1004-P: The Influence of Body Mass Index in Mixed-Ancestry South Africans on the Ability of A1C and Glycated Albumin to Detect Dysglycemia

Abstract

As A1C and glycated albumin (GA) correlate in opposite directions with body mass index (BMI), combining these two biomarkers has led to improved detection of dysglycemia. As this has not been studied in Africa, we assessed in 1274 mixed ancestry South Africans in Cape Town with no prior history of diabetes (male 26%; obese 40%; mean age 48±16y), the correlation of A1C and GA with BMI and the sensitivity of A1C and GA separately and jointly in detecting OGTT diagnosed dysglycemia. A1C and GA levels were obtained. OGTT performed. Dysglycemia was defined as either newly detected diabetes (FPG≥7.0 mmol/L and/or 2hPG≥11.1 mmol/L) or prediabetes (6.1≥FPG<7.0 mmol/L and/or 7.8<2hPG<11.1 mmol/L). Dysglycemia occurred in 24% of participants (prediabetes: 17%, diabetes:7%). By robust analysis, the correlation between BMI and A1C was positive [r=0.34 [95%CI: 0.29-0.39)] and negative for BMI and GA [-0.08 (-0.13- -0.03)]. The C-statistic for A1C diagnosis of dysglycemia in the total cohort was 0.77 (95%CI: 0.73-0.80) and for GA was 0.67 (0.64-0.71). In the nonobese the C-statistic for A1C was 0.71 (0.65-0.76) and for GA was 0.68 (0.62-0.73). In the obese the C-statistic for A1C was 0.79 (0.74-0.83) and for GA was 0.69 (0.64-0.74). Optimal thresholds by Youden index for dysglycemia diagnosis by A1C was 6.0% (95%CI: 5.8-6.2) and for GA was 13.44% (12.72-14.71). At these thresholds, sensitivities for A1C-alone, GA-alone, and A1C+GA in the total cohort were 63% (95%CI: 57-68), 53% (47-59) and 76% (71-81); in the nonobese were 51% (42-60), 55% (46-63) and 69% (60-76); and in the obese 72% (65-78), 52% (44-59) and 82% (75-87), resp. In short, this is the first demonstration in Africa of the positive correlation between A1C and BMI and the negative correlation between GA and BMI. The inverse correlations of A1C and GA with BMI may explain why there is improved detection of dysglycemia in the nonobese when the two biomarkers were combined and has important implications for screening. Disclosure A. Kengne: None. T. E. Matsha: None. D. B. Sacks: Research Support; Self; Sebia, Trinity Biotech plc. A. E. Zemlin: None. R. T. Erasmus: None. A. E. Sumner: None.