Epidemiology and Outcome of Early-Onset Acute Kidney Injury and Recovery in Critically Ill COVID-19 Patients: A Retrospective Analysis

Background: The clinical significance of early-onset acute kidney injury (EO-AKI) and recovery in severe COVID-19 intensive care unit (ICU) patients is poorly documented. Objective: The aim of the study was to assess the epidemiology and outcome of EO-AKI and recovery in ICU patients admitted for SARS-CoV-2 pneumonia. Design: This was a retrospective single-centre study. Setting: The study was carried out at the medical ICU of the university hospital of Clermont-Ferrand, France. Patients: All consecutive adult patients aged ≥18 years admitted between 20 March 2020 and 31 August 2021 for SARS-CoV-2 pneumonia were enrolled. Patients with chronic kidney disease, referred from another ICU, and with an ICU length of stay (LOS) ≤72 h were excluded. Interventions: EO-AKI was defined on the basis of serum creatinine levels according to the Kidney Disease Improving Global Outcomes criteria, developing ≤7 days. Depending on renal recovery, defined by the normalization of serum creatinine levels, EO-AKI was transient (recovery within 48 h), persistent (recovery between 3 and 7 days) or AKD (no recovery within 7 days after EO-AKI onset). Measurements: Uni- and multivariate analyses were performed to determine factors associated with EO-AKI and EO-AKI recovery. Main Results: EO-AKI occurred in 84/266 (31.5%) study patients, of whom 42 (50%), 17 (20.2%) and 25 (29.7%) had EO-AKI stages 1, 2 and 3, respectively. EO-AKI was classified as transient, persistent and AKD in 40 (47.6%), 15 (17.8%) and 29 (34.6%) patients, respectively. The 90-day mortality was 87/244 (35.6%) and increased with EO-AKI occurrence and severity: no EO-AKI, 38/168 (22.6%); EO-AKI stage 1, 22/39 (56.4%); stage 2, 9/15 (60%); and stage 3, 18/22 (81.8%) (p < 0.01). The 90-day mortality in patients with transient or persistent AKI and AKD was 20/36 (55.6%), 8/14 (57.1%) and 21/26 (80.8%), respectively (p < 0.01). MAKE-90 occurred in 42.6% of all patients. Conclusions: In ICU patients admitted for SARS-CoV-2 pneumonia, the development of EO-AKI and time to recovery beyond day 7 of onset were associated with poor outcome.


Introduction
Since the beginning of the outbreak of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), many studies have been published to better understand and The SarSCoV-AKI study was a retrospective analysis of prospectively collected data. It was a single-centre observational study performed at the medical ICU of the university hospital of Clermont-Ferrand, France. All consecutive adult patients aged ≥18 years admitted between 20 March 2020 and 31 August 2021 with a positive SARS-CoV-2 polymerase chain reaction test result, pneumonia confirmed by computed tomography and hypoxemia requiring >6 L/min supplemental oxygen were enrolled. Exclusion criteria were patients referred from another ICU, patients with chronic kidney disease stages 4 and 5, kidney transplant recipients, patients who refused to participate and patients with a hospital stay shorter than 72 h since they could not be classified according to our definitions. If there were multiple admissions, only the first admission to the ICU was included.

Ethics
The SarSCoV-AKI study received approval from the ethics committee of the French Intensive Care Society in 2020 (CE-SRLF20-21-French acronym for Comité éthique de la Société de réanimation de langue française) in accordance with our local regulations. All patients or close relatives were informed that their data would be included in the SarSCoV-AKI cohort study and gave written informed consent for the storage and research use of residual blood from samples collected as part of routine care. All the procedures were followed in accordance with the ethical standards of the responsible committee on human experimentation (institutional or regional) and with the Helsinki Declaration of 1975.

Data Collection
All data were obtained from medical records and electronic patient charts. Baseline patient characteristics were collected, including demographics and comorbidities, before ICU admission. The variables recorded regarding ICU admission and treatments were rela-tive to clinical presentation, reason for ICU admission, diagnosis, therapies implemented and outcomes. Blood sampling and routine biological testing were performed on the day of admission and daily according to standard laboratory protocols.

