Non-small cell lung cancer: Assessment of methodologies to combine survival curves in meta-analysis.

Abstract

Purpose. To assess the accuracy of different methods of combining published survival curves into a summary curve that can be used in disease models or to make treatment comparisons in meta-analysis. Methods. Five methods for combining survival curves were identified: (1) iterative generalized least-squares (IGLS, Dear 94), (2) meta-analysis (MA) of failure-time data with adjustment for covariates (MFD, Hunink 94), (3) non-linear regression (NLR, Shore 90), (4) log relative risk (LRR, Voest 89), and (5) weighted LRR (w-LRR). Published survival curves were scanned into a graphical package, coordinates along each curve were identified, and required data elements were extracted semi-automatically. Two separate analyses were done: (1) each method was used to combine the survival curves from eight single arm Phase II trials of chemotherapy in 918 patients with advanced non-small cell lung cancer (NSCLC); (2) each method was used to combine the survival curves from seven randomized trials in order to reproduce a MA of chemotherapy in advanced NSCLC. I compared the resulting summary curves using the Kolmogorov-Smirnoff goodness of fit test to a "gold standard" curve calculated from the corresponding individual patient data (IPD). (Abstract shortened by UMI.)