HbA1c and Diabetes: Mismatches and Misclassifications

While HbA1c, a surrogate measure of average glucose over 2–3 months, has become the mainstay for the diagnosis of diabetes and for guiding management, an increasing number of studies have demonstrated interindividual variability in the relationship between HbA1c and underlying glucose levels. This has sometimes been referred to as a “glycation gap” because the discordance involves glycated hemoglobin in red blood cells. Because the underlying mechanism has not yet been elucidated, “mismatch” between HbA1c and glucose might be a better descriptor. These differences between HbA1c and glucose levels have been observed not only in individuals with preexisting diabetes, both type 1 and type 2, but also in those with prediabetes and those without either diabetes or prediabetes. Moreover, having a higher HbA1c/glucose mismatch appears to be associated with certain demographic characteristics, including older age, black race, and higher body mass index. There have also been reports that mismatches are associated with increased risk for diabetes complications (1), although such an association was not found to be independent of HbA1c levels in an analysis of the Diabetes Control and Complications Trial (2).

The underlying etiology of mismatches is not well understood. As the HbA1c assays have become more standardized, differences in assays and issues with accuracy are unlikely to be major contributors. Additionally, whether using fasting plasma glucose levels (3), 7-point glucose self-monitoring during the day (4), or 24-hour continuous glucose monitoring data over 12 weeks (5) to represent the average glucose levels, mismatches between HbA1c and glucose persist. We are left then with genetic and physiologic variability as potential explanations, such as differences in red blood cell lifespan or differences in intracellular glucose glycation rates.

The current literature mostly describes differences in HbA1c relative to glucose levels found in individuals with diabetes, and related implications for diabetes management. However, few studies have reported on the potential impact of HbA1c/glucose mismatches on the diagnostic classification of diabetes and prediabetes. In this issue of The Journal of Clinical Endocrinology & Metabolism, findings reported by Hsia and colleagues (6) from the Vitamin D and Type 2 Diabetes study add to our understanding by describing differences in HbA1c levels, fasting plasma glucose levels (FPG) and 2-hour postload glucose levels (2hrPG during 75-g oral glucose tolerance tests [OGTTs]) in individuals with screen-positive prediabetes. The Vitamin D and Type 2 Diabetes study, a National Institutes of Health-sponsored, multicenter, randomized control trial, recruited subjects with prediabetes, defined as meeting 2 of the 3 American Diabetes Association-recommended criteria (FPG, 2-hour OGTT, HbA1c). Using the approach of Hempe et al (3), the Hemoglobin Glycation Index (HGI) served as the measure of HbA1c/glucose mismatches, where the linear regression of the measured HbA1c and FPG of the screening cohort (n = 6,829) was used to derive the predicted HbA1c for each subject according to his or her FPG. The HGI for each enrolled subject (n = 3,945) was calculated as the difference between the predicted HbA1c and the observed HbA1c, and the cohort was then grouped into HGI tertiles: low, moderate, and high.

They found that with increasing HGI tertiles, HbA1c levels tended to be higher (5.7%, 5.9%, and 6.1%, respectively), whereas FPG levels tended to be lower (112, 106, and 101 mg/dL), with no differences in 2hrPG. When stratified by HGI tertiles, use of FPG or HbA1c for diagnosis led to significant differences in diabetes/prediabetes classification, whereas classification using the 2hrPG was unaffected. When the diagnosis was based on HbA1c levels, the moderate and high HGI tertile groups were mostly classified as having diabetes or prediabetes, and classification as having normal glucose metabolism was infrequent. The low HGI group, however, was mostly classified as having normal glucose metabolism or prediabetes, and classification as having diabetes was infrequent. However, when the diagnosis was based on FPG, the moderate and high HGI groups were mostly classified as having normal glucose metabolism or prediabetes, and classification as having diabetes was infrequent, whereas the low HGI group was mostly classified as having diabetes or prediabetes, and classification as having normal glucose metabolism was infrequent. These findings provide compelling evidence that mismatches between HbA1c and glucose levels can lead to differences in diagnostic classification when the diagnosis is based on use of HbA1c alone compared with use of glucose levels.

Two other studies have examined the effect of HbA1c/glucose mismatches on diagnostic classification (7, 8). In both, when based on HbA1c levels, individuals with higher HbA1c/glucose mismatches tended to be classified as having diabetes and prediabetes, whereas those with lower HbA1c/glucose mismatches tended to be classified as having normal glucose metabolism and prediabetes. These findings were generally similar to those reported by Hsia et al despite racial/ethnic differences in the cohorts (26% black (6) vs. 65% black (7) vs. 100% Spanish (8)), and differences in the definition of mismatches (HbA1c/FPG (6) vs. HbA1c/average glucose [from an OGTT and a glucose challenge test] (7) vs. HbA1c/fructosamine (8)). However, the present and previous studies differ in a few respects: (1) use of different diagnostic standards for diabetes/prediabetes (glucose levels [FPG, 2hrPG] alone (7, 8) vs. glucose and HbA1c levels in the present study); (2) use of data from screening studies (7, 8) vs. data from a population targeted as high risk for diabetes, in the present study; and (3) use of separate regressions by glucose classification to determine HGI (because of observed differences between normal glucose metabolism, prediabetes, and diabetes groups) (7) vs. a single regression (8) in the present study.

The report by Hsia et al is an important addition to our understanding of the clinical relevance of HbA1c mismatches and provides further evidence that HbA1c/glucose mismatches can lead to misclassification of diabetes/prediabetes when using the HbA1c test alone for diagnosis. The findings from Hsia et al, together with the previous reports, raise critical questions with regard to current practice. Since the Diabetes Control and Complications Trial, HbA1c levels have been widely used to guide management decisions in individuals with diagnosed diabetes, and over the past 10 years, have become a preferred and convenient screening test for both prediabetes and diabetes. However, because it is now clear that HbA1c/glucose mismatches are real and could lead to errors in both diagnostic classification and related management decisions, it will be important to improve our understanding of the mechanistic basis for mismatches, and how to account for mismatches in our approaches to diagnosis and management. Future developments should include methods for identifying patients who are likely to have mismatches, and practical tools to quantify and adjust for mismatches in individual patients. With the growing recognition of HbA1c/glucose mismatches and their implications, we will need to combine use of both glucose and HbA1c measurements to ensure accurate diagnoses and optimal management—to move toward a more individualized, “precision medicine” approach to improving care and outcomes.

Additional Information

Disclosure Summary: M.R. declares the potentially competing interests of having funding from the National Institutes of Health and the Veterans Administration, as well having research support from Boehringer-Ingelheim.