Classification of methods for evaluation of treatment selection markers: A systematic review

Article type
Authors
Tajik P1, Mol B2, Bossuyt P1
1Department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical Center, University of Amsterdam
2Department of Obstetrics and Gynecology, Academic Medical Center, University of Amsterdam
Abstract
Background: There is an increasing interest in the evaluation of biomarkers useful for treatment selection. However, the literature on the methods for the evaluation of these markers is diverse and there is no consensus about their classification and nomenclature.

Objectives: To systematically review and generate a classification of study methods to identify and/or validate treatment selection markers, and to standardize the nomenclature.

Methods: We searched MEDLINE (1950 to Jan 2011), EMBASE (1980 to Jan 2011), Cochrane Methodology Register and MathSciNet (up to the Jan 2011), and handsearched the citations of included studies. We included methodology reports that described one or more study design or analysis plans related to identification and validation of treatment selection markers.

Results: Our search identified 1758 abstracts, among which 82 were found to be probably eligible. After full-text review, 36 studies were included. The following dimensions were identified for the classification of methods: trial phase, patient flow, randomization type, sample size requirements, timing, the number of evaluated markers, analysis plan, and summary statistics. For each study design advantages and limitations were described and the efficiency of methods were compared with each other. Using identified dimensions, all methods were classified and identical methods received a unique terminology.

Conclusions: We propose a standard nomenclature and classification system for study designs of treatment selection marker evaluation. Using this classification system could help researchers in designing future studies and reduce reporting inconsistencies.