PIK3CA Somatic Mutation Status in Relation to Patient and Tumor Factors in Racial/Ethnic Minorities with Colorectal Cancer

Phosphatidylinositol 3-kinase (PI3K) is a lipid kinase critical in the initiation of signaling pathways for cell proliferation, migration, and survival (1, 2). Mutations in the gene encoding the catalytic subunit of PI3K [i.e., the phosphatidylinositol-4,5 3-kinase, catalytic subunit alpha (PIK3CA) gene] can result in constitutive activation of PI3K signaling and, thus, dysregulated cell proliferation contributing to the development of cancer (3). In particular, somatic PIK3CA mutations have been noted in approximately 10% to 20% of colorectal cancers (2, 4–12). Studies characterizing the clinical profile of patients with PIK3CA-mutated colorectal cancer have suggested that tumors exhibiting these mutations are more likely to be located in the proximal colon (6, 9, 13) and to exhibit KRAS mutations (4–9, 13) in comparison with PIK3CA-wildtype colorectal tumors. Such biologic differences may translate to differences in the epidemiology of colorectal cancer according to PIK3CA mutation status.

Epidemiologic studies characterizing PIK3CA-mutated colorectal cancer are limited; however, multiple studies have noted that individuals with PIK3CA-mutated colorectal cancer, particularly those with tumors that are also KRAS-mutated, experience poorer survival than those with PIK3CA-wildtype disease (7, 13, 14). Other studies have reported an interaction between PIK3CA mutation status and aspirin use, such that aspirin use is associated with more favorable colorectal cancer survival in individuals with PIK3CA-mutated colorectal cancer (15). In another recent analysis, it was noted that the prevalence of PIK3CA somatic mutations was markedly higher among non-white adults with colorectal cancer than among white adults with colorectal cancer (13); however, few other studies have reported the distribution of PIK3CA somatic mutations in non-white populations (2, 16, 17). In light of the suggested poorer prognosis of PIK3CA-mutated colorectal cancer, any disproportionate burden of these somatic mutations in minorities with colorectal cancer could have serious implications for public health and for disparities in colorectal cancer survival.

We used data from the Colon Cancer Family Registry (C-CFR) to characterize the prevalence of PIK3CA somatic mutations in African American and Asian American adults with colorectal cancer, and to further evaluate the relationship of PIK3CA mutation status with other tumor characteristics, patient attributes, and colorectal cancer survival.

The study population included men and women diagnosed with incident invasive primary colorectal cancer who were identified through the population-based Surveillance Epidemiology and End Results (SEER) cancer registries serving the Seattle-Puget Sound region, Greater San Francisco Bay Area, or Hawaii, and who were enrolled into the C-CFR (18). The C-CFR is an international resource representing a collaboration between six study centers in Canada, the United States, and Australia; the present analysis was restricted to C-CFR minority case participants enrolled through the aforementioned SEER registries. Information on participant race/ethnicity was available from cancer registry records and was self-reported during study interviews. Recruitment protocols and eligibility criteria for these C-CFR study sites and others have been described elsewhere (18, 19).

Cases included in the present analysis were diagnosed with invasive colorectal cancer between 1997 and 2008, with ages at diagnosis ranging from 21 to 85 years (N = 385). Cases completed surveys within 5 years of diagnosis for the collection of risk factor information, including family history, demographic and anthropometric factors, medical history, smoking history, and use of NSAIDs (19). Most cases were interviewed within 2 years of diagnosis (85%).

Vital status and, as applicable, date of death were determined regularly via linkage to SEER and the National Death Index.

DNA was extracted from paraffin-embedded formalin-fixed diagnostic tumor tissue. For eligible cases with available extracted tumor DNA (n = 379), pyrosequencing was used to detect mutations in PIK3CA in three hotspots: codons 542 and 545 in exon 9 and codon 1047 in exon 20. These hotspots account for approximately 80% of all PIK3CA mutations (20, 21). Pyrosequencing was performed using the Pyromark Q96-MD and Q24 systems (Qiagen), with an optimized dispensation order to maximize the detection of known variants in the exon 9 and exon 20 hotspots. Cases for whom testing repeatedly failed or test results were equivocal for mutations in any of these regions were classified as having unknown PIK3CA mutation status (n = 2).