Definitions
AKI was defined according to the classification of Kidney Disease Improving Global Outcomes (KDIGO) [28] (Annex 1), using only the serum creatinine (SCr) component. "Baseline SCr" was the best outpatient SCr value between 7 and 365 days before ICU admission or, if unavailable, was estimated using the Modification of Diet in Renal Disease (MDRD) equation [28]. Early-onset AKI was defined as AKI occurring within 7 days after ICU admission: only the first AKI episode was taken into account in the study. Patients were stratified according to the highest AKI stage attained during this first episode.
EO-AKI recovery was based on a sustained (≥48 h) and complete reversal of AKI by KDIGO criteria, and therefore a minimum of 48 h of renal recovery was necessary to separate two distinct AKI episodes. EO-AKI was classified as transient, persistent and AKD according to the Acute Dialysis Quality Initiative (ADQI) [25]. "Transient" AKI was defined as renal recovery within 48 h of AKI onset, and "persistent" AKI as renal recovery occurring ≥3 days and <7 days of EO-AKI onset. Acute kidney disease (AKD) was characterized when AKI stage 1 or greater persisted ≥7 days after EO-AKI onset [25] (Annex 2). Chronic kidney disease (CKD) staging was defined according to the KDIGO definition [28].
The outcome of major adverse kidney event at 90 days (MAKE-90) was the composite outcome of death, dialysis dependence or a glomerular filtration rare (eGFR) <60 mL/min 90 days after ICU admission [25,29,30]. GFR was estimated using Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation (31). SCr at 90 days was retrieved from the medical health records and if unavailable, the patient's general practitioner was contacted.
Pulmonary bacterial co-infection was defined by the presence of a community-acquired or hospital-acquired bacterial pneumonia associated with SARS-CoV2 pneumonia on ICU admission. The presence of pulmonary bacterial co-infection was defined by the presence of radiological and/or scanographic condensation, bacteriological documentation (a positive quantitative culture of lower respiratory tract samples collected as recommended: bronchoalveolar lavage, >10 4 CFU/mL, plugged telescoping catheter, >10 3 CFU/mL, endotracheal aspirate, >10 6 CFU/mL) and/or presence of positive antigenuria, as defined by the European Centre for Disease Control and Prevention [31]. If bacterial pneumonia occurred at least 2 days after intubation, it was classified as ventilator associated pneumoniae (VAP) [32]. The period at risk for VAP begins from 48 h after intubation, until removal of the tracheal tube and weaning from invasive ventilation, and thus ends with extubation.

Endpoint
The primary outcome was the incidence of AKI within 7 days after ICU admission, its staging, and the proportions of transient AKI, persistent AKI and AKD.
The secondary endpoints were the determination of the risk factors of transient and persistent AKI, and AKD, and the renal prognostic evaluation assessed with MAKE at the end of ICU and hospital stays and 90 days after ICU admission.

Statistical Analysis
Patient characteristics were expressed as number (percentage) for categorical variables and median (interquartile range (IQR)) for continuous variables. Comparisons were made with exact Fisher tests for categorical variables and Wilcoxon tests for continuous variables.
Univariate followed by multivariate logistic regression analyses were performed to identify risk factors of EO-AKI. We used conditional stepwise variable selection with 0.2 as the critical p-value for entry into the model and 0.05 as the p-value for removal. Interactions and correlations between the explanatory variables were carefully checked. Longitudinal data were categorized according to their median. Data are given as odds ratio (ORs) and their 95% confidence intervals (CI).
A competing risk analysis was performed to assess cumulative renal recovery. The concomitant competing risks taken into account were discharge alive from the ICU and ICU mortality. The Fine and Gray model [33] was used to assess risk factors of renal recovery. We also used a conditional stepwise variable selection with 0.2 as the critical p-value for entry into the model and 0.05 as the p-value for removal. It was pre-planned to force the "AKI stage" in the model if this variable was not selected as sensitivity analysis. Data are given as subdistribution hazard ratios (sHRs) and their 95% CI. Proportionality of the risks was also carefully checked.
All reported probability values are two-tailed, and p-value < 0.05 was considered to be statistically significant. Missing data were imputed linearly. All analyses were performed with SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA) or R Statistical Software (v4.2.1; R Core Team 2021, Vienna, Austria).

