#103 The effects of freeze-thaw cycles on serum measurement of insulin and glucose in epidemiologic studies

Abstract

PURPOSE: We conducted a validation/calibration study to examine how measured values for serum concentrations of insulin and glucose change after one freeze-thaw cycle. METHODS: We selected frozen serum samples from 50 participants in the Polyp Prevention Trial, a multi-centered dietary intervention study of polyp recurrence. We shipped one vial of serum from each subject to “Laboratory A” for glucose and insulin assays, re-froze the remaining serum, and then repeated the glucose assay . Glucose and insulin assays were also performed for each subject on an identical second vial that previously had been shipped to “Laboratory B” for an unrelated series of assays. We compared results from the non-freeze-thaw assays to both freeze-thaw assays using Wilcoxon rank sum tests and linear regression . RESULTS: Glucose and insulin results rose in almost every case after one freeze-thaw cycle with average increases of 11.8% (“A”-“A” samples) and 14.0% (“B”-“A” samples) for glucose and 19.2% for insulin (“B”-“A” samples). The regression analyses indicated an upward bias in assay results for both insulin and glucose after one freeze-thaw cycle with R-square values ranging between 0.95-0.97. CONCLUSION: Insulin and glucose measures in serum that has undergone a freeze-thaw cycle, compared to measures in serum that has not undergone such a cycle, will be substantially higher. The consistency of this upward bias, however, would still allow one to draw valid inferences based on these measurements if the glucose or insulin levels were analyzed categorically or if the continuous measures were calibrated according to the linear relation obtained from the validation sample. Thus the re-freezing and re-thawing of serum specimens (in order to preserve biologic material) does not compromise epidemiologic associations for insulin and glucose.