6.17 Cowden’s syndrome

Cowden’s syndrome (OMIM 158350), named after Rachel Cowden, is an autosomal dominant inherited cancer syndrome characterized by multiple hamartomas involving organ systems derived from all three germ cell layers and a risk of breast and thyroid cancers (1, 2). Endocrinologists may make the diagnosis of Cowden’s syndrome when they are presented with these patients’ endocrine lesions, chief of which are multinodular goitre, thyroid adenomas, and epithelial thyroid cancer. The Cowden’s syndrome susceptibility gene, PTEN, is located on chromosome sub-band 10q23.3 (3, 4).

Cowden’s syndrome has not been well recognized; as of 1993, there were approximately 160 reported cases in the world literature. From an informal population-based study, the estimated gene frequency is one in a million. Multiple case reports have appeared after the identification of the gene in 1997, resulting in incidence estimates of 1:300 000. Because of the variable, protean, and often subtle external manifestations of Cowden’s syndrome, many cases remain undiagnosed. Despite the apparent rarity, the syndrome is worthy of note from both scientific and clinical viewpoints. Because Cowden’s syndrome is probably underdiagnosed, a true count of the fraction of isolated cases (defined as no obvious family history) and familial cases (defined as two or more related affected individuals) cannot be performed. From the literature and the experience of both major US Cowden’s syndrome centres, most cases are isolated. As a broad estimate, perhaps 10–50% of cases are familial.

Inheritance patterns in families with Cowden’s syndrome implicate an autosomal dominant pattern. Expression is variable and true penetrance is unknown. Based on the only population-based clinical epidemiology study to date, some believe that the penetrance is 90% after the age of 20 years (3). The precise penetrance will be clarified after further study of the susceptibility gene within families and affected individuals. Cowden’s syndrome was mapped to 10q22–23, without genetic heterogeneity (3). Further germline and somatic genetic analysis helped place the putative gene between the markers D10S215 and D10S541, a region of less than 1 cM (3, 5).

A candidate tumour suppressor gene PTEN/MMAC1/TEP1 was located precisely in the Cowden’s syndrome critical interval (3, 6). The gene comprises 1209 coding bp in nine exons and predicted to result in a 403-amino acid protein (6, 7). The protein has a tyrosine phosphatase domain and homology to tensin and auxilin (6, 7). Hence, this new gene was dubbed PTEN for phosphatase and tensin homologue deleted on chromosome ten. In vitro, PTEN has been shown to act as a dual specificity phosphatase, meaning it is both a lipid phosphatase and a protein phosphatase as well as a tyrosine phosphatase and serine–threonine phosphatase (8–11). Subsequent in vivo work in mouse models has suggested that PTEN plays a role in the PI3Kinase/Akt cell survival/apoptosis pathway (12–14). Because of its homology to the focal adhesion molecules tensin and auxilin, it was hypothesized that PTEN may also play a role in cell migration and focal adhesion. When PTEN was overexpressed in NIH 3T3 cells, it appeared that cell migration was inhibited while antisense PTEN enhanced migration (15). Evidence for PTEN interaction with focal adhesions kinases (FAK) was given when integrin-mediated cell spreading and focal adhesion formation were down-regulated by wild-type but not mutant PTEN; PTEN must interact with FAK to reduce its tyrosine phosphorylation (15). This leads to the hypothesis that PTEN functions as a phosphatase by negatively regulating cell interactions with the extracellular matrix. Although the genetic evidence that points to PTEN as a tumour suppressor—broad spectrum of mutations scattered throughout the gene, truncating mutations, and location of the gene in a region of loss of heterozygosity (see below)—is strong, functional demonstration was still required. When functional wild-type PTEN was transfected into a series of glioma cell lines which carry endogenous PTEN mutations or are PTEN null, growth suppression was observed (16–18). Multiple in vitro studies where wild-type and mutant PTEN were overexpressed in a broad variety of cancer cell lines, including those of the breast, thyroid, and prostate, demonstrate that PTEN-phosphatase- and PI3K-dependent G1 cell cycle arrest and/or apoptosis result in growth suppression (19–25). This is in vitro functional evidence that PTEN acts as a tumour suppressor.

