Early Predication Of Persistent Acute Kidney Injury In Critical Ill Adult Patients

Acute kidney injury(AKI)is a syndrome and defined as an abrupt decrease in kidney function due to various causes.AKI occurs in approximately 10-15% of patients admitted to hospital,while its incidence in intensive care has been reported in more than 50% of patients.AKI duration is associated with short and long term prognosis.Early recognition of AKI without short-term recovery might help to optimize renal management and implement kidney protection,which then can improve the outcome.Several indicators for early recognition and prediction of persistent AKI have been proposed and validated,such as fractional excretion of urea,neutrophil gelatinase-associated lipocalin(NGAL),tissue inhibitor of metalloproteinases-2 and insulin-like growth factor-binding protein 7([TIMP-2]×[IGFBP7]).However,these indicators perform poorly.Doppler-based renal resistive index(RRI)measurement is a rapid,repeatable and noninvasive tool that was proposed for assessing kidney perfusion,predicting kidney dysfunction and differentiating between transient and persistent AKI in ICU patients.However,large heterogeneity between these studies was noted and the predictive role of RRI for persistent AKI remained controversial.CCL14 is a member of the chemokine family of small molecules that were initially recognized for roles in leukocyte chemotaxis and are implicated in tissue injury and repair processes.A more recent study demonstrated that an elevation in urinary CCL14 predicted the persistence/progression of AKI well and showed an obvious advantage comparing with other biomarkers.This was the first time that CCL14 as a new predictor was reported and it provided a new tool for predicting kidney dysfunction.However,it is unknown whether the same diagnostic efficacy of urinary CCL14 exists for patients with different AKI stages across heterogeneous patients.In addition,the context of a diagnostic test is a critical component for the interpretation of its result.The renal angina defined using a combination of AKI risk factors and clinical symptom of kidney injury aimed to help with AKI risk stratification,provide an appropriate context for diagnostic tools or biomarkers and improve their effectiveness.Based on the concept of “renal angina”,persistent AKI risk index(PARI)has been proposed,which had a good performance in predicting persistent AKI.However,PARI has not been validated in other studies and whether it could improve prediction provided by a combination with biomarkers has not been assessed.In summary,we first performed a retrospective study based on data from a prospective study in a single-center general ICU between November 2017 to October 2018 to evaluate the performance of RRI,clinical indicators,and their combinations to predict short-term kidney prognosis in septic shock patients.Next,we conducted a prospective observational study between June 2021 and January 2022 to assess the predictive ability of urinary CCL14 and RRI in predicting persistent AKI among patients with different AKI stages.In addition,a second analysis of this prospective observational study was implemented to validate the performance of a modified persistent AKI risk index(PARI)and its combinations with RRI or urinary CCL14 for the prediction of persistent AKI in critically ill adults.Part One Performance of the renal resistive index and usual clinicalindicators in predicting persistent AKIObjective: To evaluate the performance of renal resistive index,clinical indicators,and their combinations to predict short-term kidney prognosis in septic shock patients.Methods: We performed a retrospective study based on data from a prospective study in a single-center general ICU between November 2017 and October 2018.Patients with septic shock were included.Clinical indicators were evaluated immediately at inclusion,and renal RI was measured within the first 12 hours of ICU admission after hemodynamic stabilization.Persistent AKI was defined as AKI without recovery within 72 hours.A multivariable logistic regression was used to select significant variables associated with persistent AKI.The discriminative power was evaluated by a receiver operating characteristic curve analysis.Results: Overall,102 patients were included,39 of whom had persistent AKI.Renal RI was higher in the persistent AKI patients than in those without AKI: 0.70±0.05 vs.0.66±0.05;p=0.001.The performance of RI to predict persistent AKI was poor,with an area under the receiver operating characteristic curve(AUROC)of 0.699 [95% confidence interval(CI)0.600-0.786].A clinical prediction model combining serum creatinine at inclusion and the nonrenal SOFA score showed a better prediction ability for nonrecovery,with an AUROC of 0.877(95% CI 0.797–0.933,p=0.0012).The addition of renal RI to this model did not improve the predictive performance.Conclusions: The Doppler-based renal resistive index performed poorly in predicting persistent AKI and did not improve the clinical prediction provided by a combination