5.3 The genetics of inherited cancers 456

5.3 The genetics of inherited cancers 456

ESSENTIALS All cancer can be termed ‘genetic’ as the disease is caused by somatic cell mutations (alterations in the DNA code), which result in abnormal cellular growth and/​or proliferation. Most of these mu- tations are sporadic (only occurring in the cancer cell), but some are due to the inheritance of a germline mutation in a cancer predisposition gene. Cancer predisposition genes can be rare and confer a high cancer risk (about 10-​fold lifetime relative risk), or common and confer a low to moderately increased risk (from just over onefold, up to
two-​ to threefold). They have been shown to be involved in causing some of the most common cancers, as well as some rare cancers. Mechanisms of inherited cancers Cancer predisposition genes are usually (1)  tumour suppressor genes—​for example, retinoblastoma caused by mutations in RB1—​ when, although the mutations are recessively inherited at the cel- lular level, they tend to manifest with a dominant inheritance pattern because the chance of a mutation being inherited by the offspring is 50%, and a sporadic mutation of the remaining normal allele oc- curs in a somatic cell during the lifetime of the germline mutation ­carrier to lead to cancer development; (2) oncogenes—​for ­example, the RET oncogene in the multiple endocrine neoplasia type 2A syndrome—​when gain-​of-​function mutations act in a dominant manner; (3) mismatch repair genes—​for example, causing genetic instability leading to the hereditary non​polyposis colorectal cancer (Lynch) syndrome. Clinical features Genetic predisposition to cancer should be suspected when can- cers: (1) occur at a younger age than is seen in the general popula- tion; (2) occur in more than one site or at multiple times at the same site in an individual (multiple primary tumours); or when (3) rare cancers are seen in clusters in a family; or (4) common cancers are seen in clusters in a family, often at a young age or with multiple primaries. Genetic predisposition to common cancers—​this includes (1) breast—​BRCA1 and BRCA2 mutations confer 80–​85% lifetime risk of breast cancer by 80 years (and also a significantly increased risk of ovarian cancer); TP53 (Li–​Fraumeni syndrome) mutations confer 90% risk of breast cancer by 60 years; (2) colon—​mutations in the
APC gene cause familial adenomatous polyposis and a virtually 100% risk of colon cancer by the age of 40 years; hereditary non​polyposis colorectal cancer, which is also associated with other cancers in addition to colon cancer, particularly endometrial cancer (15–​60% lifetime risk) and ovarian cancer (9–​12% lifetime risk). Rare inherited cancer syndromes—​there are many of these, including hereditary retinoblastoma, neurofibromatosis type 1 (optic nerve glioma, sarcoma, phaeochromocytoma), neurofibromatosis type 2 (acoustic neuroma and other tumours of the central nervous system), multiple endocrine neoplasia type 1 (parathyroid adenomas, pancre- atic islet tumours and anterior pituitary tumours), multiple endocrine neoplasia, type 2A and 2B (medullary thyroid cancer, phaeochromo- cytoma, parathyroid adenomas), Cowden’s syndrome (breast and other cancers), tuberous sclerosis (childhood brain tumours, car- diac rhabdomyomas), Gorlin’s syndrome (multiple basal cell naevi/​ carcinomas), Von Hippel–​Lindau syndrome (cerebellar and spinal haemangioblastomata, renal cell carcinoma, phaeochromocytoma, pancreatic tumours). Clinical management Patients and/​or families known or suspected to carry cancer pre- disposition gene mutations require genetic counselling and risk as- sessment, which may lead on to (1) cancer screening—​for example, colonoscopy for some individuals at increased risk of colon cancer; (2)  lifestyle changes—​for example, avoidance of known cancer-​ causing factors such as sunlight in Gorlin’s syndrome; (3) preven- tion strategies—​for example, prophylactic total colectomy in the familial adenomatous polyposis syndrome; (4)  cancer treatment considerations—​for example, tumours with a particular genetic ab- normality may respond to particular treatments; and (5)  genetic testing—​which may either be diagnostic (the detection of a mutation in an individual affected by cancer) or predictive (the detection of a mutation in a clinically unaffected individual). Future prospects—​gene alterations that predispose to cancer affect prognosis and treatment, hence genetic information is increasingly recognized as important in oncological practice. Cancer genetics will become part of mainstream clinical pathways for cancer care in the next decade and is likely to contribute to healthcare that is tailored to individual patients. 5.3 The genetics of inherited cancers Rosalind A. Eeles

5.3  The genetics of inherited cancers 457 Introduction Cancer is a common disease; it affects up to one-​half of the popula- tion during their lifetime. All cancer can be termed ‘genetic’ as cancer is caused by somatic cell mutations (alterations in the DNA code), which result in abnormal cellular growth and/​or proliferation. Most of these mutations are sporadic (occurring only in the cancer cell) and only a proportion of these cases is due to the inheritance of a germline mutation in a cancer predisposition gene. In these latter cases, the genetic alteration is in all cells of the body with the excep- tion of the gametes where, on average, the genetic alterations are in one-​half of the gametes. It used to be thought that such alterations were rare, but each conferred a high cancer risk (about 10-​fold life- time). However, recent studies have shown that there are also more frequent alterations in cancer predisposition genes with each of such mutations conferring a slightly increased risk (with just over a one- fold, up to a two-​ to threefold relative risk). This has implications for the role of genetic predisposition to cancer in general medical and oncological practice, as a larger proportion of cancer cases may harbour these latter alterations in the genetic code. Identification of such alterations will become important in the genetic profiling of the population to aid targeted cancer screening and prevention. There is emerging evidence that gene alterations that predispose to cancer af- fect prognosis and treatment and thus their significance is becoming incorporated into the clinical pathway for cancer care. Cancer gen- etics will become part of mainstream cancer care in the next decade and is likely to contribute to healthcare that is tailored to individual patients. Historical perspective Since Roman times, cancer has been known to run in families. In some families, the pattern of cancer incidence among family mem- bers is consistent with the inheritance of a mutated gene and car- riers of this mutated gene have a high risk of cancer. The chance that cancer will develop if an individual has a mutation in a cancer predisposition gene is called the penetrance. Most cancer predis- position genes have incomplete penetrance (i.e. the cancer risk is <100%). There are several types of evidence that inherited susceptibility plays a role in the development of cancer (also see Table 5.3.1): • In some inherited syndromes, which are rare in the general popu- lation, there is an increased risk of cancer in carriers of genetic mutation(s) which give rise to the syndrome, for example, neuro- fibromatosis type 1 (an autosomal dominant genetic syndrome—​ see later) which confers an increased risk (of a few per cent) of sarcoma and phaechromocytoma. Such syndromes can be ac- cessed using the database initiated in the early 1960s by Dr Victor McKusick as a catalogue of mendelian traits and disorders, first entitled Mendelian Inheritance in Man, now the website Online Mendelian Inheritance in Man (OMIM). • Some rare cancers cluster in families and form a ‘cancer family syndrome’, for example, the association of tumours in multiple endocrine neoplasia type 2 (MEN2:  the association of medul- lary thyroid cancer, phaeochromocytoma, and hyperparathyr- oidism). Such cancers are rare in the general population and so the occurrence of such rare cases either in relatives or in one indi- vidual is highly indicative of a genetic predisposition. • The observation that families exist which have several cases of ‘common’ cancers. Even though these cancers are prevalent in the general population, the number of such cases in these families far exceeds the number predicted by population rates. Often these cancers occur at ages earlier than seen in the general population (see Fig. 5.3.1) and family members have an increased occurrence of synchronous and metachronous lesions. • Epidemiological studies in the general population which show that there is an increased risk of cancer to relatives of cases and this risk markedly increases as the proband or index case with cancer is affected at a younger age or with bilateral cancers. • Genes have now been identified which, when mutated, are associ- ated with an increased risk of cancer. These may be rare mutations which confer a high (about tenfold) or moderate (just over two-​ to threefold) cancer risk, or common lower-​penetrance genes (which confer an increased risk of just over onefold up to about twofold). Historically, it was thought that genetic predisposition to cancer was a rare phenomenon and was predominantly observed in rare syn- dromes, such as multiple endocrine neoplasia, or was a rare compo- nent of other genetic diseases (such as neurofibromatosis). However, the advances in the Human Genome and HapMap projects (see next) have challenged this view and have shown that in fact genetic vari- ants which are common in populations form an important contri- bution to cancer risk. Inheritance, mechanisms of cancer predisposition, and the retinoblastoma story Inheritance of germline mutations in cancer predisposition genes may be either dominant, recessive, or X-​linked. We all carry two copies (alleles) of every gene, one copy from each parent, and as only one allele can be passed down to the next generation, there is a 50:50 chance as to which allele we inherit. In dominant inheritance the presence of a single mutated allele is usually sufficient to cause the associated disease and approximately 50% of all offspring develop the disease. In recessive inheritance the presence of a single mutated allele is insufficient for disease expres- sion and two mutated alleles are required. Usually both parents have to carry the mutated allele for the creation of an offspring affected by disease, but they themselves are unaffected, as their ‘normal’ allele overrides the effects of the mutated one. Two parents with a reces- sive mutated allele therefore have a 25% chance of having an affected child. An example of such a condition predisposing to colon cancer is the MutYH syndrome where adenomas occur in the colon. In such families colon cancer tends to cluster in siblings, as it is recessive (the colon cancer risk in mutation carrier parents is not thought to be raised above the general population). Most cancer predisposition genes are recessively inherited at the cellular level, but dominantly inherited in families (i.e. there is a 50:50 chance that the mutated allele will be inherited, but in the cancer cell both copies of the allele have to be altered for cancer to occur). In X-​linked inheritance the mutated gene is carried on the X chromosome. Females have two X chromosomes, and can therefore be carriers of the condition but are not usually affected. Males have