Extracted tumor DNA was also tested for the BRAF V600E mutation and deficiencies in DNA mismatch repair. Testing for the V600E (c.1799T>A) BRAF mutation was conducted using a fluorescent allele-specific PCR assay as described previously (22). Testing for DNA mismatch repair status was conducted in one of two ways: (i) a 10-marker panel for microsatellite instability (MSI) was evaluated in tumor DNA and in DNA extracted from normal surrounding tissue (18, 23), or (ii) expression of four DNA mismatch repair proteins was evaluated using immunohistochemistry (IHC; ref. 24). Tumors were classified as having deficient DNA mismatch repair (dMMR) if instability was observed in ≥30% of markers evaluated on the MSI panel or if at least one marker on the IHC panel was negative for protein expression. Tumors were classified as having proficient DNA mismatch repair (pMMR) if instability was observed in <30% of markers on the MSI panel or if all IHC markers were positive for protein expression. A high level of concordance between these two assays has previously been demonstrated (25).

Tumor site information was obtained from SEER. Tumors located in the cecum through the splenic flexure were classified as proximal colon cancers (ICD-O-3 codes C180, C182, C183, C184, and C185; ref. 26). Tumors in the descending (C186) and sigmoid colon (C187) were classified as distal colon cancer, and tumors in the rectosigmoid junction (C199) and rectum (C209) were classified as rectal cancer.

We compared the distribution of demographic factors, lifestyle factors, and tumor attributes across colorectal cancer case groups defined by PIK3CA mutation status (wildtype vs. mutated) and mutation site (exons 9 vs. 20) using χ² tests. Using a logistic regression model, we further evaluated differences in the presence and location of PIK3CA mutations according to race/ethnicity, with adjustment for age at colorectal cancer diagnosis, sex, study site (Seattle-Puget Sound/Hawaii/Greater Bay Area), and family history of colorectal cancer in first-degree relatives (yes/no); to account for differences in study site–specific case recruitment protocols, we also included an adjustment term for the interaction between study site and family history.

We used Cox proportional hazards regression to evaluate the association between PIK3CA-mutation status and overall survival after colorectal cancer diagnosis. The time axis for analysis was defined as days since colorectal cancer diagnosis with an outcome of death due to any cause. Participants were left-censored until the date of study enrollment. All regression models were adjusted for age at diagnosis and study site, with further adjustment for sex, race/ethnicity, family history of colorectal cancer, and interaction between family history and study site in multivariable models. In light of the previously described relationship between PIK3CA mutation status and aspirin use, we also conducted a separate analysis adjusting for self-reported prediagnostic aspirin use at study enrollment and analyses stratified according to aspirin use (defined as having ever used aspirin at least twice a week for more than one month; ref. 19). Proportional hazards assumptions were assessed by testing for a nonzero slope of the scaled Schoenfeld residuals on ranked failure times (27). All analyses were conducted in STATA SE version 13.1.

PIK3CA somatic mutations were identified in 11% of colorectal cancer cases (n = 42 of 377 cases), of whom 64% (n = 27) exhibited a mutation in exon 9 and 36% (n = 15) exhibited a mutation in exon 20. In total, 15 nucleotide changes were identified in PIK3CA exon 20 (codon 1047), and 43 nucleotide changes were identified in exon 9 (codons 542 and 545), such that 48% (N = 13) of cases with exon 9 mutations had multiple nucleotide changes in this region.

The prevalence of PIK3CA mutations differed by study site, with a significantly lower mutation prevalence among cases from the Hawaii C-CFR (4% vs. 16% and 13% for Seattle-Puget Sound and the Greater Bay Area, respectively; Table 1). Significant differences in mutation prevalence were also noted according to tumor site and MMR status: PIK3CA mutations were more common among those with proximal colon cancer (19%) than in those with distal colon (7%) or rectal cancers (5%), and more common among those with dMMR tumors (32%) than in those with pMMR tumors (9%; P = 0.01 and P < 0.001, respectively). PIK3CA mutation prevalence was also significantly higher among cases with Lynch syndrome (defined by the presence of a germline MMR mutation; 55%, P < 0.001); this difference in the prevalence of PIK3CA mutations according to Lynch syndrome persisted in analyses restricted to cases with dMMR tumors (23% vs. 55% prevalence in dMMR cases without vs. with Lynch syndrome, P = 0.06). Conversely, the mutation prevalence was lower among those reporting any family history of colorectal cancer in first-degree relatives as compared with those with no reported family history (6% vs. 14%, P = 0.03). There was no clear evidence for differences between those with PIK3CA-mutated and PIK3CA-wildtype colorectal cancer with respect to age at diagnosis, sex, race/ethnicity, or self-reported regular use of aspirin before diagnosis (P > 0.15). Similarly, there was no evidence of a significant difference in PIK3CA mutation status according to tumor grade (P = 0.24).

