Revaluation of biological variation of HbA1c using an accurately designed protocol

Background. HbA1c has a key role for diagnosing diabetes and monitoring glycemic state. Knowledge of its biological variation (BV) is critical for result interpretation. As recently published results showed a marked heterogeneity of available BV data, we revaluated HbA1c BV using a well designed protocol. Methods. We took five blood specimens from 18 apparently healthy subjects (9 men and 9 women, ages 26-52 years) on the same day every two weeks for two months. Samples were stored at -80°C until analysis and assayed in duplicate in a single run by Roche Tina-quant Gen.2 immunoassay on the Integra 400 platform. Data were analyzed by the ANOVA. To assess the assay traceability to the IFCC reference method, we preliminarily carried out a correlation experiment on four blood samples. Results. The regression equation (Tina-quant = 1.05 IFCC reference method – 1.6 mmol/mol) confirmed the alignment of the employed assay. There were no differences in HbA1c values between men and women (global mean, 36 mmol/mol). Within- (CVI) and between-(CVG) subject BV were 2.5% and 7.1%, respectively. Desirable analytical goals derived from BV for imprecision (0.50 CVI), bias [0.25 (CVI 2 + CVG 2)1/2] and total error [1.65 (0.5 CVI) + 0.25 (CVI 2 + CVG 2)1/2] were 1.25%, 1.88%, and 3.94%, respectively. HbA1c had marked individuality, therefore the use of population-based reference limits could be inadequate for test interpretation. The reference change value, useful to interpret changes in serial results, was ~9%. Conclusions. Our data show that strict analytical goals are needed for the clinical application of HbA1c measurements.