Clinical Study On The Application Of Predictive Score In Initiation Timing Of Renal Replacement Therapy For Acute Kidney Injury

The incidence of acute renal injury(AKI)is high in critically ill patients,and some AKI patients will eventually start renal replacement therapy(RRT)."Early" and "Delayed" initiation timing of RRT is a current clinical hot topic,but the standard of "early" and "delayed" initiation of RRT in different centers is confusing,and there is no absolute consensus in clinical practice.In the 17 th acute disease quality initiative(ADQI)international consensus conference,a new treatment concept was put forward around the theme of "precision RRT".It is agreed that RRT needs to be initiated when the metabolic capacity and fluid regulation capacity of the kidney are not matched with the needs of the patient,but there is no answer on how to quantify this "mismatch".Future studies on the "precision" timing of RRT should specify the mismatch between the metabolic capacity and fluid regulation capacity of the kidney and the needs of patients.RRT should be initiated based on objective,comprehensive,and easily accessible individual clinical characteristics(such as disease severity)of the patient.The development of clinical prediction models is a good evaluation strategy for the initiation of RRT.This study aims to develop,validate and optimize a clinical prediction model for initiating RRT based on the clinical data of AKI patients admitted to the intensive care unit,and explore its application value in the timing of initiating RRT,aiming to make it a powerful tool for clinical "precise RRT".Part One A clinical prediction model for the demand for renal replace- ment therapy in patients with acute kidney injuryObjective: To establish a clinical prediction model for initiation of RRT in AKI patients admitted to intensive care unit(ICU).Methods: A single-center,retrospective study was conducted.Patients admitted to ICU and diagnosed with AKI from January 2020 to December2021 were enrolled.The demographic characteristics,anthropometric data,diagnosis,data on the day of ICU admission,data on the day of AKI diagnosis and within 24 hours before RRT,time node of hospitalization and transfer,and prognosis of the patients were collected through the electronic medical record system.least absolute shrinkage and selection operator(Lasso)regression was used to screen the independent variables by taking whether RRT was started during ICU stay as the dependent variable.Multivariate binary logistic regression analysis was performed on the selected independent variables to establish a prediction model for RRT initiation and draw a nomogram.The area under curve(AUC)of receiver operating characteristic(ROC)curve was used to evaluate the discrimination of the model,and the Bootstrap method was used to internally validate the prediction model.In order to facilitate clinical application,a RRT predictive score table was established based on the regression coefficients of the independent variables included in the model.The ROC curve was used to evaluate the score table,and the best critical value of the score table to predict the initiation of RRT was determined.Results:1.In this study,488 cases met the inclusion criteria,105 cases were excluded,and 47 related factors were included as independent variables.Independent variables screened by Lasso regression analysis were malignant tumor,septic shock,APACHE-II score,arterial blood PH value on the day of AKI diagnosis,urine volume 24 hours before RRT,surgical treatment received,SOFA score on the day of AKI diagnosis,type of surgery,and myoglobin value on the day of AKI diagnosis.2.The results of multivariate binary logistic regression was carried out on the above 9 independent variables,and the results indicated that: Malignant tumor [OR=0.27,95%CI:(0.15-0.47),P<0.05],total urine volume 24 hours before RRT [OR=0.9988,95%CI:(0.9984-0.9991),P<0.05],arterial blood PH on the day of diagnosis of AKI [OR= 0.0002,95%CI:(0.0000-0.0064),P<0.05] was the protective factor for patients receiving RRT.Septic shock[OR=2.31,95%CI:(1.32-4.02),P<0.05] and APACHE-II score [OR= 1.10,95%CI:(1.04-1.15),P<0.05] were independent risk factors for patients with RRT.A total of 5 independent variables are finally included in the prediction model.3.Based on the five independent variables selected by the final multivariate binary logistic regression,a risk assessment model for RRT in AKI patients was