O3-14-6 - Exploratory Analysis of a Prognosis Predictive Model for Metastatic Colorectal Cancer Treated with Chemotherapy

Abstract

Background: Progression free survival (PFS) has been recognized as a surrogate outcome in the therapy for metastatic or recurrent colorectal cancer (mCRC). As molecular target therapies have become to be widely used, new surrogate outcome such as early tumor shrinkage (ETS) and deepness of response (DpR) are proposed. However, their validity has not been fully proven. We have tried to develop a new prognosis predictive model consisting of the three factors, DpR, ETS and PFS, in the first-line chemotherapy of mCRC.Method: ETS was defined as the estimated relative change in the sum of longest diameters of target lesions at week 8 compared to baseline. DpR was defined as the relative change in the sum of longest diameters of target lesions at the nadir compared to baseline. We retrospectively evaluated mCRC patients who started any chemotherapy from January 1, 2005 to December 31, 2010. PFS, ETS, DpR in the first-line chemotherapy and OS were examined following the previous reports.Result: 47 patients were enrolled. 30 patients were KRAS wild type, 15 were KRAS mutant and 2 were unknown. In 39 patients excluding survival cases, chemotherapy regimen were oxaliplatin-based plus BV; 21, irinotecan-based plus BV; 6, oxaliplatin-based; 9, irinotecan-based; 3. Anti-EGFR therapies were not performed in the first-line chemotherapy. Median of DpR, PFS and OS were 34.3 %, 257 days, 587 days respectively. PFS showed a correlation with OS, while DpR and ETS were not correlated with OS. DpR and ETS were highly correlated to each other. Compared with prognosis predictive models consisted of PFS alone, the new model consisted of the three factors did not show increased R-squared value suggesting that the new model could not provide more accurate prediction.Conclusion: The present prognosis predictive model including DpR and ETS may not be adapted for the whole mCRC patients. Specific patient population may possibly be suitable for prognosis prediction by this model.