The spectrum of tumour cell lines with PTEN mutations, its putative function as suggested by structural motifs, and its location within 10q23.3 all argued strongly that PTEN was an ideal candidate for the Cowden’s syndrome susceptibility gene. Therefore, to determine if germline PTEN mutations could be aetiological for Cowden’s syndrome, five families, with a high prior probability of having mutations, were chosen for initial analysis (3, 4). Two families had nonsense mutations, Arg233Xaa and Glu157Xaa, while two unrelated families shared an identical missense mutation, Gly129Glu, which is a nonconservative amino acid alteration occurring in one of the conserved glycines of the phosphatase signature motif (see above). No unaffected family member carried these mutations. In each family, the family-specific germline PTEN mutation segregated with disease but not in unaffected family members nor normal controls. Given these data, PTEN was most likely the susceptibility gene for Cowden’s syndrome. Further support that PTEN is indeed the susceptibility gene came when multiple other groups confirmed that germline mutations in PTEN are associated with Cowden’s syndrome.

In the single largest series of Cowden’s syndrome cases ascertained under the strict operational diagnostic criteria of the International Cowden Consortium (3, 26), 37 unrelated families were examined for frequency and spectrum of germline intragenic PTEN mutations (27). Of these, 30 (81%) had germline mutations. The 30 mutations were scattered along the length of the gene. Forty-three per cent of all mutations were found in exon 5, which encodes the phosphatase core motif, although exon 5 only represents 20% of the entire coding sequence.

An exploratory genotype–phenotype association analysis was performed in these 37 Cowden’s syndrome families. Two potential associations were noted. The first is the association between the presence of detectable germline mutation in PTEN and the presence of malignant breast disease. The second is the association between the presence of missense mutation and/or position of mutation within the phosphatase core motif or 59 of it and the development of multiorgan disease. Because most missense mutations occur within the core motif, it is unclear whether the nature and/or position of the mutation is significant. One could imagine that while missense mutations could disrupt phosphatase activity, the ability to bind substrate is maintained. In this scenario, substrates are sequestered but not dephosphorylated. Conceivably, this could lead to multiorgan involvement. Obviously, given the relatively small numbers, a second larger independent cohort needs to be accrued for genotype–phenotype analyses. If proven true, these preliminary associations might be helpful in tailoring medical management with regard to surveillance. It is also suspected that with a larger series, other associations might be found as well.

Another 10% of Cowden’s syndrome individuals not found to have intragenic mutations have germline mutations in the promoter of PTEN (28). These promoter mutations result in decreased transcription of PTEN as well as decreased translation of PTEN due to altered RNA secondary structure (28, 29). Based on small numbers, it would appear that women with germline promoter mutations are at further increased risk of developing breast cancer than those with intragenic mutations.

Bannayan–Riley–Ruvalcaba syndrome (BRRS) (OMIM 153480) is a rare autosomal dominant disorder characterized by macrocephaly, lipomatoses, hamartomas, hemangiomas, and speckled penis (30). Unlike Cowden’s syndrome, however, malignancies have not previously been rigorously shown to be components of BRRS and onset is usually at birth or shortly thereafter. Because of sharing of some, but not all, features of BRRS and Cowden’s syndrome, it was postulated that BRR and Cowden’s syndrome might be allelic. Initially, two of two BRRS families were shown to have germline PTEN mutations (31). Multiple PTEN mutations have now been described in both familial and isolated cases of BRRS, such that approximately 60% of BRRS individuals have been found to harbour intragenic PTEN mutations (26, 32). Amongst those without intragenic mutations, another 10% have been found to harbour large deletions encompassing or including PTEN (28). Since identical mutations (e.g. Arg233Xaa) have been found in Cowden’s syndrome as well as BRRS individuals, genetic and nongenetic modifiers must play a role in helping dictate the ultimate phenotype. Overall, the mutational spectrum of Cowden’s syndrome cases appears to favour the 5′ two-thirds of PTEN while that of BRRS the 3′ two-thirds of the gene (26).

In addition to Cowden’s syndrome and BRRS, germline PTEN mutations were found in variable subsets of several seemingly unrelated clinical syndromes. For example, up to 20% of individuals with Proteus syndrome have germline PTEN mutations (33). Approximately 10–20% of individuals with autism spectrum disorder and macrocephaly harbour germline PTEN mutations (34–37). Single cases of VATER and megalencephaly and hemimegencephaly have been reported to carry germline PTEN mutations as well (38, 39). The concept of PHTS to encompass any clinical disorder with germline PTEN mutation was proposed because it is clinically useful (32). Finding a germline PTEN mutation should trigger cancer risk management and genetic counselling similar to those used for Cowden’s syndrome.