To account for differences in targeted recruitment practices between study sites, we further evaluated racial/ethnic differences in PIK3CA mutation status through multivariable-adjusted models. Case counts were limited, especially when those with PIK3CA mutations were stratified by exon; however, there was no indication of an association between race/ethnicity and PIK3CA mutation status among the included minority case groups (Table 2).

A total of 133 cases (35%) died during study follow-up, including 31% of cases with PIK3CA-mutated colorectal cancer (N = 13) and 36% of those with PIK3CA-wildtype disease. The median duration of follow-up was 5.6 years (interquartile range, 2.1–9.0 years). In adjusted analyses, there was no evidence of differences in overall survival by PIK3CA mutation status (HR, 0.77; 95% confidence interval, 0.43–1.39). This association was negligibly impacted by adjustment for regular aspirin use. Analyses stratified by aspirin use were limited by small numbers, but again indicated no evidence of an association between PIK3CA mutation status and overall survival after colorectal cancer diagnosis (Table 3).

In this cohort of racial/ethnic minorities with incident invasive colorectal cancer, we found the prevalence of somatic PIK3CA mutations among African American and Asian American individuals with colorectal cancer to be consistent with the 10% to 20% mutation prevalence reported in previous studies of non-Hispanic white individuals with colorectal cancer (4–9, 13). Somatic PIK3CA mutations were more common among those with proximal colorectal cancer and those with dMMR tumors. Although both tumor site and MMR status are associated with colorectal cancer survival, we found no evidence of an association between PIK3CA mutation status and survival after colorectal cancer diagnosis. Thus, despite the distinct molecular features of PIK3CA-mutated colorectal cancer, it is unlikely that differences in PIK3CA mutation status contribute to racial/ethnic disparities in colorectal cancer survival.

Few prior studies have evaluated the distribution of somatic PIK3CA mutations in racial/ethnic minorities (2, 16, 17, 28, 29). In our prior analysis of a nonoverlapping cohort, we noted a significantly higher prevalence of PIK3CA mutations among non-white versus white women with colorectal cancer (44% vs. 11%; ref. 13); however, that analysis was based on only 18 non-white colorectal cancer cases. To our knowledge, only one other study has separately evaluated the prevalence of PIK3CA mutation status in African American adults with colorectal cancer (17): Kang and colleagues reported no significant difference in the prevalence of PIK3CA somatic mutations between African American versus white colorectal cancer cases (14% vs. 11%, respectively), but did note a nonsignificant higher prevalence of exon 20 mutations (12.5% vs. 5.5%, respectively). We found that approximately 38% of PIK3CA mutations in African American cases were located in exon 20, which is comparable with previous estimates in non-Hispanic whites (13). Similarly, few studies have evaluated the prevalence of PIK3CA somatic mutation status in Asian American colorectal cancer patient populations; however, studies in Chinese and East Asian case groups yielded prevalence estimates similar to those noted in our analysis (7.5% to 12.3%; refs. 2, 16, 29). Thus, the overall prevalence of PIK3CA mutations does not appear to vary widely across these racial/ethnic groups. Other racial/ethnic groups (e.g., Hispanic populations) could not be evaluated in the present study due to small numbers, but may merit further study.

Although there is some inconsistency in the literature, several previous studies, primarily in non-Hispanic white case populations, have suggested that individuals with PIK3CA-mutated colorectal cancer experience a slightly poorer survival relative to those with PIK3CA-wildtype disease (5, 7, 13, 14, 30), with some indication that this association may be limited to individuals with KRAS-wildtype colorectal cancer (7, 13). Although we did not have complete information on KRAS mutation status, based on the magnitude of our observed point estimates for survival among colorectal cancer cases overall (i.e., regardless of KRAS mutation status), it is unlikely that poorer survival would have been noted in our study population even if analyses could have been restricted to KRAS-wildtype cases (i.e., 60%–70% of cases). Consistent with our findings, Ogino and colleagues recently reported that PIK3CA mutation status was not associated with colon cancer outcomes (31). Other recent analyses have reported an interaction between aspirin use and PIK3CA mutation status in relation to colorectal cancer survival such that aspirin use confers a survival benefit for those with PIK3CA-mutated but not PIK3CA-wildtype colorectal cancer (15, 32). Small numbers of PIK3CA-mutated cases precluded us from replicating this previous finding, as did a lack of information on postdiagnostic aspirin use; however, we found no evidence of an association between PIK3CA mutation status and survival regardless of self-reported prediagnostic aspirin use.

