Design options for molecular epidemiology research within cohort studies

Past discussions of the relative strengths of nested case-control and case-cohort designs have not fully considered cohorts with stored biological samples in which biomarker analyses are planned. Issues related to biomarker analyses can affect an investigator's choice of design and the conduct of these two designs. The key issues identified are effects of analytic batch, long-term storage, and freeze-thaw cycles on biomarkers. In comparison with the nested case-control design, the case-cohort design is less able to handle these challenges. Problems arise because most implementations of the case-cohort design do not allow for simultaneous evaluation of biomarkers in cases and reference group members, and there is no matching. By design, the nested case-control study controls for storage duration and the batching of biological samples from cases and controls is logistically simple. The allowance for matching also means that subjects can be matched on the number of freeze-thaw cycles experienced by the biological sample. However, the matching generates complex data sets that can be more difficult to analyze, and the costly biomarker data generated from the controls has few uses outside of testing the specific hypotheses of the study. In addition, because the same subject can serve as a control and a case, or multiple times as a control, biomarker analyses and sample batching can be more complex than initially anticipated. However, in total, of the two designs, the nested case-control study is better suited for studying biomarkers that can be influenced by analytic batch, long-term storage, and freeze-thaw cycles.