458 SECTION 5  Principles of clinical oncology Table 5.3.1  Some of the ‘rare’ syndromes associated with an increased risk of malignancy and their mode of inheritance (for further details, see text) Neoplasia or syndrome Malignancy Riska Modeb Gene Location NF1 Plexiform Neurofibroma Optic glioma Sarcoma <4% <15% <5% AD NF1 17q11 NF2 Bilateral acoustic Neuroma Meningioma Spinal tumours Other brain tumours 85% 45% 26% <10% AD NF2 22q12 Gorlin’s syndrome Basal cell carcinoma Medulloblastoma Meningioma 90% 5% 1% AD PTCH 9q22 Tuberous sclerosis Renal cancer Subependymal giant cell astrocytoma 4% 14% AD TSC1 TSC2 9q34 16p13 Cowden’s syndrome Breast cancer Endometrial cancer Renal cancer Thyroid cancer Colon cancer Melanoma 30% by age 50 28% 33% 15% 9% 6% AD PTEN 10q23 Li–​Fraumeni syndrome Brain tumours Breast cancer Sarcomas Leukaemia Adrenocortical cancer Other cancers Childhood cancer All adult tumours: 90% by
60 in women, 74% in men 24% by 20 AD TP53 17p13 MEN1 Pituitary tumour Pancreas Parathyroid Carcinoid 95% AD MEN1 11q13 MEN2A Medullary carcinoma of thyroid Phaeochromocytoma 70% 50% AD RET 10q11 Retinoblastoma Retinoblastoma Osteosarcoma Soft tissue sarcomas Melanoma Lung cancer Lymphoma Bladder cancer Uterine cancer Breast cancer Brain tumours Cancers in the mouth or nose 90% <10% AD RB1 13q14 Ataxia telangiectasia Lymphoma Leukaemia 60% 27% AR ATM 11q22 Bloom’s syndrome Leukaemia Solid tumours Rarely live to >40 AR BLM 15q26 Werner’s syndrome Various—​thyroid cancer, melanoma, soft tissue sarcoma, and osteosarcoma are reported Increased but not quantified AR RECQL2 8p12 Rothmund–​Thomson syndrome Various including osteosarcoma Increased but not quantified AR RECQ4 8q24 Fanconi anaemia Leukaemia Head and neck cancer Oesophagus/​cervix/​anus 22% at the age of 36 28% at the age of 49 AR FANC genes Xeroderma pigmentosum Skin cancer 90% often <20 yrs AR XP genes NF, neurofibromatosis type; MEN, multiple endocrine neoplasia; NA, not available. a Lifetime risk of cancer. b Mode of inheritance is classified as autosomal dominant (AD) or autosomal recessive (AR).

5.3  The genetics of inherited cancers 459 only one copy of the X chromosome, so if they inherit a mutated gene on the X from their mother, they will inherit the condition. A carrier female therefore has a 50% chance of passing the condition on to each of her sons, and a 50% chance that her daughters will be carriers. X-​linked familial prostate cancer has been observed in a few families, although the causal locus has not yet been refined. Mechanisms of action Cancer predisposition genes are usually tumour suppressor genes, oncogenes, or mismatch repair genes. Very rare instances have been described where alterations in the germ line have a downstream effect which is called an ‘epigenetic’ effect; in most such cases the germline change results in alteration of methylation of genes, which alters their expression and this results in increased cancer risk. Tumour suppressor genes are normal genes in which mutation tends to cause a ‘loss of function’ effect in the control mechanisms of growth and/​or cellular proliferation pathways. The first example was retinoblastoma, a tumour of the eye, usually in children, caused by mutations in the tumour suppressor gene RB1. Most cancer pre- disposition genes are tumour suppressor genes and are recessively inherited at the cellular level. However, they tend to manifest dom- inant inheritance (the chance of a mutation being inherited by the offspring is 50%). A sporadic mutation of the remaining normal al- lele occurs in a somatic cell during the lifetime of the germline mu- tation carrier to lead to cancer development. This two-​stage process in the development of cancers (where one stage is germline and the other is somatic) is known as Knudson’s two-​hit hypothesis. Oncogenes or proto-​oncogenes are mutated normal genes in which a mutation in only one allele tends to cause a ‘gain of function’ effect, resulting in increased growth or proliferation of the affected cells. They act in a dominant manner. They rarely cause predispos- ition to cancer, but examples of those causing cancer include the RET oncogene in the multiple endocrine neoplasia 2A syndrome and the MET oncogene in familial papillary renal cancer. Mismatch repair genes maintain the integrity of the genome and mutations in them allow acquired genetic damage to accumulate, re- sulting in the creation of a cancer cell. They classically predispose to a colorectal cancer syndrome called Lynch syndrome (named after a famous cancer geneticist, Dr Henry Lynch, who was one of the first to recognize that predisposition to common cancers could be in- herited), also known as hereditary non​polyposis colorectal cancer (HNPCC; see next). Hereditary cancer predisposition genes have also been classified into ‘gatekeeper genes’ and ‘caretaker genes’. Gatekeeper genes are those that regulate progression through the cell cycle. Disturbance of their function leads to an imbalance of cell division over cell death. This cellular proliferation is followed by the accumulation of mul- tiple somatic genetic events causing tumour development. Examples of gatekeeper genes include TP53 and RB1. Caretaker genes main- tain the integrity of the genome. Mutations occurring in these genes result in genetic instability, and it is this that results in mutation in other genes, including gatekeeper genes. The DNA mismatch repair genes in HNPCC are examples of caretaker genes. The multistep pathway of carcinogenesis The development of cancer is thought to be due to a multistep pathway involving several genetic changes. This is likely to be the ex- planation for incomplete penetrance, as not all genetic changes will occur in every individual who inherits the first genetic change. In in- herited predisposition to cancer, the first change is inherited in most cases. Less commonly, the first change is still in the germline but the mutation has occurred de novo in the germ cells (i.e. the carrier of the germline genetic mutation is the first individual in the family to har- bour the mutation; the rate at which this occurs is termed the ‘new mutation rate’). This is more common in some types of inherited predisposition to cancer than others. For example, the new mutation rate in HNPCC is about 50%, but in familial breast/​ovarian cancer due to BRCA1 or BRCA2 it is extremely rare. In the latter example, analyses of genetic variation flanking the region of the BRCA1 gene and knowledge of the genetic recombination rate have enabled the occurrence of mutations in this gene in the Ashkenazi Jewish popu- lation to be dated to pre-​Roman times. The most classic multistep model which has been published is that of the progression from colorectal adenomatous polyp to invasive colo- rectal carcinoma; a multistep pathway which involves at least five gen- etic changes (the so-​called colorectal adenoma–​carcinoma sequence, proposed by Vogelstein). This is shown diagrammatically in Fig. 5.3.2. See Chapter 5.2 for further discussion of the nature and develop- ment of cancer. Research approaches for the identification of cancer predisposition genes There are several approaches to locate a cancer predisposition gene. Once it is located and characterized, genetic testing can then be offered in the clinical setting. Cytogenetic alterations Gross chromosomal changes can be analysed by a karyotype or cyto- genetic analysis from a blood sample. Rarely, karyotypic abnormal- ities in an individual who has an unusually early onset of cancer and other unusual phenotypic features have indicated the location of a cancer predisposition gene. The chromosomal study of a man with mental retardation and polyposis led to the finding of a loss of part of chromosome 5, subsequently found to be the location of the polyp- osis gene APC, which predisposes to familial polyposis. 0 0 10 20 30 40 50 60 70 80 90 100 10 20 30 40 Age at diagnosis (years) Proportion of cases due to a breast cancer predisposition gene 50 60 70 80 90 Single first degree relative affected Two first degree relative affected Fig. 5.3.1  Graph showing the probability that breast cancer is due to a predisposition gene by age at diagnosis of breast cancer. Source data from Claus EB, Risch N, Thompson WD (1991). Genetic analysis of breast cancer in the cancer and steroid hormone study. Am J Hum Genet, 48, 232–42.

