The components of progression as explanatory variables for overall survival in the RECIST database.
10602 Background: Progressive disease (PD) according to RECIST 1.1 (Eisenhauer et al, 2009) is defined as one or more of (1) PD of measurable target lesions, (2) the presence of a new lesion (NL) or (3) the unequivocal PD of non-target disease (NT-PD). We explored the impact of these components, varying over time, on predicting overall survival (OS) in the RECIST database residing at EORTC (Bogaerts et al, 2009). Methods: Data was selected from 12 randomized clinical trials (3530 patients with breast, lung or colorectal cancer). A maximum of 5 target lesions was used to determine the sum of diameters. The following were calculated or assigned at each measurement time t: best response (BR) was best % improvement from baseline up to t (0% if no improvement - 100% if complete response (CR)); tumor growth (TG) was the weekly rate of increase from nadir to t (0 if no increase; irrespective of prior shrinkage), presence of NL (yes/no), and occurrence of NT-PD (yes/no); categories were not mutually exclusive. OS was analyzed by tumor type using a Cox regression model, adjusting for baseline sum, and including BR, TG, presence of NL and NT-PD as time dependent covariates. Results: Thirty-six percent of patients had NL, 26% had NT-PD, 11% achieved CR and 14% did not have shrinkage of target lesions, while 46% experienced TG (median strongest growth per patient of 0.5 mm/week). Regardless of tumor type, the presence of NL (Hazard Ratio (HR) ranging 1.4-2.5), NT-PD (HR 1.2-2.5) and TG (per 1mm/week increase; HR 1.1-1.4) were associated with worse OS, while achieving CR was associated with a longer OS (HR 0.2-0.8). Further analyses exploring the functional shape of the association between BR and TG, and OS are ongoing. This includes putting TG in contrast with the more usual % cutoff defining target PD. Conclusions: All three components of PD according to RECIST are independently strongly associated with OS. Quantification such as this will enable a better understanding of the role of each component (e.g. clinical aspect of NT-PD assessment) in PD evaluation. Work is ongoing to incorporate this information into an updated RECIST with enhanced prediction of subsequent survival.