Evaluation of NQO1 as a predictive biomarker and therapeutic target in metastatic colorectal cancer

Abstract

BackgroundA knowledge of the biology of colorectal liver metastases would be invaluable to inform clinical decision making; however, deriving this information from metastatic lesions is not feasible until after resection. We aimed to use proteomic analysis to establish the degree of biological similarity across disease sites and identify and validate biomarkers in the primary tumour that predict response to neoadjuvant chemotherapy in liver metastases. MethodsTissue from primary colorectal tumour and liver metastases (n=16) were subjected to proteomic analysis with isobaric tagging for relative quantification. Data were analysed with ProteinPilot software (Ab Sciex, Framingham, MA USA) with stratification of patients into low or high response to chemotherapy. These biomarkers were investigated by immunohistochemistry on a tissue microarray of 56 patients. Their therapeutic potential was investigated by dosing of SW480 cells with irinotecan or fluorouracil with or without inhibition by small interfering RNA (siRNA) or a competitive inhibitor (dicoumarol). FindingsWe identified 5766 discrete proteins, of which 2·54% were differentially expressed between primary and metastatic tumours. There were 170 potential response biomarkers in the primary tumour and 27 in the metastases. Lambda-crystallin homolog and NQO1 were common to both tissue types and showed consistent dysregulation. Immunostaining of NQO1 in metastatic tissue was lower in patients responding than in those not responding to chemotherapy (p=0·041), with a significant correlation between primary and metastatic disease sites (r=0·44, p=0·001). Knockdown of NQO1 with siRNA followed by treatment with irinotecan or fluorouracil reduced the IC50 from 100·1 to 49·8 μM and from 200·1 to 25·0 μM, respectively. Treatment of cells with dicoumarol before incubation in irinotecan or fluorouracil reduced the IC50 from 100·0 μM to 50·0 μM and from 183·7 μM to 49·9 μM, respectively. InterpretationWe show that proteomic sequencing of matched metastatic colorectal cancer samples is feasible with high coverage. The high degree of similarity between the primary and secondary tumours suggests that the primary tissue is predictive of the metastatic phenotype. NQO1 expression in the primary tumour predicts response to neoadjuvant chemotherapy in liver metastases, and inhibition of this protein at both genetic and functional levels improves chemosensitivity. FundingCancer Research UK.