Extended spectrum of cancers in PTEN hamartoma tumor syndrome

PTEN hamartoma tumor syndrome (PHTS) is a term encompassing individuals with disease-causing germline PTEN variants regardless of clinical diagnosis^1,2. PHTS includes Cowden syndrome (CS), Bannayan-Riley-Ruvalcaba syndrome (BRRS), PTEN-related Proteus syndrome (PS), and PTEN-related Proteus-like syndrome¹. Even prior to the discovery of PTEN as the first susceptibility gene for CS, it was recognized that the syndrome was associated with elevated lifetime risks of breast and thyroid cancers^3,4. Subsequent studies following the identification of germline PTEN variants in individuals with CS led to the expansion of the spectrum of cancers associated with CS and the establishment of more accurate cancer risk assessment and surveillance strategies^5,6,7.

Studies in adult patients with PHTS indicate that germline PTEN variants are associated with increased lifetime risks of breast, thyroid, endometrial, kidney, and colon cancers, as well as melanoma (collectively referred to as PHTS component cancers)^6,8,9,10. However, isolated reports and anecdotal evidence from our clinical practice (PTEN Multidisciplinary Clinic and Center of Excellence, Cleveland Clinic, Cleveland, Ohio, USA) show that individuals with PHTS can also have other cancers beyond those known to be associated with the syndrome (i.e., non-component cancers or NCC). Importantly, the spectrum of non-component cancers has not been systematically studied in PHTS. In addition, it is not known whether these cancers are incidental (i.e., sporadic) or whether they are bona fide components of PHTS that were missed in earlier studies due to limited patient series. In this study, we report the spectrum of non-component cancers from a prospective international series of pediatric and adult patients with PHTS to inform prospective cancer risk assessment and surveillance recommendations.

Patient characteristics and germline PTEN variant spectrum

We prospectively accrued 701 research participants with germline PTEN variants from September 1, 2005, through January 6, 2022. Of the 701 research participants, 340 (49%) received cancer diagnoses (any cancer type), with 144 (42%) of those diagnosed with second primary malignant neoplasms (SMN). Of the 340 individuals with PHTS and cancer, 101 (30%) had at least one diagnosis with a non-component cancer (NCC) (Table 1). These cancers include any cancer type other than those known to be associated with PHTS, the latter including breast, thyroid, endometrial, kidney, and colon cancers, as well as melanoma^6,11. Amongst those with cancer (n = 340), PHTS patients with NCC (n = 101) showed an overrepresentation of male biological sex compared to patients without NCC (n = 239; 29% versus 15%, P = 0.003). In addition, patients with NCC were of older age at study enrollment compared to patients without NCC (median [IQR] age at consent, 56 [47–64] versus 47 [40–57] years; P = 0.0005). However, there were no differences between those with and without NCC when we compared the ages at cancer diagnosis, and the Cleveland Clinic (CC) score, a semi-quantitative surrogate of overall phenotypic burden (Table 1)⁵. Interestingly, 71 (70%) of PHTS patients with NCC also had a diagnosis of at least one other PHTS component malignancy (Table 1). These patients had an older age at enrollment compared to PHTS patients diagnosed with NCC in the absence of any other component cancer, although this observation was statistically non-significant (median [IQR] age at consent, 57 [50–64] versus 49 [38–63] years; P = 0.06). Of the 71 PHTS patients diagnosed with both NCC and component cancers, 42 had a full set of cancer diagnosis dates to identify the chronological sequence of cancer occurrence events; 16 (38%) had at least one NCC diagnosed before a component cancer, with a median (IQR) duration of 10 (5–16) years between both events, and 26 (62%) had at least one component cancer diagnosed before an NCC, with a median (IQR) duration of 10 (5–17) years between both events.

PTEN variants were classified into tier 1 including all pathogenic and likely pathogenic variants (215 variants [63%]), and tier 2 including variants of uncertain significance (VUS) and those with conflicting interpretation of pathogenicity with at least one VUS designation (125 variants [37%]). No differences were observed with respect to the PTEN overall tier 1 versus tier 2 variant classifications, as well as the PTEN variant types between individuals without versus those with NCC (Table 1). Considering only individuals with Tier 1 PTEN variants (n = 215), 63 (29%) had at least one diagnosis with an NCC.

