644P - Independent clinical validation of a gene expression profile to predict benefit of 5-FU in metastatic colorectal cancer

No biomarkers predict efficacy of 5-FU for metastatic colorectal cancer (mCRC). A gene expression based drug response predictor (DRP) had been shown to predict benefit of 5-FU in the PETACC-3 cohort [1]. The current study aimed to validate the biomarker in a cohort of mCRC patients. 100 mCRC patients treated with 5-FU and leucovorin (FLV) at Sahlgrenska University Hospital from 1996-2013 had mRNA isolated from pretreatment formalin-fixed paraffin embedded tumor tissue and run on Affymetrix HG-133_Plus_2 microarray. The DRP is based on correlations of in-vitro cytotoxicity to 5-FU and consequently correlations to mRNA expression of 3.500 tumors and evaluated blinded to outcome. Primary outcome was progression-free-survival (PFS) at 3 years after treatment with FLV. In addition to variables in the table, right vs left colon cancer vs rectal cancer, age, gender and surgery elective or acute were assessed in the multivariate model. Treatments in the metastatic setting included oxaliplatin and irinotecan from year 2000. To correct for possible selection bias for FLV-treatment, we corrected for treatment year 2000 in the analysis plan. The DRP was able to predict benefit of FLV in patients treated after 2000 (n=73) with a univariate hazard ratio (HR) of 0.60 (95% confidence interval (CI) 0.38-0.93, p=0.023) for a 50% difference of the DRP. For patients treated before 2000 results were not statistically significant. In the Cox model including all variables, a DRP score of 90 corresponds to a 50% likelihood of 9 months PFS and a score of 32 corresponds to a PFS of 6 months.Table644P Multivariate analysis of PFSTableVariableLevelHR95%-CIP-valueDRP, difference of 50Treatment before 20001.240.52-3.030.63Treatment after 20000.470.28-0.780.004Grade of differentiationMedium + high vs low0.480.29-0.800.005Mucinous vs low0.320.13-0.830.019Performance status2 vs 0-12.651.41-4.970.002 In patients treated with 5-FU after 2000 the DRP appears to predict benefit. There are no obvious reasons for the lack of prediction before 2000 but options include mRNA degradation, change in handling of samples, or the selection of older patients after 2000. [1] PLOS ONE; 11(5): e0155123. Oncology Venture. Oncology Venture. I.K. Buhl: Full / Part-time employment, Shareholder (immediate family): Oncology Venture. B. Gustavsson: Shareholder / Stockholder / Stock options: Isofol. I.J. Christensen: Advisory / Consultancy, Shareholder / Stockholder / Stock options: Oncology Venture. A. Hansen: Sh