Molecular Landscape of ERBB2/ERBB3 Mutated Colorectal Cancer

Abstract

Background: Despite growing therapeutic relevance of ERBB2 amplifications in colorectal cancer (CRC), little is known about ERBB2/ERBB3 mutations. We aimed to characterize these subsets of CRC.

Methods: We performed a retrospective analysis of 419 CRC patients from MD Anderson (MDACC) and 619 patients from the Nurses' Health Study (NHS)/Health Professionals Follow-Up Study (HPFS) with tissue sequencing, clinicopathologic, mutational, and consensus molecular subtype (CMS) profiles of ERBB2/ERBB3 mutant patients. A third cohort of 1623 CRC patients with ctDNA assays characterized the ctDNA profile of ERBB2 mutants. All statistical tests were two-sided.

Results: ERBB2 mutations occurred in 4.1% (95% confidence interval [CI] = 2.4% to 6.4%), 5.8% (95% CI = 4.1% to 8.0%), and 5.1% (95% CI = 4.0% to 6.2%) of MDACC, NHS/HPFS, and ctDNA patients, respectively. ERBB3 mutations occurred in 5.7% (95% CI = 3.7% to 8.4%, 95% CI = 4.0% to 7.8%) of patients in both tissue cohorts. Age, stage, and tumor location were not associated with either mutation. Microsatellite instability (MSI) was associated with ERBB2 (odds ratio [OR] = 5.98, 95% CI = 2.47 to 14.49, P < .001; OR = 5.13, 95% CI = 2.38 to 11.05, P < .001) and ERBB3 mutations (OR = 3.48, 95% CI = 1.51 to 8.02, P = .002; OR = 3.40, 95% CI = 1.05 to 10.96, P = .03) in both tissue cohorts. Neither gene was associated with TP53, APC, KRAS, NRAS, or BRAF mutations in tissue. However, PIK3CA mutations were strongly associated with ERBB2 mutations in all three cohorts (OR = 3.68, 95% CI = 1.83 to 7.41, P = .001; OR = 2.25, 95% CI = 1.11 to 4.58, P = .02; OR = 2.11, 95% CI = 1.25 to 3.58, P = .004) and ERBB3 mutations in the MDACC cohort (OR = 13.26, 95% CI = 5.27 to 33.33, P < .001). ERBB2 (P = 0.08) and ERBB3 (P = .008) mutations were associated with CMS1 subtype. ERBB2 (hazard ratio [HR] = 1.82, 95% CI = 1.23 to 4.03, P = .009), but not ERBB3 (HR = 0.88, 95% CI = 0.45 to 1.73, P = .73), mutations were associated with worse overall survival.

Conclusions: MSI and PIK3CA mutations are associated with ERBB2/ERBB3 mutations. Co-occurring PIK3CA mutations may represent a second hit to oncogenic signaling that needs consideration when targeting ERBB2/ERBB3.

Figure 1.

Summary of protein coding alterations noted in ERBB2 and ERBB3 mutant colorectal cancer patients from the MD Anderson Cancer Center and the Nurses’ Health Study/Health Professionals Follow-Up Study. aa = amino acid.

Figure 2.

Molecular profile of ERBB2 and ERBB3 mutant colorectal cancer patients from the MD Anderson Cancer Center, the Nurses’ Health Study/Health Professionals Follow-Up Study, and a ctDNA cohort. A) Patients with tissue-based sequencing. B) Patients with ctDNA-based assays. MSI = microsatellite instability.

Figure 3.

Impact of ERBB2 and ERBB3 mutations on clinical outcomes in colorectal cancer. Overall survival was compared in patients with metastatic colorectal cancer and (A)ERBB2 or (B)ERBB3 mutations compared with wild-type patients, while disease-free survival was compared in early-stage colorectal cancer patients with (C)ERBB2 or (D)ERBB3 mutations compared with wild-type patients. Groups were compared with the log-rank test and Cox proportional hazards models. All P values are two-sided. CI = confidence interval; HR = hazard ratio.

Figure 4.

Box and whisker plot demonstrating the relative allele frequency of detected ERBB2 and ERBB3 mutations stratified by whether mutations are recurrent or confirmed to be functionally activating. Patients with tissue-based assays and detected (A)ERBB2 or (B)ERBB3 mutations are shown on the top row, while patients with ctDNA detected ERBB2 variants are shown in (C). Values represent the median, interquartile range, and range. Functional characterization is based on the literature review in Supplementary Tables 1 and 2 (available online). Groups were compared using the Mann-Whitney test, and P values are two-sided.