Affiliations: Bar Ilan University and Gertner Institute for Epidemiology and Health Policy Research, Israel | [x] University of Michigan Medical Center, Ann Arbor, MI 48109-0930, USA | [y] CHS National Cancer Control Center, Department of Community Medicine and Epidemiology, Carmel Medical Center and B. Rappaport Faculty of Medicine, Technion Haifa, Israel
Correspondence:
[*]
Present address: Biostatistics Unit, Gertner Institute for Epidemiology and Health Policy Research, Tel Hashomer 52161, Israel. Tel.: +972 3 5305390; E-mail: [email protected].
Abstract: Even the most common malignancies have a low probability of occurrence over a restricted time interval. Therefore controlled intervention studies that use incident cancer as an outcome must be large, lengthy and, hence, costly. Studies with surrogate outcomes – biomarkers of pre-clinical carcinogenesis – are attractive because they are potentially smaller, shorter, and less expensive than their counterparts with cancer outcomes. Despite their potential, however, surrogate outcomes require validation to ensure that they provide sufficient quality of evidence on intervention effects. We review methods that have been proposed over the past 15 years for such validation. The two main approaches are those based on the Prentice criterion, which require data from a single study, and those based on meta-analysis, which require data from many studies. The former approach has fallen out of favor, for reasons to be explained. The latter approach is more popular, but so demanding of resources that it may prove impractical for cancer chemoprevention in all but a few instances. Researchers may have to resign themselves to more limited use of surrogate outcomes, not as replacements for traditional outcomes, but as outcomes for Phase II studies designed to decide which interventions to pass for Phase III testing.
Keywords: Intermediate endpoints, meta-analysis, Phase II chemoprevention studies, Phase III chemoprevention studies, Prentice's criterion, surrogate outcomes, surrogate endpoints
