[Meta-analyses of randomized trials in oncology: pros and cons]

A meta-analysis of trials is a quantitative synthesis of the results of trials addressing the same question. Such a synthesis is useful when the consideration of the results of the trials does not produce a clear answer. To perform a meta-analysis, one must identify all the randomized trials addressing the question, whether their results have been published or not. A meta-analysis can be based on summary data, or it can be based on individual patient data which allows extensive data checking, corrections, and an intent to treat analysis. A meta-analysis provides a test for the efficacy of the treatment under study, and an estimation of this efficacy. It allows also the study of the heterogeneity in efficacy between trials according to their characteristics or between sub-population of patients. Lastly, the results of a meta-analysis may be of help in the design of future trials. The quality of a meta-analysis depends essentially on the completeness of the collection of trials. Apart from this problem, the limits of a meta-analysis are the limits of the trials it includes, the addition of ill designed and ill conducted trials can only lead to a bad meta-analysis. The moderate effect of the treatments available for solid tumours in adults explains the widespread use of meta-analysis in oncology.