Dynamic OGTT-Derived Indices Outperform Fasting HOMA-2 for Detecting Dysglycemia in Transfusion-Dependent...

Screening for dysglycemia with an annual oral glucose tolerance test (OGTT) is recommended in transfusion-dependent β-thalassemia (β-TDT), but adherence in routine practice is low. Fasting surrogate indices of β-cell function and insulin sensitivity could offer a less burdensome alternative.

To evaluate in adult β-TDT patients with normal fasting plasma glucose (FPG < 100 mg/dL): (i) the performance of fasting HOMA-2 indices (HOMA2-IR, HOMA2-%β, HOMA2-%S) and their disposition index as predictors of dysglycemia, and (ii) the comparative value of basal versus dynamic OGTT-derived markers (30-min and 1-h plasma glucose, insulinogenic index [IGI], IGI × ISI _Matsuda index and IGI/HOMA2-IR), as predictors of hyperglycemia.

A single-centre retrospective analysis of 42 β-TDT patients (19 males/23 females; mean age 29.2 ± 5.9 yr) with FPG < 100 mg/dL who underwent a standard 75-g 2-h OGTT between January 2011 and September 2025. Patients were classified as those with normal glucose tolerance (NGT; n = 19) or those with impaired glucose tolerance (IGT: n= 20) or thalassemia-related diabetes mellitus (Th-RDM: n = 3). Fasting and dynamic indices were compared by ANOVA and Mann-Whitney U test; associations were tested by Pearson/Spearman correlation and by three sequential exploratory multivariable linear regression models with 2-h plasma glucose as the dependent variable. ROC analysis with Youden's index identified optimal hypothesis-generating cut-offs.

Fasting HOMA2-IR, HOMA2-%β and HOMA2-%S did not differ significantly between NGT and hyperglycemic patients and were not independent predictors of 2-h plasma glucose. In contrast, in the total group, 30-min PG, 1-h PG, IGI, IGI × ISI _Matsuda index and IGI/HOMA2-IR all differed significantly between groups (p: ≤ 0.0082) and were inversely correlated with 2-h PG. Receiver operating characteristic (ROC) analysis and area under the curve (AUC-ROC) were used to assess diagnostic performance of the most significant variables. The ROC-AUC cut-offs for predicting dysglycemia were 122.5 mg/dL for 30-min PG (sensitivity 0.91, specificity 0.52), 134 mg/dL for 1-h PG (0.95, 0.10), 5.7 for IGI × ISI _Matsuda index (0.87, 0.47) and 1.33 for IGI/HOMA2-IR (0.87, 0.57).

Fasting HOMA-2 indices alone are not reliable predictors of dysglycemia in β-TDT patients with PG < 100 mg/dL. The dynamic IGI/HOMA2-IR ratio and the 30-min/1-h post-load plasma glucose may help risk-stratify patients with β-TDT and normal fasting glucose; however, the proposed cut-offs should be regarded as hypothesis-generating, and prospective multicenter validation is required before these indices can be used to modify current OGTT screening recommendations.