Performance of urinary C-C motif chemokine ligand 14 and the renal resistive index in predicting persistent AKI Authors: You Fu, Ze Zhang, Cong He, Lijing Jia, Chen Ge, Ling Long, Yinxiang Bai, Na Zhang, Limin Shen, Quansheng Du, Heling Zhao Published: 2024 ABSTRACT AbstractObjectivesThe purpose of this study was to evaluate the performance of urinary C-C motif chemokine ligand 14 (CCL14) and the renal resistive index (RI) in predicting persistent AKI in unselected critically ill patients.MethodsThis prospective observational study was conducted in a tertiary hospital’s general intensive care unit (ICU). Consecutive adults who were admitted to the ICU were enrolled, with a primary endpoint of AKI lasting 48 h or longer. Urinary CCL14 was evaluated upon inclusion, and the renal RI was determined within 12 h of ICU admission. The individual discriminative ability of urinary CCL14 and the renal RI to predict persistent AKI was evaluated by the area under the receiver operating characteristic curve (AUC).ResultsOverall, 166 patients were included, of whom 56 had persistent AKI. Urinary CCL14 showed good ability to predict persistent AKI, with an AUC of 0.817. However, the overall performance of the renal RI was fair, with an AUC of 0.739. Forty-nine patients presented with mild AKI at inclusion, and the values of CCL14 were significantly lower than those of patients with moderate or severe AKI (0.205 [0.125–0.300] vs. 0.302 [0.157–0.501]; p = 0.034). In the subgroup analysis, although the diagnostic performance of CCL14 was excellent in patients with moderate or severe AKI, it was fair in patients with mild AKI [AUC = 0. 738; 95% confidence interval (CI) 0.593–0.853].ConclusionUrinary CCL14 was an excellent predictor of persistent AKI in patients with moderate or severe AKI, but its performance was not good in patients with mild AKI. The renal RI cannot discriminate between transient and persistent AKI. Introduction Acute kidney injury (AKI) is a common complication that occurs in as many as 25% of hospitalized patients and more than 50% of patients in the intensive care unit (ICU) [1]. Although preventive measures are constantly improving, AKI remains a syndrome with devastating clinical outcomes; no specific therapies have been approved, and only supportive care [2]. Persistent AKI is associated with increased short- or long-term mortality and greater risk of composite kidney-specific clinical outcomes than transient, self-limiting AKI [3,4]. Early detection of persistent AKI may prevent further damage to kidney function, as it may lead to the use of the AKI bundle to prevent further worsening of AKI then can improve patient outcome [5,6]. Urinary C–C motif chemokine ligand 14 (CCL14) is a member of the chemokine family of small molecules that were initially recognized for their roles in leukocyte chemotaxis and are implicated in tissue injury and repair processes. Recently, increased urinary CCL14 levels were shown to outperform several other AKI biomarkers [e.g. plasma cystatin C, urine neutrophil gelatinase-associated lipocalin (NGAL), urinary liver fatty acid binding protein (L-FABP), and urinary kidney injury molecule-1 (KIM-1)] in predicting the development of persistent severe AKI in the RUBY study. Moreover, this study emphasized the role of CCL14 in macrophage trafficking, which may be related to the development of persistent kidney damage, maladaptive repair, and the risk of nonrecovery of kidney function [7]. However, the RUBY study enrolled only patients with stages 2–3 AKI, as classified by the Kidney Disease Improving Global Outcomes (KDIGO). Although urinary CCL14 is a new biomarker of persistent AKI, whether it shows the same efficacy in unselected critically ill patients with different stages of AKI is unknown. The Doppler-based renal resistive index (RI) has been proposed to predict the short-term prognosis of AKI in ICU patients [8–10]. Nevertheless, a multicenter prospective cohort study showed contradictory results, with poor performance [11]. As a rapid, bedside and noninvasive tool, the renal RI is still constantly studied in critically ill patients. Two recent studies—one comparing the renal RI with urine tissue inhibitor of metalloproteinases-2 and insulin-like growth factor-binding protein 7 ([TIMP-2] × [IGFBP7]), two