Although based on small numbers, we did note a significantly higher prevalence of somatic PIK3CA mutations among cases with Lynch syndrome (55%) relative to those without a germline MMR mutation (10%). To our knowledge, this pattern has not previously been reported, although other mutations in the EGFR-signaling pathway (e.g., BRAF, KRAS) are relatively rare among those with Lynch syndrome (33). Our finding of a higher PIK3CA mutation prevalence among Lynch syndrome colorectal cancer cases is consistent with previously reported associations between PIK3CA mutation status and MMR deficiency in colorectal cancer (4, 5, 9); however, even among individuals with dMMR colorectal cancer, we found the prevalence of PIK3CA mutations to be elevated in those with Lynch syndrome (55% vs. 23% in those with sporadic dMMR colorectal cancer). These findings merit further evaluation in larger study populations. If replicated, the observed high prevalence of PIK3CA mutations reinforces our understanding that Lynch syndrome–associated colorectal cancer has a distinct molecular profile.

Findings from this analysis should be considered in the context of study limitations. Several analyses and stratified comparisons were limited by small numbers. In particular, although previous studies have suggested a difference in the pathology of colorectal cancer with exon 9 versus exon 20 somatic mutations, exon 20 mutations were observed in only 15 individuals and exon 9 mutations in only 27 individuals; thus, we were limited in our ability to evaluate differences in the exon distribution of PIK3CA mutations. Small numbers also precluded more detailed stratification of racial/ethnic categories and analyses of racial/ethnic differences in the association between PIK3CA mutation status and survival. In addition, differences in case sampling strategies may affect the interpretation and generalizability of our findings. Specifically, in contrast with Seattle-Puget Sound and Greater Bay Area C-CFR sites, the Hawaii study site limited its recruitment to individuals with a family history of colorectal cancer. This may explain the lower prevalence of PIK3CA mutations observed among cases from the Hawaii study site, and could have affected findings from analyses of all sites combined; however, we did adjust for study site, family history, and an interaction term between study site and family history in our analytic models. We also replicated primary analyses excluding cases from Hawaii and noted no significant difference in observed results.

Despite the aforementioned limitations, the present analysis offers several important strengths. To our knowledge, this analysis marks the most comprehensive evaluation to date of PIK3CA mutation status among racial/ethnic minorities in the United States. Despite the limitations of small sample size, the detailed data collection of the C-CFR made it possible for us to evaluate differences in the distribution of PIK3CA mutation status and mutation location according to family history of colorectal cancer, Lynch syndrome status, other tumor attributes, and colorectal cancer survival, as well as race/ethnicity. Centralized testing for PIK3CA using pyrosequencing ensured high standards of quality control. Lastly, despite the selection of cases from the Hawaii site on the basis of family history, the use of population-based cancer registries for the identification of study cases further supports the generalizability of study findings.

In conclusion, our findings indicate that the prevalence of PIK3CA mutation status in African American and Asian American individuals with colorectal cancer is similar to that previously reported among non-Hispanic white individuals with colorectal cancer, but may differ according to other relevant etiologic and clinicopathologic factors, including Lynch syndrome status, tumor site, and tumor MMR status. These findings underscore the importance of PIK3CA mutation status in colorectal cancer epidemiology.

D.J. Ahnen is a consultant/advisory board member for EXACT Sciences Inc. and Cancer Prevention Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

The content of this article does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the Cancer Family Registry (C-CFR), nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government or the CFR.

Conception and design: A.I. Phipps, D.J. Ahnen

Development of methodology: A.I. Phipps, D.J. Ahnen, T. Burnett

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.J. Ahnen, I. Cheng, P.A. Newcomb, A.K. Win, T. Burnett

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.I. Phipps, P.A. Newcomb, A.K. Win

Writing, review, and/or revision of the manuscript: A.I. Phipps, D.J. Ahnen, I. Cheng, P.A. Newcomb, A.K. Win, T. Burnett

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.I. Phipps, P.A. Newcomb

Study supervision: A.I. Phipps, D.J. Ahnen

This work was supported by grant UM1 CA167551 (to P.A. Newcomb, A.K. Win, T. Burnett, and D.J. Ahnen) from the National Cancer Institute, National Institutes of Health and through cooperative agreements with members of the C-CFR and Principal Investigators. Collaborating centers include Seattle Colorectal Cancer Family Registry (U01/U24 CA074794, to P.A. Newcomb) and the University of Hawaii Family Registry of Colon Cancer (U01/U24 CA074806, to T. Burnett). This research was also supported by National Cancer Institute grants K07CA172298 (to A.I. Phipps) and K05CA152715 (to P.A. Newcomb).

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.