Factors associated with acute kidney injury (AKI) and mortality in COVID-19 patients in a Sub-Saharan African intensive care unit: a single-center prospective study

Abstract Introduction Acute kidney injury (AKI) is a complication of severe coronavirus disease 2019 (COVID-19). Kidney damage associated with COVID-19 could take specific features due to environmental and socio-cultural factors. This study evaluates the incidence of AKI, the associated factors, and mortality in COVID-19 patients in a Sub-Saharan African intensive care unit. Methods In a prospective cohort study conducted in the intensive care unit (ICU) of the Centre Médical de Kinshasa (CMK), consecutive patients admitted for COVID-19 were screened for the presence of AKI between 27 March, 2020 and 27 January 2022. AKI was defined according to Kidney Disease Improving Global Outcomes (KDIGO) guidelines. The primary outcome was occurrence of AKI. The secondary outcome was 48 days’ mortality and recovery of the renal function at intensive care unit (ICU) discharge. Survival (time-to death) curves were built using the Kaplan Meier methods. Multivariate analyses were performed by logistic regression to identify factors associated with AKI and Cox regression to explore the association between AKI and in-hospital mortality. The significance level of the p-value was set at 0.05. Results The median(IQR) sequential organ failure assessment score (SOFA) score and mean age of patients (215) including in our cohort were respectively 3(2-4) and 58.9 ± 14.9 years. The incidence of AKI was 28.4% with stages 1, 2, or 3 AKI accounted for 39.3%, 11.5%, and 49.2%, respectively. Hemodialysis was required in 16 out 215 (7.4%) patients. Dyspnea (adjusted odds ratio (aOR):2.27 [1.1–-4.57] p = 0.021), SOFA ≥5 (aOR:3.11[1.29–7.53] p = 0.012), AST/ALT ratio (aOR: 1.53 [1.09–1.79] p = 0.015), N/L ratio (aOR:2.09 [1.09–3.20] p = 0.016), mechanical ventilation (aOR: 3.20 [1.66–10.51] p = 0.005) and Amikacin (aOR: 2.91 [1.37–6.18] p = 0.006) were the main factors associated with AKI. Patients with AKI had a mortality rate of 52.5% and 67.2% of the survivors did not recover kidney function at the end of hospitalization. Adjusted Cox regression analysis revealed that COVID-19-associated AKI was independently associated with in-hospital death (HR:2.96 [1.93–4.65] p = 0.013) compared to non-AKI patients. Conclusions AKI was present in three out of ten COVID-19 patients. The most significant factors associated with AKI were dyspnea, SOFA ≥ 5, AST/ALT and N/L ratio, mechanical ventilation and Amikacin. AKI has been associated with an almost threefold increase in overall mortality and seven out of ten survivors did not recover kidney function after AKI.


Introduction
In December 2019, the world started facing a new pandemic associated with the coronavirus disease 2019 .This severe acute respiratory syndrome coronavirus (SARS-COV-2) has infected more than 500 million people around the world with more than six million deaths, fueled by the different waves of infection carried by the appearance of many variants that are sometimes more contagious as illustrated by the delta variant [1].
Initially described mainly as a respiratory pathology, SARS-COV-2 infection has proven to be a more serious disease that can be responsible for multi-visceral damage, including kidney damage [2,3].Although the frequency of acute kidney injury (AKI), with or without proteinuria, was negligible in the first studies on COVID-19, recent series report a higher frequency, particularly in patients admitted to the intensive care unit (ICU) [4][5][6].Gradually, the various risk factors of acute kidney injury associated with COVID-19 (COV-AKI) are being identified, including patient-specific factors (age, ethnicity, comorbidities [chronic kidney disease (CKD), diabetes, hypertension…], genetic predisposition such as the the presence of apolipoprotein 1 (APOL-1) risk alleles, the severity of the disease well reflected by the levels of inflammatory markers that are indirect witnesses of the cytokine storm phenomenon or by hypoxemia, as well as exposure to certain therapeutic interventions (mechanical ventilation, vasopressors, antibiotics, etc.) [7].At the same time, COV-AKI has been shown to be a predictor of mortality [8] and for survivors, renal recovery varies according to the operational definition used in different studies, but remains lower when compared with other causes of AKI [9].AKI was found to be also a predictor for readmission and mortality in COVID-19 patients discharged early from the hospital [10].
Africa has also been affected by COVID-19, although the predicted magnitude was not consistent with reality [11].Multiple case report and small size studies reported an association between COVID-19 and kidney damage.Diana et al. were the first to describe AKI in a large cohort of 1102 COVID-19 patients in the Republic of South Africa [12].These authors observed a frequency of COV-AKI of 33.9% and identified risk factors similar to those reported in Western and Asian literature [12].In the Democratic Republic of the Congo, it is likely that COVID-19-associated kidney damage could exhibit specific characteristics because of the African genetic background on one hand, and the local environmental and sociocultural factors on the other hand.For instance, the prevalence of APOL-1 risk alleles for kidney disease, is high in this population [13].This high prevalence could exacerbate the severity of AKI [14].Regarding the environmental and sociocultural factors, the coexistence of infectious diseases such as malaria and the cultural attachment to traditional medicine, could further modify the presentation of AKI during COVID-19 in African communities.This study aimed at determining the incidence of COV-AKI, the associated factors, and the outcome in a cohort of Sub-Saharan African.

