Abstract
Glycated haemoglobin was first identified in 1968 as a novel haemoglobin variant associated with diabetes. 1 However, it was not until 1977 that its potential as an indicator of glycaemic control was first recognized 2 During the last 30 years, HbA1c (the major glycated fraction) has played a vital role in the monitoring of diabetes with a firm evidence base from the Diabetes Control and Complications Trial (DCCT) 3 and the United Kingdom Prospective Diabetes Study (UKPDS). 4 This has been facilitated by the development of high-throughput methods with between-laboratory coefficients of variation (CVs) of less than 3%. At the same time, alignment of results to those in DCCT and UKPDS has been facilitated by the US National Glycohemoglobin Standardisation Programme and, more recently, a strong metrological standardization anchor has been provided by the International Federation for Clinical Chemistry (IFCC). After a period of dual reporting, all UK laboratories were requested to switch to IFCC mmol/mol units from October 2011. 5
While the role of HbA1c in monitoring diabetes has intensified, that of glucose has diminished. Even though methodologies for home blood glucose monitoring have evolved to high analytical standards, its use is now restricted to special patient groups such as those on treatment with insulin, owing to its cost and inconvenience and relative lack of evidence base for clinical efficacy compared with that for HbA1c.
The only area where glucose measurement appeared until recently to be sacrosanct was in diagnosis. In 1997, the American Diabetes Association (ADA) acknowledged that blood glucose concentration was no longer believed to be bimodal, and re-focused the definition of diabetes on levels of glycaemia associated with increased risk of microvascular disease, exemplified by retinopathy. 6 There is a clear threshold in the curve describing retinopathy risk in relation to glucose, below which risk is low and above which risk increases sharply. Epidemiological studies showed an equally clear relationship between HbA1c and retinopathy rate, but variation in methodology and standardization precluded its use for diagnosis. There was also concern about the confounding effect of factors affecting erythrocyte turnover rate. This position was reiterated by the ADA in its revised diagnostic criteria of 2003 7 and by the World Health Organization (WHO) in 2006. 8
There are, however, many potential advantages to use HbA1c for diagnosis. 9 HbA1c has lower biological variation (intraindividual CV < 1%) than fasting glucose (>5.7%) and particularly two-hour oral glucose tolerance test (OGTT) (16.7%) glucose concentrations. There is no need for special patient preparation (fasting and calorie regulation before OGTT), and it is relatively unaffected by acute stress or hyperglycaemia. HbA1c has high preanalytical stability (one week at 4°C) whereas glycolysis consumes glucose even in fluoride preservative for the first two hours after blood is collected. Analytical precision for HbA1c now approaches that for glucose. Standardization of HbA1c measurement is better than for glucose, which has no internationally recognized reference method.
During the last 10 years, the diagnostic potential of HbA1c has been re-examined. Most recently, the DETECT-2 collaboration analysed cross-sectional data from 47,364 people in 12 countries and reported an HbA1c threshold of 6.3% (45 mmol/mol) as the threshold for diabetes-specific retinopathy, and receiver operating characteristic curve analysis indicated an optimal decision limit of 6.5% (48 mmol/mol). 10
It must be stressed that HbA1c is not a surrogate for glucose, but an alternative screen for microvascular risk. Not surprisingly, therefore, HbA1c detects a different set of subjects, but this is not necessarily a problem given that it performs equally well as a risk predictor. In most populations, HbA1c produces a lower diabetes prevalence, although interestingly the magnitude of the difference varies quite widely, raising questions about interracial differences in haemoglobin glycation rates that could call into question the universality of any given cut-off point. 11
On the basis of the above, the ADA now recommends HbA1c ≥ 6.5% (48 mmol/mol) as an alternative to glucose-based diagnosis. 12 After more detailed analysis, the WHO supported this view in a 2011 addendum to its 2006 diagnostic criteria. 13 Both organizations refer to the importance of avoiding HbA1c in patients with conditions associated with abnormal erythrocyte turnover, which include kidney failure 14 as well as haematinic deficiencies and haemolysis. Confirmation by repeat testing should use the same method (i.e. HbA1c must not be used second line in patients with non-diagnostic initial glucose test results). HbA1c is not recommended for gestational diabetes. In the UK, the WHO statement was immediately welcomed by Diabetes UK, and endorsed by some, but not all, experts. The UK Department of Health looks set to recommend incorporating HbA1c into current diagnostic criteria provided that satisfactory analytical performance is in place and conditions that may affect the normal relationship between HbA1c and glycaemic status are not present.
Where does all this lead? HbA1c is now accepted as a diagnostic test for diabetes. There will continue to be discussion about the optimal cut-off point, but this is unlikely to change from 6.5% (48 mmol/mol) unless long-term prospective studies establish different thresholds. HbA1c offers a simple diagnostic test that is very acceptable to clinicians and patients. However, education in its appropriate use and in particular to assist in the identification of patients with states that affect glycation rates is important.
The higher cost of HbA1c compared with glucose is a major issue not only in the developing world but also in the UK. However, in terms of the whole patient pathway, the added cost is small. Reliance solely on HbA1c removes the need for the OGTT, although this is only required in a small proportion of cases outside pregnancy and, crucially, it must be stressed that HbA1c is not recommended as a second step after an initial fasting plasma glucose concentration in the range 6.1–6.9 mmol/L. HbA1c is a poor predictor of the two-hour OGTT glucose concentration in patients with impaired fasting glycaemia (IFG) 15 and thus cannot overcome the need for an OGTT in this group. 16
What is the future for impaired glucose tolerance (IGT) and IFG? Both are risk factors for progression to diabetes mellitus and IGT, at least, is an independent cardiovascular risk factor. HbA1c 5.7–6.5% (39–48 mmol/mol) has been suggested by the ADA to denote intermediate hyperglycaemia 12 and WHO has suggested that individuals with HbA1c between 6.0% and 6.5% (42–48 mmol/mol) should be considered for diabetes prevention interventions. 13 It is unclear how these data could be factored into cardiovascular risk calculation.
HbA1c has finally come of age as a fully fledged analyte of dual value in diagnosis and monitoring of diabetes. Like any complex, but powerful, test it comes with a health warning, requiring education of requestors in order that it is used efficiently and effectively.
DECLARATIONS