, but its role remains mysterious and may simply reflect changes in extracellular osmolarity. Interestingly, CT scanning be- fore fluid and insulin replacement has demonstrated subclinical cerebral oedema in children with diabetic ketoacidosis. Swelling of the brain within the cranium causes coning, leading to cardiorespiratory arrest. It presents as a decline in consciousness, usually rapid, and often when the patient’s metabolic state has been stabilized. Papilloedema may be present, and CT or MRI will show characteristic swelling, with loss of cortical features and squashing of the ventricular system (Fig. 13.9.1.13). It is usually fatal (in >90% of established cases), but intravenous mannitol (0.2 g/​kg over 30 min, repeated hourly if there is no improvement) may help by raising the Fig. 13.9.1.13  Cerebral oedema in a patient recovering from diabetic ketoacidosis. The CT shows generalized swelling and loss of cortical detail with squashing of the cerebral ventricles.

13.9.1  Diabetes 2509 osmolality of extracellular fluid and drawing free water out of the brain; there is no firm evidence to support the use of dexamethasone. Adult respiratory distress syndrome is due to accumulation of fluid in the alveoli, perhaps due to ionic and water shifts or to ex- cessive leakiness of the pulmonary capillaries. Hypoxia is severe, and chest radiography shows an appearance like left ventricular failure but with a normal heart size. Risk factors include rapid fluid replacement. It carries a poor prognosis, but ventilation with high-​ concentration oxygen may be useful supportive treatment. Acute gastric dilatation (gastroparesis) presents with vomiting and may produce a succussion splash and a ground-​glass appear- ance on abdominal radiograph. Nasogastric drainage may be needed to prevent aspiration, especially in the unconscious patient. Hypotension may persist or develop during treatment and gen- erally reflects inadequate fluid replacement. Alternative causes of hypotension including septic shock and cardiogenic shock should also be considered. Polyuria secondary to continuing high glucose levels may occasionally give false reassurance that the patient’s fluid replacement status is adequate. Persisting acidosis despite correction of blood sugar levels and a fall in plasma (and later urine) ketones raises the possibility of lactic acidosis secondary to sepsis or metformin use (see next). Blood lactate levels should be measured. Plasma sodium levels may rise despite fluid replacement as the initial high glucose levels may have resulted in an erroneously low initial reading. Falling sodium levels may reflect the need for more saline and less dextrose-​based fluid replacement. Hypothermia indicates a poor outcome. It may respond to rewarming with a space blanket. Subsequent management When the patient can eat and drink, intravenous fluids and insulin can be discontinued. There is no need for an insulin infusion; in- stead, the patient can be restarted on their usual subcutaneous insulin regimen (or a new regimen, if newly diagnosed). The intra- venous insulin infusion should be maintained until the first injec- tion containing soluble or short-​acting analogue insulin has had time to act (the intravenous infusion should not be stopped if only basal insulin has been given). The causes of the episode must be determined if possible, and efforts made to prevent a recurrence. from happening again. The patient’s understanding of diabetes, including the ‘sick-​day’ rules (Box 13.9.1.1), must be checked and reinforced if necessary. Recurrent diabetic ketoacidosis is a feature of brittle diabetes, and these patients need careful monitoring and counselling. Hyperglycaemic hyperosmolar state (HHS)—​formerly hyperosmolar non​ketotic state (HONK) HHS is distinguished from diabetic ketoacidosis by the absence of marked hyperketonaemia (<3.0  mmol/​litre) and metabolic acidosis (pH >7.3). Hyperglycaemia can be greater than in dia- betic ketoacidosis (typically >30 mmol/​litre) and, together with a rise in urea due to dehydration and prerenal failure, may elevate the plasma osmolality to well over 350 mosmol/​kg (usually >320 mosmol/​kg). HHS may be the first presentation of type 2 dia- betes. (Guidelines for treatment can be found at https://​abcd.care/​ joint-​british-​diabetes-​societies-​jbds-​inpatient-​care-​group.) Ketosis does not develop because circulating insulin