460 SECTION 5  Principles of clinical oncology Linkage analysis The concept of genetic linkage was first recognized by William Bateson, who noted that certain characteristics of his experi- mental plants tended to be coinherited, a phenomenon that had been described by the monk Gregor Mendel, 34 years previously. The explanation for this became clear once Morgan recognized that chromosomes contain the genetic material and two traits are coinherited (linked) only if the corresponding genes for them reside close together on the same chromosome. The search for cancer pre- disposition genes using linkage relies on collections of families with numerous cancer cases of the same cancer type. Coinheritance of specific genetic markers with the disease is said to show evidence of linkage if the coinheritance is greater than would be expected by chance. This is expressed as a ‘LOD score’ which is similar to a P value in clinical trials. A LOD score of more than 3 is considered statistic- ally significant and equivalent to odds of linkage of 1000 to 1. Phenotypic features A physical characteristic associated with a cancer predisposition syn- drome may give a clue as to its location. An example of this is the coexistence of aniridia and genitourinary abnormalities with Wilms’ tumour in the Wilms tumour-​aniridia syndrome (WAGR) syndrome. This is caused by a contiguous gene deletion on chromosome 11. Association studies Over 2000 disease susceptibility loci have been identified using genome-​wide association studies (http://www.ebi.ac.uk/gwas​). In these case–​control studies, allele frequencies are compared be- tween affected individuals and controls. It is important in such studies to have controls from the same ethnic/​racial group to avoid false as- sociations. Advances in the knowledge of the structure of the human genome have identified single nucleotide polymorphisms (SNPs or single base changes) throughout the genome (the HapMap pro- ject) and large-​scale genomic analysis using chip array technology (e.g. the Illumina or Affymetrix systems) has enabled millions of such SNPs to be analysed in each DNA sample at once. Such studies have identified common variants associated with disease risk, which are present in at least 5% of the population: in some cases they are found in over one-​half of the population. Although each SNP is as- sociated with a small increased risk (usually less than twofold), the risks can be multiplicative and so a combination of SNPs can give different ‘risk profiles’ in different individuals. This therefore opens up the possibility of SNP profiling to determine the risk of cancer in defined populations. Direct sequencing and the potential role of whole genome sequencing As the Human Genome Project has identified a large proportion of the genetic code, parts or all of this can be subjected to direct analysis by sequencing of genes to look for base pair changes (point mutations or nonsense mutations), or insertions or deletions of bases (frameshift mutations). Most genetic tests for cancer predis- position now involve such sequencing technology of specific genes from blood DNA (e.g. the current breast cancer predisposition genetic tests sequence the BRCA1 and BRCA2 genes). One of the main current research activities is the sequencing of whole genomes (e.g. the ‘1000 genomes project’ which has completed the sequence of 100 000 human genomes to find genetic variation associated with phenotypes in the United Kingdom). Although at present a research tool, it is envisaged that this could become a routine investigation. The difficulty will arise in the interpretation of all the genetic vari- ants, their potential interaction effects, and the accurate prediction of disease risk. Detection of other mechanisms of gene alteration Not all genetic changes involve alterations of DNA bases, as already mentioned here. Some mutations can be due to larger deletions or gene rearrangements and these can easily be overlooked by conven- tional sequencing methods. Most are detected by multiple probe li- gation analysis (MLPA) or dosage analyses of sequence reads. For example, the standard BRCA1/​2 genetic test for breast cancer pre- disposition also includes MLPA as well as sequencing, as up to 3% of BRCA1/​2 mutations in some populations can be caused by alter- ations which are only detected by MLPA. Epigenetic changes are often due to methylation of genes. Rarely, alterations in the germ line can cause downstream methylation of cancer predisposition genes; such changes have recently been described in Wilms tumour and some cases of HNPCC. Cancer risks associated with cancer predisposition genes Cancer risks depend on the presence of mutations in a cancer pre- disposition gene and its penetrance (Table 5.3.2). Penetrance may be affected by external factors, such as lifestyle, and other environ- mental effects; it may also depend on the ethnic origin of an indi- vidual due to population-​specific mutation risks. For example, for BRCA1 or BRCA2, which predispose to breast and ovarian cancer, using data from the Breast Cancer Linkage Consortium based on breast and ovarian cancer families identified from a worldwide population of high-​risk families with breast cancer, the risk of breast cancer is estimated to be 85% by 80 years, but the risk associated with the Icelandic founder mutation in the BRCA2 gene is estimated Normal epithelium Genetic alteration APC loss Genetic instability ?TP53 loss Activation of KRAS by point mutation SMAD4 loss ?TP53 loss Further genetic mutations Hyperproliferative epithelium Early adenoma Intermediate adenoma Late adenoma Carcinoma Metastasis Fig. 5.3.2  The colorectal adenoma–​carcinoma sequence.

5.3  The genetics of inherited cancers 461 to be as low as 37% by this age. The ethnic population differences may be due to a founder mutation dependent risk, the effect of other modifying genes in a population, or the added effect of environ- mental influences, which may be shared within specific populations. Therefore it is important to ascertain the ancestry of the patient be- fore genetic counselling is initiated. The estimate of penetrance can be confounded by the presence of phenocopies when research into the identification of a cancer predisposition gene is undertaken. Table 5.3.2  Genes with high penetrance for the common cancers Neoplasia or syndrome Malignancies Risk of cancera Location Gene Breast/​ovary Breast/​ovary cancer syndrome Breast Ovary/​Fallopian tube Other cancers, e.g. pancreas, prostate Breast Ovary/​Fallopian tube Prostate, pancreas Other cancers, e.g. melanoma/​bile duct 80–​85% 40–​60% <10% 80–​85% 27% <10% 10–​14% by 74 years 17q21 13q12 BRCA1 BRCA2 Colon Familial adenomatous polyposis Bowel cancer Duodenum/​periampullary Hepatoblastoma/​ thyroid/​brain Desmoid c.100% <5% 5q21 APC HNPCC Colon Endometrium Ovary Colon Endometrium Ovary Colon Endometrium Ovary Colon Endometrium Ovary Other cancers, e.g. renal tract, pancreas,
bile duct, gastric, brain By age 70 45–​75% 30–​40% 10% 35–​65% 20–​25% 10–​15% 20–​70% 25–​70% 1% 15–​20% 15% 1% <5% 2p21 3p21 2p16 7p22 MSH2 MLH1 MSH6 PMS2 MSH2; MLH1; PMS2 Muir–​Torre syndrome As HNPCC (see above) with skin lesions 2p21 hMSH2 Peutz–​Jeghers Ovarian cancer (sex cord) Gastrointestinal <10% 19p13 STK11 Juvenile polyposis Colon 70% 18q21 10q23 10q22 SMAD4 PTEN BMPR1A MutYH Colon 70% 1p34 MUTYH (homozygotes) Turcot’s Colon 70% can be <20 APC/​HNPCC (biallelic) Gastric cancer Diffuse gastric cancer Stomach 90% 16q22 CHD1 Melanoma Melanoma Melanoma 65% 9p21 CDKN2A Renal Renal cancer (papillary) Papillary renal (type 1) 70% 7q31 MET Von Hippel–​Lindau Cerebellar Haemangioblastoma Retinal angioma Renal cell carcinoma 60–80% 85% 40% 3p25 VHL WAGR Wilms’ tumour is part of syndrome 11p13 WT1 Birt–​Hogg–​Dube Renal carcinoma 15% 17p11 FOLLICULIN Hereditary leiomyomatosis Renal carcinoma 10–16% 1q42 FH a The risk is either the ‘lifetime risk’ quoted to age 80 years unless otherwise stated. Note: The common cancers—​colon, breast, ovarian, prostate, lung, pancreas, testicular, melanoma, and lymphoma—​have had numerous lower-​risk variants identified by genome-​wide association, and rare more moderately penetrant genes have been found by panel candidate genetic mutation analysis in some of these cancers (see text).

