Increased Cancer Predisposition in Family Members of Colorectal Cancer Patients Harboring the p.V600E BRAF Mutation: a Population-Based Study

Colorectal cancer (CRC) is a heterogenous disease that can arise through several different molecular pathways. The two most common are the chromosomal-instability and microsatellite instability (MSI) pathways, both of which promote tumorigenesis by causing genomic instability. Chromosomal instability is observed in ∼85% of colorectal tumors and is characterized by aneuploidy and mutations of APC, KRAS, and p53. The MSI pathway occurs in tumors with DNA mismatch repair (MMR) deficiency caused either by germline MMR gene mutations (Lynch Syndrome) or by inactivation of MLH1 by aberrant promoter methylation (1).

More recently, the serrated pathway of CRC carcinogenesis has emerged as a novel molecular pathway, which is characterized by a distinct pathologic, clinical, and molecular phenotype for which there is evidence of a genetic predisposition (2). The serrated pathway is tightly associated with the p.V600E BRAF mutation and with aberrant DNA methylation (3, 4), which is commonly called the CpG island methylator phenotype (5). The p.V600E BRAF mutation is a T-to-A transversion at nucleotide 1796 that causes constitutive activation, and which promotes proliferation and inhibits apoptosis through the Ras/Raf/MEK/MAPK signal transduction pathway (6). The mutation is reported in approximately 10% to 18% of all colorectal tumors and in 30% to 50% that are microsatellite unstable (MSI-H; refs. 7-9). This mutation is acknowledged as an early and primary genetic event in the serrated pathway of colorectal carcinogenesis (6, 10).

The precursor lesions of the serrated pathway include sessile serrated adenomas, traditional serrated adenomas, as well as other lesions (11). Sessile serrated adenomas are relatively common colonic neoplasms (9% of all colonic neoplasms) that are strongly associated with BRAF, CpG island methylator phenotype, female sex, and proximal tumor location (12-14). Colorectal tumors expressing the BRAF mutation are associated with MSI-H, CpG island methylator phenotype, MLH1 promoter methylation, poor differentiation, mucinous morphology, and infiltrating lymphocytes (10, 15). However, it seems that the inactivation of MLH1 is not critical for carcinogenesis in the serrated pathway because 5% to 10% of microsatellite-stable (MSS) tumors also display BRAF (7-9). Furthermore, CRC tumors with the BRAF mutation rarely have KRAS mutations and do not occur in the context of Lynch syndrome, which supports the notion that the serrated pathway represents a distinct molecular pathway of carcinogenesis (3). In addition to the serrated pathway, the BRAF mutation has been recognized as a marker for poor outcomes (9, 16) and resistance to anti–epidermal growth factor receptor therapies (17). Inhibition of the mutated BRAF gene represents a novel and attractive therapy for patients with BRAF^Mut CRC, and several clinical trials are under way to evaluate this approach (18).

There is evidence for a genetic predisposition to tumorigenesis through the serrated pathway (2, 12). In addition, there is evidence to suggest that CRC patients with the somatic BRAF mutation have a stronger family history of cancer compared with patients with BRAF wild-type tumors (9, 19, 20). However, the evidence is limited, and patients who have MSI-H BRAF mutation tumors are generally regarded as sporadic cases (11).

As only one population-based study (9) has examined the familial basis of BRAF mutation CRC, we examined the clinicopathologic features and cancer family histories of CRC patients according to the MSI and BRAF status of their tumors, using a series of population-based CRC patients.

We identified 1,173 incident CRC patients (ICD-9 code; colon: 153.0-153.9, excluding 153.5; rectum: 154.0-154.1) from the Newfoundland Cancer Registry that were diagnosed before the age of 75 years, in the 5-year period between January 1, 1999 and December 31, 2003. Of these, 750 (64%) patients (or their proxies) consented to take part in the study. Index patients were invited to complete a family history questionnaire, which enabled pedigrees to be constructed. From the family history questionnaire, we recorded the type of cancer, age at diagnosis, and age at death or age at last follow-up for all first-degree relatives (FDR). Cancer diagnoses in consenting family members were confirmed by medical records whenever possible. Five hundred fifty-two patients (47% of all eligible patients) completed a family history questionnaire and all of the molecular analyses. Ethics approval was obtained from the Human Investigation Committee of Memorial University.

