Objective: Haemoglobin A1c (HbA1c), a measure of long-term glycaemic control, is at the centre of the clinical management of diabetes mellitus. However, the reproducibility of HbA1c measurements from whole blood samples which have been in long-term storage is unknown. We undertook this study to assess the reproducibility of HbA1c measurements from whole blood samples that had been in storage at -70 degrees C for over a decade.
Research design and methods: Three hundred and thirty-six samples of frozen whole blood from the Atherosclerosis Risk in Communities (ARIC) Study, stored at -70 degrees C for 11-14 years assayed for HbA1c using a dedicated ion-exchange HPLC assay (Tosoh A1c 2.2 Plus HPLC) were compared with measurements on these same samples conducted prior to storage (in 1990-92) using a Diamat (Bio-Rad) HPLC instrument.
Results: HbA1c measurements from long-term stored samples were strongly correlated with values obtained prior to long-term storage (r=0.97). The difference between HbA1c from long- and short-term stored samples had a mean of 0.35% HbA1c (sd=0.35) and a CV of 5.8%, which was approximately three times that of duplicate assays (CV 1.3 to 2.5%).
Conclusions: These data demonstrate that highly correlated but more variable and slightly higher HbA1c results were obtained from frozen whole blood samples that have been in storage for more than a decade. This highly reproducible assay performance would lead to comparable ranking of individuals and unbiased estimates of relative risks and odds ratios in epidemiological studies (case-control and cohort designs), but results should be realigned when the absolute value is of interest. These results have important implications for epidemiological studies and clinical trials which have stored whole blood specimens.