462 SECTION 5  Principles of clinical oncology Phenocopies are people who have developed the disease of interest but are found not to carry mutations in the disease predisposition gene, so that the disease has occurred by chance alone or may have been due to environmental influences. Phenocopies are a particular problem in the analysis of syndromes associated with common can- cers such as breast, prostate, ovarian, or colon cancer. When quoting cancer risks it is important to quote the risk by a specific age as the profile of risk may alter over time; for example, the risk of breast cancer from a deleterious mutation in BRCA1 is a sigmoid curve, starting to rise from the age of 30; the steepest part of the curve is in the 40s and there is still some risk until the age of 80 years. An unaffected woman with a BRCA1 mutation will there- fore have a lower residual cancer risk if she is aged 70 than she will at the age of 40 years. Genetic predisposition to the common cancers Breast cancer Breast cancer is the most common non​cutaneous cancer in women in the Western world. Several genes predispose to high risks of breast cancer, most notably BRCA1 and BRCA2 (breast cancer 1 and 2 genes) which were isolated in 1994 and 1995. These genes, when mutated, also predispose to ovarian cancer, and also have a small (<10%) risk of causing other cancers (e.g. pancreas, bile duct, melanoma, male breast cancer, prostate cancer). They are highly penetrant for female breast cancer (80–​85% risk by 80 years) and ovarian cancer (40–​60% risk for BRCA1 and 27% risk for BRCA2). The profiles of the penetrance curves are slightly different (in gen- eral, those for BRCA2 start to rise at an older age for both breast and ovarian cancer) and this is taken into consideration when consid- ering the timing of preventative surgery (see next). A rarer breast cancer predisposition gene is TP53 which usually predisposes to the Li–​Fraumeni syndrome, the association of early onset sarcoma with cancer at less than 45 years in at least two close relatives. Often this syndrome is associated with childhood cancer. The penetrance of breast cancer is 90% by age 60 and this gene can cause breast cancer at particularly young ages (in the 20s). Other rarer cancer predispos- ition genes in the DNA repair pathway have been shown to be asso- ciated with increased breast cancer risk (PALB2, ATM, CHEK2), but these risks (about two-four fold) are not as high as those from muta- tions in BRCA1/​2 or TP53. Cowden’s multiple hamartomatous syn- drome has an increased risk of female and male breast cancer and also thyroid and uterine cancer. It is associated with gynaecological and brain abnormalities and bowel polyps, but there is debate as to whether there is also an increased risk of bowel cancer. The path- ology of the breast cancer is characteristic in some of these condi- tions; in BRCA1 mutation carriers, it is often hormone receptor and HER2 negative (so-​called ‘triple negative’) and has cellular features of the basal type. There is an increased risk of lobular breast cancer in association with diffuse gastric cancer, which is due to mutations in the CDH1 E-​Cadherin gene. Colon cancer It is thought that at least a proportion of colon cancers arises from polyps in the bowel. The colon cancer syndrome with the highest bowel cancer risk is associated with the presence of thousands of such polyps in the large bowel (familial adenomatous polyposis or FAP; Fig. 5.3.3). This is due to mutations in the APC gene. The APC protein is a negative regulator of β-​catenin, a critical component of a signal transduction pathway that regulates cell–​cell adhesion, cel- lular polarity, and tissue architecture. The penetrance is high, with a virtually 100% risk of colon cancer by the age of 40. There is also a risk of other cancers, such as hepatoblastoma, periampullary, thy- roid and brain cancer, sarcoma, and desmoid tumours. Polyps can also occur in the upper gastrointestinal tract and pigment can be present in the retina. The polyps are so extensive that the mainstay of prevention is a colectomy once the polyps appear on sigmoido- scopic monitoring, of individuals who have APC mutations, from the age of 11 years. Lynch’s syndrome or HNPCC has fewer polyps (usually <100) and classical HNPCC conforms to a definition also known as the Amsterdam criteria, so-​called because Vasen in Amsterdam found that if these criteria were used then over half of families had mu- tations in the mismatch repair genes hMLH1 (chromosome 3p21), hMSH2 (2p21), hMSH6 (2p16), and PMS2 (7p22). There is con- troversy as to whether PMS1 (2q31) increases colon cancer risk. These criteria are colon cancer in at least three individuals in two generations, with at least two being first-​degree relatives of each other and at least one of the colon cancers occurring at less than 50 years of age. Classically the cancers tend to occur more often in the right side of the colon whereas sporadic colon cancer is more often left-​sided. This condition is also associated with other cancers, particularly endometrial cancer, ovarian cancer, and smaller risks of biliary and pancreatic cancer, cancer of the renal tract (particularly transitional cell carcinoma), and upper gastrointestinal tract cancer (Table 5.3.2). Extracolonic cancers are more common in hMSH2 mutation carriers, and hMSH6 is particularly associated with uterine cancer. There are less stringent clusters which are also due to mutations in these genes and so more loose criteria have been developed (the so-​called Bethesda criteria, but as the criteria get less stringent, the mutation frequency decreases). The mismatch repair genes are involved in DNA base excision repair and, if mutated, give rise to genetic instability, particularly of repeats in the DNA (micro- satellite instability). This can be analysed in tumour specimens to Adenomatous polyp Fig. 5.3.3  Large bowel with numerous adenomatous polyps (arrowed) due to familial adenomatous polyposis.

5.3  The genetics of inherited cancers 463 determine if there is likely to be an underlying mismatch repair gene defect. Protein products of these genes can also be detected by immunohistochemical staining and so lack of staining in colonic tumours is used as a triage to determine which gene may be mu- tated. More rarely, brain tumours can occur in association with very early onset (often <20 years) colorectal cancer and these cases have been found to harbour homozygous (both gene copies are altered) mutations in mismatch repair genes (Turcot’s syndrome). The pres- ence of sebaceous adenomas or keratoacanthoma should raise the possibility of Muir–​Torre syndrome, which is HNPCC with these additional skin features and is also due to mutations in mismatch repair genes. Genes predisposing to other rarer colorectal cancer syndromes have also been described. STK11 predisposes to Peutz–​Jeghers syn- drome (autosomal dominant) where hamartomatous polyps are associated with pigmentation of the lips and buccal mucosa. This syndrome also has an increased risk of breast, ovarian, uterine, pan- creatic, and testicular cancer. Juvenile polyposis is an autosomal dominant condition and causes diffuse hamartomatous polyps of the colon, small bowel, and stomach which develop at an early age (<10 years) or older (c.55 years). About half of patients have mutations in the SMAD4 gene (which, interestingly, is mutated in the multistep pathway of colorectal cancer; see Fig. 5.3.2, within the cancer cells) or PTEN or BMPR1A. Rarely families with mutations in the TGFBRII gene have been described; this is in the SMAD4 pathway. If a family has a recessive pattern of inheritance (i.e. disease in siblings but not the parents) of multiple colonic polyps, then MutYH should be considered. This gene usually has mutations at specific sites and so the genetic test can examine these regions specifically, at least in the first instance. Upper gastrointestinal cancer and pancreatic cancer There are rare reports of familial gastric cancer where the cancer can occur at a very young age (<20 years) and is of diffuse type. These are associated with mutations in the CDH1 gene. The treatment is prophylactic gastrectomy. Duodenal cancers can occur as part of FAP. Familial pancreatic cancer is described but the genetic basis is still being researched; rare instances are due to mutations in the BRCA2 and other genes. Ovarian cancer The main genes to consider are BRCA1 (on chromosome 17q21) and BRCA2 (on chromosome 13q12). These account for the ma- jority of families with two or more cases of ovarian cancer and are predominantly associated with the serous type of ovarian adenocar- cinoma. Peutz–​Jeghers syndrome can cause unusual ovarian lesions such as mucinous tumours or sex cord tumour with annular tubules. Ovarian cancer can also be part of HNPCC but the penetrance is lower (about 12%) than with BRCA1/​2. Mutations in BRIP1, RAD51C, and RAD51D have recently been reported and confer ovarian cancer risks high enough to warrant offering preventative removal of the ovaries and fallopian tubes. Melanoma Mutations in the multicancer tumour suppressor gene CDKN2A (P16) have been associated with melanoma. Mutations in this gene should only be considered if there are multiple melanomas in an individual, and/​or at least three cases of melanoma in a family, often with early onset disease (≤40 years). Prostate cancer Prostate cancer is the most common non​cutaneous cancer in men in the Western world. Over 80% of cases occur at more than 65 years of age. Rare cases (<5%), particularly at young age (<60 years), are due to mutations in the DNA repair genes, particularly BRCA2 which predisposes to more aggressive disease with a poorer prognosis. Sequencing of a linkage region on 17q showed a recurring mutation in HOXB13, G84E which is more common in younger onset cases or those with a family history. Most of the prostate cancer loci (>100) have been found by genome-​wide association studies. Variation at these loci resulting in slightly increased risk is common. A man in the highest 1% of the risk profile from SNP combinations will have nearly six times the average risk of the general population. As fur- ther SNPs are found, genetic profiling will be possible in the popu- lation. It is uncertain at present if these variants will help to identify more aggressive disease and further research is needed in this area. Common lower-​risk variants Genome-​wide association studies are revealing genetic changes (SNPs) associated with disease. These are more common than most of the genetic changes mentioned earlier, and often the SNPs are not in genes and so are presumably exerting their effect by altering gene function in another part of the genome. These types of alter- ations are currently the predominant types of variants predisposing to prostate and lung cancer. In the latter disease, one of the SNPs is on chromosome 15q25 which contains the nicotinic acetylcholine receptor subunit genes CHRNA3 and CHRNA5, suggesting that sus- ceptibility may be mediated through smoking behaviour. Rare inherited cancer syndromes Hereditary retinoblastoma: a classical example of a cancer predisposition syndrome The first cancer predisposition genes were identified by studying rare but striking clusters of conditions that occur as part of recog- nized clinical syndromes (Box 5.3.1). Hereditary retinoblastoma is a classical example of this phenomenon. Retinoblastoma is a cancer of the retinal cells and mainly occurs at in children less than 5 years of age (most occur at <2 years). One in 13 500–​25 000 children are affected, with an equal sex distribution. About 10% of patients have a family history of the disease, with an autosomal dominant pattern of inheritance. Inherited forms of the disease are due to mutations Box 5.3.1  Features of inherited cancer predisposition syndromes • Earlier onset than sporadic cancer • Bilateral or multifocal cancer • Rare cancers either alone or more often in clusters in families than is expected by chance • Phenotypic abnormalities indicating a disorder of tissue formation/​ regulation (e.g. overgrowth syndromes or skin manifestations) • New germline mutations may account for new cases where there is no family history

