The role and quality of hb a1c: a continuing evolution.

Since glycohemoglobin was first described in the red cells of diabetes patients in 1968 (1 ), the test has evolved to become an essential tool for prognosis, monitoring, treatment, and diagnosis of diabetes mellitus. An increasing clinical focus together with the development of diabetes management guidelines based on glycated hemoglobin (Hb A1c) 2 necessitated a standardization of the method initially achieved in schemes such as the National Glycohemoglobin Standardization Program (NGSP) (2 ) and later refined by the IFCC (3 ). Now that we know the reference values we should be aiming for, the IFCC task force has investigated approaches for setting the analytical quality targets for Hb A1c (4 ) so that we can know how close to those reference values we need to be; the task force’s report is published in this issue of Clinical Chemistry (4 ). The task force has acknowledged the importance of the Stockholm hierarchy, in which clinical outcome studies are the preferred method of defining analytical quality (5 ). In Milan during November 2014, the European Federation of Laboratory Medicine revisited the Stockholm consensus and retained clinical decision making at the forefront of the rationales for defining analytical quality, ahead of biological variation and state-of-the-art approaches (6 ). Hb A1c is ideally placed for using clinical rationales to define quality specification, because we know so much about diabetes outcomes in relation to Hb A1c, and we correspondingly have agreed-on clinical protocols to manage diabetes. The IFCC task force authors have borrowed the excellent example of the UK National Institute for Health and Care Excellence (NICE) guidelines for the management of type 2 diabetes that focus on a change in Hb A1c of 5 mmol/mol (0.46%) to define the total allowable error (TAE) according to clinical decision making. The IFCC Task Force found that the clinically defined TAE of 5 mmol/mol (0.46%) was achievable by 77% of laboratories participating in proficiency testing, whereas only 48% of laboratories were able to achieve the minimum biological variability goal of 3.4 mmol/mol (0.31%). The underlying principle of using biological variability as the basis for analytical quality specifications is that measurement noise should not mask the underlying variations we are trying to detect. The simplest Cotlove criterion, that the imprecision of analysis (CVA) should be less than half the intraindividual biological variation (CVI) (7 ), has been refined to include minimal (CVA 0.75CVI), desirable (CVA 0.5CVI), and optimal (CVA 0.25CVI) criteria. The critical step in these definitions is to define CVI, which is easier said than done: not only does Hb A1c CVI vary from study to study, as highlighted in the IFCC task force article, but the variability of Hb A1c itself seems to vary, from the stable situation of health (CVI 1.0%) to the increasing instability of type 2 diabetes (CVI 4.3%) and type 1 diabetes (CVI 8.8%) (8 ). It is important to acknowledge that the higher Hb A1c in each of these states is a reflection not only of higher average glucose concentrations but also the higher Hb A1c variability of blood glucose values due to increasing insulin resistance and deficiency (9 ). The task force chose to use the single CVI value of 1.85% from the Westgard QC Ricos database (10 ) while acknowledging that different quality targets could be obtained using different values for CVI. The variations in CVI across varying states of health and types 1 and 2 diabetes have corresponding implications for clinical decision making using Hb A1c. The clinical challenge of controlling type 2 diabetes, even in the NICE guidelines (6 ), depends on the clinical context, as the target value is 48 mmol/mol (6.5%) in type 2 diabetes treated with a single oral hypoglycemic medication and 58 mmol/mol (7.5%) in a type 2 patient treated with more medication or insulin (similar to type 1 diabetic treatment). Therefore, not only does the biological variability of Hb A1c change in various clinical conditions, but the clinical decisions themselves vary according to the patient category. In an adolescent with poorly controlled type 1 diabetes and Hb A1c of 86 mmol/mol (10.0%), the clinical challenge may be to improve control and reach a Hb A1c value 64 mmol/mol (8.0%). These large differences can be measured with confidence using relatively imprecise assays, which is why point-ofcare technologies are well suited to clinics that manage 1 Melbourne Pathology, Melbourne Victoria, Australia. * Address correspondence to this author at: Melbourne Pathology, 103 Victoria Parade, Collingwood, Melbourne VIC 3066, Australia. E-mail [email protected]. Received March 2, 2015; accepted March 4, 2015. Previously published online at DOI: 10.1373/clinchem.2015.239319 © 2015 American Association for Clinical Chemistry 2 Nonstandard abbreviations: Hb A1c, glycated hemoglobin; NGSP, National Glycohemoglobin Standardization Program;NICE,National Institute forHealth andCare Excellence; TAE, total allowable error. Clinical Chemistry 61:5 689–690 (2015) Editorials