Glycated haemoglobin and diagnosis of diabetes mellitus: now well established but beware the pitfalls

In 2011, the World Health Organization (WHO) sanctioned the use of glycated haemoglobin (HbA1c) as a diagnostic test for type 2 diabetes mellitus. ¹ HbA1c ≥ 48 mmol/mol is the recommended diagnostic threshold, based on the observed increased risk of diabetic retinopathy above that level. However, in asymptomatic patients, a repeat test is required to confirm the diagnosis. Whether this step is adhered to in practice is questionable. Several considerations limit the diagnostic use of HbA1c, and it seems appropriate at this juncture to review these and to discuss the role of the laboratory in ensuring its appropriate use.

The deployment of HbA1c as a diagnostic tool has been a popular one with clinicians. Its pre-analytical stability, lack of diurnal variation and low biological variation make it a more robust test than glucose. ² Within-subject biological variation for HbA1c is approximately 3.6% set against fasting glucose 5.7% and oral glucose tolerance test (OGTT) 2 h glucose 16.7%. ³ The OGTT is increasingly confined to the diagnosis of gestational diabetes, amidst previous concerns that correct procedure was not always followed. The removal of the requirement to fast or undergo an OGTT is preferable for patients and more practical for clinicians. There is, of course, increased laboratory cost when switching to HbA1c for diagnosis. In our local service, requests have risen by 16% in the last year. However, early diagnosis and prompt control of hyperglycaemia are pivotal to reducing long-term complications of diabetes,^3,4 and a recent UK study showed that HbA1c is a cost-effective screening modality. ⁶ There remains debate regarding whether population screening is appropriate, but in 2016, the NHS Diabetes Prevention Programme will roll out. This is likely to use HbA1c as a screening tool in high-risk people, thus increasing workload again.

Whilst HbA1c is now widely used for diagnosis, its limitations may not be well understood. It is imperative that the patient is not labelled incorrectly as diabetic or equally that their diagnosis is not missed. The cautions regarding its use can be broadly divided into: (a) indications for use; (b) medical conditions affecting the test accuracy (via either alteration in red cell lifespan or glycation); (c) analytical interference. ⁷

The first of these is relatively well recognised by clinicians: HbA1c should not be used in situations when hyperglycaemia has developed rapidly, including type 1 diabetes, children and young adults, symptoms less than three months, acutely ill patients, medication that may cause rapid rise in glucose (such as steroids), acute pancreatic damage or pancreatic surgery. ¹ It should not be used in gestational diabetes as the sensitivity is not sufficient to replace OGTT in this cohort.

HbA1c must not be used in the presence of medical conditions which affect red cell lifespan or glycation of haemoglobin. Examples of these conditions and their effects on HbA1c are given in Table 1. ¹

Table 1.

The effects of disease states on HbA1c.

	Increased HbA1c	Decreased HbA1c	Variable HbA1c
Red cell lifespan	Increased red cell survival: Previous splenectomy Iron deficiency anaemia Erythropoietin B12 treatment Iron treatment Erythropoiesis reduction	Red cell destruction: Splenomegaly Rheumatoid arthritis Haemolysis Drugs such as ribavirin or antiretrovirals	Haemoglobinopathies Hb F MetHb
Glycation	Alcoholism Chronic renal failure	Aspirin Vitamin C and E Haemoglobinopathies	Genetic heterogeneity
Analytical interference	Hyperbilirubinaemia High dose aspirin Opiates Carbamylated Hb	Hypertriglyceridaemia	Haemoglobinopathies

Whilst the analytical implications of haemoglobin variants have been well characterised, their effect on red blood cell survival or glycation may not always be recognised. Haemoglobinopathies comprise a diverse group of inherited conditions that result in a structural abnormality in globin. They are usually inherited in an autosomal recessive fashion and approximately 7% of the world population are carriers. ⁸ Homozygous patients exhibit changes in red cell lifespan caused by either aggregation of blood cells (such as in sickle cell disease) or haemolysis (such as with Hb E). They can also cause an altered synthetic rate resulting in a thalassaemia phenotype or in ‘hyperunstable’ variants Hb is degraded rapidly. ⁹ However, even in heterozygotes, mean red cell survival can be reduced. ¹⁰ The clinician may be alerted to potential confounding effects from altered red cell lifespan when anaemia is present, but a full blood count is not normally requested alongside HbA1c and patients may have normal haematology.

Glycation occurs throughout the haem molecule. HbA1c refers specifically to glycation of the valine residue on the N terminus of beta globin chains. If the amino acid substitution (in haemoglobinopathy) is near the glycation site, the rate of glycation may be reduced. ⁷ This may have an effect on HbA1c even in asymptomatic heterozygotes, it raises the question of whether analytical methods should be able to detect variants even when they do not cause analytical interference.

Analytical interferences include those caused by haemoglobin variants, elevated haemoglobin F and chemical modifications to HbA1c such as which occurs in uraemia.^1,2 Methods may be divided into those that separate haemoglobin species on the basis of charge differences (electrophoresis and ion exchange chromatography) and those that separate based on structural differences (boronate affinity and immunoassay). Haemoglobin variants may co-elute with HbA1c on chromatograms, affecting the accuracy of charge-based methods. In HbA1c immunoassay, amino acid substitutions near the N terminus of the haemoglobin beta chain may alter its antibody affinity. Boronate affinity is affected less by all interferences as m-aminophenylboronic acid reacts with the cis-diol groups of glucose bound to Hb at all its sites. It measures total glycated haemoglobin rather than HbA1c. However, there is still potential for interferences, for example, with Hb Marseille-Long Island, ¹² in which there is an amino acid substitution close to the N terminal of the beta chain, or in beta thalassaemias where beta chain production is decreased (with a compensatory increase in HbF). In both cases, formation of the major glycation product, HbA1c, is reduced resulting in a slower overall glycation rate with lower HbA1c results.^11,12

Engaging the laboratory user and ongoing education are required to ensure that these pitfalls are not forgotten. However, the role of the laboratory goes beyond this. The change to IFCC approved units following standardisation has already left a legacy where many clinicians have lost their ability to intuitively interpret the new units. The laboratory must consider how it can better aid the physician. Different test codes for HbA1c on electronic ordering systems could be used to differentiate diagnosis and monitoring, thereby reflexing different comments, diagnostic thresholds or treatment targets. Comments can prompt the physician to repeat the HbA1c in asymptomatic patients or link to guidelines. With the advent of patient access to reports and especially with the increased use of preventative programmes, there may also be a role for more educational or motivational comments. Exploring ways to communicate the result pictorially or graphically could help patient understanding.

Improved reporting is an important area for future development. All clinical biochemists should have an awareness of the analytical impact of interferences on their methods, but it is important to realise that the pitfalls are diverse and the need to educate clinicians regarding the impact of pre-analytical effects on glycation rates or red cell survival is equally vital.