464 SECTION 5  Principles of clinical oncology in the RB1 gene on chromosome 13q14; the new mutation rate is also high. The other cases are apparently sporadic but many of these have a germline mutation which has occurred de novo (see earlier). Knudson calculated that only one additional mutation is necessary for tumour development, leading to the two-​hit hypothesis. All bi- lateral cases where retinoblastoma occurs in both eyes should there- fore be considered to be gene mutation carriers. Eighty-​five per cent (85%) of retinoblastomas presenting at less than 6 months of age af- fect both eyes and are likely to be the inherited form. The proportion of bilateral cases declines to 6% by 24 months, when most cases are of the sporadic type, with a much lower risk of genetic transmis- sion (<5% of cases will be mutation carriers). The penetrance of RB1 mutations is 90%. Individuals with hereditary retinoblastoma are at an increased risk of developing a variety of other cancers (espe- cially osteosarcoma and bladder cancer). Retinoblastoma illustrates the cardinal clinical features of inherited cancer predisposition syn- dromes: early onset, bilateral cancer, familial clustering, and pre- disposition to multiple tumours both within the eye and also at multiple sites. Many genetic alterations in RB1 are gene deletions and insertions. Of interest, point mutations may be associated with a lower penetrance, suggesting a genotype–​phenotype effect (the gen- etic and physical effects, respectively). Neurofibromatosis type 1 This is one of the most frequent single gene disorders with a fre- quency of 1 in 2500–​3300 individuals. The diagnosis is clinical and the clinical features of neurofibromatosis type 1 (NF1, von Recklinghausen’s disease) require at least two of the following: • At least six café-​au-​lait spots larger than 5 mm (if prepubertal) or 15 mm (if postpubertal) • At least two neurofibromas or one plexiform neurofibroma • Axillary or inguinal freckling • Optic nerve glioma • At least two iris hamartomas (Lisch nodules) • An osseous lesion such as sphenoid dysplasia or thickened long bone cortex ± pseudarthrosis • A first-​degree relative with NF1 Learning difficulties can also occur in about 30%. About 3–​5% of cases have a malignancy which is usually an optic nerve glioma, sar- coma, or phaeochromocytoma. Recent analyses of other tumour risks have also reported a moderately increased breast cancer risk. The NF1 gene (on chromosome 17q11) is very large (60 exons) and it encodes a guanosine triphosphatase activating protein known as the NF1-​GAP-​related protein, or neurofibromin. GAP proteins negatively regulate RAS to control cell proliferation. Mutations are numerous and varied in type, probably because the gene is so large. There is no genotype–​phenotype correlation of the mutation with the NF1 features. Some patients have numerous neurofibromas while others have very few. The phenotype is thought to be controlled by other genes. The new germline mutation rate is high (30–​50%), again probably because of the extremely large size of target locus. Neurofibromatosis type 2 Neurofibromatosis type 2 (NF2) is less common than NF1 (1 in 33 000 births) and has an autosomal dominant pattern of inherit- ance. The neurological effects of neurofibromas predominate in NF2. There is a predisposition to development of tumours of the central nervous system, particularly schwannoma of the eighth cranial nerve (‘acoustic neuroma’, which occurs bilaterally in 85% of cases), men- ingioma, spinal cord schwannoma, and malignant gliomas. Deafness and tinnitus due to acoustic neuromas, as well as muscle weak- ness and wasting due to spinal cord compression are not unusual. Criteria for diagnosis are bilateral acoustic neuromas or a family his- tory of NF2, plus unilateral acoustic neuroma at less than 30 years or any two of the following:  meningioma, glioma, schwannoma, and juvenile posterior subcapsular lenticular opacities. There is a milder and a more severe type with presentations at under and over 20 years of age, respectively. The NF2 gene, located at chromo- some 22q12, encodes a protein named schwannomin (or Merlin, for Moesin Ezrin Radixin-​like protein) which communicates between the extracellular matrix and cytoskeleton. About half the cases are de novo. There seems to be a genotype–​phenotype correlation with milder phenotypes associated with point mutations and more severe phenotypes associated with nonsense and frameshift mutations. It is noteworthy that about 20% of apparently de novo cases will not have germline mutations, and the NF2 mutation is only at the tumour site as a result of mosaicism (only some cells in the body have the mu- tation). Such patients have a lower risk of transmission to offspring as the chance of their gametes being involved in the mosaicism is less. Patients should be managed in specialist centres as complex screening and interdisciplinary management is needed. Multiple endocrine neoplasia type 1 This syndrome (MEN1) is associated with parathyroid adenomas, pancreatic islet tumours, and anterior pituitary tumours (for ease of recall, the 3Ps) and is autosomal dominant. Carcinoid can also occur. The MEN1 gene on 11q13 codes for menin, which acts as a growth suppressor protein. Mutations can occur throughout the gene and there is no genotype–​phenotype correlation. Only about 10% of cases are de novo. Multicentric pancreatic tumours and hyperparathyroidism at young age (<30 years), with or without pi- tuitary tumour, should raise the possibility of MEN1 and genetic testing can be undertaken using gene sequencing, although in 20% of classical cases no mutation is found. The penetrance is high (95% by 70 years). Multiple endocrine neoplasia types 2A and 2B and familial medullary thyroid cancer Three disorders are due to activating mutations in the RET tyro- sine kinase-​linked cell surface receptor encoded by the RET gene at 10q11.2, which is an oncogene (only one mutated copy is necessary for disease development). Medullary thyroid cancer is common to all three conditions and is histologically associated with C-​cell hyperplasia which should be looked for in the pathology report. When no other features are present, the condition is termed familial medullary thyroid cancer. In multiple endocrine neoplasia (MEN) types 2A and 2B, additional unusual tumours arise including phaeo- chromocytoma (in 50%) and parathyroid adenomas in 20–​30% (particularly in MEN2A). MEN2B is associated with marfanoid habitus, intestinal ganglioneuromas (causing Hirschsprung’s dis- ease), and mucosal neuromas, often in the lips, causing them to be prominent. Particularly in MEN2B, medullary thyroid cancer can occur at less than 10 years of age. In 95% of cases of MEN2A, muta- tions affect cysteine residues in the extracellular binding domain of