novel stress/damage biomarkers of AKI [12], and the other evaluating the use of the RI in sepsis-associated AKI—showed that the renal RI had an absolute advantage in predicting persistent AKI [13]. However, only patients with AKI were included in these studies. Therefore, our study aimed to evaluate the performance of this new biomarker—urinary CCL14—and reassess the ability of the renal RI to predict persistent AKI in unselected critically ill patients. Methods Patients This prospective, observational study was performed at the general ICU of Hebei General Hospital, China. This study was approved by the Ethics Committee of Hebei General Hospital (No. 202109). Written informed consent was obtained from all the participating patients or their delegates. From June 2021 to January 2022, consecutive adults who were admitted to the ICU were enrolled. The exclusion criteria were as follows: expected ICU stay < 48 h; chronic kidney disease stage IV–V; recovery from AKI according to our definition; renal replacement treatment (RRT) administered; pregnancy; renal transplant; anuria; or no indwelling catheter. Patients with obstructive renal disease, previously known renal artery stenosis or renal vein thrombosis, cardiac arrhythmia or a mass of abdominal gas precluding renal Doppler measurement were also excluded. Finally, patients with actual ICU stays < 48 h or those for whom the nature of AKI recovery was nonevaluable at 48 h after inclusion were excluded secondarily. Definitions AKI was defined according to the KDIGO classification as an increase in serum creatinine (sCr) of at least 26.5 μmol/L or to 1.5 times from baseline or a urine volume (UO) < 0.5 mL/kg/h for 6 h. AKI severity was determined separately and classified as stage 1, stage 2, or stage 3 based on the changes in the creatinine concentration or urine output according to the KDIGO guidelines [14]. Baseline serum creatinine was determined from available laboratory test results from the previous 12 months. When the baseline sCr level was unknown, this variable was defined as the minimum serum creatinine level during hospitalization or the estimated value according to the Modification of Diet in Renal Disease (MDRD) equation [15]. An investigator determined the final value based on the patient’s clinical manifestations and serum creatinine measurements during hospitalization. This investigator was blinded to the study’s primary objective and the value of the renal RI or biomarkers. Transient AKI was defined as AKI with recovery occurring within 48 h. Recovery from AKI was defined as the absence of any stage of AKI according to either the sCr or UO criteria. Persistent AKI was defined as persistent oliguria and/or stable or higher AKI KDIGO stages beyond 48 h from AKI onset [16]. Sepsis and septic shock were diagnosed following the sepsis 3.0 criteria [17]. Diuretic use was defined as the use of diuretics at any time during the first 24 h in the ICU. Data collection and study protocol Urine samples were collected at enrollment. After centrifugation at 3000 rpm for 10 min at 4 °C, the urine supernatant was stored and frozen at −80 °C until analysis. CCL14 was measured by enzyme-linked immunosorbent assay according to the manufacturer’s protocol (Aviva Systems Biology, San Diego, CA, USA). The technicians measuring the CCL14 levels were blinded to the clinical data, and the kit instructions were strictly followed. The responsible physicians were blinded to the CCL14 test results. Serum creatinine was measured at inclusion and then daily until ICU discharge; urine output was recorded hourly based on our standard procedures. The nature—transient or persistent—of AKI was assessed at 48 h after inclusion. The RI was measured within 12 h of ICU admission by two investigators who were not involved in patient management and had received training on the Doppler evaluation of kidney perfusion from the Chinese Critical Ultrasound Study Group. Previous studies have demonstrated that both junior and senior operators perform well in RI calculations [10]. Moreover, the intraobserver and interobserver reproducibility of the RI measurement were also excellent [18]. A CX50 CompactXtreme ultrasound machine (Philips Medical Systems, Bothell, WA, USA) with a 2–5 MHz curved-array multifrequency