SARS-COV2 Renal Impairment in Critical Care: A Retrospective Study of 42 Cases - Kid COVID Study

Background The new coronavirus (SARS-CoV-2) infection leads to 5% to 16% hospitalization in Intensive Care Units (ICU) and is associated with 23% to 75% of kidney impairments, including acute kidney injury (AKI), as a major prognosis factor. The current work aims to characterize the renal impairment associated to SARS-CoV-2 in ICU patients, to evaluate its risk factors and its relationship with morbidity and mortality.


Abstract Background
The new coronavirus (SARS-CoV-2) infection leads to 5% to 16% hospitalization in Intensive Care Units (ICU) and is associated with 23% to 75% of kidney impairments, including acute kidney injury (AKI), as a major prognosis factor. The current work aims to characterize the renal impairment associated to SARS-CoV-2 in ICU patients, to evaluate its risk factors and its relationship with morbidity and mortality.

Methods
Forty-two patients consecutively admitted to the ICU of a university hospital (Paris, France) who tested positive for SARS-CoV-2 between March 25, 2020 and April 29, 2020 were included and classi ed in categories according to their renal function. Complete renal pro les and their evolution during ICU stay were fully characterized in 34 patients. Factors associated with AKI were identi ed through univariate analysis.

Conclusions
Critical SARS-CoV-2 is associated with persistent intrinsic renal injury and AKI, which is a risk factor of mortality. Identifying SARS-CoV-2 patients at risk of AKI will help in modifying clinical practice in ICU.

Trial registration
In accordance with the French law on biomedical research, this study obtained the approval of an Institutional Review Board ("Comité d'Éthique de la Recherche en Anesthésie-Réanimation" under the reference IRB 00010254 -2020 -106). Patients were all informed of the possible use of their data in researches as well as their right and terms of objection. Data were collected and integrated anonymously into a secure database in accordance with the French CNIL MR-004 methodology (registration number 20200803123416).
Background Page 3/19 About 5% to 16% (1,2) of the patients who tested positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) required hospitalization in Intensive Care Units (ICU), mainly for respiratory distress associating dyspnea, high respiratory rate, low oxygen saturation, or rapid increase in lung in ltrates (3,4). Mortality associated with ICU hospitalizations ranged from 49% to 67% (5,6). While respiratory symptoms are the cornerstone of the disease, other organs can be affected. Acute kidney injury (AKI) occurred in 23% of SARS-CoV-2 patients during their hospitalization (7), and Renal Replacement Therapy (RRT) was used for 13% of them (8). However, when urine dipstick tests are systematically performed, the incidence of renal impairment on hospital admission can be evaluated as high as 75% (9). The SARS-CoV-2 is thought to have a direct renal toxicity (10) via entry into proximal tubular cells and podocytes where angiotensin converting enzyme 2 (ACE2) receptors and transmembrane serine proteases (TMPRSS) are highly expressed (11,12). In critically ill patients, other factors may be implicated, such as cytokine storm, angiotensin II pathway activation, dysregulation of complement, hypercoagulation, and microangiopathy (7). Moreover, 14% of SARS-CoV-2 patients will develop acute respiratory distress syndrome (ARDS) (2) that is, by itself, also independently associated with AKI out of SARS-CoV-2 context (13,14). This speci c impairment could be associated with a higher mortality rate in ICU (10).
However, at this time, no study has characterized renal impairment of SARS-CoV-2 in ICU patients. Our work aims to de ne the intrinsic renal injury induced by SARS-CoV-2 infection in critically ill patients, its consequences in terms of renal function, and its relationship with morbidity and mortality.

Study design and patients
This data-based, monocentric, observational study was conducted in a university hospital (Hôpital Saint-Antoine, Assistance Publique -Hôpitaux de Paris, France). All patients admitted to the ICU who tested positive for SARS-CoV-2 by Reverse Transcription-Polymerase Chain Reaction (RT-PCR) on nasopharyngeal or tracheal swab between March 25, 2020 and April 29, 2020 were included. Data were selected from electronic medical records (see additional le 1).