A positive family history of CRC was defined as having at least one FDR affected by CRC at any age. We identified patients with a strong CRC family history if their family history fulfilled the Amsterdam 1 criteria (21). Additionally, we defined patients with a strong family history that had a later age of onset if they met the Amsterdam 1 criteria, except those that did not have a CRC for <50 years. We refer to these families as CRC-Triad.

Tumors from index patients underwent pathologic review, molecular analysis for MSI, immunohistochemistry, MLH1 promoter methylation, and the p.V600E BRAF mutation. DNA from patients whose tumors had MMR deficiency, as indicated by MSI or immunohistochemistry, underwent MMR gene sequencing to identify mutations in MLH1, MSH2, MSH6, and PMS2. Patients with a personal or family history of polyposis were tested for APC mutations. DNA from most patients (97.6% of those that provided a blood sample) was screened for MUTYH mutations. Six patients with a germline mutation in either APC or MUTYH were excluded from this study.

The protocol for MSI analysis and immunohistochemistry staining of the MMR proteins (MLH1, MSH2, MSH6, and PMS2) in colorectal tumors has previously been described in detail (22). Briefly, formalin-fixed, paraffin-embedded tissues were sectioned, deparaffinized, and rehydrated using xylene and alcohol. The slides underwent microwave antigen retrieval followed by incubation with the appropriate monoclonal antibodies.

For MSI analyses, DNA was extracted from thick sections cut from paraffin-embedded tissue blocks that were determined by histology to contain predominantly tumor tissue. Five microsatellite markers were used for the analysis: BAT25, BAT26, D2S123, D5S346, and D17S250. MSI was detected by the presence of additional bands in the PCR-amplified product from the tumor tissue, compared with those from DNA from matched normal colon tissue. For some samples, MSI status was determined using the five mononucleotide markers—BAT-25, BAT-26, NR-21, NR-24, and MONO-27—supplied in the MSI Analysis System kit Version 1.1 (Promega Corp.), following the manufacturer's instructions. MSI status was assigned as MSI-high (MSI-H, >30% of markers tested unstable), MSI-low (MSI-L, >0% and <30% of markers unstable), or MSS (<30% markers unstable); however, because only 25 tumors were identified as MSI-L, we combined them with the MSS tumors and thus assigned tumors as either MSI-H (>30% of markers unstable) or MSS (<30% of markers unstable).

Methylation was detected using the MS-MLPA kit ME001B (MRC-Holland). This kit can generate two methylation-dependent signals from the MLH1 promoter: a 166-bp fragment is produced if the Hha1I site at position −7 (relative to the ATG start codon) is methylated, and a 463-bp fragment is generated if the Hha1I sites at −378 and −401 are both methylated. We scored the tumor DNA sample as methylated if either of these fragments was present at a normalized ratio of 0.15 of the peak area of the sample before digestion with Hha1I (23). To identify the c.1799T>A (p.Val600Glu) variant in BRAF, we used a protocol previously described (24). This is an allele-specific PCR assay that includes a set of primers for GAPDH as an internal positive control.

Alterations of MLH1, MSH2, and MSH6 were determined by sequencing all 45 exons and intron/exon boundaries. Genomic DNA from probands who had tumors deficient in an MMR protein was sequenced, except those cases with tumors that were deficient due to MLH1 promoter methylation. Automated sequencing was done on an ABI 3700 DNA Analyzer (Applied Biosystems). Sequence information of the coding region was derived from RefSeq NM_000249.2 (MLH1), NM_000251.1 (MSH2), and NM_000179.1 (MSH6). Primer sequences and intronic nucleotide information were derived from genomic sequences from the National Center for Biotechnology Information—AC011816.17 (MLH1), AC079775.6 (MSH2), and AC006509.15 (MSH6). Primer sequences are available from the authors upon request. The PMS2 (RefSeq NM_ 000535.4) variants were detected as previously described (25), with the modifications explained in Clendenning et al. (26).