Cancer spectrum and lifetime cancer risks

We observed a wide spectrum of diagnosed cancer types in the NCC category. These malignancies were generally categorized into hematological, gastrointestinal, genitourinary, germ cell, soft tissue, skin, and other cancer types (Table 2). While most cancers occurred in single individuals or in a limited number of individuals, the most overrepresented were non-melanoma skin cancers (41/101 or 40.6%), ovarian cancer (10/101 or 9.9%), prostate cancer (10/101 or 9.9%), soft tissue sarcomas (7/101 or 6.9%), and lung cancer (6/101 or 5.9%). Ovarian cancer cases included malignant epithelial ovarian tumors, malignant ovarian germ cell tumors, and malignant ovarian stromal tumors. There were five individuals who were diagnosed with both basal cell and squamous cell carcinomas, and one individual who was diagnosed with two soft tissue sarcomas (malignant giant cell tumor of tendon sheath of the right hand at age 46, and chondrosarcoma at age 59). Compared with standard population risks, we observed elevated risks of ovarian cancer (age-adjusted SIR 11.4; 95% CI, 5.8–20.3), prostate cancer (SIR 4.1; 95% CI, 1.9–7.8), and soft tissue sarcomas (SIR 10.7; 95% CI, 3.9–23.7). These observations were consistent when only including individuals with PHTS and tier 1 germline PTEN variants, except for ovarian cancer which became insignificant and comparable to the general population (Table 3). We could not evaluate the age-adjusted SIR for non-melanoma skin cancers due to the lack of age-related population-based estimates. The median (IQR) age at diagnosis for ovarian cancer was 49 (46–57) compared to 63 years in the SEER database, 59 (56–61) for prostate cancer versus 67 years in the SEER database, and 46 (28–47) for soft tissue sarcomas compared to 62 years in the SEER database (Table 4).

Independent cases of non-component cancers

In support of these data, we performed a literature survey of clinical cases reported by independent research groups describing non-component cancers in PHTS (Table 5). Of note, we did not limit the NCC to those we observed in our patient series. All cases included individuals with pathogenic and/or pathogenic germline PTEN variants, except for one case with a VUS¹², and another case where the exact PTEN variant was not reported¹³. Relatedly, of the 33 reviewed studies, 16 performed genomic or immunohistochemical studies on tumor specimen derived from individuals with PHTS and NCC. Of these 16 studies, 13 reported somatic loss of heterozygosity or deletion at the PTEN locus and/or loss of PTEN protein expression through immunohistochemical analyses. These data corroborate the role of PTEN in the etiology of at least a subset of these non-component cancers.

In this study, we report the spectrum of non-component cancers from the longitudinal follow-up of an international prospective series of pediatric and adult patients with PHTS. Anecdotal evidence from our clinical practice shows that individuals with PHTS can also have other cancers beyond the six component cancers known to be associated with the syndrome. Indeed, independent case reports and case series corroborate this finding^{12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44}.

In our extended series, we found that approximately one-third of patients with PHTS and cancer have at least one NCC. Interestingly, those with NCC showed an overrepresentation of male biological sex. We posited that this was the case since we are comparing this group to PHTS patients with component cancers, known to be enriched with breast cancer and thus female sex. However, we found this to not be the case since most patients with NCC also had a prior or subsequent component cancer (71/101), with equal enrichment of breast cancer between the two groups. In addition, patients with NCC were of older age at study enrollment compared to those without NCC. Since most patients with NCC also had a PHTS component cancer, this observation does not support the hypothesis that these individuals had delayed enrollment due to the lack of diagnosis with cancers typical for PHTS. Indeed, individuals with PHTS and NCC alone tended to have a younger age at enrollment compared to those who had at least one other component cancer. Therefore, we acknowledge that age at enrollment is a complex phenomenon that may be influenced by multiple factors besides phenotypes of interest. Importantly, the ages at cancer diagnosis were not different between patients with PHTS with and without NCC. This corroborates the importance of clinicians being cognizant of other features that would raise a suspicion for a PHTS diagnosis (e.g., macrocephaly, pathognomonic mucocutaneous lesions), particularly in the presence of NCC. Of note, the median ages at diagnosis for ovarian cancer, prostate cancer, and soft tissue sarcomas were younger in individuals with PHTS compared to estimates from the general population. While we cannot propose guidelines for the surveillance of these cancers in the absence of age-related penetrance estimates, we call for more awareness regarding these malignancies, particularly where risk reduction strategies may be implemented. For example, in scope of the elevated risk of endometrial cancer as a component malignancy, and the consideration of hysterectomy with completion of childbearing⁷, we recommend considering bilateral salpingo-oophorectomy in women for whom a total abdominal hysterectomy is indicated.