De nitions
The diagnosis of ARDS was carried out according to the Berlin criteria (16) (see additional le 1). AKI scoring was de ned according to the 2012 Kidney Disease: Improving Global Outcome (KDIGO) de nitions (17) and AKI as KDIGO ≥ 1. We de ned the serum creatinine baseline as the serum value on ICU admission, as previously described (18). In order to characterize the renal impairment of patients, the following standard de nitions were used: i/ renal response to hypovolemia -a marker predictive of prerenal acute kidney injury-as a fractional excretion of urea (FeUrea) of less than 35%; ii/ glomerular injury as the excess of high molecular weight proteins in urine: albuminuria >0.03 g/24h accounting for > 50% of total proteinuria or associated with supraphysiologic excretion of immunoglobulin G (IGG) (> 11.3 mg/24h). Samples with macroscopic hematuria were excluded as blood can interfere with the interpretation of proteinuria; iii/ proximal tubular injury as an increase in urinary low molecular weight proteins: urinary retinol binding protein (RBP) (>1.1 mg/24h ) or urinary alpha1-microglobulin (>15.8 mg/24h) in the absence of glomerular injury; iv/ renal tubular acidosis as a positive urinary anion gap in a context of metabolic acidosis; v/ mixed intrinsic kidney injury as the association of glomerular injury and renal tubular acidosis or by the presence of urinary IGG (>11.3 mg/24h) associated with albuminuria < 50% of daily proteinuria; glomerular injury, proximal tubular injury, renal tubular acidosis and mixed intrinsic kidney injury are referred as intrinsic kidney injury. Augmented renal clearance (ARC), was de ned as a clearance greater than 120 ml.min -1 (19).

Statistical analysis
Mean and standard deviation (SD) or median and interquartile ranges (25th; 75th percentiles) were calculated for continuous variables, while numbers and percentages were calculated for categorical parameters. The normal distribution of each continuous variable was assessed with the use of the Shapiro-Wilk test. For the univariate analysis, categorical variables were compared between independent groups using the exact Fisher test or the Chi-square tests, and continuous variables were compared using the Student's t-test or the Mann-Whitney test. Chord diagrams were used to depict the relationship between organic kidney impairment and kidney function at two different time points of the ICU stay. All statistical analyses were performed on R (version 3.6.2 for Macintosh, licenses GNU GPL, The R foundation for statistical computing, Vienna, Austria). All tests were 2-sided and a p-value < 0.05 was considered for statistical signi cance.

Characteristics of the study population
During the study period, between March 25, 2020 and April 29, 2020, 42 patients were admitted in ICU for SARS-CoV-2 infection, bene ted of urine assessment and were analyzed. The characteristics of the population are summarized in Table 1  *Values are expressed as median (interquartile ranges), absolute value (percentages); ARC (augmented renal clearance),BMI (body mass index), CKD (chronic kidney disease), CRP (C reactive protein), GFR (glomerular ltration rate), PaCO2 (CO2 partial pressure in the arterial blood), PEEP (positive endexpiratory pressure), SOFA (sequential organ failure assessment score) First week kidney abnormalities (Table 2) The rst blood analysis including in ammatory cytokines and urinary assessment was performed on a median of 8 days after ICU admission (IQR 6; 10 days), see additional le 3. Complete urinary samples allowing to de ne intrinsic renal injury were available for 34 patients.
Kidney abnormalities on ICU discharge (Table 3) The rst blood analysis including in ammatory cytokines and urinary assessment was performed on discharge from ICU at a median of 20 days after admission (IQR 15.75;23.25 days), see additional le 4. Urinary samples were available for 16 patients who stayed in ICU, eight were dead and ten were transferred to another hospital without any urinary analysis. Relationship between kidney function estimated by KDIGO and intrinsic kidney injury at early and late stages are represented in Figure 1.   vs 121.04 ml.min ¹ respectively in AKI and non-AKI group, p = 0.005). Creatinine clearance signi cantly decreased from hospital admission to discharge when AKI occurred during the stay. (Table 4) Augmented Renal Clearance was observed in 23 patients (54.8%) and more frequent in the non-AKI group (29.2% and 88.9% respectively in AKI and non-AKI group, p < 0.001). See additional le 5. Other known risk factors such as nephrotoxic agents' infusion (e.g., contrast agents, diuretics, aminosides), negative uid balance and admission severity scores (SOFA and SAPSII scores) didn't show any statistical association with the occurrence of AKI.

