Serum uric acid, disease severity and outcomes in COVID-19

This retrospective study included two independent, discovery and validation cohorts of consecutive adult patients admitted to the Cliniques universitaires Saint- Luc, Brussels, Belgium, with a diagnosis of SARS-CoV-2 infection during the first (February 23, 2020 to April 18, 2020) and second (September 21, 2020 to November 21, 2020) waves of the pandemic. Patients on chronic kidney replacement therapy (hemodialysis, peritoneal dialysis or kidney transplantation) or for whom serum uric acid levels were not available were excluded. COVID-19 diagnosis was based on the detection of SARS-CoV-2 by real-time reverse transcription polymerase chain reaction on nasopharyngeal swab or broncho-alveolar lavage. The standard treatment for patients hospitalized for COVID-19 during the first wave (discovery cohort) included hydroxychloroquine 400 mg b.i.d. on the first hospital day and then 200 mg b.i.d. for four additional days in patients without contraindication, as recommended at the time of the study by the Belgian COVID-19 interim guidelines. The standard treatment for severe and critical patients hospitalized during the second wave (validation cohort) consisted of oral dexamethasone at a dose of 6 mg once daily for up to 10 days [9]. Enteral nutrition using Fresubin HP Energy Fibre (Fresenius Kabi, Schelle, Belgium) was initiated early (< 48 h) in all patients admitted to the ICU and as needed in those hospitalized in a non-ICU COVID-19 unit. We aimed for a target of 25–35 kcal/kg/day and 1.2–1.3 g/kg protein equivalents per day. All patients admitted to the ICU could be fully fed through the enteral route, and we did not use parenteral nutrition in any of them. Crystalloid solutions (Hartmann, alternatively 0.9% saline) were used. Patients were followed until death or end of study follow-up (July 31, 2020 for the discovery cohort and April 11, 2021 for the validation cohort). The flowcharts are presented in Fig. 1. A subset of 49 patients who underwent specific urinalyses to screen for the presence of proximal dysfunction were previously reported [8]. The twenty-three individuals from the discovery cohort for whom no uric acid level was available included (i) elderly patients with severe comorbidities, major therapeutic restrictions and a high mortality rate, and (ii) patients with mild or moderate disease and a very short hospital length of stay (Additional file 1: Table S1).

The study was conducted in accordance with the World Medical Association’s Declaration of Helsinki, the Belgian law related to experiments in humans dated May 7, 2004, the General Data Protection Regulation 2016/679 and the Belgian law of July 30, 2018, regarding the protection of personal data. The Ethical Review Board of Cliniques universitaires Saint-Luc/UCLouvain approved the study and waived the requirement to obtain informed consent based on the retrospective observational design of the study.

The following data were extracted from electronic medical records: demographics, symptoms at admission, vital signs, biological and imaging data, and outcome. Information about the use of the following drugs, known to interfere with uric acid metabolism, was also collected: allopurinol, febuxostat, fenofibrate, angiotensin receptor blockers and trimethoprim-sulfamethoxazole.

Laboratory measurements were performed on an automated Roche Cobas 8000 analyzer, equipped with modules ISE, c702, c502 and e602 (Roche diagnostics, Rotkreuz, ZG, Switzerland), or a UC3500 automated urine analyzer (Sysmex). Biological data were obtained at admission, and the worst (lowest or highest) levels were also recorded for some variables. Serum uric acid level was collected from medical records at the time of admission, at the time of first level under 2.5 mg/dl (if any) as a cutoff previously used for the study in SARS-CoV-1 [10] and at lowest level in both cohorts. Serum uric acid level was systematically assessed upon admission. Monitoring during hospitalization was systematic and performed on a daily basis among those admitted to the ICU, and left to the discretion of the physician in those on the general ward. In the small subset of patients requiring kidney replacement therapy, uric acid levels were only recorded before initiation of dialysis. In the discovery cohort, serum uric acid level was also collected at discharge from hospital, as well as one to twelve months before admission for COVID-19 in a subset of patients (n = 122) for whom these data were available. Inappropriate uricosuria was defined as a fractional excretion of uric acid > 10% in the presence of hypouricemia [8, 10]. In a previous report, detailed urinalyses showed that, in patients with COVID-19, hypouricemia is commonly associated with inappropriate uricosuria (fractional excretion > 10% in 18/20 [90%], and 7% and 8% in the two remaining individuals), supporting defective tubular handling and enhanced urinary losses of uric acid [8]. The definition of acute kidney injury was adapted from the KDIGO guidelines [11, 12]. As urine output was not recorded digitally in most patients (i.e., those admitted to a general ward), the definition of AKI was restricted to changes in serum creatinine, as previously done [13]. The term ‘mechanical ventilation’ refers to intubation followed by invasive mechanical ventilation.

