Different characteristics of acute kidney injuries in different stages of COVID-19

Background Kidney involvement in COVID-19 may manifest as acute kidney injury (AKI). This study aimed to analyze and compare AKIs in different stages of COVID-19. 1056 hospitalized COVID-19 patients were retrospectively evaluated and 383 of them met the inclusion criteria. Eighty-nine patients who developed AKI, but didn’t have prior kidney diseases were involved in the nal analysis. Patients were classied into three groups, those who had AKI on admission, those who developed AKI in the rst week and those who developed AKI starting from the 7 th day. Electrolytes, acid-base status and changes in the inammatory markers were compared.


Introduction
The clinical course of Coronavirus Disease 19  patients who needed hospital admission might be examined in 3 consecutive stages: stage 1 as the early infection period ( rst 3 days after being infected by the virus) stage 2 as the intermediate period (until 7 th day of the illness) with pulmonary involvement and stage 3 as the systemic hyper-in ammation phase. Stage 3 is generally accepted to start within the 2nd week of disease course [1]. While COVID-19 is mainly a respiratory illness, kidneys may also be involved. Multiple pathologic mechanisms have been proposed to explain the cause of kidney involvement including uid balance disturbances, angiotensin II pathway activation, endotheliitis with intravascular coagulation, lung-kidney and heart-kidney cross talks, cytokine release syndrome and drug nephrotoxicity [2,3].
Kidney involvement in COVID-19 can be manifested as acute kidney injury (AKI). Previous studies generally evaluated all forms of AKIs together. However, AKIs may have different characteristics depending on the timing and etiologies. This study aims to analyze the etiologies and prognosis of AKI in different phases of the disease among hospitalized COVID-19 patients without prior kidney diseases.

Materials And Methods
Setting Patients who were admitted to the designated COVID wards in Cerrahpasa Medical Faculty, a tertiary healthcare center, between 15th March and 1st July 2020 were retrospectively analyzed. This period has been the rst wave of the pandemic in our country and symptomatic patients were admitted immediately.
Hospitalized COVID-19 patients whose disease status was con rmed by a real-time polymerase chain reaction (RT-PCR) test for Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-Cov-2) were involved in the study. Kidney transplant patients and those who were younger than 18 years old were excluded from the study. As glomerular ltration rate (GFR) below 60 mL/min/1.73 m 2 was already shown to be related to mortality [4], these patients were also excluded. (Figure-1).

Evaluation of COVID -19 patients
History, physical examinations, clinical features, laboratory tests, radiologic investigations, microbiological tests and in-hospital clinical progress of the patients were retrospectively investigated.

