Screening for Turner Syndrome-Associated Hyperglycemia: Evaluating Hemoglobin A1c and Fasting Blood Glucose

Abstract

Introduction: Individuals with Turner syndrome (TS) are at increased risk of developing diabetes mellitus (DM). Currently, annual DM screening with hemoglobin A1c (HbA1c) with or without fasting blood glucose (FBG) is recommended starting at age 10. However, the optimal DM screening for individuals with TS is not known. The purpose of this study was to evaluate the correlation between HbA1c, FBG, and the 2-h oral glucose tolerance test (OGTT). A second goal was to query whether optimal HbA1c and FBG cut points for TS-associated DM and impaired glucose tolerance (IGT), as defined by the OGTT 2-h blood glucose (BG), might differ from those for the general population.

Methods: Individuals with TS ≥ age 10 from the TS: Genotype Phenotype study in the National Institute of Child Health and Human Development's Data and Specimen Hub (DASH) who had 2-h OGTT BG, HbA1c, and FBG were included. Correlations between HbA1c, FBG, and 2-h OGTT BG were evaluated. Areas under the receiver operative characteristic (ROC-AUC) curves were generated. Optimal cut points for predicting TS-associated IGT (2-h BG ≥7.77 mmol/L) and DM (2-h BG ≥11.10 mmol/L) were determined.

Results: 348 individuals had complete data (2-h OGTT BG <7.77 mmol/L, n = 239; TS-associated IGT, n = 79; DM, n = 30). ROC-AUC was poor for HbA1c to predict IGT (0.57, 0.49-0.65) but better for DM (0.81, 0.71-0.90). ROC-AUC was also poor for FBG to predict IGT (0.63, 0.56-0.70) but better for DM (0.85, 0.77-0.93). At a cut point of 38 mmol/mol (5.6%), HbA1c had 67% sensitivity (95% CI: 47-83%) and 86% specificity (95% CI: 82-90%) for identifying TS-associated DM defined by 2-h OGTT BG.

Conclusions: The correlation of HbA1c and 2-h OGTT BG is lower in TS than in other published studies regarding type 2 DM. HbA1c is fairly specific for DM in TS but lacks sensitivity, especially at currently utilized levels. Future research should focus on characterizing individuals with TS whose glycemic status is discordant, as this may provide additional insights into the pathophysiology of glucose metabolism in TS. Longitudinal assessment of glycemia as it relates to micro- and macrovascular complications in individuals with TS will further inform DM screening in this population.

Keywords: Diabetes mellitus; Fasting blood glucose; Hemoglobin A1c; Oral glucose tolerance test; Turner syndrome.

Fig. 1.

Scatterplots demonstrating the correlation of 2-hour OGTT BG and HbA1c (1a) and of 2-hour OGTT and FBG (1b). OGTT= oral glucose tolerance test, BG = blood glucose , FBG = fasting blood glucose, HbA1c = hemoglobin A1c

Fig. 1.

Scatterplots demonstrating the correlation of 2-hour OGTT BG and HbA1c (1a) and of 2-hour OGTT and FBG (1b). OGTT= oral glucose tolerance test, BG = blood glucose , FBG = fasting blood glucose, HbA1c = hemoglobin A1c

Fig. 2.

ROC curves for HbA1c (2a) and FBG (2b). Values of HbA1c (2a) and FBG (2b) are included along the ROC curve. ROC-AUC was poor for both markers in detecting individuals with IGT on OGTT. ROC-AUC improved for detecting DM, with FBG having the best performing ROC curve. ROC = receiver operator characteristic, AUC = area under the curve, FBG = fasting blood glucose, IGT = impaired glucose tolerance, DM = diabetes mellitus

Fig. 2.

ROC curves for HbA1c (2a) and FBG (2b). Values of HbA1c (2a) and FBG (2b) are included along the ROC curve. ROC-AUC was poor for both markers in detecting individuals with IGT on OGTT. ROC-AUC improved for detecting DM, with FBG having the best performing ROC curve. ROC = receiver operator characteristic, AUC = area under the curve, FBG = fasting blood glucose, IGT = impaired glucose tolerance, DM = diabetes mellitus

Fig. 3.

ROC curves using HbA1c plus additional potential predictors for 2 hour OGTT status. The black line in Figure 3a is the unadjusted ROC curve for HbA1c in predicting IGT. The red line is the adjusted ROC, which was adjusted to include FBG. Adding FBG to HbA1c had minimal impact on ROC-AUC compared to HbA1c alone (0.57 to 0.62). The black line in figure 3b is the unadjusted ROC for HbA1c in predicting IGT. The red line is the adjusted ROC in predicting IGT, which was adjusted to include age, BMI, and HOMA-IR. The purple line is the ROC of the covariates only in predicting IGT. Adding age, BMI, and HOMA-IR to HbA1c improved the ROC-AUC from 0.57 to 0.78. However, the improved ROC-AUC is driven entirely by the covariates, in that removing HbA1c from the model still resulted in an ROC-AUC of 0.78.

Fig. 3.

ROC curves using HbA1c plus additional potential predictors for 2 hour OGTT status. The black line in Figure 3a is the unadjusted ROC curve for HbA1c in predicting IGT. The red line is the adjusted ROC, which was adjusted to include FBG. Adding FBG to HbA1c had minimal impact on ROC-AUC compared to HbA1c alone (0.57 to 0.62). The black line in figure 3b is the unadjusted ROC for HbA1c in predicting IGT. The red line is the adjusted ROC in predicting IGT, which was adjusted to include age, BMI, and HOMA-IR. The purple line is the ROC of the covariates only in predicting IGT. Adding age, BMI, and HOMA-IR to HbA1c improved the ROC-AUC from 0.57 to 0.78. However, the improved ROC-AUC is driven entirely by the covariates, in that removing HbA1c from the model still resulted in an ROC-AUC of 0.78.