Cystatin C: a potential predictor for acute kidney injury in patients with community acquired pneumonia

Background: Acute kidney injury (AKI) is common in patients with community acquired pneumonia (CAP), and associated with poor prognosis. Cystatin C had been demonstrated to be a biomarker for the early detection AKI in in several other clinical settings. Therefore, in this study we evaluated whether AKI in patients with CAP could be well predicted by serum Cystatin C within 24 hours after admission. Methods: Univariate and multivariate logistic regression analyses were used to investigate independent factors of AKI in patients with CAP. Prediction performance of all independent factors for AKI was measured using area under the receiver operating characteristic (ROC) curves (AUCs). Results: 2716 patients with CAP were eligible in this study, and 766 (28%) patients developed AKI. After multivariate logistic regression analysis, Cystatin C, albumin, uric acid, platelet count, white blood cell count, CURB-65 score, and acute respiratory failure were independent factors for AKI in patients with CAP. The maximum AUC was reported for serum Cystatin C within 24 hours after admission. Cystatin C had an AUC of 0.81 (95% CI: 0.79–0.83, P < 0.001) for predicting AKI, with an optimal cut-off value of 1.37 mg/L, computing 68% sensitivity, 80% speci�city, 57% positive predictive value and 86% negative predictive value. In the strati�cation analyses, Cystatin C still had a good performance for predicting AKI in all the subgroups (AUCs: 0.69-0.84). Conclusions: The level of serum Cystatin C within 24 hours after admission appears to be a good biomarker for predicting of AKI in patients with CAP.


Background
Community acquired pneumonia (CAP) is responsible for substantial mortality, with a third of patients dying within 1 year after being discharged from hospital for pneumonia [1].Acute kidney injury (AKI) is a common complication of CAP, with the incidence rates from 18 to 34% [2][3][4].Moreover, AKI is associated with a poor prognosis in patients with CAP [2,3,5,6].Lakhmir S. et al. found that patients who were admitted to hospital for pneumonia and developed AKI had a poor effect on long-term prognosis.They were at high risk for death, dialysis, and permanent loss of renal function [6].Even among patients diagnosed with non-severe pneumonia, AKI could increase higher long-term mortality risk [3].Recently, our team also found that patients with CAP who developed AKI had worse short-term prognosis.They were more likely to require admission to intensive care unit, mechanical ventilation, invasive mechanical ventilation, non-invasive mechanical ventilation, had higher in-hospital mortality, and experienced a longer duration of hospital stay [5].
Previous study showed that Cystatin C was a good biomarker in the prediction of AKI in other clinical settings [7], as it was not in uenced by age, gender, race, muscle mass, and protein intake [8].However, few studies explored the prediction e ciency of Cystatin C for AKI in patients with CAP.Therefore, in this study we evaluated whether AKI in patients with CAP could be well predicted by serum Cystatin C within 24 hours after admission.

Patient selection
This retrospective study was approved by the Nanjing First Hospital Institutional Review Board and patients' informed consent was waived due to the retrospective nature of the study.This study was conducted according to the guidelines of the Declaration of Helsinki.We reviewed the medical records of 5851 patients, who were ≥ 18 years of age and admitted to hospital for CAP at Nanjing First Hospital from January 2014 to May 2017.Exclusion criteria were as follows: patients without Cystatin C, patients with less than two repeated serum creatinine (SCr), patients with a history of end stage renal disease or requiring dialysis, and patients lacking complete medical records.Finally, 2716 patients were enrolled in this study.(Fig. 1)

De nitions of CAP and AKI
Pneumonia diagnosed based on detection of interstitial in ltrate changes on chest radiography or CT in patients with one or more of: (a) recent presence of dyspnea, cough, or sputum; (b) core body temperature > 38.0°C; or (c) peripheral white blood cell > counts10 × 10 9 /L or < 4 × 10 9 /L.In addition, illness onset was speci cally in the community, rather than in the health-care setting [9,10].
The de nition of AKI in our study adhered to the Kidney Disease Improving Global Outcomes (KDIGO) criteria, which de ned AKI as an increase in SCr levels by ≥ 1.5-fold from baseline within 7 days of illness onset or an increase in SCr levels by ≥ 0.3 mg/dL (26.4 µmol/L) within 24 h of illness onset [11].Baseline SCr values were de ned as the lowest levels measured during hospitalization.Due to the lack of data concerning urine output, urine output standards were not considered in this study.