Classification of COVID-19 severity was based on the adaptation of the Novel Coronavirus Pneumonia Diagnosis and Treatment Guidance [14]. Mild disease was defined as SARS-CoV-2 infection with no evidence of pneumonia on thoracic computed tomography (CT) scan; moderate disease, as clinical symptoms associated with dyspnea and radiological findings of pneumonia on thoracic CT scan, and requiring a maximum of 3 L/min of oxygen, stable for at least the following 24 h; severe disease, as respiratory distress requiring more than 3 L/min of oxygen and no other organ failure, stable for at least the following 24 h; and critical COVID-19, as respiratory failure requiring mechanical ventilation, shock and/or other organ failure requiring intensive care unit admission. Disease severity was defined as the worst severity score during hospital stay and also evaluated upon admission to assess progression during the course of the disease.

SARS-CoV-2 RNA detection was performed using COVID-19 genesig® Real-Time RT-PCR assay (Primerdesign Ltd, Chandler’s Ford, United Kingdom) in a LightCycler 480 instrument (Roche Diagnostics, Mannheim, Germany). Primers and probe of this assay target the RNA-dependent RNA polymerase gene. A test with a cycle threshold less than 40 was considered positive.

Immunofluorescence was performed on kidney samples obtained from autopsies of five patients who died from COVID-19 and three control patients. Paraffin blocks were sectioned into consecutive 5-μm-thick slices on Superfrost Plus glass slides (Thermo Fisher Scientific, Merelbeke, Belgium). Before staining, slides were deparaffinized in decreasing concentrations of ethanol and antigen retrieval was performed by incubating in sodium citrate buffer (1.8% 0.1 M citric acid, 8.2% 0.1 M sodium citrate, in distillated water, pH 6.0) in a water bath for 30 min. The sections were blocked with phosphate-buffered saline containing 5% BSA and incubated for 1 h with primary antibodies. Rabbit anti-human SLC22A12/URAT1 (HPA024575, Sigma-Aldrich, Saint-Louis, MO) and rabbit anti-human AQP1 (ab2219, Millipore) were used in this study. After 3 phosphate-buffered saline rinses, fluorophore-conjugated Alexa secondary antibodies (Invitrogen, Carlsbad, CA) were applied for 30 min. Negative controls were performed by omitting the primary antibody. Sections were subsequently mounted in ProLong Gold DAPI Antifade reagent (Invitrogen) and analyzed on a Zeiss LSM800 confocal microscope (Carl Zeiss, Jena, Germany), using × 20/0.8 Plan-Apochromat (Carl Zeiss). Quantitative image analysis was performed using Zen 2 (blue edition) software (Carl Zeiss) by randomly selecting 5 visual fields per each slide that included at least three to five proximal tubules, using constant setting parameters (i.e., pinhole, laser power, and offset gain and detector amplification below pixel saturation).

Results are presented as means ± SD or median [interquartile range (IQR)] for continuous variables and as numbers and proportions for categorical variables. Continuous variables were expressed in their natural units without standardization. Comparisons between groups were performed using unpaired t-test, Mann–Whitney, Kruskal–Wallis, or Chi-square test, as appropriate. Kinetics of serum uric acid was assessed using a mixed effects model taking into account repeated measures.

Survival analyses were performed using Kaplan–Meier estimates, Cox proportional hazard regressions and a competing risk model. Kaplan–Meier estimates and Cox proportional hazard regressions assessed the time to mechanical ventilation or death, and a Fine and Gray model was applied for the time to mechanical ventilation taking into account the competing risks of death or discharge. Multivariate analyses included the following pre-specified covariates: age, gender and baseline biological parameters (CRP, LDH, total lymphocyte count) in a first model and the same parameters and biological parameters of disease severity (highest CRP and LDH levels) in a second model. Nadir lymphocyte count was not included in Model 2 because of collinearity with baseline lymphocyte count. Collinearity between variables was quantified using variance inflation factors, and variance inflation factors > 10 suggested excessive correlation between variables.