Exon deletions and duplications in MSH2 and MLH1 were detected by multiplex ligation-dependent probe amplification (MLPA; ref. 27) in DNA from patients whose tumors were deficient in either MLH1 (unrelated to MLH1 methylation) or MSH2 using kit SALSA P003 according to the protocol provided by MRC-Holland. All rearrangements identified by MLPA were confirmed in other affected relatives by MLPA and, when possible, cDNA analyses.

One representative tumor slide from each patient's tumor was reviewed and scored for several histologic features, including Crohn's-like lymphocytic reaction and tumor-infiltrating lymphocytes (TILs), as previously described (28). Tumor grade and histology were determined from the original pathology reports. Tumor location was obtained from the Newfoundland Cancer Registry. Proximal location was defined as proximal to the splenic flexure. Mucinous component was defined as the presence of any mucin dissecting into stroma surrounding a tumor gland. This definition includes tumors with a mucinous histology, but also those with histologic heterogeneity, in which any area of the tumor displays dissecting mucin. We defined an increase in stromal plasma cells to be positive when plasma cells made up >25% of stromal immune cells. Pathology was reviewed for all available tumors, which included all MSI-H tumors and 98% of MSS tumors.

Comparison of continuous variables was analyzed with independent sample t test or one-way ANOVA. Categorical data were analyzed with the Pearson's χ² test. Multinomial logistic regression analysis was used to compare the clinicopathologic features of patients with BRAF mutation tumors to patients with BRAF wild-type tumors. The analysis includes a univariate model and a multivariate model, which includes all variables as well as age at diagnosis of CRC as a covariate. A Cox regression model was used to estimate hazard ratios (HR) for developing cancer in FDRs according to the molecular classification of the index patient. Index patients were excluded from the estimates of cancer risk. Age at diagnosis or age at last follow-up was used as the time variable. HRs are adjusted for the age and sex of the index patient, and the sex of the FDR was entered as a strata variable. A univariate Cox regression analysis was used to estimate the risk of developing CRC in FDRs according to the sex and type of FDR. The cumulative lifetime risk (<80 y of age) of developing cancer in FDRs was estimated with Kaplan-Meier survival analysis. All P values were two sided, and P ≤ 0.05 was considered significant. All analyses were done with the SPSS software package, version 17.0.

In 552 population-based CRC patients, we identified the p.V600E BRAF mutation in 12% (n = 65) of all patients' tumors and MSI-H in 11% (n = 61) of all patients' tumors. Nearly half of all MSI-H tumors harbored the p.V600E BRAF mutation (n = 27, 44%) and the remaining MSI-H BRAF^Wt tumors either harbored a pathogenic MMR gene mutation (n = 17) or were of unknown etiology (n = 17).

The majority of patients' tumors were MSS BRAF^Wt (n = 453, 82%). These patients were predominantly male (36% female) with distally located tumors (33% proximal; Table 1). The p.V600E BRAF mutation was identified in 8% (n = 38) of 491 index patients with a MSS tumor. In contrast to patients with MSS BRAF^Wt tumors, MSS BRAF^Mut patients were predominantly female (63%), with proximally located tumors (81%). Similarly, patients with a MSI-H BRAF^Mut tumor were predominantly female (67%), with proximally located tumors (96%); however, in addition, they had a high incidence of multiple tumors (25%).

Seventeen patients with a MSI-H tumor were found to have a germline mutation in MLH1, MSH2, MSH6, or PMS2 (3.1% of the entire cohort). These patients are characterized by having multiple tumors (41%), early age onset CRC (mean age, 51. 1 y), and a family history of CRC (77%).