Prevalence, Illness Severity Associated with, and Risk Factors for, EO-AKI
The cumulative incidence of EO-AKI and that of EO-AKI recovery, taking into account competing events, are shown in Figure 2, panels A and B, respectively.
Other biological inflammatory factors such as serum values of D-dimers, ferritin and C-reactive protein (CRP) were not associated with EO-AKI. The median length of ICU stay was significantly longer in patients with EO-AKI (10.5 days (7-16)) than without (6 days (4-10), p < 0.01) ( Table 1).
Overall, EO-AKI patients developed greater illness severity than patients without, as shown by worse oxygenation and higher needs in invasive mechanical ventilation and vasopressors (all p < 0.01). Higher ICU and hospital mortality rates were observed in patients with EO-AKI (47 (56%) and 52 (61.9%)) than in patients without EO-AKI (32 (17.6%) and 38 (20.8%)), (all p < 0.01). The higher the EO-AKI stage, the greater the mortality ( Table 2). Of note, patients with EO-AKI who required RRT during ICU stay, compared to patients who required RRT during ICU but without EO-AKI (late-AKI) were quite similar in terms of age and comorbidities. However, patients with EO-AKI and RRT had more severe disease on ICU admission. The day-90 death rate in these two subgroups was high: 21 (87.5%) for EO-AKI and 9 (81.82%) for late-AKI (Table S2).
Multivariable analysis showed that risk factors associated with EO-AKI included age, impaired immune system, being under vasopressors on admission, higher serum procalcitonin levels and exposure to aminoglycosides prior to ICU admission. In contrast, we identified that time between symptom onset and ICU admission >10 days was a protective factor against EO-AKI (Figures 3 and S1, ACE, angiotensin converting enzyme; AKI, acute kidney injury; BMI, body mass index; ECMO, extra corporeal membrane oxygenation; GFR, glomerular filtration rate; ICU, intensive care unit; NSAID, non-steroidal anti-inflammatory drug; RRT, renal replacement therapy; SAPS II, simplified acute physiology score; SOFA, sequential organ failure assessment. 1 Aplasia (lymphocytes < 1 G/L); or corticosteroids (if treatment duration > 1 month or if treatment amount > 2 mg/kg regardless of duration); or HIV (positive serology); AIDS (positive HIV serology and clinical complications: pneumocystis pneumonia, Kaposi's sarcoma, tuberculosis, toxoplasmosis. Multivariable analysis showed that risk factors associated with EO-AKI included age, impaired immune system, being under vasopressors on admission, higher serum procalcitonin levels and exposure to aminoglycosides prior to ICU admission. In contrast, we identified that time between symptom onset and ICU admission >10 days was a protective factor against EO-AKI (Figures 3 and S1, Table S2).  Variables with a p-value < 0.1 in the univariate analyses were age, cardiovascular disease, impaired immune system, time from first symptoms to ICU admission, antimicrobial therapy, aminoglycosides, vasopressors, invasive mechanical ventilation, bacteriemia on admission and procalcitonin. Because of correlations, only the following covariates were tested in multivariate analysis: age, cardiovascular disease, impaired immune system, time from first symptoms to ICU admission, aminoglycosides, vasopressors and procalcitonin.

Figure 4.
Factors associated with recovery from AKI, multivariate analyses, sub-distribution AKI: acute kidney injury; SubHR: sub hazard ratio. Variables with a p-value <0.1 in un analyses were age, lopinavir-ritonavir, diuretics, vasopressors and KDIGO stage. These co were considered into the final model.