established.The area under the ROC curve of the nomogram shown by the model was 0.900(95%CI: 0.870-0.930),the specificity was85.0%,and the sensitivity was 79.0%.The model was internally validated using the Bootstrap method with 1000 times,and the calibration curve and the ideal curve coincide well.When the RRT predictive score ≥-4.5,the AUC of ROC was 0.895(95%CI: 0.863-0.928),the specificity was 87.9%,and the sensitivity was 75.7%.Conclusion: This study established a predictive model for RRT initiation in AKI patients in ICU.When RRT predictive score ≥-4.5,it has a high predictive value for RRT initiation,which is expected to be popularized in clinical practice.Part Two Clinical study of biomarkers urinary ccl14 and NGal for the determination of RRT in patients with AKIObjective: To verify and compare urinary neutrophil gelatinase-associated lipocalin(NGAL),urinary C-C motif chemokine ligand 14(CCL14)and the predictive value of RRT predictive score for the initiation of RRT in ICU.Methods: This was a single-center,prospective study.The study included AKI patients admitted to the ICU between January 2022 and June 2022.Blood and urine samples were collected,urinary NGAL and urinary CCL14 concentrations were measured by enzyme-linked immunosorbent assay,blood-related indexes were determined,clinical information of patients was recorded,and RRT predictive score was calculated.The ROC curves of urinary NGAL,urinary CCL14 and RRT predictive scores were plotted to predict RRT initiation,and the AUC was compared.Results: In this study,there were 101 AKI patients eligible for inclusion,a total of 22 cases were excluded,and 79 cases were finally included,30 of whom underwent RRT.Body weight,APACHE-II score,SOFA score at ICU admission,basic Scr value,the maximum hourly dosage of norepinephrine on the day of AKI diagnosis,the highest Scr and BUN value in the RRT group were higher than those in the Non-RRT group(P<0.05).The total urinary volume,non-mechanical ventilation time and non-vasoactive drug application time on the day of AKI diagnosis in the Non-RRT group were higher than those in the RRT group(P<0.05).When the urinary NGAL concentration of patients admitted to ICU for AKI diagnosis was greater than 339.94ng/ml,the ROC AUC for predicting RRT was 0.794(95%CI: 0.62-0.83,P<0.001),the sensitivity was 73.33%,and the specificity was 75.51%.When the urinary CCL14 concentration of patients admitted to ICU for diagnosis of AKI was greater than 570.16pg/ml,the area under the ROC curve for predicting RRT was 0.588(95%CI:0.452-0.724,P=0.206),the sensitivity was 30%,and the specificity was95.92%.When the predicted RRT score was ≥-4.5,the sensitivity was 80.0%,the specificity was 75.51%,and the area under the ROC curve for predicting RRT was 0.889(95%CI: 0.798-0.948,P<0.001).Comparison of AUC of ROC curve for predicting RRT initiation: urinary NGAL concentration and RRT predictive score at the time of diagnosis of AKI were not significantly different(P=0.115),but both were higher than urinary CCL14(P<0.05).Conclusion: Urinary NGAL concentration and RRT predictive score at the time of AKI diagnosis in ICU patients have high predictive value for RRT initiation,but there is no significant difference in predictive ability between them.Urinary CCL14 has poor predictive ability for RRT initiation.Part Three External validation of the renal replacement therapy prediction score and comparison with the ability of the modified renal colic index to predict RRTObjective: To validate the RRT predictive score established in part I at different times,and compare the predictive ability of the RRT predictive score and modified renal angina index(mRAI)for the initiation of RRT.Methods: This was a single-center,retrospective study.Patients admitted to the ICU from January 2022 to July 2022 were enrolled.General information of the patients,such as gender,age,body mass index(BMI),time of admission,discharge and transfer,data on the day of AKI diagnosis,urine volume and fluid balance during ICU stay,and clinical outcomes were collected.RRT predictive score and m RAI were assigned for each patient,ROC curve of two parameters was drawn and AUC was compared.Results: A total of 131 AKI patients met the inclusion criteria,60 patients were excluded,and 71 patients were finally included,of whom 17 patients eventually underwent RRT.When the RRT predictive score of AKI patients was ≥-4.5,the sensitivity to predict RRT initiation was 64.71%(95%CI: 38.3.