Frequent gastrointestinal polyps and colorectal adenocarcinomas in prospective series of PTEN mutation carriers

Background

Cowden syndrome is often considered a rare hamartomatous polyposis syndrome caused by germline alterations in the tumor suppressor gene, PTEN.¹ It is thought to occur in 1 in 200,000 individuals; however, experts believe this is likely an underestimate because of the variable expression and subtle physical manifestations.² CS is one of the disorders that comprises the PTEN hamartoma tumor syndrome.³ It is an autosomal dominant disorder that is characterized by mucocutaneous lesions, macrocephaly, and an increased risk of benign and malignant diseases of the breast, thyroid, and endometrium.^{2, 3} Although the gastrointestinal (GI) tract is affected in individuals with CS, it has not been systematically assessed.

Published case reports and highly selected small series reveal that 35–85% of Cowden syndrome patients had GI hamartomatous polyps.^4–9 Although the majority of patients have been described to have hamartomatous polyps, they have also been reported to have ganglioneuromatous polyps, colonic lipomas and lymphoid aggregates, and hyperplastic, adenomatous, and inflammatory polyps. These polyps have been reported to occur in the esophagus, stomach, duodenum, jejunum, ileum, colon and rectum, with the colon being the site most often affected.

Whether GI neoplasias, especially malignancies, are true component phenotypes of CS is not known because this association has not been systematically studied in large series.¹⁰ Various case reports of colorectal cancer in patients with CS have been published, mainly prior to the mid-2000’s and often without the advantage of PTEN mutation status.^{5, 11–13} In a series of 93 Japanese patients, nine (9.6%) were reported to have colon cancer.¹⁴ Gastric cancer has been highlighted in case reports of two individuals with CS.^{15, 16} The risk of benign and malignant GI neoplasias has yet to be characterized in a large series of individuals with PTEN mutation positive Cowden syndrome. We, therefore, sought to determine the prevalence and characteristics of the GI phenotype in our series of PTEN mutation positive subjects.

Materials and Methods

Results

Patients who met relaxed International Cowden Consortium (ICC) criteria (N=2548) or with ≥5 GI (any location) polyps, ≥1 of which was hyperplastic or hamartomatous (N=397) were prospectively recruited (see schema in Fig. 1). Of the 2945 total subjects from these two prospectively accruing protocols (see Methods for details), 127 patients having clear pathogenic PTEN mutations were eligible for this study (Fig. 1). Four of these subjects were enrolled from the cohort ascertained by presence of at least five GI polyps, at least one of which was hyperplastic or hamartomatous, and the remaining 123 were enrolled from those who met relaxed ICC criteria (Fig. 1). In our series of 127 PTEN mutation positive individuals, there were 8 individuals who might be clinically diagnosed with BRRS, comprising 5 male individuals with penile freckling and macrocephaly, 3 of whom also had lipomatosis. The remaining 3 were females, all of whom had macrocephaly, at least one lipoma (not lipomatosis) and one vascular anomaly. None of these 8 had GI malignancies. Average age at study enrollment was 34.6 years (range 1–73 years), and 63 (49.6%) were male. The majority of subjects were white (78), of which 4 were Hispanic/Latino, with the remaining black or African American (9), American Indian or Alaska Native (8), or Asian (6). Race was unknown for 26 subjects.

Of the 127 eligible subjects, GI polyps were reported in 65 (51.2%), with 24 having both upper and lower GI polyps, two only upper GI polyps and 41 only colorectal polyps. Subjects with polyps were significantly older (42.0 years) at the time of enrollment compared to those without polyps (26.6 years, p=0.0001). No gender differences were observed between those with and those without polyps. There was no clear genotype-phenotype correlations for those with or without polyps or those with or without malignancies.

Discussion

Recognition of pertinent personal medical and family history features as well as physical manifestations is critical for accurate diagnosis, risk assessment, genetic testing, and medical management for individuals with hereditary cancer syndromes. Textbooks and single case reports have noted the association of CS and hamartomatous polyps for some time, but not necessarily based on systematic analysis. However, it remains an under-acknowledged manifestation of the disorder for several reasons, perhaps because the malignant potential of these polyps is not well characterized and because there has yet to be a systematic series addressing even the frequency and characteristics of the polyp histology. In our series of all comers with germline pathogenic PTEN mutations, GI polyps occurred in 51.2%, making it the second most common CS feature. Considering that colorectal polyps occurred in 95% of our eligible subjects with germline PTEN mutations who underwent at least one colonoscopy, and approximately half of the entire series, this 51.2% prevalence of any GI polyp is likely an underestimate. Notably, in addition to the textbook-acknowledged hamartomatous polyps, we show here that hyperplastic polyps, ganglioneuromatous polyps and adenomatous polyps are important components of the CS polyp histology. In fact, at least half of our mutation carriers who were shown to have colorectal polyps had two or more histologic types.

In an effort to create uniform criteria in 1995 for the purposes of identifying the predisposition gene, the ICC developed consensus operational diagnostic criteria,^{20, 21} which were revised in 2000, in an attempt to broaden the net for clinical usage.² These criteria specify that GI hamartomas are a minor criterion because of lack of systematic study. If the ICC criteria were amended to include GI polyps as a major criterion, then an additional 21 (16.5%) subjects would have had a clinical diagnosis of Cowden syndrome at the time of study enrollment. The most recent version of the National Comprehensive Cancer Network Cowden syndrome testing criteria were updated to include multiple GI hamartomas or ganglioneuromas as a major criterion and a single GI hamartoma or ganglioneuroma as a minor criterion based on expert opinion in the context of single case reports and highly selected small series, often without PTEN mutation information.²² The observations from our prospective systematic study lend objective support to the changes to the NCCN testing criteria.

