A prospective diagnostic study investigating urinary biomarkers of AKI in major abdominal surgery (the AKI-biomas...

Post-operative acute kidney injury (AKI) is associated with increased morbidity and mortality with evidence suggesting that early identification using biomarkers of AKI may impact prognosis. Most studies in surgical patients has focussed on cardiac, vascular and transplant surgery cohorts. Evidence on the utility of biomarkers in major abdominal surgery is sparse.

This was a prospective observational single centre diagnostic study conducted on 488 patients undergoing major abdominal surgery. Urine was collected four hours post-surgery. The biomarkers for AKI NGAL, KIM-1, DKK-3 and IGFBP-7*TIMP-2 were measured and diagnostic performance assessed utilising Receiver Operating Characteristic (ROC) curve analysis to predict the development of post operative AKI using serum creatinine and urine output criteria.

242 participants developed AKI by urine output criteria (49.5%) and 43 by serum creatinine criteria (8.8%). The area under the receiver operating characteristic curve values for stage 1 AKI as determined by serum creatinine criteria for NGAL was 0.741 (95%CI 0.699-0.770, p < 0.001) and 0.871 (95%CI 0.838-0.899, p < 0.001) for stage 2. AUC values for IGFBP-7*TIMP-2 for stage 1 were 0.655 (95% CI 0.611-0.697, p0.003) and stage 2 0.803 (95%CI 0.764-0.837 p0.002). The AUC for KIM-1 was statistically significant for stage 1 (0.68, 95%CI 0.637-0.722) but not for stage 2. No AUC values for DKK-3 were statistically significant. Biomarkers performed poorly for prediction of AKI by urine output criteria.

In this large prospective study of a clinical cohort of 488 patients undergoing major abdominal surgery AKI rates are dependent on the criteria used with 49.5% of patients developed AKI by urine output criteria, compared to only 8.8% by serum creatinine. NGAL and IGFBP-7*TIMP-2 showed reasonable diagnostic performance when diagnosing AKI by serum creatinine criteria, with NGAL returning the highest AUC values.