levels are high enough to suppress lipolysis and ketogenesis; these pa- tients are therefore C-​peptide positive, with type 2 diabetes which is often previously undiagnosed. It is more common in people of Afro-​Caribbean origin. Precipitating factors include myocardial infarction, stroke, infection, and diabetogenic drugs such as gluco- corticoids and thiazide diuretics; fizzy glucose drinks may also contribute. Presentation is typically with classical hyperglycaemic symptoms (polyuria, intense thirst, weight loss, blurred vision), without the features of ketoacidosis. Confusion, drowsiness, and coma are more common than in diabetic ketoacidosis. At blood glucose levels over 30 mmol/​litre, drowsiness can lead to a cycle of deterioration as fluid loss continues due to the osmotic diuresis, but the patient is increas- ingly too lethargic to drink adequate replacement fluids. Progressive dehydration then leads to even higher glucose levels, more lethargy, and a further decline in oral intake. Complications include thrombotic events such as stroke and per- ipheral arterial occlusion, and deep venous thrombosis and pul- monary embolism, these being due apparently to increased blood viscosity. Mortality is 15 to 20%, partly because these patients are old and often have a serious precipitating illness. Biochemical features of HHS are: • hyperglycaemia:  often over 50 mmol/​litre, sometimes over 90 mmol/​litre • hypernatraemia: often over 155 mmol/​litre (may be artefactually depressed by high glucose levels) • uraemia due to dehydration, with or without renal failure • hyperosmolality: over 320 mosmol/​kg • blood and ketone levels are normal or only slightly raised (usually through anorexia) • arterial pH, venous bicarbonate, and anion gap show no features of severe acidosis Management of HHS • Saline replacement must be particularly cautious in older pa- tients, in whom cardiac disease is common; 0.9% saline should be used rather than the previously used half isotonic (0.45%) sa- line. Rehydration will lower the blood glucose level, which will lower osmolality and fluid will shift into the intracellular space. The falling glucose will cause a rise in sodium levels which is not an indication for hypotonic saline; a rising sodium is only cause for concern if the osmolality is not falling. Osmolality should be measured hourly. • Potassium levels must be carefully monitored and replaced as described. • Rehydration alone will lower glucose. A safe glucose fall is by 4 to 6 mmol/​l per hour, aiming for a glucose of 10–​15 mmol/​litre. If significant ketones are present intravenous insulin should be started immediately however if not insulin should not be com- menced immediately as it may drop osmolality dramatically and precipitate cardiovascular collapse. Once rehydration is adequate and glucose is still not falling then intravenous insulin may be started at a rate of 0.05 unit/​kg/​hour. • low-​dose heparin (5000 U subcutaneously 8-​hourly or low mo- lecular weight heparin once daily) should be given prophylactically,

section 13  Endocrine disorders 2510 but full anticoagulation should be reserved for proven thrombo- embolism as the risks of fatal gastrointestinal bleeding are high. Intercurrent illness, such as infection, must be sought and treated appropriately. After recovery, many of these patients should be converted to in- sulin however after weeks to months can be successfully weaned off insulin. Drugs and other precipitating factors must be identified and avoided if possible. Lactic acidosis Lactate is generated by glycolysis and its levels rise rapidly during tissue anoxia (e.g. during shock, cardiac failure, or pneumonia) or when the liver is prevented from utilizing it as a gluconeogenic substrate (e.g. in hepatic impairment). Lactic acidosis is best known in diabetic patients as a rare but often fatal complication of the biguanides, phenformin and metformin, which act mainly by inhibiting hepatic gluconeogenesis. The risk is about 10 times higher with phenformin than with metformin, and it is very rare during metformin treatment as long as other predisposing factors (the major organ failures) are avoided. Lactic acidosis presents as coma with metabolic acidosis (reduced arterial pH and venous bicarbonate) and a wide anion gap due to hyperlactataemia. Blood glucose levels are usually raised. Treatment is still unsatisfactory. Intravenous sodium