Cell cycle arrest biomarkers for predicting renal recovery from acute kidney injury: a prospective validation study

Annals of Intensive Care

Table 4 Biomarkers and combination models for predicting non-recovery from AKI

	AUC (95% CI)	Cutoff value	p value
[TIMP-2]*[IGFBP7] day 0 [(ng/mL)²/1000]	0.751 (0.701, 0.852)	1.05	< 0.001
TIMP-2 day 0 (ng/mL)	0.744 (0.688, 0.850)	8.50	< 0.001
IGFBP7 day 0 (ng/mL)	0.721 (0.623, 0.820)	117.60	< 0.001
[TIMP-2]*[IGFBP7] day 1 [(ng/mL)²/1000]	0.668 (0.551, 0.785)	0.89	0.028
TIMP-2 day 1 (ng/mL)	0.653 (0.499, 0.726)	7.50	0.050
IGFBP7 day 1 (ng/mL)	0.603 (0.482, 0.725)	144.90	0.163
Clinical risk prediction model	0.722 (0.640, 0.802)	0.436	< 0.001
([TIMP-2]*[IGFBP7] day 0)—clinical risk prediction model	0.852 (0.750, 0.891)	0.290	< 0.001
(TIMP-2 day 0)—clinical risk prediction model	0.822 (0.744, 0.900)	0.224	< 0.001
(IGFBP7 day 0)—clinical risk prediction model	0.805 (0.725, 0.886)	0.180	< 0.001

Clinical risk prediction model consisting of AKI diagnosed by UO criteria, AKI stage 2–3 and nonrenal SOFA score
AUC area under the receiver operating characteristic, CI confidence interval, AKI acute kidney injury, TIMP-2 tissue inhibitor of metalloproteinases-2, IGFBP-7 insulin-like growth factor-binding protein 7