Abstract Background Hemoglobin A1c is used to assess glycemic control to both diagnose and monitor diabetes mellitus. Several different methodologies are available including high performance liquid chromatography (HPLC), capillary electrophoresis (CE), and immunoassay (IA). These methods are standardized and traceable to the Diabetes Control and Complications Trial (DCCT). The aim of this study was to evaluate the difference in HbA1c results between methods and percent of reportable results in variant samples. Methods The HPLC (Bio-Rad Laboratories, Inc; D-100), the CE(Sebia Inc; Capillarys 3), and IA(Roche Diagnostics Inc; Tina-quant Generation 3) were compared using two sets of residual patient samples left after physician ordered testing was complete: 1) samples with an absence of abnormal hemoglobin peaks (i.e. non-variant) detected by HPLC; 2) samples with abnormal hemoglobin peaks (i.e. variant) characterized by hemoglobin electrophoresis performed clinically or lacking an A1c peak by HPLC/CE. The samples were tested within manufacturers stability claims. 317 non-variant samples were compared between the HPLC and CE methods. The absolute difference was calculated for each method using this equation (HbA1c%avg – HbA1c%X) where HbA1c%avg is the average of the methods that produced reportable results and X is the result from each method evaluated. Differences less than or equal to +/-0.3% were considered acceptable. 28 of the 317 non-variant and 47 variant samples were tested by all 3 methodologies. The 47 variant samples grouped into 3 cohorts: common heterozygous variants (HbC, HbD, HbE, HbS and HbF, n=10), uncommon heterozygous variants (n=29) and homozygous variants lacking HbA (n=8). The absolute differences in variant sample results were calculated for samples that did not surpass manufacturer’s claims for producing an accurate result. The frequency of producing a numeric result in variant samples was calculated for each method. Results The HPLC/CE non-variant sample comparison (N=317) had results spanning the measuring range (3.8% - 14.0%) with average(range) of absolute difference=0.1(-0.4to0.6%) with 293/317(92.4%) samples within +/-0.3%. The HPLC/CE/IA non-variant sample comparison (N=28) had a result range=4.0%-9.4%, the average(range) of difference=0.1(-0.2to0.5%) with 25/28 (89.2%) of samples within +/-0.3%. Common heterozygous variant samples gave numeric results with no disqualifying flags in 9 of 10(HPLC), 8 of 10(CE), and 9 of 10(IA) samples tested. The average(range) difference from average=0.0(-0.3,0.5%) for HPLC, 0.0(-0.2,0.3%) for CE, and -0.1(-0.6,0.2%) for IA. Uncommon variant samples gave results in 4 of 29 (HPLC), 20 of 29 (CE), and 26 of 29 (IA) of samples tested. The average(range) difference from average=-0.1(-0.3,0.1%) for HPLC, 0.0(-1.0,1.1%) for CE, and 0.0(-1.1,1.0%) for IA. The homozygous variant (no HbA) samples gave results in 0 of 8 (HPLC), 0 of 8 (CE), and 4 of 8 (IA) of those tested. Conclusion All three methods meet their claims for accurately measuring hemoglobin A1c from non-variant and common hemoglobin variants. The advantage of HPLC and CE is the ability to identify hemoglobin variants that may impact HbA1c quantitation. The IA method provides the most numeric results. IA gave results even when HbA was absent. The lab should understand the capabilities and limitations of their HbA1c method on the population served.
Rosemark et al. (Wed,) studied this question.