Considering only individuals with Tier 1 PTEN variants (n = 215), 63 (29%) had at least one diagnosis with an NCC, which is higher than the 5–10% estimates from European adult individuals with PHTS^45,46. However, the latter studies did not systematically study NCC, and it is unclear whether the reported estimates also included individuals with component cancers in addition to the NCC^45,46. It is important to note that we focused our comparisons within individuals with PHTS and cancer since it is impossible to ascertain whether participants without cancer at the time of enrollment (particularly children, adolescents, and young adults), will go on to develop cancer later in life. Compared with standard population risks and including individuals with both tier 1 and tier 2 PTEN variants, we observed elevated risks of ovarian cancer, prostate cancer, and soft tissue sarcomas. These cancers were also associated with a younger median age at diagnosis compared to standard population-based estimates, supporting a non-incidental (i.e., sporadic) finding in the context of PHTS. This is also supported by the finding that additional somatic PTEN alterations, loss of heterozygosity at the PTEN locus, and/or loss of PTEN protein expression have been found in at least a subset of various non-component cancer specimens derived from individuals with PHTS (Table 5)^{12,18,19,20,21,25,27,30,31,36,40,41,44}. Relatedly, PTEN somatic alterations and expression changes, as well as downstream PI3K-AKT signaling have been associated with sporadic carcinogenesis in the ovaries, prostate, and soft tissues^{47,48,49,50,51,52,53,54}.

This study has limitations. There is a possibility that patients lost to follow-up may have developed cancer since the time of accrual or the time of their last clinical visit. This may result in an incomplete spectrum and underestimation of NCC frequencies. Relatedly, the heterogeneous spectrum of non-component cancers resulted in a limited number of individuals per cancer type, likely reflecting the lack of capture of these cancers in earlier studies⁶. The limited number of individuals per cancer type precluded us from performing age-related penetrance estimates within PHTS, and cancer risk estimates of all reported non-component cancers. We also could not evaluate the age-adjusted SIR for non-melanoma skin cancers due to the lack of age-related population-based estimates. Relatedly, there is a possibility that at least a subset of NCC may have arisen secondary to component cancer treatments. However, this scenario is unlikely, especially when the NCC is the first reported cancer and given the wide spectrum of observed NCC. Moreover, it is possible that since these patients receive more frequent and specialized care, providers and patients may be more judicious on working up possible cancer, including NCC. This may have resulted in overdiagnosis of NCC, although these patients do not receive any additional routine screening for NCC under current PHTS surveillance guidelines and are unlikely to have increased detection. In addition, we could not perform functional studies to evaluate somatic PTEN second hits, LOH, and/or loss of expression in tumor specimen from individuals with NCC in our study. However, with awareness regarding the presence of NCC in PHTS, we anticipate independent systematic studies to validate and characterize these risks in other PHTS populations around the world.

According to the presented data, we recommend increased awareness regarding the existence of NCC in individuals with PHTS. The elevated lifetime cancer risks of prostate cancers and soft tissue sarcomas in carriers of tier 1 PTEN variants will inform prospective validation studies to characterize age-related penetrance and accordingly, the appropriate ages for any warranted enhanced surveillance and risk-reducing strategies, in an adequate representative sample of patients.