Study design, setting, and population
This prospective, observational, cohort study was conducted at the ICU of the Centre Médical de Kinshasa (CMK), a private facility appointed by the Ministry of Public Health as one of the few reference centers for the management of the COVID-19.CMK is located in the commune of Gombe, the epicenter of the COVID-19 disease in Kinshasa, capital of the Democratic Republic of Congo (DRC).Most of the patients admitted to the CMK had healthcare coverage fully provided by the employer, whereas few were nonaffiliated and payed out of their own pocket.At the time of COVID-19, CMK had two intensive care units, one of which was specially dedicated to COVID-19 patients with 15 intensive care unit beds and a ventilator to bed ratio of 1:1.The medical team included two nephrologists who were responsible for prescribing and monitoring hemodialysis sessions.The COVID-19 intensive care unit was equipped with one hemodialysis machine.Intermittent hemodialysis was the only one modality of renal replacement therapy.
To be recruited, patients had been at least 18-year-old, have a positive reverse transcription polymerase chain reaction (RT-PCR) COVID-19 test, and hospitalized and/or followed up at the ICU of the CMK during the first 4 COVID-19 waves from 27 March 2020 to 27 January 2022.Wave was defined according to the Congolese Technical Secretariat of the Multisectoral COVID-19 response committee (First wave: week 16/2020-week 34/2020; second wave: week 47/2020-week 13/2021; third wave: week 20/2021-week 32/2021; fourth wave: week 45/2021-week 2/2022) [14].At the time of the study, the Democratic Republic of Congo had one of the lowest COVID-19 vaccination rates, with less than 4% of the total population fully vaccinated [14].Only patients who demonstrated signs of moderate, severe and critical illness were admitted to hospital and systematically treated with macrolides and chloroquine in accordance with national guidelines.The positive experience of using this conventional treatment in many African countries, including the DRC, justifies the continuation of this therapeutic protocol for all 4 waves [14,15].The sample size was not predetermined and all patients meeting the inclusion criteria were recruited consecutively.Patients undergoing chronic dialysis (hemodialysis or peritoneal dialysis) and those with less than two serum creatinine (SCr) measurements were excluded.

Data collection
The clinical (demographic characteristics [age, gender and race], the existence of any chronic conditions [hypertension, diabetes mellitus, obesity, chronic kidney disease , thick blood smear, electrolytes and urinary analysis [proteinuria, hematuria, pH, specific gravity]), and radiological (thoracic computerized tomography scan [CT] score) data were collected from the admission.In accordance with the local protocol for COVID-19 patients' management at the CMK, specific biological parameters such as SCr, CRP, PCT, LDH, troponin and BNP were also systematically sampled on day 3 and 7, and beyond if the hospital admission was prolonged.In addition, we collected information about the treatment received (administration of antibiotics, corticosteroids, mechanical ventilation or hemodialysis), complications, and outcomes during the hospital admission.The date on which a complication was noticed or the treatment was initiated or the one reported as the day of exposure was considered as day 1.Chest CT-scans were performed on a 128 Slice HITACHI.SCr was measured by enzymatic method using a COBAS C 111 Automaton-Roche technology (Mannheim, Germany).Proteinuria and hematuria were assessed immediately at the admission from a sample of a 20 mL of urine by a semiquantitative and visual method using Combur 10 Test-M urinary strips from Cobas-Roche technology, Switzerland.