5.3  The genetics of inherited cancers 465 RET, resulting in inappropriate disulphide bond formation, dimer- ization, and activation of the RET tyrosine kinase. Familial medul- lary thyroid cancer results from mutations which similarly involve cysteine residues in most cases, but at different sites. The mutation found in MEN2B is distinct and involves a methionine to threonine substitution in the adenosine triphosphate (ATP) binding site of the receptor tyrosine kinase, leading to excessive receptor activity. There is a strong genotype–​phenotype correlation with certain mu- tations particularly associated with phaeochromocytoma. Rarely, other mutations may occur in other parts of RET and some of these are associated with a lower penetrance. As medullary thyroid cancer can occur in early childhood, current optimal management is gen- etic testing and prophylactic total thyroidectomy in gene mutation carriers. Many geneticists will undertake predictive tests from birth in families where there is a known RET mutation. Screening for phaeochromocytoma and monitoring of parathyroid hormone and calcium levels should be undertaken; the disease is penetrant by age 70. All cases of medullary thyroid cancer should be referred to a cancer geneticist to offer genetic testing for RET mutations; if no mutation is found in such a case then the chance that it is inherited is less than 5%. Cowden’s syndrome This autosomal dominant condition is a multiple hamartomatous syndrome. It has characteristic skin and tongue hamartomatous lesions, gynaecological abnormalities, and intestinal hamartomas. Craniomegaly and mental subnormality occur in about 50% of af- fected individuals. The pathognomonic mucocutaneous lesions include trichilemmomas, acral keratoses, papillomatous papules, hyperkeratosis, and oral fibromas. Breast cancer occurs in 30% of female gene carriers by age 50 and multiple painful fibroadenomas of the breast are common, often necessitating prophylactic mastec- tomy. Thyroid cancer, male breast cancer, and endometrial cancer can also occur. Glial masses may present as cerebellar ataxia and seizures (Lhermitte–​Duclos disease). The gene concerned, ‘phos- phatase tensin homologue deleted in chromosome 10’ or PTEN, is located on 10q23. The PTEN phosphatase, by operating in oppos- ition to the phosphoinositol-​3-​kinase pathway, inhibits cell survival and growth. Tuberous sclerosis This is a disease of variable severity characterized by the develop- ment of multiple hamartomas involving many organs; it is auto- somal dominant. Characteristic lesions—​facial angiofibromas (adenoma sebaceum), shagreen patches, and ungual fibromas—​ along with epilepsy and learning difficulties often suggest the diagnosis. There is an association with cardiac rhabdomyomas. Often there is no family history since as many as 60% of cases are due to a de novo mutation. There is a 5–​15% incidence of child- hood brain tumours in affected individuals, mostly subependymal giant cell astrocytomas. In addition, a weak association with renal cell cancer has been reported. A wide variety of benign tumours, including hamartomas, angiofibromas, and renal lesions occur. The renal tumours are characteristically angiomyolipomas, which can cause renal haemorrhage or compress the normal kidney leading to renal failure. Linkage studies have identified two genes, TSC1 at 9q34 and TSC2 at 16p13. TSC1 encodes a protein called hamartin. Most mutations described within this gene result in a truncated protein. TSC2 encodes tuberin, a protein showing some homology to GTPase activating proteins. Li–​Fraumeni syndrome This is a rare but important autosomal dominant syndrome. It is named after the two epidemiologists who noticed an increased cancer risk in first-​degree relatives of patients with rhabdomyo- sarcoma. The key feature is sarcoma, particularly at young age (<45 years). The definition of classical Li–​Fraumeni syndrome is sarcoma in the proband before age 45 years and cancer before age 45 years in two close relatives, one of whom is a first-​degree rela- tive. The tumour spectrum is wide and multiple tumours can occur in one individual; there is also a high (24% by age 20) risk of tu- mours in childhood. The lifetime penetrance by age 60 years is 90% in women and 74% in men; penetrance is increased by exposure to carcinogens, particularly smoking. The spectrum of early onset tu- mours particularly includes bone and soft tissue sarcoma (excluding Ewing’s sarcoma), breast cancer, brain tumour, leukaemia, and adrenocortical carcinoma. Approximately 75% of Li–​Fraumeni syn- drome families have germline mutations within the TP53 gene lo- cated at 17p13. TP53 has been referred to as the ‘guardian of the genome’ because of a critical role in arresting the cell cycle in the presence of DNA damage. It can act as a tumour suppressor and also as a dominant oncogene. Radiation may increase the risk of second tumours (57% rate of second tumours over 30 years) and so should be avoided if possible. Basal cell naevus syndrome (Gorlin’s syndrome) This condition should be considered in any patient presenting with a basal cell carcinoma before the age of 30 years, or with a per- sonal or family history of multiple basal cell naevi/​carcinomas. It is associated with abnormalities of skin, bone, and tooth forma- tion, including polyostotic bone cysts, odontogenic keratocysts (jaw cysts), bifid ribs, ectopic calcification (lamellar calcification as seen on a posteroanterior skull radiograph is pathognomonic), and palmar or plantar pits. An increased incidence of other cancers, including medulloblastoma, ovarian carcinoma, and sarcomas, may also occur. The incidence has been estimated at 1 in 31 000. The gene responsible for the majority of cases, PTCH on 9q22.3, is a homo- logue of the Drosophila patched gene that encodes a transmembrane receptor for an extracellular ligand (Hedgehog). This pathway con- trols the fate of cells, body patterning, and growth by forming gradi- ents in embryonic tissues. The most important management feature to note is that such patients should avoid therapeutic radiation as this induces further tumours. Renal cancer and syndromes Mutations in several genes have been demonstrated to predis- pose to renal cell carcinoma. These include the VHL gene (as- sociated with von Hippel–​Lindau syndrome), FOLLICULIN (associated with Birt–​Hogg–​Dubbe syndrome), FH (associ- ated with leiomyomas), the succinate dehydrogenase genes, and rare reports of disruption of the TRC8 gene (by a translocation). Translocations involving chromosome 3 have also been found con- stitutionally in renal cancer cases. Hereditary papillary renal cell carcinoma has been related to mutations in several genes, particu- larly the oncogene MET.

466 SECTION 5  Principles of clinical oncology Von Hippel–​Lindau disease Von Hippel–​Lindau syndrome (VHL) is a dominantly inherited fa- milial cancer syndrome predisposing to a variety of malignant and benign neoplasms, most frequently retinal angioid streaks, cere- bellar and spinal hemangioblastomas, renal cell carcinoma, phaeo- chromocytoma, and pancreatic tumours. There are numerous liver, pancreatic, and renal cysts which can be seen on imaging. The in- cidence in the United Kingdom is 1 in 36 000, with near complete penetrance by 70 years. There are two types, VHL type 1 and type 2 (without and with phaeochromocytoma respectively; type 2 is div- ided into three further types A–​C depending on the combinations of associated features). The VHL gene at 3p25–​p26 contains three exons that encode a 213-​amino-​acid protein. There is a genotype–​ phenotype correlation with missense mutations more often associ- ated with phaeochromocytoma. The VHL protein plays a role in the transduction of growth signals generated by changes in oxygen ten- sion, promoting the translation of target genes that include vascular endothelial growth factor. VHL is a classical tumour suppressor gene, with a second, somatic mutation required for the development of cancer. Mutations in VHL are common in sporadic renal clear cell carcinoma within the tumour cells only. Management of VHL pa- tients should be in a multidisciplinary clinic as several systems need monitoring (eye, neurological, and urological). Specialist urological management is necessary as renal tumours are often multifocal, so nephron-​sparing surgery is used. Familial papillary renal cell carcinoma Any case of the less common renal cancer type, papillary renal cell carcinoma, should be considered for genetic analysis of the MET oncogene (on chromosome 7q31) which predisposes to familial papillary renal cancer. It codes for a transmembrane tyrosine kinase receptor for hepatocyte growth factor or scatter factor, a peptide with essential roles in embryogenesis, cell motility, and tumour in- vasion. Germline MET missense mutations in cysteine residues, homologous to those involved in aberrant dimerization and activa- tion of the RET receptor, are associated with familial papillary renal cell carcinoma. Monoallelic activating mutations in MET are also found in 15% of cases of the sporadic form of the disease. The spec- trum of mutations found in sporadic papillary renal cell carcinoma is wider and includes activating mutations in the MET tyrosine kinase domain. Rarely other loci may be associated with papillary renal cancer, but MET is the gene to be most considered. Birt–​Hogg–​Dubbe syndrome This syndrome is a rare inherited genodermatosis character- ized by hair follicle hamartomas, kidney tumours (usually a renal oncocytoma or chromophobe renal cancer), and spontaneous pneumothorax; fibrofolliculomas on the face are its hallmark and trichodiscomas (tumour of the hair disc) and acrochordons (‘warts with a thin neck’; skin tags) are associated features. Onset is in adulthood. It is due to mutations in the FOLLICULIN gene on chromosome 17p11. Hereditary leiomyomatosis and renal cell cancer Multiple skin and uterine leiomyomas (fibroids) associated with renal cancer are associated with mutations in FH (on 1q42) causing fumarate hydratase deficiency. Wilms tumour (nephroblastoma) Wilms tumour is a poorly differentiated tumour of the kidney, usually in childhood. It occurs in 1 in 10 000 children and accounts for 8% of childhood cancers. It is associated with aniridia, hemihypertrophy, and developmental abnormalities of the genitourinary tract (the WAGR syndrome). Males and females are equally affected and usu- ally present early in childhood, most often with an abdominal mass. Two sites of loss of heterozygosity have been identified in Wilms tumours, WT1 at 11p13, and WT2 at 11p15.5. There are also rare familial cases in which linkage to neither 11p locus has been es- tablished (referred to as the WT3 group). In 10–​30% of patients, the disease is bilateral or multifocal, but less than 1% of all cases are truly familial. Most cases of bilateral nephroblastoma are due to new germline mutations in WT1. The protein encoded by the WT1 gene is a ‘zinc finger’ DNA-​binding transcription factor. WT1 interacts with another tumour suppressor, TP53, to bind and sup- press transcription from the epidermal growth factor receptor and insulin-​like growth factor 2 gene promoters. When WT1 function is compromised, transcription from these growth-​ and survival-​ promoting proteins is increased, initiating tumour development. WT1 is not, however, a strictly Knudson-​type tumour suppressor. Statistical analysis of age at diagnosis and proportion of bilateral and unilateral tumours does not follow the pattern described for retinoblastoma. Furthermore, the children of patients who sur- vive Wilms tumour are at lower risk of the disease than would be expected from a dominant-​acting tumour suppressor gene. There is evidence that ‘genomic imprinting’ may explain some of these anomalies. Imprinting is a process of gene inactivation through DNA methylation that preferentially favours expression from genes inherited from one or other parental lineage. There is a report in some Wilms tumour cases of genetic alteration of a methylation centre in the genome, which is an example of a germline change that has epigenetic effects. Chromosome fragility syndromes All these syndromes are autosomal recessive and they cause other abnormalities of phenotype, such as short stature, autoimmune and immunodeficiency disease, and other features. Although they are very rare, they are important as such patients are sensitive to DNA-​damaging agents, which is an important consideration when treating the associated cancers with such agents. Ataxia telangiectasia This is a rare recessive condition (1 in 30 000–​100 000). Ataxia tel- angiectasia patients who are homozygous for ATM mutations have telangiectases in the eye, progressive ataxia due to cerebellar de- generation, general neuromotor dysfunction, and immune defects. There is a 30–​40% lifetime risk of malignancy including epithelial tumours, chronic T-​cell leukaemia, and lymphoma. ATM heterozy- gotes do not exhibit any of these defects, but have a two-​ to threefold increase in the risk of cancer, particularly female breast cancer. The ATM gene (11q22) encodes a 350-​kDa protein which contains a do- main sharing homology with members of the phosphatidylinositol-​ 3-​kinase family and which is a signal transduction protein that regulates cell cycle checkpoints.

