International Expert Committee Report on the Role of the A1C Assay in the Diagnosis of Diabetes

In 2008, an Expert Committee was convened to redefine the diagnosis of diabetes (in non-pregnant individuals) using glycated haemoglobin (HbA_1c). Whilst Committee Members were appointed from the American Diabetes Association (ADA), the European Association for the Study of Diabetes (EASD) and the International Diabetes Federation (IDF), it should be noted that the recommendations of the Committee have yet to be endorsed by these associations. The recommendations¹ were published in the July edition of Diabetes Care with an accompanying editorial² and were presented at the 2009 ADA 69th Scientific Sessions held in New Orleans in June. A summary of the recommendations is presented below.

The Committee presents background information regarding the history of diabetes diagnosis and the relationship between glucose levels and microvascular diabetes complications. They note that whilst type 1 diabetes has a typically abrupt clinical onset with significant elevations of plasma glucose, making specific diagnostic glucose cutpoints unnecessary for most clinical settings, type 2 diabetes commonly has a more insidious onset. It is for this reason that specific glucose values have traditionally been identified in an attempt to distinguish patients with type 2 diabetes from a non-diabetic population.

Diagnostic criteria have evolved but the basis for all has been a timed glucose sample such as a fasting sample, casual sample independent of prandial status, or, after a standardised oral glucose load (75 g). The need for timed glucose measurements has been identified as inconvenient for patients and also potentially a source of missed screening opportunity. Earliest attempts to standardise the definition of diabetes relied on the 75 g oral glucose tolerance test (OGTT) and the 2h value was established as the diagnostic value of choice. This was in spite of the fact that studies assessing performance and interpretation of the test were inconsistent and subject numbers small.

In 1979, the National Diabetes Data Group (NDDG) defined the diagnostic criteria³ that were used until they were superseded in 1997. The NDDG criteria relied on the distributions of glucose levels rather than on the relationship of glucose levels and complications as the basis of diagnosis. The NDDG acknowledged that no clear distinction between diabetics and non-diabetics existed and thus an arbitrary decision was made. The diagnosis of diabetes was made when 1) classic symptoms were present, 2) venous fasting plasma glucose was >7.8 mmol/L, or 3) the 2h venous plasma glucose during the OGTT was >11.1 mmol/L (with the 1h sample >11.1 to exclude the possibility of delayed absorption). An intermediate category of “impaired glucose tolerance” (IGT) was also defined (fasting plasma glucose <7.8 mmol/L and 2h glucose value between 7.8 and 11.1 mmol/L).

An Expert Committee re-examined the diagnostic criteria in 1997.⁴ Attention was refocused on the relationship between glucose levels and the presence of complications as the foundation for diabetes. Additionally, the Committee summarised data questioning the 2h glucose value as the gold-standard test for diagnosing diabetes. Epidemiological studies assessing retinopathy in almost 5000 patients with diabetes were examined. These studies demonstrated within populations glycaemic levels below which there was little prevalent retinopathy and above which prevalence increased linearly. This inflection point was demonstrated for fasting glucose, 2h glucose and HbA_1c. Glycaemic values were similar among populations. Prevalence of retinopathy began to increase at HbA_1c values of between 6.0 and 7.0%. It became apparent that the previous fasting diagnostic glucose of >7.8 mmol/L was too high and the Committee recommended lowering to >7.0 mmol/L. The 2h OGTT value remained unchanged. It was felt that a fasting glucose of 7.0 mmol/L represented a similar degree of hyperglycaemia to a 2h OGTT value of 11.1 mmol/L. However, it was also acknowledged that they were not perfectly concordant and that an individual could be diagnosed with diabetes using one test but not the other. In addition to changing the fasting glucose value, the Expert Committee in 1997 also promoted the fasting level rather than an OGTT as the diagnostic test of choice, citing convenience for patients, cost and time savings and superior test reproducibility. HbA_1c was not recommended as a diagnostic test mainly due to lack of assay standardisation at that time. A new term “impaired fasting glucose” (IFG) was also introduced to categorise those patients with an intermediate fasting glucose (6.1–7.0 mmol/L). IGT was now used to describe those patients with an intermediate 2h glucose value during an OGTT (7.8–11.1 mmol/L). The World Health Organization adopted most of these recommendations but continued to promote the OGTT as the gold-standard diagnostic test, especially for individuals with IFG.