Discussion
To our knowledge, this study is one of the first to characterize EO-AKI in critically ill COVID-19 patients and to assess the relationship between time to recovery and outcome.
First, we confirmed that EO-AKI is a frequent complication of COVID-19 in critically ill patients [15]. The timing of studies according to the outbreak wave and WHO treatment recommendations, the diversity of the case-mix populations (restricted or not to patients on invasive oxygen therapy, for instance), the differences in the definitions of AKI (based on both diuresis and SCr criteria or on SCr criteria alone) and of AKI onset make comparisons between studies on this topic difficult.
In our study, we observed a −31% prevalence of AKI, which is lower than the rates reported in most recent European studies, such as those of Lumlertgul et al. in the United Kingdom [10], who found an AKI prevalence of 76.6%, or of Arrestier et al. [14] in France with an AKI prevalence of 64.3%. In the study by Geri et al. [34], half of the cohort of 379 patients developed AKI in the first 7 days of admission. These studies used a longer time window of up to 21 days, which could be an explanation for the differences in AKI rates. In contrast, in a cohort study performed by the COVID-19 Critical Care Consortium and using the maximum SCr on day 1-2 of invasive mechanical ventilation to define EO-AKI, the rate of EO-AKI was 20.9%. It is noteworthy that most patients in the SarSCoV-AKI cohort were recruited after August 2020, hence after the first wave in France. As a consequence, they received the recommended treatment of steroids and reinforced anticoagulation, and in most cases the intubation strategy was performed later than during the first wave [35,36]. The patients in our cohort were also probably less severely ill at ICU admission than in cohorts reporting higher rates of AKI with a lesser need for vasopressors, 11.4% compared to 44.3% in Arrestier et al.'s study [15] or 44.2% in Lumlertgul et al.'s study [11]. We also used only SCr results whereas other studies used both SCr and urine output criteria to define AKI. Collectively, the results might have led to an underestimation of AKI prevalence in our study [37].
Older age, a higher procalcitonin level, a higher SOFA score and use of mechanical ventilation were independently associated with an increased risk of EO-AKI in our cohort. These results are consistent with those of other studies in both SARS-CoV-2 ICU populations [10,15,17] and general ICU patients [38,39], in which a higher severity score, hypertension and older age were frequent risk factors for developing AKI.
Since patients who developed AKI had more antimicrobial therapy, especially aminoglycoside use, more organ failure requiring vasopressors compared to patients without AKI, but also presented more bacteriemia on admission, EO-AKI might be caused by sepsis rather than direct renal effects of the SARS-CoV-2 virus in most cases.
Severe COVID-19 is classically associated with elevated values of systemic non-specific inflammatory biomarkers including CRP, ferritin and D-dimers [40,41]. Numerous studies suggest that in critically ill patients, systemic inflammation contributes to AKI occurrence [10,13,42,43]. In our study, only increased values of serum D-dimers were associated with EO-AKI in univariate analysis, but they were no longer found in multivariate models. This absence of an independent association between systemic inflammation and EO-AKI could result from the widespread use of corticosteroids during severe COVID [44]. Further investigations regarding the relationship between the inflammatory process and EO-AKI in severe COVID are therefore warranted.
Although our mortality rate was high at 90 days in EO-AKI patients (64.5%), only 10.8% of them developed chronic kidney disease, as defined by eGFR < 60 mL/min, and none were dialysis dependent. Other studies showed renal recovery rates between 17% and 84% and dialysis dependence rates between 8% and 56.5% [9,[45][46][47]. Lumlertgul et al. [10] reported kidney recovery in 90.9% of survivors at 90 days. The absence of dialysis dependence observed in our study could have been due to the exclusion of patients with pre-existing kidney disease. However, these findings strongly suggest that developing AKD after EO-AKI is often a reversible condition in ICU survivors. In our cohort, most of the patients with AKI stages 2 or 3 and persistent AKI or AKD died, and hence, only less severely ill patients survived and were able to recover from EO-AKI.
In our study, which was performed mainly during waves 2 and 3, we observed a lower rate of transient EO-AKI (47.5%) and a higher rate of patients with non-renal recovery ≥7 days (34.5%) than those reported by Lumlertgul et al. during wave 2 (65% and 10%, respectively) [47]. However, patients in the latter study were more likely to have received mechanical ventilation. We observed a higher rate of patients with non-renal recovery ≥7 days in our cohort. We cannot exclude that these discrepancies were partly due to differences in the definition of renal recovery: we used normalization of SCr while Lumlertgul et al. used SCr < 1.5 times baseline value, a more sensitive definition [48]. Additionally, our study shows that patients who made a rapid renal recovery had a better prognosis since AKD was associated with increased mortality at 90 days and mortality was strongly associated with initial AKI stage. These findings are in line with studies in both the general ICU population [49] and COVID-19 ICU populations [15,50].
We also identified that use of diuretics on ICU admission was a risk factor for nonrenal recovery. While diuretics can help prevent lung overload, they could have induced hypovolemia and pre-renal AKI, which might increase the risk of renal failure and delay renal adaptation to injury due to SARS-CoV2, and subsequent recovery. On the other hand, we found that an interval between symptom onset and ICU admission greater than 10 days protected against AKI development. This can be explained by the fact that peak infection usually occurs after 3-7 days [51] and ICU admission often occurs before 10 days of the first symptoms [36], and hence, patients admitted to the ICU after 10 days had probably gone through the most unstable period of the disease.
We found that, after adjusting for cofounders, all EO-AKI stages were associated with higher ICU mortality, which is in line with other reports [10,18,52].
Our study shows that EO-AKI patients had a longer length of stay, a higher rate of invasive mechanical ventilation and a greater need for vasopressors. EO-AKI was strongly associated with mortality, which is consistent with the results of other studies evaluating AKI in ICU-hospitalized patients [9,10,14,15,17].
Our study has several strengths. First, we admitted patients consecutively thus limiting selection bias, and had high quality data, prospectively collected, with almost none missing. In addition, we were able to establish the incidence and outcomes of transient AKI, persistent AKI and AKD. To our knowledge, data reporting renal recovery in patients with SARS-CoV2 pneumonia are scarce [10].
However, our work also has certain limitations. First, it was a single-centre observational study, which could impede the generalization of the results. Second, we might have over-or underestimated the prevalence of AKI. True baseline SCr was known for only 55% of patients and if the baseline value was unknown, SCr was estimated using the MDRD formula. In accordance with French legislation, we did not record patient ethnicity, and back-calculated SCr on the assumption that all patients were Caucasians. In addition, we did not define AKI according to the urine output criteria of the KDIGO definition, which could also have biased the true prevalence of AKI. Third, we might have either overestimated the prevalence of AKD, since it was defined by the absence of SCr normalization following EO-AKI and not by a decrease in SCr < 1.5-fold baseline value, or underestimated it since we did not take into account the markers of kidney damage to diagnose AKD. Similarly, since markers of kidney damage were not recorded, the MAKE criterion "persistence of renal failure" was defined by the functional criterion for kidney diseases and disorders: eGFR < 60 mL/min/1.73 m 2 [28]. Finally, as previously mentioned, follow-up was only 3 months, which is a short period to evaluate renal prognosis, and further studies are needed to evaluate long-term kidney outcomes in COVID ICU survivors after EO-AKI [47,53]. Additional studies to assess long-term renal prognosis should also be undertaken.