transducer was used. After visualization of a kidney in grayscale and color Doppler modes, an interlobar or arcuate artery was selected, and measurements were obtained using pulse-wave Doppler. The Doppler gain was set to obtain a clear outline of flow waves with minimal background noise. The Doppler spectrum was optimal when at least three similar consecutive waveforms were visualized. The peak systolic velocity (Vmax) and the minimal diastolic velocity (Vmin) were recorded, and the RI was calculated with the following formula: (Vmax–Vmin)/Vmax. Three measurements were averaged to obtain the mean RI values used for the study [19]. Baseline patient characteristics (e.g. demographics, medical history, admission type and reason for admission) were collected together with baseline sCr. The following variables were recorded at inclusion: vital signs, routine blood parameters, blood biochemical parameters, blood gas analysis parameters, mechanical ventilation and tidal volume, positive end expiratory pressure, vasoactive drug application, and the exist of sepsis. Patients who were given nephrotoxic agents including angiotensin II-converting enzyme inhibitors or angiotensin receptor blockers (ACEIs/ARBs), iodinated contrast agents, aminoglycosides, and glycopeptides, before ICU admission or within 48 h after admission were also recorded. During patient management, the amount of fluid therapy received within the first 24 h and the use of diuretics were recorded. The severity of illness was determined using the Acute Physiology and Chronic Health Evaluation (APACHE II) and Sequential Organ Failure Assessment (SOFA) scores at ICU admission. The SOFA score without a renal component was calculated simultaneously. Statistical analysis Continuous variables are presented as the mean ± standard deviation (SD) or median and interquartile range (IQR) based on their distribution. Categorical variables are presented as frequencies and proportions. Categorical variables were compared using the chi-square test or Fisher exact test, as appropriate, and continuous variables were compared using the nonparametric Mann–Whitney test or Kruskal–Wallis test. Missing values were not inferred. The predictive ability of each marker was assessed using the area under the receiver operating characteristic (ROC) curve (AUC). The following values were utilized to describe the AUC: 0.90–1.0, excellent; 0.80–0.89, good; 0.70–0.79, fair; 0.60–0.69, poor; and 0.50–0.59, useless. We used the Youden index to determine the optimal cutoff for calculating specificity and sensitivity. The confidence interval (CI) for the AUC was calculated with Delong’s method. Based on our pretest results, the incidence of persistent AKI was 0.63 for the exposed group (urine CCL14 levels above the threshold) and 0.18 for the nonexposed group (urine CCL14 levels below the threshold), using a two-sided test with an a-risk (type I error) of 5% and a statistical power of 90%, the sample size calculated by PASS 15 software (NCSS, LLC, Kaysville, Utah, USA) was 48. The same method was used to calculate sample sizes for the RI, which was 117. Considering the loss rate to follow-up (about 20%), the estimated total sample size was 141. All the statistical analyses were performed using SPSS 21 (IBM, Armonk, NY, USA) and MedCalc version 16.8 (MedCalc Software, Ostend, Belgium), with a two-sided p value < 0.05 considered statistically significant. Results Study population Four hundred seventy-eight patients were screened, and 166 patients were ultimately included in the primary analysis (Figure 1). The values of urinary CCL14 from 9 patients were missing. The average age of the patients was 65 years, and approximately two-thirds of the patients were male. More than half of the patients (57.8%) were admitted to the ICU after the operation, and 32 (19.3%) patients were admitted from the emergency room, which was less than the number of patients who were admitted from the ward (22.9%). The main reasons for ICU admission were as follows: shock (70 patients, 42.2%), coma (39 patients, 23.5%), respiratory failure (30 patients, 18.1%), and other causes (36 patients, 21.7%). The mean APACHE II and SOFA scores were 20.3 ± 7.5 and 7.0 [4.0–9.0], respectively. One hundred forty-nine patients required mechanical