Accuracy of glutamic acid decarboxylase antibodies for the identification of adult-onset type 1 diabetes mellitus: a systematic review and meta-analysis

Introduction The glutamic acid decarboxylase autoantibody (GADA) test is a widely used marker to differentiate type 1 diabetes and type 2 diabetes in adults. We conducted a systematic review to estimate the sensitivity and specificity of the presence of GADA for T1D in adults with newly diagnosed diabetes. Methods We conducted a PubMed and Embase search for studies that report both the GADA result and type 1 versus type 2 diabetes in adults diagnosed with diabetes. We calculated the sensitivity and specificity for each study. Results We identified 19 studies involving 11,760 patients from diverse geographic settings. Across these studies, the sensitivity of GADA for identifying adult-onset type 1 diabetes varied widely (range: 0.27 to 0.83), with a pooled estimate of 0.53 (95% CI: 0.46–0.60). In contrast, specificity was consistently high, with a pooled estimate of 0.93 (95% CI: 0.89–0.96). The positive and negative likelihood ratios were 7.3 (95% CI 4.8 to 11.3) and 0.51 (0.44 to 0.58), respectively. Discussion Our review demonstrates that GADA has high specificity and moderate sensitivity for identifying adult-onset type 1 diabetes. As a limitation, factors such as assay choice and cut-off values, as well as heterogeneity of both the type 1 and type 2 diabetes groups with regard to unmeasured genetic influences may contribute to the variability in antibody prevalence between studies. Our pooled likelihood ratios for GADA results might be useful for developing clinical and algorithmic tools to distinguish adult-onset type 1 diabetes from type 2 diabetes.