Using the OncoMap platform, we assayed for 385 known and clinically relevant somatic mutations in colorectal cancers from 427 Asian, Black and White patients to evaluate possible differences in the frequency of somatic mutations in colorectal cancer. In our study we found the mutation rate for cancers from Asians to be significantly lower than either the Black or White cohort. We found significant differences in the frequency of BRAF
mutations with BRAF
V600E mutations occurring most frequently in White patients. We find BRAF
mutations to be more common in cancers from White patients, women, older patients, and in the proximal colon consistent with other reports [23
]. In addition, while not significant using the more stringent criteria (Bonferroni correction), we found an enrichment of KRAS
mutations in the Black cohort.
While the differences in BRAF
mutation frequency between White, Black and Asian cohorts in colorectal cancer have not been previously reported, differences between other population groups have been noted. For example, in a recent study, Rozek et al [18
] found the BRAF
V600E mutation to be more common in persons of Ashkenazi Jewish descent and less common in those of Arab descent. Furthermore, English et al [24
] found Australians of Anglo-Celtic descent to have a higher incidence of colorectal cancer and significantly more BRAF
V600E mutations than those of Southern European descent. These findings demonstrate again that the frequency of important oncogenic mutations can vary in populations of different ancestral backgrounds. Whether and to what degree these variants have an impact on incidence, response to treatment or survival is yet to be determined.
The differences in the rates of BRAF
mutation may reflect differences in the underlying etiology of genomic instability across these populations. BRAF
mutations occur preferentially in the ascending or proximal colon in precursor lesions referred to as serrated adenomas. In the setting of BRAF
mutation, these adenomas generally progress to cancer through an acquired DNA mismatch repair deficit caused by hypermethylation of MLH1
in the setting of the CpG island methylator phenotype (CIMP) [25
]. Thus these tumors tend to be characterized by microsatellite instability (MSI) and have high rates of somatic mutations, especially small insertions and deletions at small repeat regions. KRAS
mutations, by contrast, occur more often in colorectal cancers that harbor chromosomal instability, the more common pattern of genomic aberrations in sporadic colorectal cancer.
The lower rates of BRAF mutant colorectal cancer seen in the Asian cohort could thus indicate that BRAF mutations themselves may be less common in this group. Alternatively, it is possible that there is a reduced propensity for pathologic DNA hypermethylation in the Asian population studied, thus reducing the ability of BRAF mutant precursor lesions to progress towards cancer. We did not have sufficient genomic material to pursue MSI testing or methylation testing for all specimens, and thus we were not able to directly measure if overall rates of CIMP-positive tumors were lower in the Asian population. However, for 62 of 83 Asian cancers assayed for MSI, only 3% (2/62) were MSI-high. This rate is lower than the normally observed rate (10-15%) in sporadic colorectal cancers, and consistent with a hypothesis that tumors driven by these mechanisms may be less common in the Asian population. Further work integrating more comprehensive genomic information with MSI assays and methylation status should clarify this.
The V600E mutation is an acquired activating mutation that results in constitutively active BRAF
kinase activity leading to activation of the MAPK
pathway. In melanoma, approximately 40% of patients have BRAF
mutations of which 69% are V600E. The V600E has also been found in benign nevi [26
], suggesting that V600 mutants may be early or precursor events. Similarly, BRAF
mutations have been found in colorectal adenomas suggesting an early event in tumorgenesis [27
]. Ogino et al [28
], looking at clinical trial results (CALGB 89803) from 1264 stage III colorectal cancer patients, clarified predictive and prognostic effects of BRAF
mutations and found BRAF
mutants to be associated with worse outcome. Although the frequency of BRAF
mutations is low (Asian, 4%; Black, 7%; White, 17%), compared to KRAS
(Asian, 39%; Black, 53%; White, 43%), BRAF
may be an important target for patients whose tumors harbor mutations.
Our finding of a lower mutation rate in our Asian cohort is of interest. The Asian cohort is comprised of significantly more lower stage patients and the cancers are predominantly distal (Asian, 40%; Black, 25%; White, 21%) rather than proximal. There does not appear to be an association between cancer location and mutation rate. If we look at mutation rates by cancer location we find that in the Asian cohort 48% of patients with distal cancers have no mutations compared to 19% of the Black and 22% of the White cohort. Of the Asian patients with distal cancers with mutations, there are 23 mutations in 17 patients (1.35 mutations per patient), less than either the Black or White cohorts at 1.63 and 1.69 mutations per patient, respectively. However, the average stage for distal, proximal and rectal cancers is about the same. Moreover, while this finding may truly represent a lower mutation rate for the Asian cohort, it also may be that the spectrum of mutations in this population is not well represented in OncoMap.