ventilation (89.8%), 59 patients required vasopressors (35.5%) at inclusion, and 12 patients required RRT (7.2%) during the ICU stay. The median ICU stay was 4.7 days, and the ICU mortality rate was 18.7%. Flow chart. AKI: acute kidney injury; ICU: intensive care unit; RRT: renal replacement therapy. According to our definitions, 76 patients (45.8%) presented with AKI at ICU admission, including 56 patients with persistent AKI and 20 patients with transient AKI. AKI severity was determined to be stage 1 in 49 patients (64.5%), stage 2 in 21 patients (27.6%), and stage 3 in 6 patients (7.9%). Table 1 summarizes the main characteristics of patients across the groups with no AKI, transient AKI, and persistent AKI. Age, admission to the ICU with shock or coma, sepsis, mean arterial pressure (MAP), arterial oxygen partial pressure to fractional inspired oxygen, lactate, and illness severity scores differed significantly among the three groups. The use of vasoactive drugs and fluid accumulation were associated with AKI and persistent AKI. All patients with stage 3 AKI at inclusion developed persistent AKI. Table 1.Patient characteristics according to renal dysfunction and short-term renal recovery.VariablesNo AKI(n = 90)Transient AKI(n = 20)Persistent AKI(n = 56)P valueSex, male, n (%)61 (67.8)15(75.0)33 (58.9)0.354Age, yr, mean ± SD63 ± 1559 ± 2071 ± 150.002Mode for ICU admission, n (%) 0.192 Emergency room13 (14.4)7 (35.0)12 (21.4) Operating room58 (64.4)8 (40.0)30 (53.6) Ward19 (21.1)5(25.0)14 (25.0) Reason for ICU admission, n (%) Shock21 (23.3)8 (40.0)41 (73.2)<0.001 Respiratory failure19 (21.1)5 (25.0)6 (10.7)0.196 Coma31(34.4)4(20.0)4(7.1)<0.001 other23 (25.6)6(30.0)7 (12.5)0.095Risk factors for AKI, n (%) Hypertension50 (55.6)13 (65.0)27 (48.2)0.404 Diabetes mellitus15 (16.7)5 (25.0)12 (21.4)0.543 Chronic kidney disease2 (2.2)1 (5.0)2 (3.6)0.539 Nephrotoxic agents28(31.1)5 (25.0)15 (26.8)0.785 Sepsis15 (16.7)6(30.0)36 (64.3)<0.001 Trauma12(13.3)1(5.0)3(5.4)0.251Clinical indicators at inclusion Heart rate, beats/min, mean ± SD84 ± 2090 ± 1894 ± 230.043 MAP, mmHg, mean ± SD89 ± 1386 ± 1184 ± 150.032 PaO2/FiO2,mmHg, mean ± SD269 ± 117224 ± 112222 ± 1100.017 PaCO2,mmHg, mean ± SD39.0 ± 8.139.3 ± 11.240.2 ± 14.60.936 Lactate, mmol/L, median (IQR)1.8 [1.4-3.2]2.3 [1.5-3.0]2.6 [1.8-5.6]0.005 APACHEII score, mean ± SD18.3 ± 6.019.0 ± 5.323.9 ± 9.00.001 SOFA score, mean ± SD5.0 [4.0-7.0]7.5[6.0-9.0]8.5 [7.0-11.8]<0.001 nonrenal SOFA score, mean ± SD5.0 [4.0-7.0]7.0 [6.0-9.0]8.0 [7.0-11.0]<0.001Renal function Baseline sCr, μmol/L, mean ± SD57.7 ± 18.061.4 ± 16.367.8 ± 17.50.001 SCr at inclusion, µmol/L, mean ± SD58.1 ± 18.280.9 ± 24.6118.3 ± 52.3<0.001Organ support at inclusion Vasoactive drugs, n (%)17 (18.9)7 (35.0)35 (62.5)<0.001 Mechanical ventilation, n (%)79 (87.8)18 (90.0)52 (92.9)0.621 Tidal volume, ml, mean ± SD511 ± 118541 ± 130533 ± 1710.535 PEEP, cmH2O, median (IQR)5[5-5]6[5-8]5[5-8]0.005Use of diuretics at 24 h, n (%)16(17.8)2 (10.0)14 (25.0)0.315Fluid balance at 24 h, L, median (IQR)−0.1[-0.7,0.6]0.6[-0.3,1.2]0.5[-0.5,1.9]0.010AKI stage at inclusion, n (%) No AKI90(100)00 Stage 1018(90.0)31(55.3) Stage 202(10.0)19(33.9) Stage 300(0.0)6(10.7) RRT in ICU, (%)0 (0.0)2 (10.0)10(17.9)<0.001Length of ICU stay, days, median (IQR)4.2 [2.5-9.4]7.5[5.2-12.6]4.3 [2.3-10.9]0.052Death in the ICU, (%)11 (12.2)4 (20.0)16 (28.6)0.051AKI: acute kidney injury; APACHE II: Acute Physiology and Chronic Health Evaluation II; IQR: interquartile range; ICU: intensive care unit. MAP: mean arterial pressure; PaO2/FiO2: arterial oxygen partial pressure to fractional inspired oxygen; PaCO2: arterial partial pressure of carbon dioxide; PEEP: positive end expiratory pressure; RRT: renal replacement therapy; SCr: serum creatinine; SOFA: Sequential Organ Failure Assessment.Values represent means ± SDs, n (%), or medians [IQR]. Patient characteristics according to renal dysfunction and short-term renal recovery. AKI: acute kidney injury; APACHE II: Acute Physiology and Chronic Health Evaluation II; IQR: interquartile range; ICU: intensive care unit. MAP: mean arterial pressure; PaO2/FiO2: arterial oxygen partial pressure to fractional inspired oxygen; PaCO2: arterial partial pressure of carbon dioxide; PEEP: positive end expiratory pressure; RRT: renal replacement