bicar- bonate may paradoxically aggravate intracellular acidosis, although forced ventilation to blow off carbon dioxide may help (see earlier). Haemodialysis may both clear lactate and hydrogen ions, and cor- rect any sodium overload following bicarbonate administration. Sodium dichloroacetate, which stimulates pyruvate dehydrogenase to metabolize lactate, is undergoing evaluation. Mortality remains high (>30%), partly because of the organ fail- ures that commonly coexist. Hypoglycaemia Hypoglycaemia is an inevitable side effect of antidiabetic drugs that raise circulating insulin levels, namely insulin itself and sulphonylureas; it does not occur with metformin or thiazolidinediones alone, or with dietary restriction. Common contributory factors are: • accelerated insulin absorption (e.g. due to exercise or hot surroundings) • unfavourable timing of insulin injection: injecting too soon be- fore eating can cause late postprandial hypoglycaemia, while long-​ acting insulins injected in the early evening often cause nocturnal hypoglycaemia • too much insulin injected: dosage errors are quite common, par- ticularly in older people • inadequate food intake: missed, delayed, or small meals; vomiting, including gastroparesis • exercise:  this hastens insulin absorption while enhancing in- sulin action; delayed hypoglycaemia may occur many hours later because muscle continues to take up glucose to replenish glycogen • alcohol:  this inhibits hepatic gluconeogenesis, preventing the increase in hepatic glucose output that is crucial for restoring euglycaemia • impaired awareness of early warning symptoms (see next) Progressively more frequent or severe attacks may be caused by various conditions, which should always be considered: • weight loss, including anorexia nervosa and appetite disorders (relatively common in young women with type 1 diabetes) • loss of counterregulatory hormones:  Addison’s disease, hypo­ thyroidism, hypopituitarism, blunted glucagon secretion in long-​ standing type 1 diabetes • intestinal malabsorption, notably coeliac disease (more common in type 1 diabetes) • renal failure, which impairs the clearance of insulin • deliberate inappropriate injection of insulin, often in the context of ‘brittle’ diabetes Manifestations Clinical features of hypoglycaemia are due to an autonomic dis- charge, predominantly sympathetic, together with the cerebral effects of neuroglycopenia. Falling glucose levels are sensed by glucose-​sensitive neurons, which are found in the periphery (vagal sensory endings in the portal vein) and medulla as well as the hypo- thalamus. This triggers a powerful sympathetic discharge that re- leases adrenaline from the adrenal medulla and noradrenaline from sympathetic nerve endings, causing the familiar ‘flight or fight’ response. Features include pallor (cutaneous vasoconstriction), sweating (which can be very profuse), tremor (a β2-​adrenergic effect on skeletal muscle), and tachycardia; systolic blood pressure rises due to increased cardiac output while pulse pressure widens—​giving the typical bounding pulse—​because β2-​mediated vasodilatation in skeletal muscle causes peripheral resistance to fall. Hypoglycaemia also triggers the secretion of counterregulatory hormones, namely glucagon and adrenaline (both crucial to re- storing euglycaemia), growth hormone, and cortisol. Collectively, these inhibit insulin secretion and raise blood glucose by enhancing hepatic glycogenolysis and gluconeogenesis, causing glucose to pour out of the liver. Defects in glucagon or adrenaline release (which occur in long-​standing type 1 diabetes, for example), or in the ability of the liver to produce glucose (e.g. the presence of ethanol which inhibits gluconeogenesis, or a recent glucagon in- jection which depletes liver glycogen) will delay recovery of blood glucose. The physiological and neurological features of hypoglycaemia usually develop in a fixed sequence when blood glucose is lowered in a controlled fashion in the laboratory. However, this hierarchy may not be apparent in real life, and some patients specifically lose their awareness of the early warning symptoms (see next). Key events as glucose falls are: • at c.3.8 mmol/​litre: increased glucagon and adrenaline secretion • at c.3.0 mmol/​litre: onset of hypoglycaemic symptoms • at c.2.8 mmol/​litre: neuroglycopenia and cognitive impairment • less than 2 mmol/​litre: coma Symptoms of hypoglycaemia The symptom complex can be extremely variable, and hypogly- caemia should be suspected as the cause of any ‘funny turn’ in patients treated with insulin or sulphonylureas. Autonomic mani- festations include sweating, tremor, tachycardia, and hunger, while neuroglycopenia can cause drowsiness, confusion, incoordination,