Operational definitions
AKI was defined and staged according to the KDIGO guidelines: Stage 1: increase in SCr by 0.3 mg/dL within 48h or a 1.5-1.9times increase in SCr from baseline within 7 days; Stage 2: 2.9 times increase in SCr within 7 days; Stage 3: 3 times or more increase in SCr within 7 days or new initiation of renal replacement therapy (RRT) [16].The lowest outpatient serum creatinine values between 7 and 365 days before Hospital admission was considered as baseline creatinine.If no previous results were available, baseline creatinine was the lowest creatinine value between that calculated from an estimated glomerular filtration rate (eGFR) of 75 mL/min/1.72m 2 and creatinine reported on admission.Baseline serum creatinine was available prior to admission only in 92/215 (42.7%) patients.The peak serum creatinine value during hospitalization was used to determine the AKI stage.Urine output criteria to define AKI were not used.The COVID-19 patient discharging criteria were defined by at least two negative RT-PCR-COVID-19 tests (within 48 h) in addition to the absence of fever for three days, and a regression of respiratory symptoms and the chest computer tomography (CT) Scan lesions [17].Proteinuria and hematuria was defined by the presence of semi-quantitative proteinuria or hematuria on the strip ≥1 cross.Renal recovery was defined according to Acute Disease Quality Initiative (ADQI) consensus as the return of serum creatinine (SCr) within 0.3 mg/dL of baseline SCr [18].The clinical and radiological severity of patients was assessed on admission using the Sequential Organ Failure Assessment (SOFA) [19] and the chest CT severity score, respectively [20].Acute respiratory distress syndrome (ARDS) was defined and staged according to the Berlin criteria: mild hypoxemia (200 mm Hg < PaO 2 /FIO 2 ≤ 300 mm Hg), moderate hypoxemia (100 mm Hg < PaO 2 /FIO 2 ≤ 200 mm Hg), and severe hypoxemia (PaO 2 /FIO 2 ≤ 100 mm Hg) [21].Dyspnea was defined as a subjective experience of respiratory discomfort reported by the patient on admission [22].

Outcomes
The primary outcome was the occurrence of AKI.The second outcome included need for renal replacement therapy (RRT), AKI recovery at discharge, survival defined as time to death and the vital status until the 48 th day.

Statistical analysis
Data were collected and analyzed using Excel and Stata (Stata Corporation version 15.0, Texas, USA 2017), respectively.Descriptive statistics were presented as mean and standard deviation for quantitative data with Gaussian distribution; median and interquartile range (IQR) for quantitative data with non-Gaussian distribution.Proportions were used for categorical data and percentages are based on the total number of non-missing values.Pearson's chi-square test or Fisher's exact test (for small numbers in one or more subgroups) was used to compare these proportions.For continuous variables, the comparisons were made using the student's t-test (variables normally distributed) or Mann-Whitney's test (variables not normally distributed).Kaplan Meier curves were used to describe patient survival (time to death).Patients who survived at the end of the study or shifted to another center were censored.A comparison survival curves was made using the Log Rank test.Univariate logistic regression analysis was used to identify factors associated with AKI.In order to control the effect of potential confounding, we further performed multivariable logistic regression analysis (using ascending step-by-step approach) to search factors independently associated with AKI.The adjusted ORs and their 95% CI were calculated finally to determine the degree of association between AKI and the independent variables.Only factors with p < 0.05 on univariate analysis were entered into the logistic multivariate regression analysis to define factor associated with AKI.The variance inflation factor (VIF) coefficient was used to identify multicollinearity between factors.Two multivariate mathematical models were used to limit the cumulative effect of independent variables.Multivariable cox regression model was used to identify independent factors associated with mortality.The significance level retained was then p < 0.05.

