Broadening risk profile in familial colorectal cancer type X; increased risk for five cancer types in the national Danish cohort

Heredity is estimated to explain ~ 20% of the colorectal cancer diagnoses and covers a complex genetic landscape [35]. Though several rare high-risk alleles have been identified, a large fraction of families with seemingly inherited colorectal cancer diagnoses remains genetically undefined. In here families, who meet the Amsterdam I criteria for Lynch syndrome, but with no signs of mismatch repair (MMR) deficiency, i.e. a mismatch-repair stable phenotype and/or retained MMR protein expression, are referred to as familial colorectal cancer type X (FCCTX) [21, 38]. This subset constitutes 40% of the families that fulfil the Amsterdam I criteria and belong to the hereditary non-polyposis subgroup of hereditary cancer [34]. The genetic aetiology of FCCTX is most likely heterogenous and may include rare pathogenic germline variants in e.g. heterozygous MUTYH, CHEK2, BRCA2, POLE, POLD1, SEMA4A, BMPR1A, RPS20 or OGG1 or modifying single nucleotide polymorphisms in SEMA4A, EXO1, TGFBR1, or NUDT1 [1, 5, 9, 10, 12, 24, 25, 29, 40].

Besides the MMR proficient molecular phenotype, FCCTX-associated colorectal cancers have been distinguished from the genetically defined cancer syndrome, Lynch syndrome, by a predilection for tumour development in the distal colon and the rectum, a high adenoma/carcinoma rate and a lower risk of synchronous and metachronous colorectal cancer [6, 19, 22, 30]. The risk of colorectal cancer is lower than in Lynch syndrome with a relative risk (RR) of 0.5, but higher than the general population with a standardized incidence ratio of 2.3 [2, 21]. Current literature suggests that FCCTX is a colorectal cancer-only syndrome, which provide the basis for current recommendations of surveillance with regular colonoscopy starting 5–10 years prior to the youngest case in the family in families classified as FCCTX [2, 21, 22, 36].

We challenged this notion through risk assessment of 30 different extra-colorectal cancer types in the Danish FCCTX cohort compared to the Danish general population and to the national Lynch syndrome cohort and found increased risk of five extra-colorectal cancer types, i.e. urothelial cancer, female breast cancer, gastric cancer, pancreatic cancer, and ocular melanoma.

The national Danish hereditary non-polyposis colorectal cancer (HNPCC) register contains ~ 6000 families with suspected or verified hereditary colorectal cancer reported to the register by genetic counsellors, surgeons, pathologists, and genetic diagnostic laboratories. Families have been included based on a suspicious family history of colorectal cancer, fulfilment of the Amsterdam I or II criteria [38], or identification of disease-predisposing variants in genes linked to hereditary colorectal cancer. Based on family history, the register subclassifies families according to genotypic and phenotypic subsets [20]. The Danish HNPCC register identifies all family members in the Danish Civil Registration System, regardless of cancer history, based on data collected from clinical files and health care registers.

The FCCTX cohort comprised 213 families, including 646 individuals with a colorectal cancer, 1982 first-degree relatives and 1044 second-degree relatives. These individuals contributed with 110,767 person-years at-risk and during this time 966 individuals developed 1078 cancers, including 493 extra-colorectal cancers and 585 colorectal cancers (Supplementary Table 1). The most prominent cancer types observed were breast cancer (N = 103), prostate cancer (N = 51), urothelial cancer (N = 45), and lung cancer (N = 40). Prospective analyses were based on 223 individuals in surveillance from 109 families, who contributed with 903,448 person years at-risk. In this subgroup that considered only cancers that developed after the initiation of surveillance in the family, 160 cancers, including 26 breast cancers, 21 urothelial cancers and 19 prostate cancers, were diagnosed (Supplementary Table 1).

Compared to a population-based cohort, the FCCTX cohort revealed significantly increased risks for five cancer types, i.e. breast cancer, urothelial cancer, pancreatic cancer, gastric cancer, and ocular melanoma with variable peak incidence ages identified in the different tumour types (Fig. 1, Table 1, Supplementary Table 2). Significantly increased IRRs were observed for breast cancer in the age groups from age 30 until 69 years (IRR for age 30–49: 1.71, 95% CI 1.02–2.68, p = 0.0070 and IRR for age 50–69: 1.54, 95% CI 1.06–2.14, p = 0.0030), while gastric cancer showed an IRR of 5.87 (95% CI 1.57–15.02, p = 0.0007) for the age group 30–49 years. Urothelial cancer and pancreatic cancer developed at significantly increased IRRs of 2.10 (95% CI 1.26–3.30, p = 0.0003) and 2.21 (95% CI 1.12–4.38, p = 0.0023), respectively, in the oldest age cohort, above age 70. In total, five eye tumours developed in age group 50–69 years, all of which were classified as ocular melanomas, giving an IRR of 7.73 (95% CI 1.76–21.45, p = 0.0006) (Fig. 1, Table 1).