of serum creatinine at inclusion and the nonrenal SOFA score in patients with septic shock.Part Two Performance of urinary C-C motif chemokine ligand 14 andrenal resistive index in predicting persistent AKIObjective: To evaluate the performance of urinary C-C motif chemokine ligand 14 and renal resistive index in predicting persistent AKI among patients with different AKI stages.Methods: A prospective observational study included 166 adult patients admitted to the general ICU of Hebei General Hospital between June 2021 and January 2022.Urine samples used were collected at enrollment for CCL14 testing and the RRI was determined within the first 12 h.Patient data including their demographic data,medical history,mode for ICU admission,the use of nephrotoxic agents,baseline serum creatinine were extracted from the medical records.General clinical information and organ support were also recorded.Persistent AKI was defined as persistent oliguria and/or elevated serum creatinine beyond 48 h from AKI onset.The reversibility of renal dysfunction was assessed within 48 h.The AUROC was recorded,together with its 95% CI.AUROCs were compared in De Long’s test.Logistic regression analysis was performed to build clinical prediction model.The nested models(clinical models and urine CCL14 or RRI)were compared by computing the integrated discrimination improvement(IDI)and the category-free net reclassification improvement(cf NRI)together with respective 95% CIs and P values.Results: 166 patients were enrolled of whom 56(33.7%)developed persistent AKI.CCL14 had a good performance on predicting persistent AKI with an AUROC of 0.817(95%CI 0.748-0.874)and it improved the prediction provided by clinical model(IDI 6.7%,P<0.05).The performance of RRI to predict persistent AKI was fair with an AUROC of 0.739(95%CI 0.665-0.804)for a cutoff equal to 0.709.The addition of RRI to the clinical model did not improve the predictive performance(IDI 0.05%,P> 0.05).The prediction of CCL14 was also assessed in the subgroup of patients presenting with mild AKI(KDIGO stage 1)at enrollment,and its performance was not good with an AUROC of 0.738(95%CI 0.593-0.853).27 patients developed severe AKI(KDIGO stage 2 or 3)at enrollment,We also evaluated the predictive ability of CCL14 in this subgroup,and the performance of CCL14 was excellent,with an AUROC of 1.000(95% CI 0.863-1.000).Conclusions: CCL14 had a good performance on predicting persistent AKI among patients with different AKI stages,however,its predictive ability was fair in the subgroup of patients presenting mild AKI(KDIGO stage 1)at enrollment.RRI still did not show an advantage in predicting persistent AKI.Part Three Assessment of a risk index based on renal angina forpersistent AKI prediction in critically ill adultsObjective: To evaluate the performance of a modified persistent AKI risk index(PARI)and its combinations with RRI or urinary CCL14 for the prediction of persistent AKI in critically ill adults.Methods: This was a secondary analysis of Part Two.Adult patients who had their serum creatinine measured at 2 points(within 3 days before ICU admission and at ICU admission)were included.Patient data including their serum creatinine(baseline,before ICU admission and at ICU admission)and general clinical information were recorded.Urine samples used were collected at enrollment for CCL14 testing and the RRI was determined within the first12 h.Persistent AKI was defined as persistent oliguria and/or elevated serum creatinine beyond 48 h from AKI onset.We identified the factors significantly associated with persistent AKI using univariable or multivariable logistic regression and assigned risk scores according to their odds ratio.The area under the receiver-operating characteristic curve was recorded,together with its 95% CI.AUROCs were compared in De Long’s test.Results: 154 patients were enrolled of whom 53(34.4%)developed persistent AKI.The PARI included: presence of sepsis,pressor use,fluid overload percentage within the first day of ICU admission and percentage change in serum creatinine(s Cr)at ICU admission in reference to s Cr within 3days before ICU admission.The PARI had a good performance to predict persistent AKI,with an AUROC of 0.790(95%CI 0.717-0.851).The sensitivity,specifcity,positive predictive value,and negative predictive value of the PARI at a cut-off point of ≥6 were 60.4%,88.1%,72.7% and 80.9%respectively.The incidence of persistent AKI was significantly increased in patients with PARI score of ≥6(72.7%vs19.1%,P<0.001).Combinations with PARI increased the RRI for predicting persistent AKI from 0.736(95%CI0.659-0.803)to 0.97(95% CI: 0.93,1.00)(z statistic=2.837,P=0.004).Incorporation of PARI into urinary CCL14 did not improve its predictive performance(z statistic =0.735,P=0.462).Conclusions: A PARI including presence of sepsis,pressor use,fluid overload percentage and percentage change in serum creatinine performed well in predicting persistent AKI and provided context to direct measurement of RRI and improving its prediction of persistent AKI.