A multinomial logistical regression analysis was conducted to investigate the clinicopathologic features of patients according to the MSI and BRAF mutation status of their tumors' (Table 2). Patients with MSS BRAF^Wt tumors were assigned as the reference category. Patients with MSS BRAF^Mut tumors are significantly more likely to be female [odds ratio (OR) = 3.1; 95% confidence interval (95% CI), 1.4 -6.8] and have proximally located tumors (OR = 6.3; 95% CI, 2.6-15.1), which have a mucinous component (OR = 2.9; 95% CI, 1.1-7.4). Similarly, patients with MSI-H BRAF^Mut tumors are significantly more likely to have proximally located (OR = 27.5; 95% CI, 3.5-213), mucinous tumors (OR = 5.8; 95% CI, 1.8-18.4), which presented with TILs (OR = 5.2; 95% CI, 1.8-15.5). However, the association with the female sex did not reach statistical significance (OR = 2.7; 95% CI, 0.9-2.1). The tumors of patients with Lynch syndrome are characterized by having TILs (OR = 5.0; 95% CI, 0.9-28.6) and a mucinous component (OR = 3.9; 95% CI, 1.0-14.5); however, the association with TILs is not statistically significant.

Approximately 5% (n = 2) of patients with a MSS BRAF^Mut tumor reported a strong family history by meeting the Amsterdam 1 criteria; however, none fulfilled the definition of a CRC-Triad. In patients with a MSI-H BRAF^Mut tumor, a family history that satisfied the CRC-Triad was reported by 22% (n = 6) of patients, and no patients satisfied the Amsterdam I criteria.

We estimated the risk of developing cancer in FDRs according to the molecular classification of the index patients' tumor with a Cox regression model. The hazard of developing CRC in FDRs of patients with a BRAF^Mut tumor was significantly greater compared with the FDRs of index patients with a MSS BRAF^Wt tumor (HR = 1.93; 95% CI, 1.35-2.75; Table 3; Fig. 1). Furthermore, the lifetime risk of developing CRC was significantly greater in the FDRs of patients with a BRAF^Mut tumor compared with the FDRs of patients with MSS BRAF^Wt tumors (lifetime risk (%) = 19% versus 11%; Log-rank, P < 0.001). The risk of CRC in FDRs of patients with Lynch syndrome was significantly greater compared with all other molecular classifications.

We investigated the risk of developing other common malignancies in FDRs according to the molecular classification of the index patients' tumor (Table 4). The risk of nonmelanoma skin cancer was significantly elevated in the FDRs of patients with a MSS BRAF^Mut tumor (HR = 2.61; 95% CI, 1.15-5.92) and MSI-H BRAF^Mut tumor (HR = 2.81; 95% CI, 1.07-7.41) compared with the FDRs of patients with MSS BRAF^Wt tumors. The incidence of other extracolonic tumors did not differ between FDRs of patients with or without the p.V600E BRAF mutation.

To evaluate the inheritance pattern of CRC in FDRs, we compared the incidence of CRC between parents and siblings according to the molecular classification of the index patients' tumor (Table 5; Fig. 1). In the FDRs of index patients with a MSS BRAF^Mut tumor, the risk of CRC was significantly greater in siblings compared with parents (HR = 3.28; 95% CI, 1.09-9.89). This was not observed in the FDRs of patients with MSI-H BRAF^Mut tumors (HR = 1.23; 95% CI, 0.65-3.06).

In this population-based study, we found evidence to suggest that FDRs of CRC patients with BRAF^Mut tumors have a significantly elevated predisposition to develop CRC and nonmelanoma skin cancer compared with patients with BRAF^Wt tumors. The risk of developing CRC was significantly greater in FDRs of index patients with BRAF^Mut tumors compared with index patients with MSS BRAF^Wt tumors (HR = 1.93; 95% CI, 1.35-2.75). Furthermore, the risk was significantly elevated in the family members of index patients with either a MSI-H BRAF^Mut or MSS BRAF^Mut tumor. We also observed a significant increase in the risk of nonmelanoma skin cancer in the FDRs of patients with BRAF^Mut tumors, compared with the FDRs of patients with BRAF^Wt tumors. In addition, this analysis provides further evidence that CRC patients with the somatic p.V600E BRAF mutation are characterized by a distinct clinical, molecular, and pathologic phenotype.