The increased risks of extra-colorectal cancers applied to all FCCTX individuals with similar risk levels in first-degree and second-degree relatives compared to the relatives affected with colorectal cancer (Supplementary Figure 1). In the prospective analysis, significantly increased risks applied to urothelial cancer and ocular melanoma with an IRR of 2.94 (95% CI 1.34–5.54, p = 0.0004) above age 70 for urothelial cancers and an IRR of 13.81 (95% CI 1.68–49.32, p = 0.0015) for ocular melanomas (Table 2).

At least one individual with breast cancer was observed in 85 FCCTX families. If the criteria for genetic testing for hereditary breast and ovarian cancer were considered, 35 of these families fulfilled at least one of the criteria previously described [41]. Likewise, gastric cancer was found in 18 FCCTX families, of which two families fulfilled the clinical criteria for genetic testing previously published [4, 26]. As we did not have genetic data on variants in the BRCA1/BRCA2 genes predisposing for hereditary breast and ovarian cancer or the CDH1 gene predisposing for hereditary diffuse gastric cancer, we excluded these families from the FCCTX cohort. The sensitivity analysis decreased the IRRs to non-significant levels for breast cancer, gastric, pancreatic and urothelial cancer, except for breast cancer age 30–49 where the IRR significantly decreased to 0.19 (95% CI: 0.02–0.67, p = 0.0002), from a previous increased IRR. Eye tumours remained significantly increased with similar IRR (Table 3).

Compared to Lynch syndrome, FCCTX families showed significantly lower risks in at least one age group for eight different tumour types, i.e. endometrial cancer, ovarian cancer, urothelial cancer, kidney cancer, gastric cancer, cancer of the small bowel, non-melanoma skin tumours and brain tumours (Table 4). For urothelial cancer and skin cancer this observation was consistent from age 30+. The risk of urothelial cancer in FCCTX showed IRRs of 0.06–0.31 (p-values < 0.0009) compared to Lynch syndrome and the risk of skin cancer showed IRRs of 0.11–0.24 (p < 0.005) (Table 4). For the other tumour types this difference particularly applied to the peak ages of these cancer types in Lynch syndrome, i.e. age 30–69 years (Table 4). No difference in risks of breast cancer, eye tumours and pancreatic cancer were found between the FCCTX and Lynch syndrome cohorts for all age groups (Supplementary Table 4.).

Increased awareness of hereditary colorectal cancer and improved access to genetic diagnostics implies that a growing number of families with a phenotype suggesting hereditary cancer with an undefined genotype are identified. FCCTX represents one of these subsets where refined risk estimates are relevant to develop evidence-based surveillance recommendations. Lindor et al., described a standardized incidence ratio for colorectal cancer of 2.3 in a cohort of 71 FCCTX families and did not identify any significantly increased risk of extra-colorectal cancer [21]. Based on this observation, FCCTX is considered a colorectal cancer-only syndrome with surveillance generally recommended to be confined to colonoscopy with 5-year intervals starting 5–10 years prior to the first case in the family. Surveillance programs for colorectal cancer have been optimized with documentation of more efficient detection of precursor lesions and early-stage tumours (RR 0.2–0.3) [14]. Reduced risk of mortality from colorectal cancer and increased life expectancy implies that individuals with FCCTX may be at risk of extra-colorectal tumour types during this increased lifetime. Our data, based on all 213 FCCTX families in the national Danish HNPCC-register, challenges the present view on FCCTX as a colorectal cancer-only syndrome and demonstrate significantly increased incidence rates for five extra-colorectal cancer types with urothelial cancer remaining significant in the colorectal cancer-surveilled cohort (Fig. 1, Table 1, Table 2).

We demonstrate an increased risk of urothelial cancer from age 70 in FCCTX with IRRs of 2.1 in the entire FCCTX cohort and 2.9 in the surveilled subset compared to the risk in an age- and sex matched Danish population (Tables 1 and 2). The risk of urothelial cancer in FCCTX was significantly lower than in Lynch syndrome with IRRs of 0.1–0.3 (Table 4). Except for Lynch syndrome, urothelial cancer has not been linked to hereditary colorectal cancer [11]. One possibility would be undiagnosed Lynch syndrome cases e.g. MMR gene variants that allow for retained MMR function and normal MMR protein expression. Alternatively, a subset of FCCTX could harbour mutations in genes linked to urothelial cancer development, e.g. FGFR3, TP53 or HRAS [43]. To this point, 42/45 urothelial cancer in the FCCTX cohort developed in the urinary bladder, which stands in contrast to a predilection for tumours in the upper urinary tract in Lynch syndrome [17]. Unfortunately, it was not possible to discriminate between upper and lower urinary tract cancers in the population-based cohort since the Nordcan database does not differentiate between these sites.