De nitions
To de ne AKI, Kidney Disease Improving Global Outcomes (KDIGO) criteria were used; an absolute increase of 0.3 mg/dl in creatinine levels in 48 hours or 50% increase in creatinine levels in the last 7 days or when urine output is less than 0.5 mL/kg/h for the previous 6 hours. [5].
We observed the progression of creatinine values in all patients who were admitted with COVID-19 diagnosis. In patients with an increase in creatinine levels, we directly applied KDIGO criteria. The rst calculated creatinine level after being admitted to hospital was taken as the baseline creatinine level for these patients. For patients with a decrease in their creatinine levels following hospital admission, KDIGO criteria were applied according to patients' previous creatinine levels. When there was no previous data 7 to 365 days prior to hospital admission, baseline creatinine levels were backwards calculated using the MDRD 75 formula [5,6].
Stage of the AKI was also de ned according to KDIGO criteria; 1.5 -1.9 times baseline creatinine or 0.3 mg/dl absolute increase as stage 1 AKI; 2.0 -2.9 times baseline creatinine as stage 2 AKI and more than 3.0 times baseline creatinine or increase to more than 4.0 mg/dL as stage 3 AKI.
Estimated glomerular ltration rate (eGFR) was used to de ne the kidney functions and it was calculated by Chronic Kidney Diseases Epidemiology Collaboration (CKD-EPI) formula.
State of consciousness was evaluated by Glasgow Coma Scale and a drop of more than 2 points was a reason to call intensive care unit (ICU) team to evaluate the patient for a possible ICU admission. Mean arterial pressure was calculated as: [(systolic blood pressure) + (2 x diastolic blood pressure) / 3].
Hematuria was de ned as the presence of more than three red blood cells per high power eld in the urine sediment. Proteinuria was detected semi-quantitavely by a fully automated urine dipstick test. The level of proteinuria was graded as +1, +2 or +3; indicating levels between 30-100 mg/dL, between 100-300 mg/dL and over 300 mg/dL respectively.
We de ned three groups on the basis of the timing of AKI; those seen on admission, those developed in the 1st week and those developed after the 1 st week.
Etiologic evaluation of AKI were carried out according to following criteria: -Transient pre-renal AKI was de ned when creatinine levels could be reversed to baseline levels with relevant uid resuscitation in 24 to 72 hours.
-AKI was attributed to rhabdomyolysis in patients with at least ve times elevated CK (i.e, >950 U/L) above upper normal limit with concomitant increase in LDH and transaminases.
-Hypoxemia related kidney damage was noted in patients who had disrupted gas exchange. Such disruption was documented with partial oxygen pressure lower than 60 mmHg despite the use of high ow oxygen therapy or mechanical ventilation (either non-invasive or invasive) or when respiratory acidosis developed with increasing levels of CO 2 retention.
-In ammation mediated AKI is considered in patients who have increasing levels of ferritin reaching to 5 times above the normal upper level (>750 ng/mL) or increasing levels of D-dimer that reached at least ten times of the upper normal limit (>5 mg/L).
-Secondary bacterial infections were diagnosed by blood cultures or urine cultures.
-AKI was attributed to drug toxicity if AKI developed after the patient was exposed to nephrotoxic drugs or agents.

In ammation mediated injury assumption
It's known from before that immune system dysregulation, complement system activation and hypercoagulopathy were all linked with each other [7]. We have observed a similar phenomenon in our etiologic analysis. It may not be always possible to de ne which has started before and caused the others. That is why, AKIs in patients either with increasing D-dimer levels or cytokine release syndrome that manifests with increasing levels of ferritin were associated with the hyper-in ammation state of COVID-19 [8].

Severity of COVID-19
Clinical picture of COVID-19 patients were classi ed according to a scale that included following -Critical (acute respiratory distress syndrome, shock) [9].
Patients were admitted to intensive care unit, if arterial partial oxygen pressure was persistently below 50 mmHg despite the use of venturi mask, when there was accumulation of CO 2 rising above 55 mmHg, when the patient had loss of consciousness or when the patient was consistently hypotensive despite uid resuscitation.

Acquisition of Data
Hospital electronic health records and patient les were used to collect the data. Admission day to the COVID ward was accepted as the day zero of the patient follow-up.

Statistical Analysis
Data were expressed as means ± standard deviation. Continuous variables were compared by independent samples t-test. Categorical variables were compared either by Pearson chi-square or Fisher's exact test. For the comparison of three groups that were created according to the timing of AKI, ANOVA test was performed. Tukey HSD test was used for post-hoc analysis. All tests were applied using SPSS for Windows, version 22.0 software (SPSS Inc. Chicago, IL, USA). P values less than 0.05 were accepted as statistically signi cant.

Results
A total of 1056 patients were admitted in this speci ed period. The mean age of COVID-19 patients hospitalized in our center was 55 ± 15 years, 582 (55,2%) of them were males while 474 (44,8%) of them were females. 427 patients were con rmed by RT-PCR. 104 of the PCR con rmed COVID-19 patients experienced AKI (24,3%). 89 patients who developed AKI with an eGFR of over 60 ml/min/1.73 m 2 were included in the nal analysis ( Figure-1). Patients who were included in our study were older (62,4 ± 14,2 years) than other patients in the general COVID-19 cohort and there was a male predominance (67 males, 75%).
Twenty-nine (32%) of the patients had AKI on admission. 33 of them (37%) developed AKI during the rst week of admission and 27 patients (30%) developed AKI starting from the second week of admission. For patients who developed AKI later than hospital admission date, AKI developed on the 6.7th ± 5.4th day of the admission. Initial laboratory values on hospital admission day and in-hospital prognostic indices of all 89 patients can be found in the supplementary document.