The p.V600E BRAF mutation is recognized as a marker of the serrated pathway (29), and previous studies (7-9) have reported that approximately 10% to 18% of all CRCs and up to 50% of MSI-H tumors harbor the BRAF mutation. In our study, we observed the BRAF mutation in 12% of all tumors and in 47% of MSI-H tumors, which is comparable with previous reports (7-9). The increased incidence of cancer in FDRs of patients harboring the p.V600E BRAF mutation may be explained by an inherited predisposition to develop cancer through the serrated pathway of carcinogenesis. Evidence for an inherited predisposition to develop CRC through the serrated pathway is linked to two CRC predisposition syndromes, namely hyperplastic polyposis and serrated pathway syndrome (30). The latter was described by Young et al. (19), who provided evidence for an autosomal dominant predisposition in several families, which develop advanced serrated lesions and MSI-variable CRC that were associated with the BRAF mutation, female sex, and right-sided tumors. Interestingly, in our study, some patients with BRAF^Mut tumors exhibited a family history that is consistent with autosomal dominant disease. Six index patients with a MSI-H BRAF^Mut tumor fulfilled the definition of a CRC-Triad, and two patients with MSS BRAF^Mut tumors reported a family history that fulfills the familial colorectal type X (FCCTX) criteria (31). Further evidence for a genetic predisposition to the serrated pathway comes from a recent study (12), which reported that patients with sessile serrated adenomas, compared with patients with other colonic lesions, were more likely to have a family history of CRC (42% versus 25%; P > 0.05) and a greater polyp burden (P < 0.001).

A small number of studies (9, 19, 20) have investigated the association between CRC patients with BRAF^Mut tumors and a family history of cancer, only one of which used a population-based approach (9). A correlation between a family history of CRC and patients with MSI-H BRAF^Mut tumors has been previously reported (20). However, that study (20) used a selected cohort of familial colorectal patients, and only eight MSI-H tumors were evaluated. Samowitz et al. (9) used a large number of unselected population-based CRC patients and reported a significant association between patients with MSS BRAF^Mut tumors and a family history of CRC (OR = 4.23; 95% CI, 1.65-10.84). However, no association was found among patients with MSI-H BRAF^Mut tumors (OR = 0.64; 95% CI, 0.18-2.19). To our knowledge, our study is the first to report an association between patients with MSI-H BRAF^Mut tumors and a family history of CRC, using an unselected series of population-based CRC patients. In our study, the risk of developing CRC was significantly elevated in the FDRs of index patients with either MSS BRAF^Mut or MSI-H BRAF^Mut tumors.

The incidence of Lynch syndrome in this population (3.1%) is consistent with estimates that attribute 3% to 5% of the total CRC burden to Lynch syndrome (32). Lynch syndrome is a highly penetrant autosomal dominant condition that is characterized by early-age, right-sided CRC, as well as various other extracolonic malignancies (33). Our findings are consistent with studies reporting that MMR mutations are highly penetrant (34) because the FDRs of Lynch syndrome patients in this study were affected at an early age, and ∼50% were diagnosed with CRC by age 80 years.

In FDRs of patients with MSS BRAF^Mut tumors, the risk of CRC was found to be significantly greater in siblings compared with parents, whereas in FDRs of patients with MSI-H BRAF^Mut tumors, the risk was similar for parents and siblings. These results suggest that the genetic factors that cause BRAF^Mut CRC predisposition may be recessively inherited in families with MSS BRAF^Mut CRC. In contrast, the incidence of CRC in families with MSI-H BRAF^Mut CRC is more suggestive of dominant or multifactorial disease. It has been postulated that the burden of CRC arising from the serrated pathway could be explained by the presence of common codominant or recessive alleles causing defective epigenetic regulation (30). In a codominant model of inheritance, an inherited predisposition to develop sessile serrated adenomas and CRC may be attributable to carriers of one allele, and the more severe phenotype associated with hyperplastic polyposis may be the result of homozygous carriers or a carrier paired with a recessive allele. Studies (35, 36) have shown that CRC risk is significantly greater among siblings than for parent-offspring, which is suggestive of recessive inheritance. Interestingly, in a large population-based study (35), the greatest CRC risk was associated with an affected sibling who had a right-sided colon cancer. Our results would suggest that MSS BRAF^Mut CRC and the serrated pathway could account for some of the excess CRC risk observed in the siblings of patients with right-sided CRC.