Research participants

Research participants were prospectively accrued from September 1, 2005, through January 6, 2022, as part of a prospective follow-up study approved by the Cleveland Clinic institutional review board (IRB protocol 8458)¹¹. The study was conducted in agreement with the principles of the Declaration of Helsinki. All participants provided informed written consent to participate. Participants were evaluated at the PTEN Multidisciplinary Clinic and Center of Excellence at the Cleveland Clinic (Cleveland, Ohio, USA). This study included both pediatric and adult patients with PHTS accrued from community and academic medical centers throughout North and South America, Europe, Australia, and Asia. Inclusion criteria for the Cleveland Clinic PTEN study (IRB 8458) included meeting, at minimum, the relaxed operational diagnostic criteria of the International Cowden Consortium (ICC), meaning individuals fulfilling pathognomonic criteria, or at least 2 criteria; either major or minor^55,56. Reported cancer diagnoses were documented through pathology reports or verified cancer genetics visits. For the purposes of this focused study, we prioritized individuals from the parent PTEN study with confirmed germline PTEN testing results and regardless of specific cancer phenotypes. Specialist genetics staff reviewed all checklists, and if necessary, corresponded with the enrolling center to obtain primary documentation of medical records and pathology reports for phenotypic confirmation with patient consent. Baseline information including any (component and non-component) cancer history was recorded at the time of consent. Between July 2021 and July 2022, we obtained updated phenotypic information in those who had not routinely seen us in genetics clinic within 3 years¹¹. We reviewed cancer-related health records of patients with PHTS internal to the Cleveland Clinic health system.

Germline PTEN variant classification

PTEN variants were classified into tier 1 (n = 514/701 or 73%), including all pathogenic and likely pathogenic variants, and tier 2 (n = 187/701 or 27%), including variants of uncertain significance (VUS) and those with conflicting interpretation of pathogenicity (n = 9; at least one classification should be a VUS). PTEN variant classifications were ascertained by clinical genetic testing reports where available, ClinVar database classifications, and/or the ClinGen gene-specific criteria for PTEN variant curation⁵⁷. Analyses were performed for the entire series of patients (tier 1 and tier 2 PTEN variants) and for those with tier 1 PTEN variants only, separately.

Published case reports and case series

We conducted a literature search to identify case reports, case series, and other studies that report on non-component cancers in PHTS. Our literature search used Ovid MEDLINE and EMBASE databases of all articles published between 2002 and 2023. Search terms (alone or in combination) included: “PTEN Hamartoma Tumor Syndrome”, “Cowden Syndrome”, “Bannayan-Riley-Ruvalcaba Syndrome”, “Proteus syndrome”, “Proteus-like syndrome”, “Lhermitte Duclos”, “PHTS”, “PTEN or PTEN Phosphohydrolase”, “Neoplasms”, “(overgrowth or hamartom* or fibrom* or macrocep*)”, and “(carcin* or cancer* or neoplas* or malign*)”. Identified articles were imported into Covidence, a web-based collaboration software platform that streamlines the literature review process, where studies were deduplicated and screened. The literature review process took place between September 2022 and March 2023. We excluded any article without English translation, those that go back to more than 20 years, and those that cover PHTS component cancers. We also excluded studies from our research group to avoid the reporting of duplicate cases. We only included studies where the germline PTEN variant status was known and included either tier 1 or tier 2 PTEN variants to be consistent with the study design. Abstract screening and full text review were performed in parallel by two independent reviewers. Any discrepancies were resolved through consensus or discussion with a third reviewer. Following this process, an independent PubMed search was performed in September 2023 to cover any articles missed since the time of the Covidence literature search.

Statistical analysis

Clinical and demographic characteristics among participants with germline PTEN variants and cancer (n = 340) are described and comparisons between NCC versus without NCC were tested using a Chi-square test for categorical variables or t-test for continuous variables. In the whole series of 701 participants, we calculated standardized incidence ratios (SIR) using age-specific incidence data from the Surveillance Epidemiology and End Results database (SEER incidence data, November 2022 submission, 1975–2020). We evaluated four different cancer types separately and did not adjust for multiple comparisons. Analyses were performed using R software version 4.2.1 (R Project for Statistical Computing) and OpenEpi version 3.01 (Open Source Epidemiologic Statistics for Public Health) software. All statistical tests were two-sided, and P values < 0.05 were deemed significant.