Etiologic evaluation
Patients who had AKI on admission Twenty-nine patients had AKI on admission.
Twelve (41,3%) of these patients had transient pre-renal AKIs as their kidney functions were rescued by relevant uid resuscitation. Two (6.8%) of the patients had rhabdomyolysis related AKI. Both of these patients also had increasing levels of either ferritin (>750 ng/mL) or D-dimer (>5 mg/L). Six patients (20,6%) had respiratory disruptions (persistent hypoxemia or hypercapnia). These patients were considered to have hypoxemic kidney injury. Four of these patients also had high levels of ferritin (>750 ng/mL) or D-dimer (>5 mg/L) levels. A total of eight patients (27.5%) had hyperferritinemia and/or high Ddimer levels without rhabdomyolysis. AKIs in these patients were attributed to hyper-in ammation. One of the patients in this group developed proteinuria concomitant with AKI, despite normal levels of in ammatory and coagulation markers. COVID severity of the patients who had AKI on admission was mainly moderate and just 7 of them (24%) had severe or critical disease.
Patients who had AKI in the 1 st week.
Thirty-three patients experienced AKI during the 1 st week.
Rhabdomyolysis was noted in four patients (12.1%). All of these four patients also had concomitant hyperferritinemia (>750 ng/mL). A total of 13 patients (39.3%) had high ferritin or D-dimer levels without ndings of rhabdomyolysis. The kidney injury was attributed to hyper-in ammation in these patients. Three patients (9%) either had hypoxemia or increasing levels of CO 2 . AKI was attributed to hypoxemia in these patients. AKI in three patients (9%) of this group was related to drug toxicity (contrast agents in two patients and non-steroid anti-in ammatory drug in one patient). When COVID severity of the patients was evaluated, 14 of these 33 patients (42%) were classi ed as severe or critical.
Patients who had AKI after 1 st week: Twenty-seven patients had AKI after the 1 st week of their admission.
One patient (3.7%) had transient pre-renal AKI. Six patients (22.2%) had rhabdomyolysis. All of the patients who had rhabdomyolysis also had either high D-dimer or ferritin levels. Two patients (7.4%) had severely disrupted gas exchange without concomitant high ferritin or D-dimer levels. AKIs in these patients were attributed to hypoxemia. Sixteen patients (59.2%) had very high levels of either ferritin (>750 ng/mL) or D-dimer (>5 mg/L). AKIs in these patients were associated with hyper-in ammation. Two patients (7.4%) had contrast agent induced AKI. Clinical evaluation pointed out to severe or critical illness in 22 of these 27 patients (81%).

Urine analysis:
Urine analysis was available in a total of 35 patients. Hematuria was the most prominent nding, which was seen in 21 of them. Proteinuria was documented in 9 patients and they were all 1+ semiquantitavely. Proteinuria was going along with hematuria in 7 patients while two patients had isolated proteinuria.

Imaging studies
Chest CT to investigate pulmonary involvement was performed in all patients. COVID pneumonia was detected in a total of 82 patients (92.1%).
Kidney imaging (urinary ultrasonography or abdominal CT) was available in 14 patients. Eight of them were reported to be completely normal. Three patients had nephro-urolithiasis, one patient had pelvic ectasia, one had prostatic hypertrophy and one had the ndings of cystic kidney diseases. Imaging studies neither yielded obstruction ndings nor could explain the AKI etiology.

Treatment modalities
Although there is no speci c validated treatment for COVID-19 yet, some antiviral therapies were applied in accordance with the ministry of health (MoH) treatment guidelines. These include different combinations of hydroxychloroquine, favipiravir and lopinavir. Anti IL-6 receptor antibody tocilizumab or steroids were used in patients who had high in ammatory response. Low-molecular-weight heparin were prescribed for all patients in line with the MoH guidelines [10]. Continuous renal replacement therapy (CRRT) in ICU setting was performed with Prisma ex® system in a citrate anti-coagulated circuit, aiming a blood ow of around 20 mL/kg/hour.