13.9.1  Diabetes 2511 dysarthria, and automatic or disinhibited behaviour; distinct neurological deficits include aphasia, diplopia, and hemiparesis. Non​specific malaise and headache afterwards are also common. Nocturnal episodes may pass completely unnoticed by the patient, or may cause sweating and restlessness (often obvious to the patient’s partner), vivid nightmares, nocturnal epilepsy, or a hungover feeling the following morning. Awareness of hypoglycaemic symptoms Diabetic patients rely on the early autonomic symptoms (sweating, shaking, and hunger) to warn them of an impending hypoglycaemic attack, when corrective action can be taken. In some patients, the early warning symptoms are attenuated or not noticed at all; this clumsily named ‘hypoglycaemia unawareness’ is potentially dan- gerous because severe neuroglycopenia (confusion, fitting, ir- rational behaviour, coma) may suddenly incapacitate the patient. Reduced awareness of hypoglycaemia occurs particularly in two set- tings, which may coexist: • Long-​standing type 1 diabetes. Some 30 to 50% of patients with diabetes of more than 20 years’ duration have decreased awareness of symptoms, and many also show a flat glucagon and adrenaline response to hypoglycaemia. Blunted recognition of hypogly- caemia by the central nervous system may be responsible. • Excessively tight glycaemic control impairs awareness of hypogly- caemia; for unknown reasons, even a single episode can blunt per- ception of symptoms and counterregulatory hormone responses for some days. Conversely, relaxing control and avoiding hypogly- caemia completely for several weeks can partially restore aware- ness of warning symptoms. The use of human insulin has been suggested to impair aware- ness of hypoglycaemic symptoms. Human insulin is relatively lipophilic—​hence its faster subcutaneous absorption—​which could theoretically promote its entry into the brain. Insulin may act directly on the brain to affect various autonomic processes, but detailed comparisons of human and animal insulins, both in the laboratory and in real life, have not shown any species differences in counterregulatory responses or the intensity of hypoglycaemic symptoms. Sequelae of hypoglycaemia Even the most dramatic neurological manifestations of acute hypoglycaemia—​including aphasia, hemiparesis, fitting, and unconsciousness—​usually resolve rapidly when blood glucose is normalized. Recovery from profound coma may take many hours or even days, and this is probably due to cerebral oedema. Patients who survive severe and prolonged hypoglycaemic coma may show permanent neurological damage, including memory loss, aphasia, and a vegetative state. There are concerns that repeated mild attacks, especially in children and perhaps particularly at night, can cause cumulative intellectual impairment, but this is not yet proven. Severe hypoglycaemia has been implicated in precipitating myo- cardial infarction or stroke, particularly in older people; rises in blood pressure and increased coagulability of the blood following sympathetic stimulation may contribute. Like any convulsions, hypoglycaemic fits may cause injury, including limb and vertebral crush fractures. Prolonged severe hypoglycaemia can be fatal and is one of the most common causes of death in young type 1 patients. Post-​ mortem studies show neuronal damage and necrosis in the hippo- campus and cerebral cortex. Hypoglycaemia has been suspected as a cause of death in patients found unexpectedly dead in bed; however, it is thought arrhythmias secondary to autonomic dysfunction may be responsible. Diagnosis and detection of hypoglycaemia Hypoglycaemia is easy to diagnose but is also easily missed; dif- ferential diagnoses include transient ischaemic attacks, psychosis, drunkenness, epilepsy, and migraine. Symptoms may be instantly recognizable to some patients, but may present atypically. If sus- pected, the blood glucose levels should be checked, taking care to avoid under-​reading artefacts with reagent test strips. Urinalysis is obviously of no