For patients who are suspicious for the PTEN hamartoma tumor syndrome, it is important to assess previous history of GI polyps. Efforts should be made to confirm the polyp burden in these patients with medical records. Additionally, given the inaccuracies with polyp histology,²³ consideration should be given to having the pathology re-reviewed by a dedicated GI pathologist. For patients suspicious for Cowden syndrome or a PTEN hamartoma tumor syndrome who have not previously undergone an endoscopy, the risks and benefits of a baseline upper endoscopy and colonoscopy should be considered in the diagnostic workup.

Colorectal cancer occurred in 7.1% of our entire series and 13% of eligible subjects who underwent at least one colonoscopy (age- and gender-adjusted SIR=224). Currently, colorectal surveillance is not routinely recommended for individuals with Cowden syndrome beyond that for the general population. However, in our series, all nine subjects were diagnosed with colorectal cancer prior to age 50, with the youngest age at diagnosis being 35 years. Therefore, had our subjects initiated screening at age 50, their malignancies would likely not have been detected until an advanced stage. Individuals with PTEN mutations may benefit from earlier colonic surveillance. One group has advised a “vigorous” screening approach to patients with Cowden syndrome that includes colonoscopy beginning at age 15 with follow up every 1–2 years.²⁴ This approach is rather aggressive. Based on our current observations, we recommend considering a baseline colonoscopy at age 35, or sooner if symptoms develop, with follow-up time based on polyp burden.

Based on our series, ~15% of PTEN mutation carriers with colorectal hyperplastic polyps developed colorectal cancer; and almost 20% of mutation carriers with colorectal adenomas had colorectal cancer. The patients who developed colorectal carcinomas also tended to have multiple polyps, based on small numbers. Therefore, if these observations can be reconfirmed in an independent series, then indications for colorectal surveillance should be offer to any PTEN mutation carrier with multiple lower GI polyps, and/or the presence of hyperplastic and/or adenomatous polyps.

There were 2 (13%) individuals with sessile serrated adenomas, both of whom were amongst the 16 individuals who met the criteria for hyperplastic polyposis syndrome. There are 2 interpretations to this observation: that this is a true representation, or there was uneven “calling” of this histology amongst the pathologists. We tried to gauge this by culling out our research subjects enrolled in another study (who may or may not have PTEN mutations) during a similar period, and we found that of those who met the diagnosis of HPS, ~50% also had sessile serrated adenomas (p=0.07).

Only one individual in our series developed gastric cancer. Although the formal adjusted SIR is elevated, we cannot draw genetic counseling or clinical conclusions given that this single subject was a white male who developed gastric cancer at 67 years old. It is nonetheless interesting to note the co-existing diffuse hyperplastic polyposis in the stomach and duodenum, almost mirroring the situation in the colon. In our series of PTEN mutations carriers, upper GI polyps do occur with some frequency, and, for a subset of patients, they do experience symptoms. Excluding the five individuals with unknown polyp histology in the upper GI tract, 15 of the 19 with polyps in the upper and lower tracts had concordant histologies. Interestingy, fundic gland polyps, often said to be indicative of CS, were only found in two mutation carriers. Notably, 20% of those with GI examinations had glycogenic acanthosis. Although no consensus guidelines exist regarding upper GI management in CS, Schreibmann et al. recommend upper endoscopy and an upper GI series with small bowel follow-through beginning at age 15 with follow up every two years.²⁴ Again, based on our current observations, we believe that this approach is rather aggressive and recommend upper GI surveillance only for symptom management, at least in the white population. Based on anecdotal reports on Asian populations, we suspect that GI features, perhaps even upper GI malignancies, may be more prominent in the Asian population.

One of the major limitations of this study is that not all medical records were available to confirm all of the reported histories, although these remain in the minority. Therefore, for some patients, we do not have detailed information about their precise GI history. Further, since patients were recruited from multiple medical centers, there is wide variability in the medical reports and pathology expertise available. The most ideal study would be to offer at least baseline colonoscopy, by a limited number of endoscopists, and to have the pathology reviewed by a single pathologist with expertise in hereditary GI disease in order to accurately capture the frequency with which GI polyps occur in these patients. Nonetheless, despite these limitations, the conditions of the study do somewhat simulate the data available at a high-risk GI consultation.

Based on the observations from our prospectively accrued series of PTEN mutation carriers and the literature, we conclude that both upper and lower GI polyps are common component features of the PTEN hamartoma tumor syndrome. The presence of non-adenomatous polyps, and especially comprising mixed histologies, should signal a healthcare provider to refer such individuals to high-risk assessment. Furthermore, the presence of both macrocephaly and non-adenomatous GI polyps together should predict, to a high probability, for the presence of germline PTEN mutations. Colorectal adenomas are usually not considered part of the CS phenotype, but our series suggests that they most likely are, with almost 20% of PTEN mutation carriers with colorectal adenomatous polyps developing colorectal adenocarcinomas. Similarly, multiple colorectal polyps of any histology and hyperplastic polyps, especially a hyperplastic polyposis syndrome phenotype, may be red flags for PTEN mutation carriers developing colorectal cancers. Importantly, colorectal adenocarcinomas have an increased prevalence in PTEN mutation carriers, all occurring prior to the age of 50 years, and so, routine colonic surveillance should be considered.