Table 4.

The AUCs, sensitivity, and specificity of plasma miRNAs for detecting prediabetes and type 2 diabetes.

Target gene	AUC	95 % CI	p-value	YI	Criterion	Sensitivity	Specificity
Normal vs Prediabetes
miR-9	0.676	0.438–0.913	0.123	0.436	≥1.87	55.6	88.0
miR-29a	0.627	0.391–0.862	0.266	0.324	≥3.36	44.4	88.0
miR-192	0.747	0.556–0.937	0.032	0.498	≥3.01	77.8	72.0
miR-375	0.676	0.471–0.881	0.123	0.489	≥2.57	88.9	60.0
miR-9+29a+192+375	0.782	0.555–1.00	0.013	0.618	≥0.33	77.8	84.0
Normal vs Diabetes
miR-9	0.827	0.677–0.977	0.000	0.630	≥2.02	75.0	88.0
miR-29a	0.950	0.864–1.000	0.000	0.877	≥4.19	91.7	96.0
miR-192	0.890	0.787–0.993	0.000	0.673	≥5.61	83.3	84.0
miR-375	0.887	0.781–0.992	0.000	0.673	≥11.01	83.3	84.0
miR-29a+375	0.960	0.892–1.000	0.000	0.877	≥0.25	91.7	96.0
Prediabetes vs Diabetes
miR-9	0.662	0.421–0.903	0.214	0.334	≥2.18	66.7	66.7
miR-29a	0.833	0.632–1.000	0.009	0.639	≥6.16	75.0	88.9
miR-192	0.648	0.406–0.890	0.256	0.277	≥4.77	83.3	44.4
miR-375	0.787	0.570–1.000	0.028	0.556	≥6.28	100.0	55.6
miR-9+29a+375	0.852	0.671–1.000	0.007	0.695	≥0.37	91.7	77.8

The AUCs for each plasma miRNA were determined using non-parametric methods. The 95 % confidence intervals for AUCs were determined using the binomial exact confidence interval method. The Youden index (YI) values were calculated using the formula: YI= Sensitivity-(1-Specificity). The sets of combined microRNAs included under each study group represent the combinations that yielded the highest AUCs out of all possible combinations. All tests were two-tailed and a p-value <0.05 was considered statistically significant. ROC: Repeater Operating Characteristic, AUC: Area under the curve, CI: Confidence interval, YI: Youden index.