therapy; SCr: serum creatinine; SOFA: Sequential Organ Failure Assessment. Values represent means ± SDs, n (%), or medians [IQR]. Urinary CCL14 and renal RI in patients with AKI of different durations/severities Table 2 and Figure S1 show the urinary CCL14 and renal RI values at ICU admission in patients with AKI of different durations. Patients with persistent AKI had higher CCL14 levels (0.255 [0.181–0.466] vs. 0.112 [0.059–0.220]; p < 0.001) and RIs (0.71 ± 0.08 vs. 0.67 ± 0.08; p = 0.049) than patients with transient AKI did. The values of CCL14 remained greater in patients with moderate or severe AKI (stage 2 or 3) than in those with mild AKI (stage 1) (0.302 [0.157–0.501] vs. 0.205 [0.125–0.300]; p = 0.034), but the value of the RI did not differ (0.69 ± 0.09 vs. 0.70 ± 0.07; p = 0.487) (Table 3 and Figure S2). Table 2.Urinary CCL14 and the renal RI of patients with AKI of different duration. No AKI(n = 90)Transient AKI(n = 20)Persistent AKI(n = 56)P value(transient vs. persistent)CCL14, ng/ml, median (IQR)0.119 [0.067-0.195]0.112 [0.059-0.220]0.255 [0.181-0.466]<0.001Renal RI, mean ± SD0.63 ± 0.060.67 ± 0.080.71 ± 0.080.049AKI: acute kidney injury; CCL14: C-C motif chemokine ligand 14; RI: resistive index. Urinary CCL14 and the renal RI of patients with AKI of different duration. AKI: acute kidney injury; CCL14: C-C motif chemokine ligand 14; RI: resistive index. Table 3.Urinary CCL14 and the renal RI of patients with different stages of AKI. No AKI (n = 90)Stage 1 (n = 49)Stage 2-3 (n = 27)P value(stage 1 vs. stage 2-3)CCL14, ng/ml, median (IQR)0.119[0.067-0.195]0.205[0.125-0.300]0.302[0.157-0.501]0.034Renal RI, mean ± SD0.63 ± 0.060.70 ± 0.070.69 ± 0.090.487AKI: acute kidney injury; CCL14: C-C motif chemokine ligand 14; RI: resistive index. Urinary CCL14 and the renal RI of patients with different stages of AKI. AKI: acute kidney injury; CCL14: C-C motif chemokine ligand 14; RI: resistive index. Performance of urinary CCL14 and the renal RI for predicting persistent AKI Urinary CCL14 had good performance for predicting persistent AKI, with an AUC of 0.817 (95% CI 0.748–0.874). The sensitivity and specificity were 69.4% and 79.2%, respectively, at an optimal cutoff of 0.205 ng/mL. The performance of the renal RI was fair, with an AUC of 0.739 (95% CI 0.665–0.804). An RI >0.795 predicted persistent AKI with 53.6% sensitivity and 83.6% specificity. The combination of urinary CCL14 and the renal RI also performed well, with an AUC value of 0.829 (95% CI 0.760–0.885). This value was significantly greater than the RI (z statistic = 2.920, p = 0.0035) but similar to that of CCL14 (z statistic = 0.630, p = 0.5287) (Figure 2). Area under the ROC curve for the prediction of persistent AKI by urine CCL14, the renal RI and their combination. AKI: acute kidney injury, CCL14: C-C motif chemokine ligand 14, RI: resistive index, ROC: receiver operating characteristic. The discriminative performance of urinary CCL14 was also assessed in a subgroup of patients presenting different severity of AKI at inclusion. However, its performance was not good in patients with mild AKI (stage 1), with an AUC of 0.738 (95% CI 0.593–0.853) (Table 4). Among 27 patients with moderate or severe AKI (stage 2 or 3), only two did not develop persistent AKI. CCL14 performed very well in this subgroup, and the AUC was 1.000 (95% CI 0.863–1.000) (Table 4). Table 4.Performance of urinary CCL14 for predicting persistent AKI in patients with different AKI severity.SubgroupSampleCutoffvalueSensitivity(95% CI)Specificity(95% CI)PPV(95% CI)NPV(95% CI)AUC(95% CI)Stage 1 AKI at inclusion490.114100 (88.8-100)50.0 (26.0-74.0)77.5 (61.5-89.2)100(66.4-100)0.738 (0.593-0.853)Stage 2 or 3 AKI at inclusion270.097100 (85.2-100)100 (15.8-100)100 (85.2-100)100 (15.8-100)1.000 (0.863-1.000)AKI: acute kidney injury; AUC: area under the receiver operating characteristic curve; CCL14: C-C motif chemokine ligand 14; CI: confidence interval; NPV: negative predictive value; PPV: positive predictive value. Performance of urinary CCL14 for predicting persistent AKI in patients with different AKI severity. AKI: acute kidney injury; AUC: area under the receiver operating characteristic curve; CCL14: C-C motif chemokine ligand 14; CI: confidence interval; NPV: negative predictive value; PPV: