Single-agent chemotherapy trials in small-cell lung cancer, 1970 to 1990: the case for studies in previously treated patients.
Review
Overview
abstract
PURPOSE: This review was undertaken (1) to determine the antitumor activity of agents studied in phase II trials in small-cell lung cancer (SCLC) patients, (2) to evaluate the adequacy of published trials, (3) to determine if previously treated patients are suitable for phase II trials in SCLC, and (4) to develop an improved design for phase II trials. DESIGN: English-language, single-agent efficacy trials in SCLC, published from 1970 to 1990, were reviewed. Study design and reporting of results were assessed for clinical and statistical methodology. Response rates observed in previously treated patients were compared with those observed in previously untreated patients. RESULTS: One hundred forty-one articles evaluating 57 agents in 3,042 patients were reviewed. Eleven drugs were active (defined as a response rate greater than or equal to 20% in a trial with greater than or equal to 14 assessable patients), and 12 were inactive. Due to methodologic problems with the clinical trials, the usefulness of the remaining 34 drugs (60%) remains uncertain. Deficiencies identified in trials include inappropriate sample sizes, poorly defined response criteria, and failure to report important prognostic factors. When studied in adequate trials, all agents known to be active in SCLC had an observed response rate greater than or equal to 10% in previously treated patients. CONCLUSIONS: Over the past 2 decades, phase II trials in SCLC have failed in their primary task of effectively identifying agents that warrant further clinical study and rejecting inactive agents. If only previously treated patients had been entered into these trials, no useful agent would have been missed provided that a lower observed response rate had been used as evidence of antitumor activity. We propose a two-stage sequential study design, entering previously treated patients, for future phase II trials in SCLC.
