RESPONSE RATE AS AN END-POINT FOR EVALUATING NEW CYTOTOXIC AGENTS IN PHASE-II TRIALS OF NON-SMALL-CELL LUNG-CANCER

Background: Response rate (RR) has been used as a defining endpoint of new-agent phase II trials for non-small-cell lung cancer (NSCLC). How ever, tumor responses to chemotherapy do not always result in prolonge d survival of patients with this disease. Design: Single-agent phase I I trials were identified by a MEDLINE search of the period from 1976 t o 1995. Associations between RR, median survival time (MST) and charac teristics of patients who entered the trial, including tumor extent, p erformance status and prior chemotherapy, were studied by using the lo gistic regression model. Results: A total of 183 treatment arms in 176 trials (including 10 randomized phase II trials) were identified. The overall RR in the 6768 evaluable patients was 11%. Eleven drugs, cisp latin, epirubicin, ifosfamide, edatrexate, irinotecan, vinorelbine, do cetaxel, paclitaxel, etoposide, vindesine, and 254-S, produced a RR of more than 20%. An MST of eight months or longer was obtained with 12 drugs, but there were cases in which no objective responses were produ ced by these drugs. MST was correlated with RR (r = 0.504, P < 0.0001) , but ranged broadly at a given level of RR. Multiple linear regressio n analysis showed a significant correlation between RR and MST (regres sion coefficient = 0.60, P = 0.00003) after adjustment for other varia bles. Conclusions. RR was significantly correlated with MST in single- agent phase II trials for NSCLC, but there is room for further conside ration of the endpoint of these trials.