Discussion
The current cohort is representative of critical SARS-COV-2 patients with 80% of overweight men with a median age of 61.5 years, and at least one comorbidity (including hypertension and diabetes mellitus), (20,21). All of the patients had mechanical ventilation because of moderate to severe ARDS onset, which is in line with previous retrospective studies (20). Patients were comparable in terms of severity (median SOFA score of 7) to those usually admitted for ARDS (22).
Within the rst week of ICU admission, 94.1% of patients had features of intrinsic kidney injury and all the patients had documented intrinsic kidney injury three weeks after admission. Fifty-seven per cent of patients presented AKI according to KDIGO ranking and one third of them required RRT (Table 4). Conversely, 54.8% of patients presented augmented renal clearance (ARC) during their ICU stay including 29.2% patients in the AKI group. The intrinsic renal injury differed according to the time from admission; within the rst week, patients presented mainly mixed injury (73.5%) while on discharge injuries were distributed between mixed injury, glomerular injury, and tubular injury (43.8%, 31.2%, and 25% respectively). To our knowledge, this is the rst study describing with this level of precision kidney injury during SARS-CoV2 infection leading to ICU care.
In our cohort, 82.5% of critical SARS-CoV-2 patients presented proteinuria, indicating a high rate of intrinsic kidney injury which is in line with Pei et al concerning the Wuhan pandemic (9). In ICU patients, acute tubular necrosis is well documented up to 78% patients in autopsy series (23) whereas this current population of critically ill SARS-CoV-2 patients presented mixed pattern lesions (23,24). The prognostic signi cance of proteinuria or intrinsic kidney injury is unclear: it might re ect a strong adaptative potential of the kidney, or on the opposite, might be an early symptom of renal impairment. It might also re ect a speci c effect of SARS-CoV-2 infection on kidney cells, as suggested by previous descriptions of viral inclusions and by the expression of the ACE2 (the virus' putative receptor for cell invasion) in renal glomerular and tubular cells (7). It has to be noted that all the patients had a urinary catheter that could induce proteinuria in case of traumatic catheterization, reinforcing the importance of a complete urinary pro le to describe the intrinsic kidney injury speci c to SARS-CoV2. In another study conducted by Cheng et al. (18), including ICU and non ICU patients, only 9.8% of the patients had a urinary catheter, whereas 43.9% had proteinuria. This supports the idea that SARS-CoV-2 was responsible for the proteinuria that could be as high as 6.6 g/l in the current cohort.
In the current cohort, 94.12% patients presented with either AKI (KDIGO≥1) or intrinsic renal impairment (tubular or glomerular or mixed injury with KDIGO=0). Lower molecular weight proteins excretion (such as alpha1-microglobulin) in the non-AKI group in comparison to the AKI group suggests a continuum from intrinsic kidney injury to AKI. On the other hand, we assessed the contribution of prerenal mechanisms to AKI in these patients using the fractional excretion of urea (FeUrea). FeUrea is a robust marker of the renal response to hypovolemia more relevant than other markers like fractional excretion of sodium or urine to plasma ratio of creatinine, especially in case of diuretic use. According to FeUrea values hypovolemia is frequent among COVID-19 patients in ICU but the lack of association between AKI and low FeUrea values strengthens the role of intrinsic kidney injury associated to SARS-CoV2 infection in AKI occurrence.
Although the small number of patients did not allow to perform a multivariate analysis, univariate analysis highlighted several predictive factors of AKI during SARS-CoV-2 infection. Firstly, pre-hospital kidney function, de ned as renal functional reserve (25), might be critical in renal prognosis with higher creatinine level on hospital admission in patients who will develop AKI (94 mol.l ¹ in AKI group vs 74 mol.l ¹in non-AKI group). This creatinine value depends on patient's prior renal function but also on many other factors: hypovolemia induced by prolonged fever, infection-induced digestive disorders (26) and chronic hypertension, especially if treated by ACE inhibitors (27,28). However, hypovolemia may, at least partly, play a role as up to 50% of SARS-COV-2 patients had modi ed fractional excretion of urea during their ICU stay related to ARDS management, that requires avoidance of excess uid infusion (29).
Secondly, in the current cohort, all the patients presented high in ammatory response to SARS-CoV-2 infection, including high levels of IL6 plasma concentrations, as documented in previous studies (30) and morevover those who developed AKI had signi cantly higher CRP and PCT values on ICU admission.
Thirdly, mechanical respiratory support including the level of PEEP is critical in developing AKI in SARS-COV-2. As described by Hirsch et al., SARS-CoV-2 ventilated patients are at greater risk of developing AKI than non-ventilated ones (89.7% vs. 21.7%), especially during the rst 24 hours following intubation (31).
Elevation of central venous pressure due to high intrathoracic pressures may result in an increased kidney hydrostatic pressure, which leads to glomerular ltration impairment (32,33). Patients in both groups (AKI or non-AKI) presented similar PaO2/FiO2 ratios, indicating that the severity of the lung impairment is independent of kidney injury and, thus, reinforcing the importance of a tight controlled and tailored PEEP level. Finally, even if univariate analysis did not show any association between the type of organic impairment and AKI, all of critical patients presented intrinsic kidney injury 3 weeks after admission, which should contribute to the loss of renal functional reserve.
Renal prognosis is also a key issue for SARS-CoV-2 patients, considering the incidence of intrinsic kidney injury and the proportion of AKI. In a study including 5,273 non-SARS-COV-2 patients with no preexisting CKD, who developed AKI during an ICU stay, de novo AKI was associated with increased short and longterm risk of death at one and ve years (34). At one year, AKI acquired in ICU was also independently associated with increased risk of CKD (6%) and end stage renal disease (2%). In our study, among the 42 patients, 24 patients (57.1%) presented AKI during their ICU stay. When discharged from the hospital, 45.8% of the patients who developed AKI had impaired renal function with a creatinine clearance < 60 ml.min ¹. Given the facts that patients had several known risk factors for developing chronic kidney disease (sepsis, ARDS, mechanical ventilation, and in ammatory and procoagulant responses) and that direct viral injury on kidney cells is still not yet fully characterized, long-term nephrological follow-up may be required for all those who have developed AKI, especially if they have other comorbidities or still have proteinuria.
The current study also revealed ARC in more than 50% of patients, especially in those who did not present AKI during their ICU stay (88.9%). In a systematic review, 20 to 65% of critically ill patients presented ARC with a higher prevalence in trauma patients (35). Two main mechanisms have been suggested to explain ARC: i/ the release of pro-in ammatory cytokines that would lead to a decrease in resistance and an increase in glomerular ltration rate thanks to an increase in cardiac output and ii/ the e cacy of the physiological renal reserve allowing an increased glomerular ltration rate to cope with certain pathological situations such as ICU care. However, we did not nd an increase in albuminuria in patients with ARC, suggesting that it more likely re ect just a hemodynamic adaptation. According to some authors, ARC is a good prognostic factor, which is con rmed by our study (36). This ARC highlights a potential issue in critical patients because the use of regular doses of renally cleared drugs might induce underdosage (37). This is of most importance considering antimicrobial treatment such as beta-lactam antibiotics, vancomycin or aminoglycosides where ARC may condition clinical failure or emergence of resistance if higher dosage is not used.
Eventually, another point highlighted by the current study is the signi cant association between kidney failure and mortality, as 33.3% of AKI patients died, in comparison to 0% in the non-AKI patients. This relationship was suggested in previous studies in the pandemic context (18) with a 3.5 fold higher mortality in case of AKI KDIGO stage 2 or more (38).
This study has several limitations. First, it includes only a limited number of patients (42 patients) in one university hospital, making it impossible to rule out residual confusion and bias. In addition, some clinical and biological data from the admission to discharge from ICU were missing. Also, we used the baseline creatinine level at hospital entry as the baseline creatinine level for some patients, which could lead to an underestimation of AKI.