positive predictive value. Discussion AKI without short-term recovery is associated with poor clinical outcomes and significant healthcare costs [20,21]. Therefore, predicting whether patients with AKI will recover quickly or whether AKI will persist is crucial and challenging. An early indicator for AKI nonrecovery is needed, which would allow for a window of time when interventions, such as the rapid restoration of renal perfusion, fluid restriction, avoidance of nephrotoxic drugs, or early CRRT, might prevent further injury [22]. We performed a prospective study to explore the ability of urinary CCL14 and the renal RI to predict persistent AKI in critically ill patients with different AKI stages. We found that CCL14 performed well in the whole cohort and patients with moderate or severe AKI (stage 2 or 3), but the predictive ability was not good in patients with mild AKI (stage 1). Moreover, the renal RI was unable to distinguish transient AKI from persistent AKI in these unselected patients. Urinary CCL14 was first reported as a predictor of persistent stage 3 AKI in the RUBY study, which included 331 patients with stage 2–3 AKI and reported that CCL14 had the greatest ability to predict persistent severe AKI, with an AUC of 0.83 [7]. Subsequently, Bagshaw et al. [23] externally validated the predictive ability of urinary CCL14. Similarly, diagnostic efficacy has also been reported for cardiac surgery-associated AKI. In general, previous studies have focused mainly on patients with stage 2–3 AKI at enrollment, and our present study replicated the performance of urinary CCL14 in such a population. However, the performance was not good in patients with mild AKI. These differences can be explained as follows. First, CCL14 is a critical chemokine for monocyte/macrophage recruitment and activation and may mediate kidney injury and accelerate kidney fibrosis [7]. Etiologies of AKI exist across a spectrum ranging from functional to structural, and AKI recovery depends in part on the delicate balance between the injury/inflammation and repair that are mediated by recruited monocytes and macrophages. An indicator that can assess a patient’s status on the spectrum from injury to repair would be favorable for clinical prognostication [24]. Indeed, CCL14 may be a potential predictor of AKI persistence, and its value may reflect the exist or progression of AKI. In our present study, functional AKI, meaning no damage or stress to the kidney, may occur more frequently in patients with stage 1 AKI and thus may explain the underperformance of urinary CCL14. Second, the use of biomarkers in the appropriate context or population is critical for their efficiency and effectiveness. Just as the performance of troponin testing deteriorates when used in a general heterogeneous population with a low probability for acute coronary syndrome (ACS) [25], urinary CCL14 was not effective in predicting persistent AKI in patients with mild AKI, as it may not be suitable for this population. On the one hand, mild AKI is associated with a lower risk of persistent AKI [26,27]. On the other hand, Koyner and colleagues [28] established two cutoffs of CCL14 for the prediction of persistent severe AKI: one at a lower CCL14 concentration to ensure high sensitivity to facilitate the identification of the majority of patients who would develop persistent severe AKI and a second at a higher CCL14 concentration that showed high specificity for identifying patients who should be prioritized to receive the AKI bundle recommended by the KDIGO Clinical Practice Guidelines. Finally, they reported that more than half of patients with stage 1 AKI had CCL14 values below the low cutoff (1.3 ng/ml), which was in the negative range. In our study, mild AKI was also associated with lower CCL14 values, which indicates that there is little possibility of persistent AKI. In addition, a meta-analysis suggested that urinary CCL14 may exhibit greater effectiveness in predicting persistent AKI in clinical settings characterized by a higher risk of persistent AKI than in those with a lower risk [29]. Therefore, the predictive ability of CCL14 was indeed limited among patients with mild AKI. Another finding was that the absolute value of urinary CCL14 varied greatly, and it was lower in our study