Data Analysis
Baseline characteristics were presented as means ± SDs or medians and interquartile ranges for continuous variables if appropriate, and proportions for categorical variables.Students' t tests or Mann-Whitney test were used to compare continuous variables between groups.Chi-square tests or sher exact were conducted to tell the differences of categorical variables between groups.
Univariate and multivariate logistic regression analysis was used to identify independent risk factors of AKI through stepwise selection.Prediction performance of all independent factors for AKI was measured using area under the receiver operating characteristic (ROC) curves (AUCs).We found that the maximum AUC was reported for serum Cystatin C within 24 hours after admission.To determine the optimal cut-off value for Cystatin C in discriminating AKI, we estimated the Youden index.
As a strati cation analyses to evaluate the performance of serum Cystatin C within 24 hours after admission for predicting AKI in various subgroups, we conducted multivariate logistic regression analysis to explore whether Cystatin C was still an independent predictor for AKI in various subgroups.The various subgroups were classi ed by age, gender, comorbidities (hypertension, diabetes, coronary artery disease, cardiac insu ciency, cerebrovascular disease, atrial brillation, COPD, chronic kidney disease, and tumour), laboratory investigations (albumin, anemia [for male, hemoglobin < 120 g/L; for female, hemoglobin < 110 g/L] [14], platelet count, white blood cell count, and uric acid), CURB-65 Score, and complication (acute respiratory failure).Strati cation analyses adjusted for all above factors except the strati cation factor itself.For the subgroups of intensive care unit (ICU) admission, and mechanical ventilation, strati cation analyses adjusted for all above factors (demographics, comorbid conditions, complication, and laboratory investigations).Furthermore, prediction performance of Cystatin C for AKI was also measured by AUCs in various subgroups.P values < 0.05 were considered as statistically signi cant.Statistical analysis was using SPSS software version 22 (IBM, Armonk, NY, USA), the EmpowerStats (www.empowerstats.net,X&Y solutions, Inc.Boston MA) and R version 3.6.1 (http://www.r-project.org).

Patient characteristics
Totally, 2716 patients were eligible in this study.The mean (median, range) age of the patients was 71.6 (75, 63-83) years and most of the study population consisted of males (60.9%).487 (17.9%) patients complicated with acute respiratory failure.447 (16.5%) patients required mechanical ventilation, and 597 (22.0%) patients needed admission to ICU.AKI characteristics 766 (28%) patients developed AKI.The characteristics of patients with AKI were shown in Table 1.Compared with non-AKI group, male gender (67.2% versus 58.4%; P < 0.001) and older age (78.2 years versus 69.0 years; P < 0.001) had signi cant difference between the two groups.Hypertension, diabetes mellitus, coronary artery disease, cardiac insu ciency, atrial brillation, chronic kidney disease, and cerebrovascular diseases were more common in the AKI group.However, no statistically signi cant comorbidities were in COPD, pulmonary hypertension, and tumour between the two groups.Patients in the AKI group were more commonly complicated with acute respiratory failure (39.2% versus 9.6%; P < 0.001), and had a higher CURB-65 score than the no-AKI group.In addition, patients with AKI had higher levels of uric acid, Cystatin C, and white blood cell count, while they had lower levels of albumin, hemoglobin, and platelet count.Abbreviation: AKI = acute kidney injury; CAP = community acquired pneumonia; OR = odds ratio; CI = con dence interval; CURB-65 = confusion, urea nitrogen, respiratory rate, blood pressure, and age≥65 years.

Prediction e ciency of Cystatin C for AKI in patients with CAP
We performed ROC analysis for all independent factors for AKI to determine their prediction performance, respectively.Figure 2 showed the comparisons of AUCs for all independent factors of AKI in patients with CAP.The maximum AUC was reported for serum Cystatin C within 24 hours after admission.Table 3 presented the accuracy of serum Cystatin C for detecting AKI in patients with CAP.Cystatin C had an AUC of 0.81 (95% CI: 0.79-0.83,P < 0.001) for predicting AKI, with an optimal cut-off value of 1.37 mg/L, computing 68% sensitivity, 80% speci city, 57% positive predictive value and 86% negative predictive value.
Table 3 The Abbreviation: = acute kidney injury; CAP = community acquired pneumonia; AUC = area under the receiver operating characteristic curve; CI = con dence interval; PPV = positive predictive value; NPV = negative predictive value.

Subgroup analysis
In the cation analyses, serum Cystatin C within 24 hours after admission was still an independent predictor in all the various subgroups.Figure 3 showed the ORs, AUCs, cut-off values, sensitivity, and speci city of Cystatin C in all the different subgroups.Moreover, Cystatin C still had a good performance for predicting AKI in all the subgroups (AUCs: 0.69-0.84).

