Response to Comment on Hempe et al. The Hemoglobin Glycation Index Identifies Subpopulations With Harms or Benefits From Intensive Treatment in the ACCORD Trial. Diabetes Care 2015;38:1067–1074

Drs. Riddle and Gerstein (1) suggest renaming the hemoglobin glycation index (HGI) to reflect the possibility that interindividual variation in HGI in the ACCORD trial could be an artifact of how blood glucose was measured. HGI is the difference between an individual’s observed HbA_1c and a predicted HbA_1c derived by inserting a time-matched blood glucose measurement into a regression equation describing the linear population relationship between blood glucose and HbA_1c. In our analysis of ACCORD (2), we used baseline HbA_1c and fasting plasma glucose (FPG) to show that high HGI at baseline (i.e., HbA_1c higher than predicted by FPG) was associated with worse outcomes. Riddle and Gerstein correctly note that FPG provides no information about person-to-person differences in glucose control during the day. One might thus conclude that high HGI calculated using FPG was associated with adverse outcomes in ACCORD because HbA_1c reflects person-to-person differences in blood glucose dynamics that are missed by FPG.

But this suggestion ignores prior studies showing that adverse outcomes are associated with high HGI when more comprehensive assessments of blood glucose dynamics are used to calculate HGI. For example, HGI based on mean blood glucose derived from seven-point profile measurements was positively associated with microvascular disease in the Diabetes Control and Complications Trial (DCCT) (3). The glycation gap (4), calculated the same way as HGI except that glycated serum protein replaces blood glucose in the regression equation, is positively associated with both HGI and adverse diabetes outcomes. Persistent person-to-person differences in HGI and the glycation gap have been observed in enough studies (5) that the question is not whether the phenomenon exists, but why.

Riddle and Gerstein’s concern highlights the need to determine if persistent person-to-person differences in HbA_1c measured by HGI are of analytical or biological origin. Glycohemoglobin standardization programs make HbA_1c an unlikely source of analytical bias, except perhaps as bias due to differences in erythrocyte turnover rates. Otherwise, HGI could only be an analytical artifact if the method used to estimate blood glucose concentration produced results that were persistently lower or higher than true blood glucose in some patients but not others. Although conceptually possible with FPG or patient meter data, this explanation seems unlikely in studies where blood glucose was estimated based on glycated serum protein or mean glucose from profile sets or continuous glucose monitoring.

Ultimately, either HbA_1c reflects hemoglobin exposure to glucose the same way in everyone or it does not. If it does, then HGI is an analytical artifact. If not, then it behooves us to delve more deeply into the biochemistry of nonenzymatic hemoglobin glycation in search of the underlying mechanisms. Given the historical precedent, we see no reason to change the name of the hemoglobin glycation index. We agree with Riddle and Gerstein, however, that HGI should be considered a “clinically helpful trigger for reassessment of both glycemic targets and treatment tactics for individual patients” regardless of the source of population variation in HbA_1c measured by HGI.