than previously reported values. Several points can explain these differences. First, racial disparity may influence urinary CCL14. Stanley et al. [30] reported that urine biomarkers of lupus nephritis varied across ethnicities. In addition to our study, similar values of urinary CCL14 were observed in two other studies, including patients from Asia [13,31], which differed from those reported in patients from Europe and America [7,23,28,31]. Second, the level of CCL14 varies over time. Massoth et al. [31] reported that the kinetics of urinary CCL14 levels in cardiac surgery patients with persistent AKI were characterized by a twin-peak pattern, with the first peak on Day 1 and the second peak on Day 7 after enrollment. In addition, Jiang et al. [13] found that the level of CCL14 gradually decreased within 24 h of admission to the ICU in patients with sepsis-associated AKI. Based on these findings, explanations for the changes in CCL14 values are incomplete, and identifying the dynamic changes in CCL14 in different AKI subtypes is worth exploring, which may help identify the appropriate measurement time or population. Finally, different biomarker test kits may affect the results [32]. Designing a study using the same test kit and then comparing the differences is necessary. The renal RI was not an excellent indicator for predicting persistent AKI, similar to previous studies [7,33]. This may be explained by the fact that numerous confounding factors, including vascular compliance, MAP, heart rate, hypoxemia, hypercapnia, and intra-abdominal pressure, can influence RI values [18,34–38]. Additionally, the relationship between the RI and actual renal vascular resistance (RVR) seemed linear only when vascular compliance was normal, but this relationship progressively disappeared when arterial stiffness increased [39,40]. Furthermore, better predictive power has emerged in studies including patients with AKI [12] or AKI caused by particular etiologies, such as sepsis or cardiac surgery [13,41]. Further studies may combine the renal RI with other indicators, such as urinary CCL14, among an appropriate population, which will help efficiently identify patients at high risk of persistent AKI. Our study has several advantages. First, we concomitantly evaluated urinary CCL14 and the renal RI to predict persistent AKI in critically ill patients with AKI of different severities. Additionally, our definition of persistent AKI was based on the latest consensus guidelines [16]. Finally, all the investigators who performed Doppler were not in charge of the patients and were blinded to the urinary CCL14 test results, which were batched and measured centrally. Our study has limitations, however. First, the effect of CKD on the predictive value of urinary CCL14 was not considered. Because only five patients with CKD were included, subgroup analyses could not be performed in our study. However, CCL14 has been demonstrated to be significantly associated with persistent AKI regardless of the occurrence of CKD in cardiac surgery patients [31]. Second, the AUC of CCL14 for the prediction of persistent AKI was as high as 1.00 among patients with stage 2–3 AKI, which is reasonable in theory but not in practice. This may be associated with the relatively small sample size, especially for patients with severe AKI. Third, CCL14 was measured at only one point, and kinetic CCL14 may be a better indicator, as absolute values vary due to both the passage of time and the potential layering of additional insults [24]. Finally, similar to other single-center studies, our study was conducted only in critically ill patients, which limits its generalizability to other clinical settings. However, these data support the findings of the RUBY study, which revealed that CCL14 has high predictive value for persistent AKI in patients with severe AKI. Conclusions In summary, urinary CCL14 was an excellent predictor of persistent AKI in patients with moderate or severe AKI, but its performance was not good in patients with mild AKI. The renal RI cannot discriminate between transient and persistent AKI. Combining these two parameters among an appropriate population may be a better choice for further research. Supplementary Material additional figures and tables.doc STROBE checklist .doc