Patient characteristics
Of 277 consecutive patients with COVID-19 who were admitted to the ICU of the CMK between 27 March 2020 and 27 January 2022, only 215 were included in the final cohort (Figure 1).Portions of this results were presented at the International Society of Nephrology World congress [23].The mean age and the median (IQR) creatinine of the patients on admission were respectively 58.9 ± 14.9 and 91 µmol/L [74-115].The majority were male (77%), African ancestry (80.0%), and had hypertension as the main comorbidity in 98 (45.6%) patients.The median(IQR) sofa score of patients and time from hospital admission to discharge or death were respectively 3(2-4) and 10 (9.0-11.0)days.A total of 142/215 (67.3%) had ARDS (mild: 45.1%, moderate: 45.1%, and severe ARDS: 9.8%), 47 (21.9%) were treated with mechanical ventilation, and 37 (17.2%)required vasopressor support during the hospital stay.Norepinephrine was used as the first line vasoactive agent.Urine analysis was available in a total of 170 patients.Proteinuria was the most prominent finding and was documented respectively in 92/170 (54.1%) patients.Hematuria was seen in 32/143 (22.4%) patients.The mean pH and urine-specific gravity were respectively 5.5 ± 0.7 and 1.02 ± 0.007.

AKI and Outcomes
Within our cohort, 52 (24.9%) patients died.The median (IQR) length of stay from hospitalization to death was 10 (7-14) days, while the median (IQR) time from hospitalization to discharge 27

Incidence and severity of AKI
Although it is now well established that COVID-19 can induce AKI, the incidence of COV-AKI varies widely depending on the study populations, the AKI definition criteria, the COVID-19 strain, and the improvement in support over time.
Although the first Chinese, European, and USA series reported an incidence of AKI which varied between 1 and 42% [24][25][26], the more recent series report a lower incidence with an average of 29% among patients hospitalized in the USA and in Europe or and 6% in China [27].In African developing countries, this incidence varied widely from 18% to 50% [28,29].The incidence in our cohort was 28.4% and increased slightly to 29.6% when considering mainly patients of African ancestry but no statistical difference was observed when comparing ancestry groups.
The relationship between ancestry and the incidence of AKI has been widely reported.People with black ancestry are reported to be at higher risk, and this is ascribed to differences in clinical, socioeconomic, or genetic factors.Grams et al. in 2014, showed that the risk was more related to disparities related to economic status [30].In their study, by adjusting for the impact of AKI on differences in income and/ or insurance, the authors reduced the risk of AKI occurring in African Americans.High-risk alleles of APOL-1 variants were not associated with the risk [30].The framework of our study allows for this adjustment on the economic status given that all patients admitted to the CMK are covered by some sort of health insurance and therefore shows that improving health Bold values are the statistically significant p defining the variables (factors) associated to aKi.aK: acute kidney injury; alT: alanine amino transferase, aST: aspartate amino transferase, CCB: calcium channel blockers; Ci: confidence intervals MV mechanic ventilation; n: neutrophil; l:lymphocyte; aOR: adjusted odds ratio; OR: odds ratio; SOFa: sequential organ failure assessment.coverage in developing countries should have a significant impact on reducing the incidence of AKI.In our cohort, the incidence of AKI-D was high during the first wave and decreased over the time with a slightly increase during the third and fourth wave.This trend was also reported in a retrospective analysis of the incidence rate of hemodialysis unit admission in another COVID-19 center in Kinshasa [23].This decrease in incidence could be on the one hand the result of a significant reduction in the severity of the disease with time following the acquisition of spontaneous immunity or induced by vaccination, and on the other hand an improvement in the management.As reported elsewhere [31], Otshudiema et al.
showed in a retrospective study in the DRC, that the first and second waves of the COVID-19 pandemic in DRC were more severe than the third and fourth waves [14].Freund et al reported in their cohort a low incidence of acute complication in the vaccinated hospitalized patients [32].