Comparison between the groups formed according to the AKI timing
Patients of the three groups (AKI on admission, AKI in the 1 st week, AKI after the 1 st week) were in similar age and had similar baseline mean arterial pressure, creatinine and hemoglobin levels. Co-morbidities such as diabetes, hypertension, malignancies and ischemic heart diseases/heart failure were also similar between three groups. CRP and D-dimer levels on admission didn't differ between the groups. Patients who had AKI on admission day had higher initial uric acid levels. All initial laboratory values of the patients can be found in table-1.
Duration of hospital stay, intensive care unit (ICU) requirement and mortality was higher when AKI developed later in the disease course, especially after 7 th day. Patients who develop later AKIs had lower serum albumin levels as well as lower arterial O 2 pressure and lower oxygen saturation levels. Predominant stage of AKI was stage 1; however, stage 2 & 3 AKIs, which have worse prognosis tend to increase with AKIs that occurred later (table-2). AKI related prognostic indices of the patients can be found in table-2.
While there were no signi cant differences between the initial in ammatory markers of the three groups, comparison of changes put forth signi cant differences. Nadir lymphocyte counts were signi cantly lower while peak CRP and peak D-dimer levels were signi cantly higher for patients who developed AKI later in the disease course (Table-3). Although it couldn't reach the statistical signi cance, peak ferritin levels were also higher for patients who developed AKI later.
Sodium, chlorine and potassium abnormalities were more common in patients who developed AKI later (Table-3).
Treatment modalities were not different between the groups (Table-4). RRT had to be performed in 6 patients who developed AKI later (2 among the 1 st week AKIs and 4 among the AKIs developed after the 1 st week) but none of the patients who had AKI on admission needed RRT. Anti IL-6 receptor antibody tocilizumab use was signi cantly more frequent for patients who developed AKI after 7 th day. Pulmonary involvement (i.e. COVID pneumonia) was similar between the groups and there was not a statistically signi cant difference for secondary bacterial infections (Table-2).

Comparison between survivors and non-survivors
Duration of hospital stay was not different for survivors and non-survivors. Those who died were older. Patients who survived and who didn't had similar rates of diabetes or hypertension, while concomitant malignancies were more frequent in patients who died (Table-5).
AKI had 24.7% mortality in our patients who had eGFRs above 60 ml/min/1.73 m 2 . AKI developed later in non-survivors and it lasted longer. Non-survivors had signi cantly higher initial CRP, LDH, ferritin and Ddimer levels while their hemoglobin and lymphocyte counts were signi cantly lower (Table-5).
Patients who died had more secondary bacterial infections (OR: 3.5 ; 95%CI: 1.9 -6.4). However, ferritin levels, as a marker of in ammation, were similar in patients who had secondary bacterial infections and in those who hadn't (n=24; 1120 ±691 vs n=62; 976 ± 109; p=0.548). Urea-to-creatinine ratios checked both on the day of AKI and on the day of worst kidney function, were higher in patients who died (p=0,02 and p=0,000 respectively).

