Gene signatures associated with drug resistance to irinotecan and oxaliplatin predict a poor prognosis in patients with colorectal cancer

Abstract

The identification of novel survival predictors may help to improve the appropriate management of colorectal cancer (CRC). In the present study, two gene sets associated with irinotecan or oxaliplatin resistance in CRC cell lines were first identified and subsequently applied to the clinical CRC microarray dataset GSE14333. Subsequently, a 60-gene irinotecan resistance-associated signature and a 13-gene oxaliplatin resistance-associated signature were established, which were able to classify CRC patients into high- and low-risk subgroups with varied clinical outcomes [irinotecan-resistance gene signature: hazard ratio (HR)=0.4607, 95% confidence interval (CI)=0.3369-0.6300, P<0.0001; oxaliplatin-resistance gene signature: HR=0.6119, 95% CI=0.4547-0.8233, P=0.0008]. The performance of these two gene expression signatures in predicting outcome risk were also validated in two other independent CRC gene expression microarray datasets, GSE17536 (irinotecan-resistance gene signature: HR=0.5318, 95% CI=0.3359-0.8419, P=0.0079; oxaliplatin-resistance gene signature: HR=0.5383, 95% CI=0.3400-0.8521, P=0.0114) and GSE17537 (irinotecan-resistance gene signature: HR=0.2827, 95% CI=0.1173-0.6813, P=0.0088; oxaliplatin-resistance gene signature: HR=0.2378, 95% CI=0.09773-0.5784, P=0.0023). Furthermore, the combination of these two gene classifiers demonstrated a superior performance in CRC prognosis prediction than either used individually. Therefore, this study proposed novel gene classifier models for CRC prognosis prediction, which may be potentially useful to inform treatment decisions for patients with CRC in clinical settings.

Keywords: chemotherapy resistance; colorectal cancer; gene signature; microarray; prognosis.

Figure 1.

Kaplan-Meier curves of the low-risk and high-risk subgroups defined by the irinotecan-resistance and oxaliplatin-resistance gene signatures. (A) Irinotecan-resistance signature. (B) Oxaliplatin-resistance signature. The GSE17536 dataset's (C) irinotecan- and (D) oxaliplatin-resistance signatures. The GSE17537 dataset's (E) irinotecan- and (F) oxaliplatin-resistance signatures. These two gene signatures were able to stratify patients with colorectal cancer into low- and high-risk subgroups with significant differences in terms of the prognosis in the GSE14333 training dataset.

Figure 2.

Unsupervised hierarchical clustering of the colorectal cancer samples from the GSE14333 dataset using drug resistance-associated gene signatures. The (A) irinotecan-resistance and (B) oxaliplatin-resistance gene signatures were able to hierarchically cluster the samples into two subgroups with varied distributions of outcome risk.

Figure 3.

Kaplan-Meier curves based on the combined drug resistance gene signatures for patients with colorectal cancer in the three cohorts. In all three datasets, (A) GSE14333, (B) GSE17536 and (C) GSE17537, patients predicted as high-risk by the two gene signatures had the poorest clinical outcomes.