A follow-up report in 2003⁵ from the same Expert Committee recommended a change in the definition of IFG to include a fasting glucose of >5.6 but <7.0 mmol/L. Whilst it was noted that the vast majority of HbA_1c assays within America had been standardised since the 1997 recommendations, there were still disadvantages with HbA_1c, such that the Committee once again reaffirmed the previous recommendation that it not be used as a diagnostic test. Following this report, the WHO did not alter its previous recommendations.

After providing the above historical background to the present diagnostic classification for diabetes, the current Expert Committee subsequently poses the question “Can the A1c test be used to diagnose diabetes?” It is widely accepted that no clinical diagnostic test is perfect. To be of clinical use, a test should ideally be accurate, specific, standardised, convenient and inexpensive. The Expert Committee provides some compelling arguments for the use of HbA_1c as a diagnostic test. When compared with fasting glucose or 2h OGTT glucose, the following advantages for HbA_1c are listed:

1. standardised and aligned to the DCCT/UKPDS; measurement of glucose is less well standardised

2. better index of overall glycaemic exposure and risk for long-term complications

3. substantially less biologic variability (<2% day-to-day within person variability for HbA_1c compared with 12–15% for fasting glucose)

4. substantially less preanalytic instability

5. no need for fasting or timed samples

6. relatively unaffected by acute (e.g. stress or illness related) perturbations in glucose levels

7. currently used to guide management and adjust therapy

Following on from this, the most appropriate A1c cutpoint for diabetes diagnosis is discussed. The Committee chose a value >6.5% which is likely to generate debate. The relationship between A1c and average glucose levels is not perfect. An unpublished analysis of nearly 19 000 subjects by Dr Stephen Colagiuri (a member of the Expert Committee) is discussed in the current recommendations and indicates that the optimum cutpoint for detecting at least moderate retinopathy was an HbA_1c of 6.5%. This is consistent with data reviewed in the 1997 recommendations and other recently published literature. Nevertheless, if an alternate complication outcome had been considered, it is interesting to speculate whether a different HbA_1c diagnostic value would have been identified. The Committee feels that specificity should be emphasised over sensitivity (considering the stigma and cost of incorrectly labelling an individual as diabetic greater than the risk of delaying the diagnosis in someone with an HbA_1c of <6.5%). Whilst not providing an absolute divide between normoglycaemia and diabetes, the level of 6.5% was considered by the Committee to optimise specificity whilst retaining adequate sensitivity. With a normal HbA_1c being <6.0%, a diagnostic hiatus is present in the range of 6.0–6.5%. Whilst the Committee does also recommend initiation of preventive strategies in these individuals at highest risk (HbA_1c 6.0–6.5%), there remains no consensus on the optimum test to use in predicting diabetes. The suggested diagnostic threshold of 6.5% is also different from the treatment target recommended by most guidelines (7%).

Possible limitations of A_1c as a diagnostic test are mentioned in addition to other practical issues relating to the test. The recommendation to use HbA_1c as a diagnostic test assumes that the local performance of the assay is acceptable. It should be noted that the Committee recommends against the use of point-of-care instruments, as they have not been shown to have adequate accuracy or precision. Other potential causes of spurious HbA_1c results include abnormal haemoglobins (although affinity assays are unlikely to be affected) and conditions affecting red cell turnover (haemolysis, chronic malaria, major blood loss and transfusions). Traditional glucose-based diagnostic criteria would remain appropriate in these clinical settings. Ageing and racial differences were not thought to be sufficiently well-defined to justify age or ethnicity specific values which others may debate. It is uncertain to what extent clinicians are aware of these confounders. Many other known or postulated contributors to spurious HbA_1c values were not mentioned, including iron deficiency anaemia, effects of HIV therapy, renal failure, dapsone therapy, high dose salicylates, vitamin C and E, splenectomy and aplastic anaemia.^6–10 Lastly, the authors discuss the clinical setting of type 1 diabetes, where an increase in HbA_1c may lag behind the relatively acute onset of hyperglycaemia. However, symptoms and increased random glucose should still allow diagnosis.