use—​hence all patients receiving insulin or sulphonylureas must be able to check their blood glucose. The patient’s close associates should also know how to diagnose and treat hypoglycaemia. Various experimental hypoglycaemia detectors are undergoing development including subcutaneous sensors and transcutaneous near-​infrared spectroscopy. Continuous glucose monitoring sys- tems are able to give a real-​time display of blood glucose levels every few minutes and may prove especially valuable in recurrent hypoglycaemia. Current limitations include cost (sensors require replacing every 3–​6 days) and the fact that there is a delay of ap- proximately 20 min before interstitial fluid and blood glucose levels equilibrate, which means that the sensor can underestimate the severity of hypoglycaemia when blood glucose levels are falling rapidly. Prevention and treatment of hypoglycaemia It has been shown that insulin-​treated patients fear hypoglycaemia as much as blindness or renal failure; this may prevent them from tightening their diabetic control as much as their doctors would prefer. Many doctors underestimate the impact of hypoglycaemia; asking about it and trying actively to prevent it are an essential part of diabetes care. It should be emphasized that a ‘good’ HbA1c level can be misleading in this context, as it can be the re- sult of a combination of high blood glucose levels and recurrent hypoglycaemia. Attention to the factors listed earlier should help to reduce the frequency and severity of attacks. Advice about exercise, moderating alcohol intake, and timing of insulin injections and meals are par- ticularly important. Nocturnal hypoglycaemia can be reduced by checking the blood glucose at bedtime, and by taking long-​acting carbohydrate (e.g. bread or cereal) if the level is less than 6 mmol/​ litre. Long-​acting analogue insulin and insulin pump therapy may also be of value in this situation (see earlier). Blood glucose levels of less than 3 mmol/​litre should be treated immediately (see Table 13.9.1.9). Oral glucose or sucrose or other carbohydrate should be given if the patient can swallow safely. Give 20 to 30 g (e.g. 3–​6 Dextrosol tablets or 75–​150 ml of a glucose drink such as Lucozade) initially; if possible, check the blood glu- cose 15 min later and repeat glucose administration if this has not risen. Taking too much carbohydrate—​which is an understandable reaction, given the unpleasantness of hypoglycaemia—​can cause marked rebound hyperglycaemia.

section 13  Endocrine disorders 2512 If the patient is unconscious, give either: • glucagon 1 mg (0.5 mg in children), subcutaneously or intramus- cularly; with either route, glucose should rise within 10 to 15 min. Side effects of glucagon include malaise, nausea, and abdominal discomfort. Importantly, as glucagon acts primarily by breaking down hepatic glycogen (a limited resource), a second injection may be ineffective • intravenous glucose: 15 to 20 g intravenously, as a 10% solution. Formerly, 50% glucose was used however this is associated with a painful thrombophlebitis. Glucose gels or jam can be smeared inside the mouth and cheeks in the unconscious patient, but these alone are unlikely to correct serious hypoglycaemia. On recovery, blood glucose should be checked and oral glucose given as earlier. Slow recovery from coma may be due to cerebral oe- dema, which has a high mortality (c.10%) but may respond to intra- venous mannitol and forced ventilation with high inspired oxygen concentration. Once the episode is treated, its cause must be iden- tified if possible and corrective action taken to prevent it from hap- pening again. Since it takes some time to recover cognitive function after an episode of hypoglycaemia, recent driving guidelines in the United Kingdom recommend not driving for at least 45 min after correction of hypoglycaemia. Options for intractable, recurrent hypoglycaemia Recurrent, disabling hypoglycaemia often develops in patients who have maintained very tight glycaemic control over many years. The frequent hypoglycaemic episodes that often occur with tight control themselves result in loss of hypoglycaemia awareness and physiological defence mechanisms (e.g. glucagon and adrenaline). The result is more frequent hypoglycaemia—​‘hypoglycaemia be- gets hypoglycaemia’. Options for reversing this situation include careful inspection of injection sites and avoidance