Factors associated with AKI
The consensus report of the 25th Acute Disease Quality initiative summarized all of the risk factors for AKI-COVID-19 into three broad categories (demographic risk factors, AKI risk factors at admission and during hospitalization) [7].The main risk factors for AKI in our cohort fell into these different categories.COVID-19 disease severity is a risk factor of AKI and SOFA score that can be used to measure organ failure is positively correlated with the disease severity.As in previous study, SOFA score was associated with AKI in our cohort [33,34].
The AST/ALT ratio has not only been widely used for the assessment of the progression of liver failure and the prediction of liver fibrosis but also as a factor associated with all-cause and cardiovascular mortality [35].Its association with renal damage (AKI and/or proteinuria), although little reported during COVID-19, could be the expression on the one hand of the inflammatory syndrome and/or oxidative stress and on the other hand of a hepato-renal syndrome, a consequence of splanchnic vasodilatation from liver dysfunction, as an explanatory factor of AKI in COVID-19 patient [36][37][38].
As in sepsis, AKI associated with COVID-19 is the consequence of multiple factors, among which is also the dysregulation of the systemic inflammatory response [39].This is currently well evaluated through the neutrophil/lymphocyte ratio widely used in all medical disciplines as not only a marker of the immune response to infectious diseases but also as a valid index of stress and systemic inflammation [40].
The hemodynamic disturbance associated with mechanical ventilation are those associated with a reduction in cardiac output following the reduction in venous return induced by positive end expiratory pressure (PEEP).Ottolina et al. reported an association between the use of a high PEEP and AKI with a risk 5 times greater [41].More recently, ventilator-induced lung injury has been proposed as another mechanism of AKI via inflammatory crosstalk from the lung to the kidney [42].
The association of the inflammatory marker over time (days 3 and 7) in our study highlighted the possibility of induction of inflammation by AKI (Supplementary Table 1).The association between inflammatory markers and AKI has been widely reported in COVID-19 [43,44] and recent research has shown a bidirectional interplay between AKI and the immune system.Indeed, on the one hand, increased production and decreased clearance of cytokines as well as dysfunction of immune cells, in particular neutrophils, can contribute to immune dysfunction during AKI [45].On the other hand, during viral infection, white blood cells fight against the offending agent by producing cytokines that stimulate liver to produce C-reactive protein, an inflammatory stress molecule that may promote AKI in COVID-19 with other molecules or pathways such as angiotensin II-associated hypertensive stress diabetes-related metabolic stress, cytokine storm, over-reactive TGF-B signaling, complement activation and lung-kidney cross-talk [45] AKI associated with aminoglycosides has been also widely reported in the ICU patient population and is not only related to proximal or distal tubular damage but also to glomerular and vascular damage [46,47].This toxicity is dependent on the cumulative dose and more important in the context of disturbance of the renal hemodynamics [46].During COVID-19, although AKI associated with aminoglycosides is very little reported [48,49], it is easily understood in the face of disturbances in intra-renal hemodynamics associated with inflammation, hepatic dysfunction, cardiac dysfunction, mechanical ventilation, and significant rhabdomyolysis associated with COVID-19 [50].The use of amikacin also highlights the possibility of superimposed bacterial infection in COVID-19 patients or the existence of concurrent infectious disease which could explain the occurrence of AKI.These two phenomenon was reported in the literature and suggest the need of early diagnosis to possibly prevent AKI [51,52].