Conclusions
To conclude, we found that one third of critically ill patients with severe COVID-19 developed EO-AKI. EO-AKI stage 3 or AKD were associated with fatal outcome in around two-thirds of the cases at 3 months. EO-AKI stage 1 or transient AKI were also associated with a higher death rate in half of the cases. Monitoring and preserving kidney function should be considered for all patients admitted to the ICU for SARS-CoV-2 pneumonia.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/biomedicines11041001/s1, Figure S1: Correlation coefficient between covariates on admission; Table S1: AKI prevalence and mortality rate in the ICU; Table S2: Characteristics of patients depending on occurrence of early onset and late AKI and requirement for renal replacement therapy; Table S3: Factors associated with occurrence of Early-Onset Acute Kindey Injury, uni and then multivariate logistic regression analyses; Table S4: Factors associated with renal recovery after Early-Onset Acute Kidney Injury, uni and then sub distribution, survival analyses. Funding: This study was supported by a grant from the Michelin Corporate Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Institutional Review Board Statement: The SarSCoV-AKI study received approval from the ethics committee of the French Intensive Care Society in 2020 (CE-SRLF20-21) in accordance with our local regulations. All patients or close relatives were informed that their data would be included in the SarSCoV-AKI cohort study and gave written informed consent for the storage and research use of residual blood from samples collected as part of routine care. All the procedures were followed in accordance with the ethical standards of the responsible committee on human experimentation (institutional or regional) and with the Helsinki Declaration of 1975.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data can be provided under request at the corresponding author.