[Construction of a risk predictive model of acute kidney injury based on urinary tissue inhibitor of metalloproteinase 2 and insulin-like growth factor-binding protein 7 and its early predictive value in critically ill patients].

Abstract

Objective: To establish a risk predictive model nomogram of acute kidney injury (AKI) in critically ill patients by combining urinary tissue inhibitor of metalloproteinase 2 (TIMP2) and insulin-like growth factor-binding protein 7 (IGFBP7), and to verify the predictive value of the model.

Methods: A prospective observational study was conducted. The patients with acute respiratory failure or circulatory disorder admitted to the intensive care unit (ICU) of Northern Jiangsu People's Hospital from November 2017 to April 2020 were enrolled. The patients were enrolled within 24 hours of ICU admission, and their general conditions and relevant laboratory test indicators were collected. At the same time, urine was collected to determine the levels of biomarkers TIMP2 and IGFBP7, and TIMP2×IGFBP7 was calculated. Patients were divided into non-AKI and AKI groups according to whether grade 2 or 3 AKI occurred within 12 hours after enrollment. The general clinical data and urinary TIMP2×IGFBP7 levels of patients between the two groups were compared. The indicators with P < 0.1 in univariate analysis were included in the multivariate Logistic regression analysis to obtain the independent risk factors for grade 2 or 3 AKI within 12 hours in critical patients. An AKI risk predictive model nomogram was established, and the application value of the model was evaluated.

Results: A total of 206 patients were finally enrolled, of whom 54 (26.2%) developed grade 2 or 3 AKI within 12 hours of enrollment, and 152 (73.8%) did not. Compared with the non-AKI group, the patients in the AKI group had higher body mass index (BMI), pre-enrollment serum creatinine (SCr), urinary TIMP2×IGFBP7 and proportion of using vasoactive drugs, and additional exposure to AKI (use of nephrotoxic drugs before enrollment) was more common. Multivariate Logistic regression analysis showed that BMI [odds ratio (OR) = 1.23, 95% confidence interval (95%CI) was 1.10-1.37, P = 0.000], pre-enrollment SCr (OR = 1.01, 95%CI was 1.00-1.02, P = 0.042), use of nephrotoxic drugs (OR = 2.84, 95%CI was 1.34-6.03, P = 0.007) and urinary TIMP2×IGFBP7 (OR = 2.19, 95%CI was 1.56-3.08, P = 0.000) was an independent risk factor for the occurrence of grade 2 or 3 AKI in critical patients. An AKI risk predictive model nomogram was constructed based on the independent risk factors of AKI. Bootstrap validation results showed that the model had good discrimination and calibration in internal validation. Receiver operator characteristic curve (ROC curve) analysis showed that the area under the ROC curve (AUC) of urinary TIMP2×IGFBP7 alone in predicting grade 2 or 3 AKI within 12 hours in critical patients was 0.74 (95%CI was 0.66-0.83), the optimal cut-off value was 1.40 (μg/L) ²/1 000 (sensitivity was 66.7%, specificity was 85.0%), and the predictive performance of the model incorporating urinary TIMP2×IGFBP7 was significantly better than that of the model without urinary TIMP2×IGFBP7 [AUC (95%CI): 0.85 (0.79-0.91) vs. 0.77 (0.70-0.84), P = 0.005], net reclassification index (NRI) was 0.29 (95%CI was 0.08-0.50, P = 0.008), integrated discrimination improvement (IDI) was 0.13 (95%CI was 0.07-0.19, P < 0.001).

Conclusions: The AKI risk predictive model based on urinary TIMP2×IGFBP7 has high clinical value and is expected to be used to early predict the occurrence of AKI in critically ill patients.