The rising incidence of colorectal cancer in many Asian countries in the last few decades is marked [10
] and incidence in Korea has now surpassed the rate for Asian Americans. Rates of mortality for Asian Americans, while considerably lower than either Blacks or Whites, have been steadily decreasing, likely due to the institution of screening as a standard medical test. While the drivers behind this complex picture are numerous, there is some evidence that shift from a more traditional Asian diet to a Western style diet containing more processed and red meats and alcohol [10
], may be a contributor.
The differences in the location of the cancers across our cohorts are striking and significant with White patients having more proximal cancers and Asians having more distal cancers. It has been previously observed that distal cancers are more common in the Korean [12
] population and we observe this difference in both our Korean cohort and our Asian American cohort. Looking at tumor stage, we find all groups to have stage III cancers most frequently but stage distribution is otherwise diverse. There were more stage II cancers in the Asian group and the fewest in the White group (Asian, 34%; Black, 18%; White, 11%), and more stage IV cancers in the White group and the least in the Asian (Asian, 5%; Black, 12%; White, 19%). Looking at the average stage for each group, we find Whites to have the highest average stage (Asian, 2.41; Black, 2.85; White, 3.09).
mutations are well known in numerous cancers, located most commonly at exon 2 amino acids G12 and G13, two adjacent amino acids located near the catalytic site, and have been shown to result in constitutive activation of the MAPK
signaling pathway. The prevalence of KRAS
mutations in the Black group is driven by the KRAS
G12D, however, the significance of KRAS
G12D enrichment is unclear. This observation has been previously reported by Sylvester et al [19
], however, they noted a higher frequency of G13 mutations in colon cancers from Black patients compared to the White, which we did not observe.
While there is some evidence that KRAS
exon 2 mutations may be associated with a worse prognosis [20
], and that G13 mutations may result in a less aggressive cancer, the evidence regarding the different effects of specific G12 or G13 variants is sparse. There are a considerable number of possible nucleotide changes at G12 and 13 resulting in different mutations, and each of these may produce altered downstream activity. A 2001 study by Andreyev et al [30
] found that in colorectal cancer, of all the codon 12 and 13 mutations, only GGT to GTT transversions, resulting in G12V, were associated with worse survival. In contrast, in pancreatic cancer, which has the highest incidence of KRAS
mutations of any cancer, 2 small studies [31
] showed evidence that patients with G12V (GTT) mutations have longer median survival and those with G12D (GAT) have shorter. A recent and intriguing finding by Garassino et al [33
], showed in NSCLC cell lines, that the G12D mutant had increased sensitivity to sorafenib (a MAP
Kinase pathway inhibitor) compared to G12V and G12C, suggesting that differences at the nucleotide level may have clinical implications.
However, Ogino et al [28
], in looking at KRAS
mutations in the CALGB 89803 trial, found no significant effect on survival or disease progression for KRAS
mutant colorectal cancer patients. Except for the predictive significance for efficacy of anti-EGFR
monoclonal antibodies in the setting of metastasis for EGFR
-expressing tumors [13
mutations appear to have no effect on outcome.
Colorectal cancer is generally considered to progress from adenoma to carcinoma, with APC
mutations occurring early [34
mutations in the normal epithelium and KRAS
somewhere along the transition from small to medium size adenoma [34
mutations are late events. We have 7 patients each with more than 3 mutations (), and when we look at these 7 samples, we can clearly see evidence of the generally accepted pathway of colorectal cancer progression: APC
, as the first event, is found in 5/7; KRAS
, a later event, in 6/7; and TP53
, the last step to a carcinoma, is found in half of the samples.
We found the mutation rates of oncogenes to be at the expected frequencies [16
]; however in tumor suppressor genes, our frequencies were considerably below what is expected. This is a limitation of the OncoMap platform in that it assays for known mutations and, while mutations in oncogenes tend to occur in the same locations, mutations in tumor suppressors occur more randomly across the gene. Next generation sequencing will be able to overcome most of the shortcomings inherent in OncoMap.
Genomic variations not included in OncoMap may yet explain some of the differences in our patient groups, for example, mutations in regions not covered by OncoMap, copy number or gene expression alterations, or epigenetic alterations. Further research should focus on an expanded set of genomic alterations, and integration of these data with MSI testing and/or assessment of CIMP in carefully controlled studies with extensive patient annotations where staging, diagnosis and treatment are equal.
In summary, we have identified differences in the somatic mutation frequency of known cancer genes in colorectal cancers from Asian, Black and White patients. These data argue in favor of using a genomics-driven precision medicine approach in order to elucidate the different ancestral determinants harbored by patient populations and thus, to more precisely and effectively treat colorectal cancer.