In addition to CRC, we observed an increase in the incidence of nonmelanoma skin cancer in the FDRs of patients with BRAF^Mut CRC. An association between CRC and skin cancer has recently been observed in a prospective cohort study (37), which reported a 2-fold increase in the risk of CRC following a diagnosis of nonmelanoma skin cancer. Risk factors for nonmelanoma skin cancer and BRAF^Mut CRC include environmental exposures such as UV radiation and smoking (38). Furthermore, susceptibility to both nonmelanoma skin cancer and BRAF^Mut CRC has been shown (39, 40) to be modified by the interaction between environmental exposures and polymorphisms in the base excision repair genes XRCC1 and OGG1, respectively. A recent case-control study (40) reported that ever-smokers homozygous for the OGG1 (S326C) polymorphism were twice as likely to have the BRAF mutation CRC. Polymorphisms of various DNA repair genes have been linked (41) to cancer susceptibility, and for some cancer types, the risk is exacerbated in individuals who are exposed to environmental carcinogens such as smoking. The association observed in this study might be explained by a germline variant that would increase susceptibility to both cancer types by attenuating DNA repair capability.

Relative to patients with BRAF^Wt tumors, those harboring the p.V600E BRAF mutation were characterized by females with poorly differentiated, proximally located tumors, which had a mucin component. Furthermore, patients with MSI-H BRAF^Mut tumors were older at diagnosis, more likely to have synchronous or metachronous tumors, and to have tumors that presented with increased stromal plasma cells and TILs, compared with patients with MSS BRAF^Mut tumors. These findings support the notion that BRAF^Mut CRC represents a distinct clinical, molecular, and pathologic phenotype (9, 10, 15).

There are several limitations to this study that should be acknowledged. Although we ascertained population-based patients, only those diagnosed before 75 years of age were included, which may have enriched the study cohort with patients having a greater genetic predisposition. The statistical power to detect a difference in clinicopathologic features and cancer risk may have been limited due to the small number of patients with MSI-H and BRAF^Mut tumors, and by the small number of cancer events in FDRs. We have made numerous comparisons and it is possible that some of the findings may be due to chance. Although we made every effort to verify all cancers reported in FDRs, family history was self-reported and therefore susceptible to recall bias. However, family history information was obtained before knowledge of the molecular characteristics of the index patients' tumors and, therefore, should not have systematically biased the results.

This analysis provides further evidence that patients with BRAF^Mut CRC have a distinct clinical, molecular, and pathologic phenotype. However, in contrast to the notion that these patients represent sporadic cases, we observed a significant association with a family history of cancer. Approximately 12% of all index patients were found to harbor the BRAF mutation, and the risk of CRC in FDRs of these patients was significantly elevated. In addition, we observed that the CRC risk to siblings and parents of patients with BRAF^Mut CRC was dependent on the MSI status of the patients' tumor. Furthermore, we observed an association with nonmelanoma skin cancer in FDRs of patients with BRAF^Mut CRC. Additional large population-based analyses are needed to confirm these findings, which could have clinically relevant implications for cancer screening.

No potential conflicts of interest were disclosed.

Grant Support: Canadian Institute of Health Research (#FRN-79845 and CRT-43821) for the “CIHR Team for Interdisciplinary Research on Colorectal Cancer” at Memorial University, Newfoundland, and the Samuel Research Institute, Toronto. National Genome Canada funded Atlantic Medical Genetics and Genomics Initiative (2006-2010). Cancer Institute of Canada grants 18223 and 18226.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.