5.3  The genetics of inherited cancers 467 Bloom’s syndrome This a rare autosomal recessive disease of unknown incidence, more common in Ashkenazi Jews. Features include short stature, sensi- tivity to the sun, skeletal abnormalities (a bird-​like face), and suscep- tibility to infection. An increased frequency of malignant neoplasms occurs throughout life, with dramatically reduced life expectancy; it is very rare to live into the 30s. Lymphoma and leukaemia predom- inate before the age of 25 years; those that survive into their 20s and 30s are prone to a variety of common solid tumours, particularly squamous cell carcinoma of the head and neck, breast cancer, and gastrointestinal tumours. The age at diagnosis for these carcinomas is usually 20 years earlier than in the general population. The gene responsible, BLM (15q26), is a RecQ DNA helicase, and mutations result in genetic instability with spontaneous chromosomal abnor- malities and increased sensitivity to radio-​ and/​or chemotherapeutic agents. Treatments therefore have to be appropriately tailored. Males are infertile because of a defect in meiosis. Werner syndrome Werner syndrome is recessive and is characterized by a scleroderma-​ like, multisystem premature ageing phenotype, which is also due to a RecQ helicase defect. It is associated with atherosclerosis and diabetes mellitus and short stature. The incidence is 1 in 50 000–​ 100 000. Affected individuals have an excess of neoplasms (espe- cially osteosarcoma, meningioma, and thyroid cancer). Mutation in the RECQL2 gene (8p12) leads to genetic instability. Rothmund–​Thomson syndrome This is a hereditary dermatosis characterized by atrophy, poikiloderma (marbleized pigmentation), and telangiectasia, and is frequently accompanied by juvenile cataract, saddle nose, con- genital bone defects, disturbances of hair growth, short stature, and hypogonadism. Classical features are absent radii and a rudimen- tary/​absent thumb. Survival is fairly good and can be into the 40s. This is a recessive cancer predisposition syndrome, due a defect in a different helicase (gene RECQ4 at 8q24). There is a predisposition to malignancy, especially osteosarcomas and skin tumours. Fanconi anaemia Fanconi anaemia is a collection of recessive diseases characterized by a complex variety of developmental abnormalities, progressive marrow failure, and predisposition to acute myeloid leukaemia (15 000 times that of the general population). Fanconi anaemia commonly presents in early to middle childhood with anaemia and bruising. Progressive pancytopenia and chromosome breakage, worsened by exposure to alkylating agents, is characteristic. Fanconi anaemia homozygotes may develop a wide variety of common can- cers occurring at an early age. Squamous cell carcinomas, especially of the head and neck, oesophagus, cervix, vulva, and anus, occur with increased frequency, as do liver adenomas. Life expectancy is poor, around 12 years, with most deaths resulting from marrow failure and cancer. Approximately one-​fifth of childhood aplastic anaemia is associated with Fanconi anaemia and treatment is bone marrow transplantation. Treatment using radiation and chemo- therapy for the transplant has to be carefully given (reduced doses of conditioning are used as the cells are sensitive to DNA-​damaging agents). The heterozygote frequency is estimated to be 1 in 300 to 600; the frequency is greater in Ashkenazi Jews. Fanconi anaemia homozygous cells form abnormal chromosomes when exposed to cross-​linking agents such as mitomycin C, which is one of the diagnostic tests for the condition. Spontaneous chromosome ab- errations are seen in a variety of cell types. There are several com- plementation groups and two of them are due to genes which also predispose to breast cancer in heterozygotes (BRCA2 and BRIP1). The other groups do not appear to have an increased malignancy in heterozygotes. Xeroderma pigmentosum This is a group of rare autosomal recessive disorders, with an in- cidence of 1 in 1 000 000. The classical feature is photosensitivity, which starts in childhood, and freckling and telangiectasia leading to progressive degenerative skin changes and early development of cancers of the skin (squamous, basal cell, and melanoma) and eye. Fifty per cent (50%) of these children have a skin cancer by age 14 years. About 20% have concurrent neurological abnormalities and some have impaired immune systems. There is also an increased risk of solid and haematological tumours. Benign neoplasms in- clude conjunctival papillomas, actinic keratoses, lid epitheliomas, keratoacanthomas, angiomas, and fibromas. Defects of several enzymes involved in excision repair of ultraviolet-​induced pyr- imidine dimers are responsible for this syndrome. There are sev- eral complementation groups which differ in their action in this process (damage recognition, nuclease function, DNA polymerase function). Identification and management of known
or suspected cancer predisposition gene
mutation carriers Cancer genetic counselling This involves assessment of cancer risk, discussion of screening and management options, and the offer of genetic testing if appropriate. Risk assessment This can be complex; it involves a risk estimate and this information has to be communicated to the patient in the manner most appro- priate to the individual concerned for optimum understanding and retention, but so they are not made inappropriately anxious about the associated risks. The first risk estimation is the chance that a familial cluster is due to genetic predisposition (the ‘prior probability’ of a genetic predis- position gene being present in a family). An extensive family history is important to determine this, often out to third-​degree relatives. Confirmation of diagnoses is important for some sites (e.g. abdom- inal tumours) as these can be misreported in families in about 17% of cases. The risk estimation is based upon published data or clin- ical experiences when published data are lacking, which unfortu- nately is often the case with rare genetic conditions. For example, for breast cancer clusters estimates can be made using data such as those shown in Fig. 5.3.1. There are also now computerized models for some common cancers which can aid prediction of the presence of a genetic mutation, such as the BOADICEA model for the chance of a BRCA1/​2 mutation.

468 SECTION 5  Principles of clinical oncology The second risk estimation is the chance that the individual has inherited a particular gene based upon their cancer status (affected or unaffected), their position in the family tree, and their age. This is termed the ‘posterior probability’. The final calculation is the chance that cancer will develop, which is the posterior probability multiplied by the penetrance. The ex- pression of this risk can be delivered in several formats: the optimal format is unknown. Currently, risk estimates tend to be given as a percentage risk or a ‘1 in X’ value and followed up with a written summary, incorporating this risk estimate, to the individual at- tending the genetics consultation. There are data which suggest that individuals prefer not to have, or do not remember, numerical infor- mation, but are able to report the qualitative category of their risk (low, medium, high) with reasonable accuracy. Identification of an at-​risk family A family at genetic risk of cancer must first be identified. This is usually via the general practitioner (family doctor) or a hospital oncology clinic; it is now becoming more common for family his- tory to be requested by cancer geneticists working as part of the multidisciplinary team coordinating the patient’s care. Because of the limited time available during most consultations, it is not ap- propriate to obtain a detailed family history from the patient. As a quick guideline, taking a history of all first-​degree relatives only (parents, siblings, and children) and then asking if there are any other cancers in the family will detect 95% of familial syndromes. From this quick family history it should, however, be possible to make an assessment of whether the family history warrants fur- ther investigation. Referral guidelines have been developed; for example, in the United Kingdom, there are national guidelines for familial breast cancer (http://​www.nice.org.uk). These guide- lines aim to delineate the management and referral pathway ac- cording to the Kenilworth model (http://​www.macmillan.org.uk) whereby individuals whose risk does not exceed that of the general population are managed in the primary care setting, individuals at moderate risk are managed in secondary care, and individuals at high risk are managed in tertiary care in cancer genetics centres. In the cancer genetics clinic, after a full family history, initial clin- ical examination involves looking for any dysmorphic features and congenital anomalies. The skin should be carefully examined, as many cancer syndromes are associated with dermatological fea- tures, as noted earlier. Genetic testing is increasingly being offered as part of the cancer care pathway (the mainstreaming programmes; e.g. https://​ mcgprogramme.com). The results of testing at diagnosis, which does not increase psychological distress, can be used to guide chemo- therapy and surgical management decisions (e.g. in breast cancer for the addition of platinum to the chemotherapy regimen and bilateral mastectomy for management of the primary lesion). Throughout the consultation, it is important to be sensitive to any issues relating to bereavement due to the premature death of close relatives, particularly a parent or child. Unresolved bereave- ment may make it difficult for people to accept their own risks and make decisions about their own management. Some individuals are particularly worried when they are approaching the age at which their relatives were diagnosed. Others erroneously assume that they are more likely to have inherited the cancer predisposition gene because they resemble their affected relative, either physic- ally or in temperament. Patients are sometimes unable to cope with their concerns, and referral for formal psychological counselling may be needed. Of particular concern are those individuals who have prophylactic surgery because of excess anxiety but who, while being temporarily relieved, could return at a later date with further cancer-​phobic symptoms. A psychological assessment and coun- selling should be part of the referral process before prophylactic mastectomy. Clinical management The subsequent management of an individual and their family will depend upon the final risk estimates regarding the inheritance of a cancer predisposition gene and the potential cancer risks. In general, management strategies fall into five categories:  cancer screening, lifestyle changes, prevention strategies, cancer treatment consider- ations, and genetic testing. Cancer screening Not all of the screening schedules have been proven to reduce mor- tality from the relevant cancer, but these schedules represent a prag- matic approach to the management of individuals at increased risk. There is, however, some evidence that screening individuals with HNPCC by colonoscopy reduces mortality due to colorectal cancer, as any suspicious polyps observed on colonoscopy may be removed at an early stage. The guidelines promulgated by the United Kingdom National Institute for Health and Clinical Excellence (NICE) men- tioned earlier have made recommendations for mammographic and MRI screening in certain groups at risk of familial breast cancer. Prostate and ovarian cancer screening are contentious and are cur- rently subjects of research. Lifestyle changes Lifestyle changes may involve avoidance of known cancer-​causing factors such as sunlight in Gorlin’s syndrome and X-​ray exposure in the Li–​Fraumeni syndrome. Other lifestyle changes are less well es- tablished in the prevention of cancer and are being assessed in trials. Prevention strategies Primary prevention strategies include prophylactic surgery and chemoprevention. The evidence in support of the efficacy of these measures is variable, mainly due to the rarity of the genetic muta- tions, making clinical trials difficult to perform. Established meas- ures include total colectomy in the familial adenomatous polyposis syndrome, total thyroidectomy in the MEN2 syndrome, and bilat- eral salpingo-​oophorectomy in women with BRCA1/​2 mutations. Limited retrospective data suggest that the risk of breast cancer is re- duced by 90% following prophylactic mastectomy, although there is still a small residual risk (about 1.5%) due to the inability to remove all breast epithelial tissues at mastectomy. The role of chemoprevention is much less certain. In meta-​analyses tamoxifen reduces breast cancer risk by at least 33%, but it should be noted that the type of tumour prevented is hormone receptor-​ positive, and this has a better prognosis. The oral contraceptive pill reduces ovarian cancer risk by about one-​third in those on the pill for 2 years. Recent data report a reduction in colon cancer risk in HNPCC families in individuals who have taken aspirin after 5 years.

