Identifying Pre-Diabetes with Hemoglobin A1c
Is HbA1c Sensitive Enough to Use Alone?
By Bill Malone
In January, the American Diabetes Association officially recommended hemoglobin A1c (HbA1c) testing as an option for diagnosing and monitoring diabetes, with HbA1c levels ≥6.5% for diagnosis and levels between 5.7% and 6.4% identifying those at risk. This issue of Strategies examines a new paper that compares HbA1c to fasting plasma glucose and the 2-hour glucose tolerance test for identifying those who are at increased risk of the disease.
While hemoglobin A1c (HbA1c) is widely accepted as a reliable way to monitor treatment of diabetics and predict microvascular complications, debate over the use of this marker has persisted, with worries about sensitivity and the paucity of definitive clinical trials demonstrating its benefit compared to other testing strategies. However, the burdensome requirement for patients to fast for at least 8 hours before being tested using the traditional oral glucose tolerance test (OGTT) and fasting plasma glucose (FPG) has helped push HbA1c forward as a tactic to test the most people and deal with the increasing prevalence of diabetes.
Adding to the apprehension about HbA1c in many in the clinical community, a new study found that HbA1c performed poorly compared to OGTT and FPG in identifying individuals at increased risk of diabetes based on a one-time measurement of the three variables (Diabetes Care 2010;33:2104–2109). The study involved 855 participants in the Insulin Resistance Atherosclerosis Study (IRAS). A total of 385 of 719 (53.5%) nondiabetic individuals were at increased risk of diabetes, defined as FPG 100–125 mg/dL and OGTT 140–199 mg/dL, or HbA1c 5.7–6.4%.
HbA1c alone compared unfavorably to the other measures. The proportion of individuals in the study identified as at increased risk by HbA1c was 23.6%, compared to FPG and OGTT at 69.1% and 59.5%, respectively. The combination of FPG and OGTT detected 95.8% and the combination of IFG and HbA1c detected 75.6%. The authors also found that HbA1c detected very few non-Hispanic whites at increased risk of diabetes and was more effective with worsening BMI. “Our paper shows that using the HbA1c level recommended by the American Diabetes Association detects far fewer people at risk for diabetes than if we use the old criteria of impaired fasting glucose and impaired glucose tolerance,” said Carolos Lorenzo, MD, an author of the study. “HbA1c also missed about two thirds of all the individuals in the study with diabetes, and we know from other studies as well that HbA1c is definitely inferior to OGTT for diagnosis.” Lorenzo is a professor in the department of medicine, University of Texas Health Science Center in San Antonio, Texas.
A total of 15.9% of the individuals in the study were found to have diabetes. The HbA1c cutoff of ≥6.5% only identified 32.3% of all individuals with diabetes, while FPG ≥126 mg/dL and OGTT ≥200 mg/dL detected 44.8% and 86.8%, respectively. The highest rate of detection came from the combination of FPG and OGTT, at 97.1%. The poor performance of HbA1c in the study may be attributed to the way in which each of the three tests assesses a different aspect of glucose metabolism, Lorenzo said. “Fasting glucose concentration is related to the resistance to insulin of the liver and skeletal muscle, whereas 2-hour glucose concentration after an oral glucose load depends on the resistance to insulin of the skeletal muscle,” he said. “HbA1c changes very little in the blood during the day, and from one day to another the level is the same, even if there have been marked changes in blood glucose, because it takes time for hemoglobin to become glycosylated in the blood.”
Although clearly HbA1c differs markedly in its sensitivity to certain aspects of glucose metabolism, in fact this very difference means that the question of how HbA1c indicates diabetic risk can’t be drawn solely from data such as the current paper, emphasized William E. Winter, MD, who was not involved in the study. “In this paper, we don’t know what proportion of people identified to be at risk by each of these tests actually went on to develop diabetes,” he said. “There could be a very different risk for diabetes depending on whether you have borderline elevated FPG, borderline elevated 2-hour glucose during an OGTT, or a HbA1c value of 5.7% to 6.4%. I don’t know that we should expect HbA1c to perform equally to FPG or OGTT in the case of a one-time measurement, because each test is really assessing something different in the natural history of diabetes.” Winter is a professor in the department of pathology, immunology, and laboratory medicine at the University of Florida College of Medicine in Gainesville, Fla. Winter also cautioned that laboratorians keep in mind the fact that, in comparing HbA1c to the two other tests, the state of risk was itself defined by one of the tests among the comparison, limiting the ability to extrapolate from this data.
Lorenzo highlighted the fact that diabetes should be easy to diagnose, and that professional organizations’ guidelines should be careful not to confuse doctors about the best way to do so. “It has to be clarified that diabetes is not difficult to diagnose, and I think until now has been very clear-cut, and the doctors know how to diagnose diabetes pretty well with traditional methods,” he said. “If we add more things, we may confuse the picture, and they may think that now diabetes is only people who have high HbA1c. If we are going to use HbA1c, then let’s say when we’ll use it, or when we cannot use it, because if not, it can be very confusing.” Lorenzo also noted that some countries are unlikely to implement HbA1c because of cost, making it difficult to make comparisons between countries and chart progress internationally. “It has to be clear cut from these organizations, WHO and ADA, so as not to complicate the issue,” he said.
Winter added that the sensitivity of HbA1c should not be considered in isolation from other benefits that the test has to offer, such as better access for patients. “If you take the blanket statement: is someone at risk for diabetes? And you compare the performance of HbA1c, FPG, and OGTT, then HbA1c does not apparently perform as well as the FPG or OGTT. But if we turn it around and say, what can be done, irrespective of the fasting state of the patient, to assess risk for diabetes—would you rather have no data at all or would you rather have an HbA1c?” Both Lorenzo and Winter agreed that prospective, randomized controlled trials are needed to understand the utility of HbA1c better. “We need prospective studies with follow-up data on complications and mortality ten or fifteen years later,” said Lorenzo. “With a longitudinal study, one could hope to prove what kind of screening and early detection can make a difference in complications and mortality, but this would be difficult to do and require thousands and thousands of patients.”