Whether A1c measurements are able to define a subdiabetic “high risk” state and whether they can be used to identify individuals at high risk for diabetes are also discussed. Even within glucose values considered normal, there remains a graded risk for the development of diabetes. In the same way, a continuum of risk has been identified based on HbA_1c levels. The Expert Committee suggests that this continuum of risk within a subdiabetic glycaemic range argues against the use of existing dichotomous subdiabetic classifications such as IFG and IGT. Instead the Committee recommends that individuals close to the 6.5% diagnostic threshold (i.e. HbA_1c >6.0%) should receive effective intervention. They state that individuals with HbA_1c values lower than this should not be considered to ‘not be at risk’, but rather considered to be ‘at lower risk’. For individuals with HbA_1c<6.0%, other risk factors for the development of diabetes should be considered, including hypertriglyceridaemia, hypertension, increased body mass index and a positive family history. The exact HbA_1c value used may be dependent upon the population, available resources and the type of intervention.

Finally the Expert Committee presents their recommendations and conclusions:

1. The HbA_1c assay provides an accurate, precise measure of chronic glycaemic levels and correlates with the risk of diabetes complications.

2. The HbA_1c assay has several advantages over laboratory measures of glucose.

3. Diabetes should be diagnosed when HbA_1c is >6.5%. A repeat HbA_1c test should be done for confirmation in asymptomatic patients.

4. If HbA_1c testing is not available, previously recommended diagnostic methods remain acceptable.

5. HbA_1c testing is indicated in children in whom diabetes is suspected but the classic symptoms and a random glucose >11.1 mmol/L are not found.

Advantages and Disadvantages of Proposal to use HbA_1c for Diabetes Diagnosis

Advantages Convenient for patients i.e. no fasting or other test preparation required Accurate, precise measure of chronic glycaemic levels Significant international efforts to standardise assays Lower between- and within-subject coefficients of variation and reduced possibility of pre-analytic errors compared with glucose Correlates with risk of diabetes defining complications (retinopathy) Familiar test parameter i.e. already used to guide therapeutic decisions

Disadvantages Relationship between HbA1c and glucose, whilst good, is not perfect Whilst existing glucose based diagnostic criteria remain valid, the current proposal does not advocate a confirmatory check of glucose at any stage Point of care instruments currently considered inadequate for diagnostic purposes Methods suffer from multiple interferences that clinicians may not be aware of The proposed cut-off of 6.5% is predictive of retinopathy but is it the most appropriate outcome on which to choose a diagnostic target? Not appropriate for diagnosis of gestational diabetes Upper limit of normal HbA1c (6.0%) leaves a diagnostic hiatus between 6.1 and 6.5%, in addition to a discrepancy between the typical treatment target of <7% and the diagnostic level Impact of reporting HbA1c in molar units

In summary, the Expert Committee presents a compelling argument on practical grounds for the adoption of HbA_1c as a diagnostic test. It is noted that at no point does the Expert Committee actively recommend measuring glucose, either fasting or random. In fact, they go so far as to say that mixing different methods for diabetes diagnosis should be avoided. Not discussed in the report is how proposed changes to reporting of HbA_1c will impact on the recommendations of the Expert Committee. This may ultimately add a further level of complexity to the debate regarding the appropriateness of A1c as a diagnostic test for diabetes. It should be noted that the proposed International Federation of Clinical Chemistry (IFCC) HbA_1c units of mmol/mol have already been adopted by the UK and New Zealand, where laboratories have been reporting in IFCC units since June 2009 and August 2009 respectively.