Our study included a total of 517 patients hospitalized for COVID-19, recruited during the first (discovery cohort, n = 192) or second (validation cohort, n = 325) waves of the pandemic in Belgium (Fig. 1). Considering the whole study population, median age [IQR] was 65 years [54–76], and 60% were males (Table 1). Comorbidities included hypertension, cardiovascular disease and diabetes in 229 (44%), 152 (29%) and 125 (24%) individuals, respectively. Thirty-five (7%) patients were treated with allopurinol or febuxostat; 10 (2%), with fenofibrate; and 165 (32%), with inhibitors of the renin angiotensin system. Based on the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia [14], 38 (7%) patients had a mild; 113 (22%) moderate; and 274 (53%) severe COVID-19; the remaining 92 (18%) had a disease classified as critical.

We first examined the prevalence and timing of hypouricemia in patients hospitalized for COVID-19. In the discovery cohort, the prevalence of hypouricemia increased from 6.8% (13/192) upon admission to 20.3% (39/192) within the first days of hospitalization and then dropped to 3.6% (5/140) at discharge (Fig. 2). In contrast with the high prevalence during hospitalization for COVID-19, low serum levels of uric acid were very uncommon in the same cohort, before SARS-CoV-2 infection. Indeed, in the 122 individuals from the discovery cohort for whom serum uric acid levels before admission (median of 79 days [31–192]) were available, the prevalence of hypouricemia was only 0.8% (1/122) during the pre-COVID-19 era. These observations were replicated in the validation cohort, in which the prevalence of low serum uric acid levels increased from 6.5% (21/325) upon admission to 19.4% (63/325) during hospitalization (Fig. 2). The occurrence of hypouricemia was independent of the use of drugs interfering with uric acid production, nephrotoxic medications, treatment received for COVID-19, or viral load (Additional file 1: Table S2).

Altogether, these observations show that serum uric acid levels decrease early during hospitalization for COVID-19, leading to severe hypouricemia in a significant proportion of patients, which then reverses among survivors.

To assess the clinical relevance of low serum levels of uric acid in patients hospitalized with COVID-19, we compared disease severity and outcome of patients with versus without hypouricemia.

In both discovery and validation cohorts, the presence of hypouricemia strongly correlated with disease severity and with established biological parameters of severity, including hsCRP, LDH and nadir lymphocyte count (Fig. 3, Table 2). Considering pooled data from discovery and validation cohorts, the proportion of patients with hypouricemia was 17/151 (11%) among those with mild or moderate; 32/275 (12%) with severe; and 53/92 (58%) with critical disease (Fig. 3a; Table 2). Patients with critical COVID-19 also had lower serum uric acid levels during hospitalization (median [IQR] 2.0 [1.4–3.9] vs. 3.9 [3.0–5.2], 4.3 [3.3–5.2] and 3.9 [3.1–5.3] mg/dl in patients with severe, moderate and mild disease, respectively, linear trend, P < 0.001) (Fig. 3a). The presence of hypouricemia not only associated with severity of COVID-19, but also with progression of severity during the course of the disease, including 5 (13%) patients who progressed from moderate to severe; 7 (18%) from moderate to critical; and 9 (23%) from severe to critical disease, in the discovery cohort. In the group without hypouricemia, 18 (12%) with moderate progressed to severe, 2 (1.3%) with moderate progressed to critical; and 11 (7%) progressed from severe to critical (P value = 0.006) (Additional file 1: Fig. S1).

Patients who developed hypouricemia also showed higher levels of CRP (211 [110–326] vs. 110 [59–174] mg/l, P < 0.001); higher peak of LDH (535 [374–733] vs. 392 [306–492], P < 0.001); and lower nadir count of lymphocytes (420 [260–660] vs. 740 [490–1060], P < 0.001), than patients without hypouricemia (Fig. 3b; Table 2).

Overall, patients were followed for a median [IQR] of 148 days [50–168]. During follow-up, 108 (21%) patients were admitted to the ICU, 59 (11%) died without mechanical ventilation; 61 (12%) were started on mechanical ventilation; and 397 (77%) were discharged without mechanical ventilation (Fig. 1; Table 2). In both cohorts and in the pooled analysis, there was a strong association between hypouricemia and progression to respiratory failure requiring mechanical ventilation (pooled data: 47/102 [46%] vs. 14/415 [3%], P < 0.001, in patients with and without hypouricemia, respectively) (Table 2; Fig. 3c and d). The prevalence of hypouricemia was higher in participants admitted to the ICU vs. those who were not (53/108 [49%] vs. 49/409 [12%], P < 0.001).