of areas of lipohypertrophy or potential for inadvertent intramuscular injec- tion, attempting less tight glycaemic control, the use of analogue in- sulins and carbohydrate counting as part of intensive multiple-​dose insulin therapy, and insulin pump therapy (with or without real-​time glucose monitoring). If these measures fail and recurrent hypogly- caemia continues to impact very significantly on the patient’s quality of life, islet transplantation remains an option as it generally achieves marked improvement in hypoglycaemia with a lower risk procedure than whole pancreas transplantation (see earlier). The patients are nonetheless exposed to immunosuppressive risks which need to be taken into account in weighing the risks and benefits. Chronic complications of diabetes Long-​term tissue damage is now the major burden of the disease, the greatest source of fear for diabetic people, and the most expensive item in the diabetes healthcare budget. The list of possible compli- cations is depressingly long but fortunately at least 40% of diabetic patients escape clinically significant complications, and improved diabetes care should reduce the risks even further. Microvascular complications—​retinopathy, nephropathy, and neuropathy—​are specific to diabetes and reflect the damage in- flicted on the microcirculation throughout the body. Retinopathy and nephropathy are obviously microvascular disorders; the micro- circulation of nerves (vasa nervorum) is also damaged in diabetic neuropathy, although other functional and structural abnormalities in the nerves themselves probably contribute. Macrovascular dis- ease is simply atherosclerosis. This causes typical coronary heart dis- ease, stroke, and peripheral arterial disease, but often behaves more aggressively than in people without diabetes. Other complications are due to irreversible biochemical and structural changes in tissues chronically exposed to hyperglycaemia. These include cataracts, whose formation during normal ageing is accelerated by diabetes, and specific soft tissue disorders such as limited joint mobility (diabetic cheiroarthropathy). Causes of chronic diabetic complications Role of hyperglycaemia Tissue lesions are identical in all types of diabetes, indicating that hyperglycaemia (or a closely related metabolic abnormality) is likely to be responsible. Microvascular disease in the retina, kidneys, and nerves is generally determined by the severity and duration of hyperglycaemia, although individual susceptibility varies consid- erably. By contrast, macrovascular disease does not display a clear dose–​response relationship with hyperglycaemia: instead, the risk is increased above glucose values that lie below the diabetic range (see earlier). Recent intervention studies have confirmed that improving gly- caemic control is rewarded by partial protection against micro- vascular complications but not atheroma. This principle is valid for both type 1 and type 2 diabetes, and is now embodied in their treat- ment targets (see Table 13.9.1.3). Two landmark studies are gener- ally cited, although several smaller ones have also reached the same conclusion. Type 1 diabetes The Diabetes Control and Complications Trial (DCCT) was a 12-​ year North American study of over 1400 patients that compared intensive insulin treatment (aiming for an HbA1c of 6% (42 mmol/​ mol)) with conventional (i.e. bad) regimens of once or twice-​daily injections (HbA1c about 9% (75 mmol/​mol)). Intensive treatment consisted of at least three daily injections or an insulin pump (CSII), and achieved a mean HbA1c of 7% (53 mmol/​mol). Table 13.9.1.9  Management of hypoglycaemia Immediate Patient conscious Oral glucose (20–​30 g) or sucrose Patient unconscious Intravenous glucose or Intramuscular or subcutaneous glucagon (1 mg; 0.5 mg in children)a Then Check blood glucose after 15–​20 min Confirm recovery (glucose > 5 mmol/​litre) On recovery Give long-​acting carbohydrate (e.g. sandwich, meal) Identify cause Re-​educate patient to avoid future episodes If recovery is delayed Patient unconscious Set up infusion of 10% dextrose; transfer to hospital Patient conscious Take more oral glucose a Caution with glucagon: it often causes nausea and malaise; depletes liver glycogen—​a second injection may therefore be ineffective; contraindicated in hypoglycaemia caused by sulphonylureas (glucagon stimulates insulin secretion).