Approximately 15% of classic Cowden’s syndrome and perhaps 90% of Cowden’s syndrome-like individuals, defined as having features of Cowden’s syndrome but not meeting the diagnostic criteria, remain without detectable PTEN mutations. Recently, a subset of such individuals were found to carry germline variants in SDHB and SDHD, encoding the B and D subunits of mitochondrial succinate dehydrogenase (40). In this pilot series, Cowden’s syndrome or Cowden’s syndrome-like individuals carrying a germline SDHB/D variant had higher frequencies of developing breast, thyroid, and renal cancers than even those with germline PTEN mutations.

Like other inherited cancer syndromes, multifocality and bilateral involvement is the rule. Hamartomas are the hallmark of Cowden’s syndrome. These are classic hamartomas in general and are benign tumours comprising all the elements of a particular organ but in a disorganized fashion. Of note, the hamartomatous polyps found in this syndrome are different in histomorphology from Peutz–Jeghers polyps, which have a distinct appearance. However, caution must be taken when the polyp histology is not read by a dedicated gastrointestinal pathologist as histological diagnoses are often incorrect when compared to genetic classification (41).

With regard to the individual cancers, even of the breast and thyroid, as of 2009, there has yet to be a systematic study published. Recently, however, one study has attempted to look at benign and malignant breast pathology in Cowden’s syndrome patients. Although these are preliminary studies, without true matched controls, it is, to date, the only study to examine breast pathology in a series of Cowden’s syndrome cases. Breast histopathology from 59 cases belonging to 19 Cowden’s syndrome women was systematically analysed (42). Thirty-five specimens had some form of malignant pathology. Of these, 31 (90%) had ductal adenocarcinoma, one tubular carcinoma, and one lobular carcinoma in situ. Sixteen of the 31 had both invasive and in situ (DCIS) components of ductal carcinoma, while 12 had DCIS only and two only invasive adenocarcinoma.

Benign thyroid pathology is more common in Cowden’s syndrome than malignant. Multinodular goitre and thyroid adenomas are often noted. Follicular thyroid carcinomas are much more common than papillary histology in PHTS although SDHB/D-related thyroid carcinomas are more likely to be papillary (2, 40). No systematic studies on thyroid pathology in Cowden’s syndrome have been performed.

Cowden’s syndrome usually presents by the late 20s. It has variable expression and, probably, an age-related penetrance. As with most syndromes prior to gene identification, the precise penetrance is unknown. By the third decade, 99% of affected individuals would have developed the mucocutaneous stigmata although any of the features could be present already (Boxes 6.17.1 and 6.17.2). It is believed that the penetrance is less than 10% under the age of 20 years. The most commonly reported manifestations are mucocutaneous lesions, thyroid abnormalities, fibrocystic disease, and carcinoma of the breast, gastrointestinal hamartomas, multiple, early onset uterine leiomyoma, macrocephaly (specifically, megencephaly), and developmental delay (Box 6.17.1) (2, 43). Pathognomonic mucocutaneous lesions are trichilemmomas and papillomatous papules (Box 6.17.2). Because of the lack of uniform diagnostic criteria for Cowden’s syndrome prior to 1995, a group of individuals, the International Cowden Consortium (2, 3), interested in systematically studying this syndrome arrived at a set of consensus operational diagnostic criteria (Box 6.17.2). Subsequently, when virtually all adult-onset presentations of Lhermitte–Duclos disease (LDD; dysplastic gangliocytoma of the cerebellum) were shown to have PTEN mutations, LDD was made a pathognomonic diagnostic criterion as well (44).

The two most commonly recognized cancers in Cowden’s syndrome are carcinoma of the breast and thyroid. By contrast, in the general population, lifetime risks for breast and thyroid cancers are approximately 11% (in women) and 1%, respectively. In women with Cowden’s syndrome, lifetime risk estimates for the development of breast cancer range from 25 to 50% (43, 45). The mean age at diagnosis is probably 10 years earlier than breast cancer occurring in the general population. Although Rachel Cowden died of breast cancer at the age of 31 (46) and the earliest recorded age at diagnosis of breast cancer is 14, the great majority of breast cancers are diagnosed after the age of 30–35 (range 14–65).