The kinetics of serum uric acid levels in patients who required mechanical ventilation was thoroughly assessed in the discovery cohort. Serum uric acid levels started to decrease before the start of mechanical ventilation; reached a nadir (1.7 mg/dl [1.15–2.80]) 48–72 h after intubation; then started to re-increase in the majority of patients (21/29, 72%) (Additional file 1: Fig. S2A). In this subgroup, time from admission to lowest level of serum uric acid (median [IQR] 7 days [5–10] was similar to time to the worst value of other established biomarkers, including peak LDH (6 days [2–10]), peak CRP (7 days [4–9]) and nadir lymphocyte count (8 days [4–11]) (P > 0.05 for all) (Additional file 1: Fig. S2B). Median time [IQR] to hypouricemia was 4.5 days [3–7] and, in patients requiring mechanical ventilation, 45% (10/22) had developed severe hypouricemia before mechanical ventilation had to be initiated (Additional file 1: Fig. S2A). To rule out any impact from volume resuscitation on the dilution of serum uric acid, we thoroughly assessed fluid balance in the subset of patients from the discovery cohort developing hypouricemia while in ICU (n = 25), where this information was digitally recorded. In this subset of patients, the amount of fluid infused over the 2 days before the lowest value of uric acid was minimal, with a median [IQR] positive fluid balance of 186 ml per day [− 58, 1050]. The absence of any significant impact from excessive volume resuscitation on serum uric acid levels is also supported by (i) a decrease in serum uric acid levels already upon admission, as compared with baseline values obtained for a large subset (n = 122) of patients; (ii) plasma sodium concentration and serum osmolality in the normal range, with a median of 139 mmol/l [137–142] and 286 mOsm/kg [283–301], respectively, at the time of the lowest level of uric acid recorded during hospitalization.

Survival analyses using Kaplan–Meier estimates showed reduced mechanical ventilation-free survival among patients with versus those without hypouricemia (log-rank, P < 0.001) (Fig. 3d; Table 3). In each cohort considered separately, and in pooled data, time-to-event analyses using Cox regressions supported the significant association between hypouricemia and higher risk of invasive mechanical ventilation or death (pooled data: hazard ratio [HR] 5.1; 95% confidence interval [CI] 3.6–7.4, P < 0.001) and showed the association was independent from age, gender and biological parameters of disease severity (adjusted HR 5.6, 95% CI 3.8–8.3, P < 0.001). A Fine and Gray model considering mechanical ventilation as the primary endpoint and death or discharge from hospital as competing events confirmed the association between hypouricemia and the need of mechanical ventilation, both in unadjusted and unadjusted models (sHR 18.1, 95% CI 9.7–33.8, P < 0.001; adjusted sHR 18.8, 95% CI 9.3–38.1, P < 0.001) (Table 3). The sensitivity and specificity of hypouricemia as a biomarker for COVID-19 requiring mechanical ventilation were 77% and 88%, respectively, and the likelihood ratio, 6.4. Of note, the proportion of patients who developed hypouricemia (23/27 [85%] vs. 24/34 [71%), P = 0.18) was similar among survivors and non-survivors of critical COVID-19 requiring mechanical ventilation.

The 59 (11%) patients who died without mechanical ventilation were elderly individuals (median age [IQR]: 83 years [75–89]) with severe comorbidities and/or disabilities (cardiovascular disease, 61%; chronic kidney disease, 32%; nursing home residents, 42%), for whom major therapeutic restrictions had been set.

Post-mortem examination of kidneys from patients with COVID-19 showed structural alterations in the proximal tubule and decreased expression of the multi-ligand receptor megalin, which mediates the reabsorption of low molecular weight proteins [8]. Based on these observations and on the defective tubular handling of uric acid, we investigated the expression of urate transporter URAT1 (SLC22A12), which mediates the reabsorption of uric acid at the apical membrane of proximal tubule cells in kidneys from COVID-19 patients (Additional file 1: Table S3).

In a small series of 5 kidney autopsy samples from patients who died of COVID-19, confocal microscopy suggested a reduction in the expression of URAT1 as compared to control kidneys (mean relative maximal intensity 0.21 ± 0.12 vs. 1.00 ± 0.27, P = 0.001) (Additional file 1: Fig. S3). In contrast, the expression of aquaporin-1, used as an internal control for the proximal tubule brush border, was not different between cases and controls (0.75 ± 0.47 vs. 1.00 ± 0.47, P = 0.5). The median [IQR] lowest level serum uric acid in these patients was 2.3 mg/dl [1.5–3.8].

In line with previous evidence showing defective tubular handling of uric acid, these observations suggest that altered expression of urate transporters in the kidney proximal tubule contributes to the development of hypouricemia in patients with COVID-19.