Prognostic significance of p53 mutations in non-small cell lung cancer: a meta-analysis of 829 cases from eight published studies

Mutation of the p53 tumor suppressor gene is considered a possible marker o f poor survival among patients with non-small cell lung cancer (NSCLC). Thi s report presents the results of a meta-analysis of the available data addr essing this issue. Using previously described methods, a protocol was devel oped for a meta-analysis examining the prognostic significance of p53 mutat ions in NSCLC, Two-year survival data derived from 829 patients in eight pu blished studies were analyzed using a general variance-based method employi ng confidence intervals described by Greenland (Epidemiol. Rev. 9 (1986) 1- 30). The outcome of interest was a summary relative risk (RRs) reflecting t he risk of death at 2 years associated with p53 mutation positive versus p5 3 negative disease. Prior to calculation of a RRs, an analysis for homogene ity (Q) showed Q to equal 22.3, With 8 degrees of freedom, this yielded a P value corresponding to P < 0.005, This indicated substantial heterogeneity across studies in terms of their estimate of effect, Although a RRs of 1.5 2 was found when all eight studies were combined (favoring a negative progn ostic role for p53 mutation), the validity of this estimate is questionable since the existing heterogeneity indicates that factors other than p53 mut ation account for the variability in RRs across studies. Sensitivity analys es suggested that selection bias might represent an important source of var iability in that p53 mutations may differ in their effects on biological be havior of NSCL tumors. Other possible confounders include smoking history, race. geographic location of study and socioeconomic status. The available data do not support a clear role for p53 mutation as a prognostic marker in NSCLC. It appears that multiple sources of bias may contribute to spurious association of p53 mutation status and survival. Future analyses must cont rol for possible confounders in order to determine whether certain p53 muta tions are truly associated with poor clinical outcome. (C) 2000 Elsevier Sc ience Ireland Ltd. All rights reserved.