Table 2. Receiver operating characteristic curves.

miRNA	AUC	95% CI	P-value	Cut-off	Sensitivity (%)	95% CI	Specificity (%)	95% CI	LR+	LR-
miR423-5p	0.67	0.52 to 0.87	0.05	N/A	N/A	N/A	N/A	N/A	N/A	N/A
miR30d	0.67	0.49 to 0.84	0.09	N/A	N/A	N/A	N/A	N/A	N/A	N/A
miR10b	0.69	0.50 to 0.87	0.09	N/A	N/A	N/A	N/A	N/A	N/A	N/A
miR126	0.81	0.68 to 0.95	<0.01	<0.037	69	39–91	84	64–95	4.3	2.7
miR423-5p + Age + Gender*	0.80	0.65 to 0.95	<0.01	<0.57	77	46–95	87	68–97	6.1	3.8
miR30d + Age + Gender*	0.81	0.68 to 0.95	<0.01	<0.52	64	35–87	87	68–97	5.1	2.4
miR10b + Age + Gender*	0.86	0.72 to 0.99	<0.01	<0.77	90	55–100	70	47–87	3.0	7.0
miR126 + Age + Gender*	0.80	0.66 to 0.94	<0.01	<0.56	85	55–98	72	51–88	3.0	4.7

Properties of receiver operator characteristic curves shows that miR126 levels can significantly discriminate between patients with low CFI (<0.39) versus high CFI (>0.39), with a p-value <0.01. In addition, in a multivariate logistic regression model with age and gender, each of the select miRNAs show significant predictive power to discriminate between patients with high or low collateral capacity.

*Multivariate logistic regression model. AUC, area under curve; CI, confidence interval; CFI: collateral flow index; LR, likelihood ratio; miRNA, microRNA; N/A, not applicable.