The lifetime risk for epithelial thyroid cancer can be as high as 10% in males and females with Cowden’s syndrome. Because of small numbers, it is unclear if the age of onset is earlier than that of the general population. Histologically, the thyroid cancer is predominantly follicular carcinoma although papillary histology has also been observed. Medullary thyroid carcinoma has yet to be observed in patients with Cowden’s syndrome.

Benign tumours are also very common in Cowden’s syndrome. Apart from those of the skin, benign tumours or disorders of breast and thyroid are the most frequently noted and probably represent true component features of this syndrome (Box 6.17.1). Fibroadenomas and fibrocystic disease of the breast are common signs in Cowden’s syndrome, as are follicular adenomas and multinodular goitre of the thyroid. Exponents of this field believe that endometrial carcinoma could be an important component tumour of Cowden’s syndrome as well. Other tumours that are seen in Cowden’s syndrome include renal cell carcinoma, malignant melanoma, and glial tumours. Whether each of these tumours is a true component of Cowden’s syndrome or whether some are coincidental findings is as yet unknown.

There are several ways in which Cowden’s syndrome patients can come to the attention of endocrinologists or endocrine surgeons. Sometimes, an individual with known Cowden’s syndrome is referred for management of their endocrine problems, chief of which are multinodular goitre, thyroid adenomas, and epithelial thyroid carcinomas. More commonly, such patients are not previously diagnosed and seek endocrinological attention because of abnormal thyroid function or a thyroid mass. Over two-thirds of Cowden’s syndrome patients have thyroid problems, which may occur at any age. However, finding multifocal lesions, especially in young individuals, should raise suspicion. Endocrinologists and endocrine surgeons should be especially mindful of the differential diagnosis of Cowden’s syndrome should they see patients with these thyroid lesions. A careful history and physical examination, as well as a meticulous family history to look for other component symptoms and signs of Cowden’s syndrome, are warranted.

Rarely, Cowden’s syndrome individuals present with an uncommon feature of Cowden’s syndrome, for example, hyperparathyroidism or parathyroid adenomas. When these occur together with ‘a thyroid cancer’, the initial diagnosis that endocrinologists might think of is multiple endocrine neoplasia type 2 (MEN 2) (see Chapter 6.12) (2, 47). However, it would be prudent to pursue the histology of the thyroid cancer as this might turn out to be a Cowden’s syndrome patient and not a MEN 2 case. Even more unusual, Cowden’s syndrome can present with ganglioneuromas of the gut and are referred to the endocrinologist as MEN 2B. However, in general, MEN 2B and Cowden’s syndrome, are clinically and genetically distinct (2, 47). A few MEN 2B cases can present with apparently isolated intestinal ganglioneuromatosis without the other classic stigmata of MEN 2B, yet all were found to have the MEN 2B-defining germline RET mutation Met918Thr and all developed medullary thyroid carcinoma (48).

With the identification of PTEN as the susceptibility gene for Cowden’s syndrome and the original linkage studies indicating no genetic heterogeneity, it is theoretically possible to perform direct mutation analysis of PTEN for molecular diagnosis of Cowden’s syndrome. Direct mutation analysis has advantages over linkage analysis as it can be performed even if only one individual is available. However, since the discovery of PTEN’s involvement in Cowden’s syndrome was relatively recent, the actual proportion of isolated and familial cases who carry germline PTEN mutations is unknown. If a germline PTEN mutation was detected in a previously undiagnosed individual or an individual with an unclear clinical presentation, then the diagnosis becomes obvious. If, however, no germline PTEN mutation was found in such an individual, then the result should be considered nondiagnostic. While SDHB/D are novel susceptibility alleles, these should be considered experimental until a validation series is achieved.

If a family-specific mutation is already known, then screening for that particular mutation in as yet unaffected family members would yield results which are 100% accurate, barring administrative error. If a family-specific mutation cannot be identified in a family that clearly fits the International Cowden Consortium operational diagnostic criteria for Cowden’s syndrome, then predictive testing based on direct mutation analysis is not possible. However, in the rare instances where the family is large and many affected members are available, then linkage analysis using makers within and closely flanking PTEN (D10S579, D10S1765, D10S2491/S2492, and D10S541) might be considered.