%-85.8%),the specificity was 77.78%(95%CI: 64.4%-88.0%),and the AUC to predict RRT initiation was 0.829(95%CI: 0.717-0.941,P<0.001),when the m RAI score of AKI patients admitted to ICU was >8,the AUC for predicting RRT was 0.631(95%CI: 0.508-0.743,P=0.069),and the sensitivity was 52.94%(95%CI: 27.8%-77%)with a specificity of 72.22%(95%CI: 58.4%-83.5%).The area under the ROC curve of AKI patients admitted to ICU predicted RRT initiation by RRT predictive score was significantly higher than that by m RAI score(P=0.020).Conclusions: The RRT predictive score has good performance in the external validation of the ability to predict RRT initiation,and is superior to the m RAI.Part Four Comparison of early and delayed renal replacement therapy initiation strategies for severe acute kidney injury combined with heart failure: a retrospective comparative cohort studyObjective: RRT predictive score has a good predictive value for RRT initiation,but it has no obvious advantage over urinary NGAL.Fluid overload(FO)is more likely to occur in AKI patients with HF.Delayed RRT in AKI patients with FO has been shown to increase mortality.There has been a lack of studies on the benefits of early initiation of RRT in patients with AKI and HF.This study aims to investigate the impact of the timing of RRT on clinical outcomes in patients with AKI and HF.Methods: A single-center,retrospective study included ICU patients with AKI and HF undergoing RRT.There were two grouping criteria.Grouping criterion 1: patients with AKI stage 3 with fluid overload present(FOP)and/or urgent indications for RRT were classified as Delayed RRT group,and the remaining patients were classified as Early RRT group.Grouping criterion2: patients with RRT predictive score ≥-4.5 points and the median time from AKI to RRT diagnosis was greater than or equal to 0.44 days were included in the Delayed RRT group,and other patients were included in the Early RRT group.The mortality,total ICU cost,ICU non-mechanical ventilation time,ICU non-RRT time,ICU non-vasoactive drug application time and the incidence of RRT related complications were compared between the two groups.Multivariate binary logistic regression analysis was used to adjust the confounding factors of early initiation of RRT on 90-day mortality.Results: A total of 207 patients met the inclusion criteria,56 patients were excluded,and 151 patients were finally included.Grouping criterion 1:77 patients in Early RRT group and 74 patients in Delayed RRT group.Grouping criterion 2: 99 cases in Early RRT group and 52 cases in Delayed RRT group.As for grouping criteria 1,the Early RRT group had significantly lower Scr and BUN values,APACHE-II scores,SOFA scores on the day of ICU admission than the Delayed RRT group(P<0.05).At the start of RRT,the Early RRT group had a higher PH value [7.37(7.32-7.43)vs.7.31(7.18-7.39),P<0.01].The Delayed RRT group had higher Scr and BUN values,and higher serum potassium and phosphorus levels(P<0.05).The Delayed RRT group had a lower oxygenation index and a higher proportion of peripheral edema.There was no significant difference in other parameters between the two groups(P>0.05).In grouping criterion 2,the time for diagnosing AKI to RRT in the Delayed RRT group was longer(P<0.01),and the total fluid balance and %FO values from ICU admission to RRT start in the Delayed RRT group were higher(P<0.01).There was no significant difference in ICU mortality,28-day mortality and 90-day mortality between Early RRT group and Delayed RRT group regardless of grouping criteria 1 or 2(P>0.05).As for grouping criteria 1,the non-RRT time in ICU of the Early RRT group was longer than that of the Delayed RRT group(P<0.05);As for grouping criteria 2,the non-RRT time in ICU of the Early RRT group was shorter than that of the Delayed RRT group(P<0.05).There were no significant differences between the two groups in the duration of non-mechanical ventilation,duration of non-vasoactive drug use,total length of hospital stay,length of ICU stay,and total cost of ICU stay(P>0.05).There was no significant difference in the incidence of adverse events that may be related to RRT between the two groups(P>0.05).Multivariate logistic regression analysis of confounding factors showed that early RRT initiation strategy was not an independent factor for reducing90-day mortality in AKI patients with HF(P>0.05).Conclusion: In patients with AKI combined with HF,early initiation of RRT does not improve clinical outcomes.