Conclusions
The prevalence of SARS-CoV-2 intrinsic kidney injury in critical care patients is high. A mixed injury is noticed early and persists during hospital stay. AKI is associated with mortality in excess in ICU patients and poor renal outcome. Detecting intrinsic renal injuries by analyzing complete urinary pro les on ICU admission might be recommended to adapt the clinical management of critical SARS-COV-2 patients. In accordance with the French law on biomedical research, this study obtained the approval of an Institutional Review Board ("Comité d'Éthique de la Recherche en Anesthésie-Réanimation" under the reference IRB 00010254 -2020 -106). See additional le 1.

Consent for publication
Patients were all informed of the possible use of their data in researches and publications as well as their right and terms of objection In order to guarantee the security of personal data, the investigators integrated the data anonymously into a secure database in accordance with the French Commission

Availability of data and materials
The data used and analyzed during the current study are available from the corresponding author on reasonable request.

Competing interest
All the authors declared no competing interests.

Funding
This work did not receive any nancial support. Figure 1 Chord diagrams representing characteristics of kidney injury during the ICU stay. a. Relationship between kidney function estimated by KDIGO and intrinsic kidney injury within the rst week after ICU admission. b. Relationship between kidney function estimated by KDIGO and intrinsic kidney injury on ICU discharge. c. Evolution of the intrinsic kidney injury between the rst week after ICU admission and ICU discharge.

Figures
The bottom part of the diagram represents patients sorted by their KDIGO classi cation, and the top part