Discussion
Previously, Cystatin C had been demonstrated to be a biomarker for the early detection AKI in in other clinical settings, such as patients with traumatic brain injury, neonates, patients with cardiac surgery, patients with liver, and so on [7,[15][16][17][18].To our best knowledge, this present study is the rst time to investigate the prognostic value of serum Cystatin C within 24 hours after admission for predicting AKI in the context of CAP with large samples, and it indeed proved to be a good predictor of AKI in patients with CAP.
In this study, the incidence rate of AKI was 28%.Previous studies had demonstrated similar incidence rates of AKI in patients with CAP, ranging from 18 to 34% [2][3][4].We found that Cystatin C was an important independent factor for predicting AKI in patients with CAP.Cystatin C is a 13 kilo Dalton proteinase inhibitor, and it is a member of the cystatin super family of cysteine protease inhibitors, which play an important role in intra-cellular catabolism of proteins and peptides [19].It is synthesized and released into plasma by all nucleated cells at a constant rate [20].Cystatin C can be more than 99% freely ltered through glomeruli, and do not show signi cant protein binding.It is considered to be neither actively secreted into the tubular lumen nor reabsorbed into the plasma.After ltration, it is normally completely reabsorbed by proximal renal tubular epithelial cells, and catabolized by megalin receptorinduced endocytosis [21].In addition, unlike SCr, the level of serum Cystatin C is not affected by age, sex, body muscle mass, and diet [7].Therefore, Cystatin C is considered as a good biomarker for the early detection of AKI.In our study, AUC of serum Cystatin C level to predict AKI was 0.81 (95% CI, 0.79-0.83),with an optimal cut-off value of 1.37 mg/L, computing 68% sensitivity, 80% speci city, 57% positive predictive value and 86% negative predictive value.Recently, Cigdem et al. found that serum Cystatin C showed a good predictive power for AKI in patients with COVID-19.The AUC value of serum Cystatin C to predict COVID-19 related AKI was 0.96 (0.90 to 1.0), with the best cut-off value of 1.00 mg/L [22].Albina et al. previously investigated the potential of plasma N-terminal prohormone B-type natriuretic peptide (NT-proBNP), mid-regional pro-atrial natriuretic peptide (MR-proANP) and B-type natriuretic peptide (BNP) levels to predict early AKI in hospitalized patients with CAP.The AUCs for the prediction of AKI of NT-proBNP, MR-proANP and BNP were 0.79, 0.79, and 0.74, respectively [23].The strength of this study was that we found Cystatin C also showed good prediction e ciency for AKI in various subgroups (AUCs: 0.69-0.84).Interestingly, it seemed that Cystatin C had slightly better performance for predicting AKI in the subgroups of the non-elderly, female, patients without comorbidities, and patients without serious condition.
In this study, we found that albumin, uric acid, platelet count, white blood cell count, CURB-65 score, and acute respiratory failure were also independent factors for AKI in patients with CAP.CURB-65 [13] and the pneumonia severity index (PSI) [24] were two scoring systems that assessed the severity of CAP.Ahsan R et al. had reported that PSI was an important factor for predicting AKI in patients with CAP [2] .In our study, we found CURB-65 was also an independent risk factor to develop AKI in patients with CAP.Low albumin level had been reported that it was a modi able risk factor linked to increased risk of AKI in different clinical settings [25].Recently, a study found that hypoalbuminemia was independently associated with the occurrence of AKI in COVID-19 patients with acute respiratory distress syndrome [26].Albumin could improve renal perfusion and glomerular ltration through inhibiting apoptosis in renal tubular cells by carrying protective lysophosphatidic acid and scavenging reactive oxygen species [27].Albumin could prolong potent renal vasodilation, which was induced by serum albumin reacting with the oxides of nitrogen to form S-nitroso-albumin [28].In addition, several studies suggested that exogenous albumin administration was bene cial to protect the kidneys from AKI [29,30].Hyperuricemia was associated with AKI in various statuses [31].A recent report showed that serum uric acid was an independent predictor of AKI in patients with COVID-19 [32].Platelet count and white blood cell count were two biomarkers for AKI, and low level of platelet count and high level of white blood cell count could increase risk of AKI [33][34][35][36], which was consistent with our ndings.
We acknowledge several limitations.First, it is a retrospective single-center study, and the conclusion will need to be con rmed by a multicenter prospective study with a larger patient cohort.AKI is de ned according the KDIGO criteria, based on SCr and urine output [11].However, the urine output data could not be obtained.Therefore, our analysis lacks the urine output standard for AKI.

Figures
Figures

Table 1
Patients baseline in patients with and without acute kidney injury accuracy of serum Cystatin C for detecting AKI in patients with CAP