Comment on: Soranzo et al. Common Variants at 10 Genomic Loci Influence Hemoglobin A_1C Levels via Glycemic and Nonglycemic Pathways. Diabetes 2010;59:3229–3239

I am deeply interested in the subject of glycemia/HbA_1c relationships, underlying mechanisms, and impact on the assessment of glycemic control and diabetes prevalence. In the article by Soranzo et al. (1) published in Diabetes, the abstract indicates that adjusted HbA_1c would imply the reclassification of about 2% of the subjects in a general Caucasian population.

The table provided in Fig. 3 of that article is displayed below (Table 1) with the blank spaces filled with the corresponding figures and my reading of it.

• Undiagnosed diabetes according to fasting plasma glucose 593/10,110 = 5.87%.

• Undiagnosed diabetes according to measured HbA_1c ≥6.5% = (234 + 112)/10,110 = 346/10,110 = 3.42%.

• Undiagnosed diabetes according to 7 single nucleotide polymorphism (SNP)-adjusted HbA_1c ≥6.5% = (222 + 96)/10,110 = 318/10,110 = 3.15%.

At the population level, should diabetes diagnosis be performed using HbA_1c, adjustment for 7 single nucleotide polymorphisms, would imply a net reclassification of 28 out of 10,110 subjects, an absolute difference in new diagnosis of −0.27% (3.15–3.42%) and a relative reduction of −7.9% (−0.27/3.42) × 100.