13.9.1  Diabetes 2513 The trial concluded that improved glycaemic control reduced the risks of microvascular complications. In subjects who were initially free of complications, intensified treatment for 9 years de- creased the prevalence of a defined degree of background retinop- athy by 70% (i.e. from 55% with conventional treatment to 15% see Fig. 13.9.1.14), while the risks of developing microalbuminuria or clinical neuropathy fell by 33% and 70%, respectively. In subjects who already had background retinopathy at baseline, intensified treatment reduced the overall progression of retinopathy by 50%; more importantly the risks of suffering sight-​threatening retin- opathy or requiring laser treatment were reduced by a similar degree. The development of clinical nephropathy (overt albumin- uria) and neuropathy were each decreased by about 60%. By con- trast, intensified insulin treatment did not reduce the prevalence of macrovascular disease. However, an 11-​year follow-​up of subjects after the end of the trial (the Epidemiology of Diabetes Interventions and Complications study, EDIC) showed a reduction in cardiovas- cular events in the subjects who had previously been in the inten- sive treatment arm, indicating that a prolonged period of good glycaemic control does confer a lasting cardiovascular protective ef- fect. Persistent reductions in microvascular complications were also seen in those originally in the intensive therapy group, despite the fact that HbA1c levels were similar in the extended follow period, indicating that good glycaemic control in the early years can have very long-​lasting benefits. Type 2 diabetes The United Kingdom Prospective Diabetes Study (UKPDS) was guided through its 20-​year course by the late Robert Turner, who died shortly after it was completed. This very large trial followed the outcome of over 5000 patients treated with diet and lifestyle alone (termed ‘conventional’ treatment), or together with sulphonylureas, metformin, or insulin; confusingly, sulphonylureas and insulin treatments were both described as ‘intensive’ treatment. The trial confirmed the real-​life difficulty of achieving good glycaemic con- trol, especially against the progressive deterioration of type 2 dia- betes:  very few patients achieved and maintained the intensive target fasting plasma glucose of 6 mmol/​litre. The trial has been criticized for its convoluted design (which diluted its statistical power) and both the lumping and splitting of data for outcome analysis. Nevertheless, it yielded useful messages about the import- ance of treating both hyperglycaemia and hypertension and about the natural history of the disease itself. Its conclusions were broadly similar to those of the DCCT study: improved glycaemic control de- creased the risk of microvascular complications. Lowering HbA1c from 7.9% (63 mmol/​mol, conventional) to 7.0% (53 mmol/​mol, in- tensive) decreased the lumped rate of microvascular events by 25% (see Fig. 13.9.1.15), including sight-​threatening retinopathy (20%) and the development of microalbuminuria (33%). Across a reason- ably wide variety of HbA1c, lowering HbA1c by 1% reduced the risk of microvascular disease by about one-​third. Improved glycaemic control had no overall effect on macrovascular disease, although metformin treatment significantly decreased cardiovascular events (see earlier). By contrast, blood pressure lowering had very signifi- cant benefits in terms of both micro-​ and macrovascular disease (see Fig. 13.9.1.14). Extended follow-​up results of the UKPDS study have now been reported. These show that although the difference in HbA1c and blood pressure control between conventional and inten- sively treated groups was lost during this period, intensively treated subjects had a lower myocardial infarction and death rate (for treat- ment with metformin, suphonylureas, or insulin) and continued to have lower microvascular complications rates (for subjects treated with sulphonylureas or insulin) 10 years beyond the end of the study. This extended benefit from early tight control has been referred to as a ‘legacy effect’ and similar benefits are seen in type 1 diabetes (see earlier). Note that by contrast no ‘legacy effect’ was seen with blood pressure lowering—​within 1–​2 years of blood pressures equating be- tween the conventional and intensively treated groups any benefit from earlier blood-​pressure lowering was lost. Possible mechanisms of hyperglycaemic tissue damage High glucose levels can damage the function and structure of many tissues. The mechanisms currently thought most relevant to human diabetic complications probably operate to different degrees in dif- ferent tissues. Glycation of proteins and macromolecules Glycation begins with the non​enzymatic combination of glucose and other reactive sugars with amino groups of proteins, and with acceptor groups of other long-​lived macromolecules such as nucleic acids. Glycation is initially reversible, yielding a Schiff base which undergoes molecular rearrangement to form an Amadori product. Amadori products then undergo further reactions, including cova- lent cross-​linking with the sugar groups in other glycated proteins. Onset of retinopathy 0 Conventional p < 0.001 Intensive Progression of retinopathy p < 0.01 Intensive Conventional 40 20 Frequency of retinopathy (percentage of patients) 60 50 30 10 Time (years) 0 1 2 3 4 5 6 7 8 9 40 20 Frequency of retinopathy (percentage of patients) 60 50 30 10 0 Fig. 13.9.1.14  Intensive insulin therapy and improved diabetic control in type 1 diabetes reduces the risks of developing retinopathy (upper panel) and of established retinopathy progressing (lower panel). Data from the Diabetic Control and Complications Trial (DCCT).