The increased risk of gastric cancer with an IRR of 5.9 in the age group 30–49 years could potentially signify a genetic subset that confers heredity for this cancer type. About 10–20% of gastric cancer is caused by heredity with confirmed causes in 1–3%, predominantly linked to the hereditary diffuse gastric cancer caused by pathogenic variants in the CDH1 gene [27]. Only two of the 18 FCCTX families in our cohort in which gastric cancer developed fulfilled the criteria currently applied for genetic diagnostics due to early-onset cases in the families [4, 26]. When we excluded these families from the FCCTX cohort the risk was, as expected, reduced and was not significant. The co-occurrence of gastric and colorectal cancer in Amsterdam I positive families calls for further studies but may be explained by polygenetic defects resulting in a severe cancer phenotype in some FCCTX families. Addition of CDH1 in the genetic testing of FCCTX families might genetically classify a small fraction of the FCCTX families.

The increased risks of breast cancer with IRRs 1.5–1.7 in the age group 30–69 years and pancreatic cancer with an IRR of 2.2 after age 70 support the suggestion of disease-predisposing variants in BRCA2, causing the observed malignancies in a small subset of FCCTX families [10]. We also identified an IRR of 10.2 for early-onset ovarian cancer. Various guidelines exist for referring individuals to genetic diagnostics in hereditary breast and ovarian cancer [41]. In addition to fulfilling the Amsterdam I criteria, 35 out of the 213 FCCTX families also fulfilled the clinical genetic testing criteria for hereditary breast and ovarian cancer. Exclusion of these families from the FCCTX cohort decreased the risk of breast and pancreatic cancer to a nonsignificant level for the age groups 50–69 and 70+, and the risk for breast cancer flipped to a significantly decreased IRR when comparing to the population-based cohort for the age group 30–49 years (Table 3). Whether the increased risk of breast cancer and colorectal cancer in these families can be explained solely by BRCA2 germline mutations or by polygenetic defects or environmental factors remains to be elucidated. Though the families included in this study had not been systematically screened for hereditary breast and ovarian cancer, our data support a role for hereditary breast and pancreatic cancer in FCCTX, and application of broader diagnostic genetic panels that also cover the BRCA2 gene may, based on our data, have a potential to identify disease-predisposing mutations in some FCCTX families.

The five eye tumours identified in the Danish FCCTX cohort were all malignant melanomas. Uveal melanoma is predominantly sporadic but between 2 and 5% have been estimated to be caused by familial or hereditary predisposition. Autosomal dominant inheritance of pathogenic BAP1 gene variants have been observed in 47% of uveal melanomas, while pathogenic EIF1AX variants have been found in 14–20% of the cases [13, 15]. BAP1-associated uveal melanomas are diagnosed in the age of 30–59 years and are associated with cutaneous melanomas and renal cell carcinomas, while EIF1AX gene variants are associated with thyroid and ovarian cancer [13, 15]. In our cohort, uveal melanomas presented in the age span from 54 to 69 years and 2 of 5 cases occurred in patients with previous cutaneous melanomas. These data encourage awareness of family history during the genetic counselling and diagnostic testing.

Comparison between the risk of extra-colorectal cancer in the FCCTX cohort with the national Danish Lynch syndrome cohort, revealed differences as well as similarities. Urothelial cancer and skin cancer showed significantly lower risk levels in FCCTX compared to Lynch syndrome with IRRs of 0.06–0.31 and 0.11–0.24, respectively (Table 4). Increased risks and reminiscent risk profiles applied to breast cancer, gastric cancer and pancreatic cancer. These similarities are also supported by other studies on the risk of upper gastrointestinal cancer in Lynch syndrome [7, 18, 39]. Surveillance for cancer of the upper gastrointestinal tract is not recommended in FCCTX, but the increased risk observed may suggest awareness with consideration of Helicobacter Pyroli screening and eradication in FCCTX similarly to the recommendations in Lynch syndrome [16, 38, 39].

Studies have shown that compared to Lynch syndrome, FCCTX confers a lower risk for colorectal cancer (11–20% vs 58–75%), a higher age at onset (60 vs 45 years), a different predominant tumour locations (distal vs proximal) and a worse prognosis [3, 21, 23, 42]. In FCCTX, colonoscopic screening is generally recommended with 5-year intervals starting 5–10 years before the earliest colorectal cancer diagnosis in the family, though surveillance patterns are likely more variable in FCCTX than in Lynch syndrome. The recent demonstration of excess cancer-related deaths in FCCTX compared to Lynch syndrome and short intervals to second primary colorectal cancer suggests that clinical management in FCCTX needs to be optimized [3].