Evidence of the Month

Commentaries on both new and classic studies of importance for the treatment of diabetes are posted here monthly.

Toward a Common Language of Glucose: HbA_1c in Relation to Day-to-Day Monitoring

Eberhard Standl Munich Diabetes Research Group/Diabetes Research Unit Munich, Germany	Click here to download PowerPoint slides of this commentary.
	Click here for the programme evaluation and post-test.

Background
The HbA_1c assay has for 25 years been the gold standard for measuring blood glucose in people with Type 1 or Type 2 diabetes. It serves as the basis for assessing the risk of long-term diabetic complications and, in turn, for setting and adjusting individual glucose targets to achieve risk reductions. Unlike contemporaneous readings of blood glucose, as obtained by self-monitoring of capillary glucose or continuous blood glucose monitoring, the HbA_1c assay measures the degree to which glucose has attached to haemoglobin molecules in the blood over time. It is reported as the percent of glycated haemoglobin molecules, whereas blood glucose concentrations are recorded as mg/dl or mmol/l.

A long-held goal of the diabetes community has been the ability to discuss chronic and current blood glucose findings in a common language, particularly for the benefit and education of individuals under care. Although the HbA_1c has been assumed to reflect average blood glucose levels in the preceding several months, the relationship between HbA_1c and average glucose (AG) has never been mathematically clear enough for diabetes organizations to endorse a change in HbA_1c reporting practices (Consensus Committee, 2007). The main problem in previous research has been an insufficient number of day-to-day glucose readings to reliably estimate AG; most studies have also been homogeneous, focusing mainly on Type 1 diabetes (Rohlfing et al, 2002; Svendsen et al, 1982). The international A1c-Derived Average Glucose (ADAG) study was designed to confirm the relationship by accumulating more complete glucose data, including extensive readings from continuous glucose monitoring, in a larger population of individuals with Type 1 or Type 2 diabetes or no diabetes. As reported in 2008, the study produced proof of a mathematical relationship between HbA_1c and the average 3-month glucose level and provided a formula to reliably convert HbA_1c into AG for most individuals with diabetes.

Methods and Key Results
The ADAG study included for analysis 507 participants—268 with Type 1 diabetes, 159 with Type 2 diabetes, and 80 without diabetes—from 10 centers in the United States, Europe, and Africa. Children, pregnant women, and people with anaemia or altered erythrocyte turnover were excluded, as was anyone with unstable control of diabetes. HbA_1c levels at baseline ranged from 4 % to >8.5 %, but 44% were between 6.6 % and 8.5 %. The study did not require any change in usual diabetes management.

Continuous interstitial glucose monitoring (CGM), performed for ≥2 days at baseline and then every 4 weeks. Readings had to include at least one successful 24-hour profile and be free of major gaps. Participants with fewer than 7 CGMs were excluded from final analyses.

Fingerstick capillary glucose monitoring, performed 8 times per day (pre-meal, 90 minutes post-meal, pre-bed, and 3:00 a.m.) at the time of CGM; otherwise, 7 times per day (minus the 3 a.m. reading) ≥3 days per week. Readings were electronically transferred to a central database.

HbA_1c assays, based on blood samples obtained at baseline and monthly and performed at a central laboratory. Mean HbA_1c was derived from the findings of 4 different assays.

Approximately 2700 glucose values (~2500 CGM and 230 fingerstick readings) were obtained for each individual. AG for each individual was calculated by combining the CGM and fingerstick readings (weighting CGM for proportionality with fingersticks). A priori, the ADAG investigators decided that for the study results to be acceptable, ≥90% of each individual’s calculated AG would have to fall within ±15% of the calculated study-wide AG.

The ADAG investigators applied linear and quadratic regression models to estimate the relationship between HbA_1c and AG, finding that a simple linear equation provided the tightest correlation:

AG_mg/dl = 28.7 x HbA_1c − 46.7, or AG_mmol/l = 1.59 x HbA_1c − 2.59 (R²=0.84, P<0.0001)

For the entire study group, standard deviation of the prediction error was 15.7 mg/dl (0.87 mmol/l). Fulfilling the a priori criterion, 89.95% of the individual glucose estimates fell within the ±15% range of the equation’s regression line. The investigators used the equation to calculate estimated AG as follows:

Clinical Implications
The chance for clinicians to allay confusion over HbA_1c appears to be at hand with the findings of the ADAG study. A consensus committee of the American Diabetes Association (ADA), European Association for the Study of Diabetes, International Federation of Clinical Chemistry and Laboratory Medicine, and International Diabetes Federation stated in 2007 that a change in HbA_1c reporting awaited only the confirmatory findings of the ADAG, whereupon ‘an A1C-derived average glucose value will also be reported as an interpretation of the A1C results’ (Consensus Committee, 2007).

The ADA already offers an online calculator to convert HbA_1cs to estimated AGs based on the ADAG equation (www.diabetes.org/AG). The expectation is that translation of HbA_1c into more familiar units will enhance diabetes education, allowing individuals with diabetes to grasp more intuitively how their daily glucose readings relate to their treatment goals and how close they are to reaching those goals. So, for practical purposes assuring these educational objectives, this would suggest that readings of HbA_1c and the derived AG values should be used in parallel to complement each other in improving the understanding of either parameter. It would be useful if this could be evaluated and confirmed in further studies. It should also be noted that the relationship between HbA_1c and AG was consistent across subgroups of participants in the ADAG study—so much so that the investigators concluded that for ‘many, if not most’ individuals, there are ‘no important factors’ that affect the relationship between HbA_1c and AG levels. However, African-American ethnicity may be an exception. Borderline evidence suggested that African-Americans have a lower mean glucose level for a given HbA_1c value.

The ADAG findings pertain only to individuals with stable diabetes and may not be applicable to pregnant women, children, people with anaemia, and ethnic groups that were under-represented in the study. As further evidence is gathered about these groups, the current findings appear to provide the needed mathematical foundation for bringing both scientifically correct and clinically helpful information on glucose to the diabetes setting.