Skip Internal Navigation
EB Research - Methodological Research in Epidemiology
Biomarker/Analytical Development: Pooling of Biological Specimens
Biomarkers of exposure and disease status play a critical role in epidemiological research, but fiscal limitations and the high cost of assays often require that investigators choose a subset of potential markers. Less often but equally prohibitive to research, samples may physically lack the volume necessary to perform a particular assay. Both of these issues often leave many interesting hypotheses unexplored.
Pooling biospecimens is a technique in which samples from different individuals are physically combined and measured. This reduces the amount of each sample necessary for the assay as well as makes each assay more informative, thus reducing the number of tests required and overall cost. Researchers here have shown that a wide variety of statistical analyses (hypothesis testing, regression, ROC curves, semi and fully parametric methods) can be applied to pooled data, often with only minor adjustments to standard practice. Additional benefits of pooling have been demonstrated, such as increasing efficiency and limiting the impact of the limit of detection. Moreover, when biospecimens with insufficient volume are impossible to measure, pooling can lead to measureable volume for a particular assay. A hybrid, pooled-unpooled design has also been developed here where the cost savings of pooling are realized, with the added ability to assess and address measurement error without the need for replicate assays.
Enrique F. Schisterman, Ph.D.
- Malinovsky Y, Albert PS, Schisterman EF. Pooling designs for outcomes under a Gaussian random effects model. Biometrics. 2012;68(1):45-52. PMID:21981372
- Danaher MR, Schisterman EF, Roy A, Albert PS. Estimation of gene-environment interaction by pooling biospecimens. Statistics in Medicine; 2012:31: 3241-3252. PMID: 22859290
- Whitcomb BW, Perkins NJ, Zhang Z, Ye A, Lyles RH. (2012). Assessment of skewed exposure in case-control studies with pooling. Statistics in Medicine, 31(22), 2461-2472. PMID: 22437722