5.3  The genetics of inherited cancers 469 Cancer treatment Data are emerging which report a difference in prognosis when cer- tain genetic changes are present in the germ line (e.g. ovarian cancer due to mutations in BRCA1/​2 has a better prognosis and a higher response to platinum-​based chemotherapeutic agents). Certain syndromes are associated with altered response to treatment (e.g. colonic tumours which have microsatellite instability are less re- sponsive to 5-​fluorouracil, and recent data show dramatic responses to immunological agents). In VHL and PTEN mutation carriers, agents can be offered which target the hypoxia-​inducible factor (HIF) and mammalian target of rapamycin (mTOR) pathways, re- spectively. Agents are now being developed which specifically target tumours with certain genetic defects (e.g. the poly-​ADP ribose poly- merase (PARP) inhibitors), which enforce the cancer cell to use the homologous recombinant DNA repair pathway which is deficient in BRCA1/​2-​null cells, have resulted in promising early response rates in tumours in patients with germline BRCA1/​2 mutations. Olaparib is now licensed for use in ovarian cancer in women with germline mutations in BRCA1/​2 genes. Genetic testing Genetic testing is possible for most cancer predisposition genes and is performed on DNA from venous blood or saliva after gen- etic counselling. Genetic testing may either be diagnostic (the detec- tion of a mutation in an individual affected by cancer) or predictive (the detection of a mutation in a clinically unaffected individual). Mutations in cancer predisposition genes often occur throughout the gene and the vast majority of mutations so far have only been observed in limited numbers of families, except in specific ethnic groups with known founder mutations such as the Icelandic and Ashkenazi populations with BRCA1/​2 mutations. Hence, unless an individual is a member of such a group, the specific mutation for that family must first be identified. An affected family member is usually tested first because they are the family member most likely to have the cancer-​predisposing mutation. Once a mutation is identified, it is important to check that the ‘mutation’ is likely to be cancer-​ causing and not a normal variant of the gene (polymorphism). When a pathogenic mutation is identified, predictive testing may be offered to unaffected family members for the identified mutation. Misleading results may occur if an unaffected individual has a genetic test in order to identify a mutation without first identifying it in an affected relative. A negative result (i.e. no mutation is identified in the cancer predisposition gene tested) may not be a true negative for several reasons: • The family history is caused by a gene other than the one being tested or may not be genetic at all. • The alteration may be regulatory, which means that it controls how the gene is expressed but the gene itself (and therefore the test which looks at the gene code) is normal. • The genetic test sensitivity is not 100% for the genetic coding mu- tations and may therefore have missed mutations. With the advent of next-​generation sequencing the risk of this is very low as the sensitivity is about 98%. When the specific mutation has been identified in an affected indi- vidual, if it is not found in an unaffected relative, this is then a truly negative result. The personal and wider social implications of posi- tive and negative results are issues discussed during genetic counsel- ling sessions. A positive result could have psychological implications as well as widespread repercussions involving the rest of the family. A  negative test result may have psychological consequences due to the recognized ‘survivor guilt syndrome’, which has been docu- mented in the setting of Huntington’s disease. There is a moratorium on the use of some genetic information in the United Kingdom, as detailed on the Association of British Insurers’ website http://​www.abi.org.uk. The code was reiterated in Oct. 2018 and now has no renewal date. For genes predisposing to adult-​onset cancers, testing of young children is not advised as the age of cancer onset permits the in- dividual to make their own decision to have genetic testing once they have reached adulthood, following full genetic counselling. Children are offered genetic testing when it may alter manage- ment, for example, in the MEN2A syndrome when thyroidectomy is offered before age 5, in retinoblastoma to avoid unnecessary eye examinations, or in FAP where regular colonoscopies or colectomy may be avoided. Recently genetic testing has been licensed for preimplantation genetic diagnosis for certain cases of hereditary cancer in some countries. The future The number of low penetrance, common variants discovered to be associated with common cancers is increasing, and this will result in genetic profiles being identified which can be used to stratify popu- lations into risk categories. This is being studied to identify groups at higher risk for targeted screening programmes. Cancer is a common disease and only a proportion of cases will be due to the inherit- ance of rarer higher risk mutations in specific genes that predispose to cancer. However, because cancer occurs with high frequency in the population, this represents a large number of individuals. Recent advances have led to the development of ‘panel’ genetic tests where more than one cancer predisposition gene is tested e.g. breast cancer predisposition panels; colon cancer predisposition panels etc. The developments of more rapid genome sequencing will enable cancer genetics to become part of cancer care, as more targeted treatments are developed for such individuals and targeted screening is under- taken in their relatives. FURTHER READING Bodmer WF, et al. (1987). Localization of the gene for familial aden- omatous polyposis on chromosome 5. Nature, 328, 614–​16. Claus EB, Risch N, Thompson WD (1991). Genetic analysis of breast cancer in the cancer and steroid hormone study. Am J Hum Genet, 48, 232–​42. Easton DF, Peto J (1990). The contribution of inherited predisposi­ tion to cancer incidence. Cancer Surv, 9, 395–​416. European Society of Human Genetics. www.eshg.org. Document en- titled ‘A guide to genetic tests that are used to examine many genes at the same time’.

470 SECTION 5  Principles of clinical oncology Leach FS, et al. (1993). Mutations of a mutS homolog in hereditary nonpolyposis colorectal cancer. Cell, 75, 1215–​25. Lynch HT, Lynch JF (1994). 25 years of HNPCC. Anticancer Res, 14, 1617–​24. Miki Y, et al. (1994). Isolation of BRCA1, the 17q-​linked breast and ovarian cancer susceptibility gene. Science, 266, 66–​71. Papadopoulos N, et al. (1994). Mutation of a mutL homolog in heredi- tary colon cancer. Science, 263, 1625–​9. Thavaneswaran S, et al. (2019). Therapeutic implications of germline genetic findings in cancer. Nat Rev Clin Oncol, doi:10.1038/ s41571-019-0179-3. Wooster R, et al. (1995). Identification of the breast cancer suscepti- bility gene, BRCA2. Nature, 378, 789–​92. Websites Association of British Insurers. http://​www.abi.org.uk BOADICEA. http://​www.srl.cam.ac.uk/​genepi/​boadicea/​boadicea_​ home.html Genetic Alliance UK. https://​www.geneticalliance.org.uk/​ Genomes England. http://​www.genomicsengland.co.uk Genetics Toolkit-Practice Guidelines. ASCO.org Mainstreaming Cancer Genetics Programme. http://​www. mcgprogramme.com NHGRI GWAS. http://​www.ebi.ac.uk/​gwas/​ Online Mendelian Inheritance in Man (OMIM). https://​www.omim.org/​ Policy documents and guidelines for genetic testing. ESHG.org