Abstract 2753: Change in tumor size by RECIST correlates linearly with progression-free and overall survival in phase I studies

Abstract

Abstract INTRODUCTION: The Response Evaluation Criteria in Solid Tumors (RECIST) sets cutoff values for complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD). The efficacy of novel anticancer therapies is traditionally described as the percentage of patients with responses in each of these 4 categories. However, the boundary values defining these categories are arbitrarily set. We therefore study the relationship between change in tumor size by RECIST and both progression-free and overall survival (PFS and OS), and evaluate if boundaries exist or if alternatively the relationship is a continuous parameter. METHODS: We analyzed 468 participants on 24 phase I trials at MD Anderson from 10-1-04 to 6-30-08. Best response (best change by RECIST while on study) is correlated with PFS and OS by Somers' rank correlation. Kaplan-Meier survival curves of patients binned by response are also calculated (Table). Finally, a Cox model with no covariates is fit to the survival data, and analysis of best response versus the model residual is conducted. RESULTS: Rank order testing demonstrated a correlation between best response and survival (Dxy = −0.56 for PFS and −0.34 for OS). Kaplan-Meier analysis also demonstrated a strong correlation (p < 1×10−6, Table). Residual analysis of the Cox model (Methods) demonstrated a near linear relationship between changes in tumor size from −100% to +25% and survival; patients with a change of +25% or greater did uniformly poorly. Landmark analyses at multiple time points demonstrated identical trends regardless of the landmark. CONCLUSION: Given the linear relationship between best response to treatment and both PFS and OS, evaluating specific patient responses, for example through mechanisms such as waterfall plots, may be more valuable than aggregate CR, PR, SD, and PD rates. Additionally, the cutoffs for CR, PR, SD, and PD as defined by RECIST may not have special significance regarding survival in this population. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 2753.