Discussion
Different studies reported variable AKI incidences in COVID-19 [11][12][13][14]. In the consensus report of Acute Disease Quality Initiative, AKI incidence was reported to be around 20% for hospitalized patients [15]. Same report underlines that AKI may develop in 50% of the patients who needed ICU support. AKI has been proposed as a poor prognostic factor for COVID-19 [16]. In a meta-analysis, it was found that 52% of patients who developed AKI had died [17]. Another study showed that, chronic kidney disease and male sex were independent predictors of AKI severity [18]. However, AKI studies in patients with normal kidney functions are scarce. In this study, we focused on the prognosis of AKI of otherwise normal kidneys by excluding patients whose eGFRs were below 60 ml/min/1.73 m 2 . Overall mortality was calculated as 24.7% in this group.
Consequences of all AKIs in COVID-19 might not be the same. As COVID-19 is a febrile illness and patients are experiencing gastrointestinal disturbances, pre-renal AKI is somewhat expected upon admission and should be transient. There were still AKIs related to other etiologies on admission, and this may be because of differences in the severity of the disease or relatively late referrals of some patients. On admission AKIs were mainly transient pre-renal AKIs that were responsive to uid therapy (41%). This decreased to 30% for rst week AKIs and to 3% after the 1 st week. It may not possible to differentiate between coagulopathy and cytokine release as both pathologies may be intertwined with each other [19,20]. When they were taken together, in ammation-mediated injury was around 27.5% for on admission AKIs. This increased to 39.3% for rst week AKIs and it was 59.2% for patients who experienced AKI starting from the second week. Severe COVID-19 was more common in patients who developed AKI later. The mortality of patients who experienced AKI in the early period was 13.7%, and this increased to 44% for patients who had AKI after the 7 th day.
It may be di cult to nd the exact etiology of AKI in the course of COVID-19. Kidney biopsies may give some clues. Direct virulence of SARS-CoV-2 may be responsible for kidney involvement with acute tubular injury and podocytopathies [21,22,23]. In a report of kidney biopsies in COVID-19 patients, podocytopathies and tubulo-interstitial diseases were main ndings while immune mediated glomerular diseases were also found [24]. That study didn't detect virus particles in the kidney. Another study of kidney biopsies on a series of 10 patients found acute tubular necrosis as the leading pathology of AKI. Myoglobin casts as well as thrombotic microangiopathy were also reported [25]. We didn't perform kidney biopsies, as it was neither clinically indicated nor would change treatment modalities in the vast majority of our patients. Clinical ndings, laboratory values and response to the planned interventions guided us to reach the possible etiologies. 12 of our patients had rhabdomyolysis. Exact mechanism of rhabdomyolysis in COVID-19 is not very clear. While one theory is direct viral invasion of muscles that results in rhabdomyolysis, the other postulates that cytokine storm causes muscular injury [26]. Concomitant increase in ferritin or D-dimer levels in our patients underlines the latter.
Patients who develop AKI later had higher peak CRP, D-dimer and ferritin levels. Such higher in ammatory response may point out that later AKI is more immune-mediated. Although secondary bacterial infections could be a confounding factor, ferritin levels, as a marker of in ammatory response in patients with or without secondary bacterial infections didn't differ.
Drug induced nephrotoxicity should not be overlooked in AKIs that develops later. Drugs that resulted in AKI in our patients were non-steroidal anti-in ammatory drugs, antibiotics (e.g. aminoglycosides) and contrast agents that were used for computer tomography scans.
Hyponatremia, and hypochloremia were common electrolyte abnormalities in COVID-19 patients who had AKIs, but they were at a similar rate for survivors and non-survivors. Hypernatremia tended to develop later and this might be related to hypertonic enteral feeding formulas, saline uid administrations or steroid use [27]. Mortality was increased in patients who had hypernatremia or hyperchloremia.
Both hyperpohospatemia and hypophosphatemia were related to poor prognosis. Hyperphosphatemia mainly develops as a consequence of GFR loss in patients who have AKI. Tubular injury, anti-acid drugs, malnutrition, respiratory alkalosis or CRRTs may be responsible factors for the development of hypophosphatemia. Negative impact of hypophosphatemia on prognosis might be a consequence of decreased diaphragmatic contractility [28].
Our ndings showed that high urea-to-creatinine ratio could be a marker of poor prognosis. These patients might have higher serum urea levels that point out to higher catabolic state and they may also have relatively lower creatinine levels, which is indicative of reduced muscle mass.
There are some limitations of our study. Firstly, due to the retrospective nature of the study, urine analysis and urinary imaging studies were not available for all patients. Sample size is relatively small, and this is because of including only PCR con rmed patients who have eGFRs of over 60 ml/min/1.73 m 2 . Due to reasons stated above, kidney biopsies, which might have given more information about etiologies, were not performed.

Conclusion
AKI in COVID-19 is not of one kind. When developed, AKI should be evaluated in conjunction with the disease stage. Early AKI tends to be more transient and may be more responsive to uid resuscitation. However, AKIs that develop later are more immune-related and have worse prognosis. Patients who develop later AKIs are also more prone to electrolyte abnormalities. Figure 1 Flow chart of exclusion criteria of the patients. PCR: polymerase chain reaction, eGFR: estimated glomerular ltration rate.