Interestingly, families or individuals with only breast and thyroid cancers or a Cowden’s syndrome-like phenotype that does not fulfil the diagnostic criteria of the International Cowden Consortium have a low frequency of germline PTEN mutation, approximately 2%. Having endometrial carcinoma might increase the likelihood for finding an occult germline PTEN mutation.

Although initially believed to be a locus for juvenile polyposis syndrome (OMIM 174900) (49), PTEN has been excluded (50). Further, the first susceptibility gene for juvenile polyposis syndrome has been identified as SMAD4/DPC4 on 18q21.1 and a second susceptibility gene is BMPR1A on 10q22 (51–53). Interestingly, germline deletions encompassing BMPR1A and PTEN seem to be peculiar to the so-called juvenile polyposis of infancy, clinically defined as juvenile polyposis presenting before the age of 6 years old (54).

The key to proper genetic counselling in Cowden’s syndrome is recognition of the syndrome. Families with Cowden’s syndrome should be counselled as for any autosomal dominant trait with high penetrance. What is unclear, however, is the variability of expression between and within families. We suspect that there are Cowden’s syndrome families who have nothing but trichilemmomas and, therefore, never come to medical attention.

The two most serious and established, component tumours in Cowden’s syndrome are breast cancer and epithelial thyroid cancer. Patients with Cowden’s syndrome or those who are at risk for Cowden’s syndrome should undergo surveillance for these two cancers. Beginning in their teens, these individuals should undergo annual physical examinations paying particular attention to the thyroid examination. Beginning in their mid 20s, women with Cowden’s syndrome or those at risk for it should be encouraged to perform monthly breast self-examinations and to have careful breast examinations during their annual physicals. The value of annual imaging studies is unclear since there are no objective data available. We usually recommend annual mammography and/or breast ultrasonography performed by skilled individuals in women at risk, beginning at age 30 or 5 years earlier than the earliest breast cancer case in the family, whichever is younger. Some women with Cowden’s syndrome develop severe, sometimes disfiguring, fibroadenomas of the breasts well before age 30. This situation should be treated individually. For example, if the fibroadenomas cause pain or if they make breast cancer surveillance impossible, then some have advocated prophylactic mastectomies.

Whether other tumours are true components of Cowden’s syndrome is unknown. It is believed, however, that endometrial carcinomas and possibly, skin cancers, might be true features of Cowden’s syndrome as well. For now, therefore, surveillance for other organs should follow the American Cancer Society guidelines, although proponents of Cowden’s syndrome will advise routine skin and uterine surveillance as well.

The key to successful management of Cowden’s syndrome patients and their families is a multidisciplinary team. There should always be a primary care provider, who orchestrates the care of such patients, some of whom will need the care of surgeons, gynaecologists, dermatologists, oncologists, and geneticists at some point.

Approximately 65% of all BRRS cases will have germline PTEN mutations (2, 28, 32). Since clinical epidemiological studies on already small numbers of BRRS suggest no formal association with cancer, it becomes difficult to interpret how finding such a mutation in a BRRS patient would alter medical management. If one were to extrapolate from the Cowden’s syndrome–PTEN data, then it might be conservative to suggest that all BRRS patients and all other PHTS patients be followed for cancer development similar to that practised for Cowden’s syndrome.

I am grateful to my collaborators and members of my laboratory, past and present, especially Debbie J. Marsh, PhD and Xiao-Ping Zhou, MD, PhD, for contributing to the work described in this chapter. I am deeply appreciative of Kathy Schneider, MPH, CGC for her superb coordination of the PTEN Study during its infancy, and for her continued friendship and support. My laboratory has been and is supported by the American Cancer Society, the Breast Cancer Research Foundation, the Department of Defence USARMC Breast Cancer Research Programme, the Susan G. Komen Breast Cancer Foundation, and the National Cancer Institute. CE is a Doris Duke Distinguished Clinical Scientist Awardee, an American Cancer Society Clinical Research Professor, and the Sondra J. and Stephen R. Hardis Chair of Cancer Genomic Medicine at the Cleveland Clinic.