AKI and outcome
AKI is a strong predictor of mortality in the intensive care unit [53].This mortality is due on the one hand to metabolic complications specific to renal insufficiency (hyperkalemia, acidosis, volume overload, uremic toxins, etc.) and on the other hand to multi-visceral failure induced by AKI (organ cross-talk) defining AKI as a systemic disease [54,55].In our cohort, the overall mortality of COVID-19 patients was 25.8%.This mortality increased with AKI status and severity.The death rate of AKI patients was more than 2 time higher than all patients.Considering the AKI patients with hemodialysis, this mortality was more than 3 times higher that AKI stage 1.This trend (higher COVID-19-related mortality with the presence and severity of AKI) has been widely reported by many authors with a higher death rate (69%) than ours in the study by Watanabe et al. an overall mortality of 35% by Hirsh et al. lower than ours [6,56,57] and a mortality of 91% in AKI-D group in the study by Patel et al. similar to the 94% in our study [58].The disparity found in the literature could be due to differences in the study population as well as the methodologies used.As in most studies, AKI increases the risk of all-cause death by 3 times in our study population [6,57,58].
Renal recovery after an AKI episode depends on the one hand on the nephron reserve and, on the other hand, on the severity of the kidney injury.Nearly 1/3 of patients admitted in the study by Chan et al. had not recovered from their AKI episode [59].These authors explained these results by the fact that the severity of AKI was greater (as evidenced by the peak in creatinine, and the use of dialysis).This low recovery rate also reported in a Chinese study [60] was contrary to the 80% reported in AKI studies unrelated to COVID-19 [61].Ng et al. reported a recovery from a post-COVID-19 AKI episode of around 74.1% for stages 1-3 and 66.6% for AKI-D [57].
This disparity also accounts for the difference in the definitions used but also for the follow-up period before discharge.Xu et al. comparing five different definitions of renal recovery, reported that the ADQI criteria may overestimate the extent of renal recovery [62].Kidney recovery rate is probably higher at longer follow-up as has been demonstrated by Lumlertgul et al. who observed kidney recovery in 90.9% of survivors at 90 days [63] while Schaubroeck et al.only reported 50% at 21 days or at ICU discharge [64].With a high proportion of stage 3 and the very strict definition of renal recovery according to ADQI, almost 66% of patients with AKI in our cohort had not recovered at hospital discharge with a median time from hospital admission to discharge of 10 (9.0-11.0)days.Racial disparities with black as a risk factor can be another explanation for this poor renal prognosis.COVID-19 can be associated with less renal recovery.Fisher et al. reported 42.3% of renal recovery from COV-AKI vs 68.5% from AKI patients who were negative for COVID-19 [65].

Strengths and limitations
Our study has some limitations.First, this study was carried out in a single and private center providing specialized tertiary care thus the results cannot be generalized to all COVID-19 patients and can likely represent a selected group and lead to an underestimation of COV-AKI incidence.Second, the small sample size and high level of missing data on some variables (bilirubin, cholesterol, CK, uric acid, hematuria, baseline creatinine) were not sufficiently powered to identify potential associations between variables of interest.Third, the AKI definition was based only on creatinine criteria only.This can bias the study's estimation of AKI incidence.
Nevertheless, this study has the advantage of being the first one in the DRC to examine epidemiological, laboratory, and radiological data during the course of admission to evaluate some of the risk factors associated with AKI among COVID-19 patients on admission and the temporal changes in laboratory markers from illness onset in patients hospitalized with COVID-19.

Figure 1 .
Figure 1.Flow chart of study population selection.

Figure 2 .
Figure 2. Mortality rate according aKi status and aKi stage.

Figure 3 .
Figure 3. Survival curve of COViD-19 patients study population and according aKi status.

Table 1 .
Frequency of aKi regarding wave status.

Table 2 .
Baseline demographics, severity-of-illness parameters, and treatment during iCu admission.
AKI patients at discharge was higher than the recovered group (274.1 [132.6-486.2]vs97.2[88.4-176.8])µmol/L.Mortality in the AKI requiring dialysis (AKI-D) group was 94%.In the Kaplan-Meier survival analysis of the two groups based on AKI status (Figure3), the log-rank test (<0.001)revealed a difference between the groups with a better survival in patients with No AKIDiscussionThis prospective cohort showed that AKI was common in COVID-19 patients in Kinshasa and was associated with increased mortality.Dyspnea, SOFA score, AST/ALT and N/L ratio, mechanical ventilation and use of Amikacin were

Table 3 .
Biological and radiological characteristics at iCu admission.Percentage are based on the total number of non-missing values in each category and not necessarily on the total number of participants.pvalueswere calculated by Mann-Whitney U test or χ2 test(2), as appropriate.

Table 4 .
Factors associated with aKi in univariate and multivariate logistic regression analysis.

Table 5 .
Predictors of mortality in COViD-19 patients.Bold values are the statistically significant p defining the variables (factors) associated to mortality aHR adjusted hazard ratio.aKi acute kidney injury.alT: alanine amino transferase; aST: aspartate amino transferase; BOS: blood oxygen saturation; CRP C:reactive protein; eGFR: estimated glomerular filtration rate; Hb hemoglobin; Ht: hematocrit; HR: hazard ratio RR respiratory rate.