*Article* **Evaluation of Two Rapid Antigenic Tests for the Detection of SARS-CoV-2 in Nasopharyngeal Swabs**

**Ysaline Seynaeve 1,\*, Justine Heylen 1, Corentin Fontaine 1, François Maclot 1, Cécile Meex 1, Anh Nguyet Diep 2, Anne-Françoise Donneau 2, Marie-Pierre Hayette <sup>1</sup> and Julie Descy <sup>1</sup>**


**Abstract:** (1) Background: In the current context of the COVID-19 crisis, there is a need for fast, easyto-use, and sensitive diagnostic tools in addition to molecular methods. We have therefore decided to evaluate the performance of newly available antigen detection kits in "real-life" laboratory conditions. (2) Methods: The sensitivity and specificity of two rapid diagnostic tests (RDT)—the COVID-19 Ag Respi-Strip from Coris Bioconcept, Belgium (CoRDT), and the coronavirus antigen rapid test cassette from Healgen Scientific, LLC, USA (HeRDT)—were evaluated on 193 nasopharyngeal samples using RT-PCR as the gold standard. (3) Results: The sensitivity obtained for HeRDT was 88% for all collected samples and 91.1% for samples with Ct ≤ 31. For the CoRDT test, the sensitivity obtained was 62% for all collected samples and 68.9% for samples with Ct ≤ 31. (4) Conclusions: Despite the excellent specificity obtained for both kits, the poor sensitivity of the CoRDT did not allow for its use in the rapid diagnosis of COVID-19. HeRDT satisfied the World Health Organization's performance criteria for rapid antigen detection tests. Its high sensitivity, quick response, and ease of use allowed for the implementation of HeRDT at the laboratory of the University Hospital of Liège.

**Keywords:** rapid diagnostic test; antigen detection; SARS-CoV-2; COVID-19

#### **1. Introduction**

In December 2019, a new virus belonging to the *Betacoronavirus* genus from the *Coronaviridae* family appeared in the city of Wuhan in China. It was identified as the agent responsible for a severe respiratory syndrome, hence its name: SARS-CoV-2 for Severe Acute Respiratory Syndrome Coronavirus. The virus is responsible for coronavirus disease 19 (COVID-19), so named because it appeared in 2019. Other betacoronaviruses involved in human respiratory syndromes include SARS-CoV-1—the agent of the Severe Acute Respiratory Syndrome discovered in 2002—and MERS-CoV, responsible for the Middle East Respiratory Syndrome in 2012 [1,2]. Within a few months, SARS-CoV-2 spread rapidly, leading to the World Health Organization's (WHO) announcement of a COVID-19 pandemic on 11 March 2020 [3]. As of 3 June 2021, 171.3 million people have been infected and 3.7 million people have died worldwide [4]. This pandemic has caused microbiology labs to develop diagnostic methods based on antigen and RNA detection. The WHO defined a reverse transcription polymerase chain reaction (RT-PCR) assay on respiratory specimens as the reference method for the detection of SARS-CoV-2 [5]. This coronavirus is an enveloped virus with a positive-sense, single-stranded RNA genome of ~30 kb. Its genome encodes a minimum of 29 proteins. The ORF1a and ORF1b genes encode non-structural accessory proteins (nsps), while the S, M, E, and N genes encode structural proteins—namely, spike (S), membrane (M), envelope (E), and nucleocapsid

**Citation:** Seynaeve, Y.; Heylen, J.; Fontaine, C.; Maclot, F.; Meex, C.; Diep, A.N.; Donneau, A.-F.; Hayette, M.-P.; Descy, J. Evaluation of Two Rapid Antigenic Tests for the Detection of SARS-CoV-2 in Nasopharyngeal Swabs. *J. Clin. Med.* **2021**, *10*, 2774. https://doi.org/ 10.3390/jcm10132774

Academic Editor: Robert Flisiak

Received: 7 June 2021 Accepted: 16 June 2021 Published: 24 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

(N) proteins [6,7]. These genes are of particular interest for the diagnosis of COVID-19 by RT-PCR.

The real-time RT-PCR assay provides a semi-quantitative estimation of viral concentration expressed by the cycle threshold (Ct), which represents the number of amplification cycles required to detect a fluorescence signal above the threshold. The cycle threshold is correlated to the viral load and contagiousness [8,9]. Despite the very high specificity and sensitivity of molecular techniques, RT-PCR is time-consuming and requires specific laboratory equipment and experienced technical staff [10,11]. These limitations—combined with the shortage of PCR reagents and disposables—have led laboratories to investigate alternatives. In September 2020, the WHO reported the possible use of antigenic tests as a promising complement to RT-PCR. The WHO described rapid diagnostic tests (RDT) with a sensitivity ≥80% and a specificity ≥97% compared to the reference RT-PCR method that can be used when molecular tests are not available or when rapid screening is needed [8]. Confronted with the second wave of COVID-19, laboratories were under pressure to perform RT-PCR for the detection of SARS-CoV-2. The turnaround time was longer than the acceptable range of 24 h, which resulted in delays to diagnosis that compromised patients and epidemic management [12]. In addition, the emergency department of our hospital (CHU, Liège, Belgium) needed an early diagnosis of SARS-CoV-2 in order to manage patient hospitalization in COVID or non-COVID care units. Many companies have developed RDTs for the diagnosis of COVID-19 as point-of-care testing methods. Despite the high sensitivity announced by the manufacturers, the performance of these tests is varied and often lower than expected; therefore, their validation in real conditions is important [13–18]. A validation of two RDT kits for the detection of SARS-CoV-2 compared to RT-PCR was conducted at the microbiology laboratory of CHU Liège. The two kits are both membrane-based immunochromatographic assays that detect SARS-CoV-2 nucleocapsid proteins: (1) the coronavirus antigen rapid test cassette (Healgen Scientific, LLC, Houston, TX, USA), named hereafter HeRDT, and (2) the COVID-19 Ag Respi-Strip (Coris Bioconcept, Gembloux, Belgium), referred to as CoRDT.

#### **2. Materials and Methods**

#### *2.1. Study Design*

The diagnostic accuracy of HeRDT and CoRDT compared to RT-PCR (defined as the gold standard) was determined in a two-phase validation. The first phase—with nasopharyngeal (NP) swabs selected by convenience—aimed to verify that the minimum expected performance was achieved based on the recommendation of minimum sensitivities and specificities described by the WHO—i.e., a sensitivity ≥80% and a specificity ≥97% [8]. The second phase carried out on randomized NP samples aimed to validate the RDT in real diagnostic conditions.

#### *2.2. Population and Study Period*

Validation was conducted between 22 October 2020 and 11 November 2020 at CHU Liège, a tertiary hospital with 1038 beds in Belgium. We included NP-flocked swabs placed in 3 mL of transport media (Vacuette Virus Stabilization Tube (VST); Greiner Bio-One, Kremsmünster, Austria) containing phosphate-buffered saline. The samples were kept between 2 and 8 ◦C and tested within 24 h for both RT-PCR and RDT. The NP samples were collected from patients admitted to the emergency department and from the testing center of the CHU Liège.

Phase 1: Evaluation of samples based on the predefined Ct value. The sample selection took place from 22 October to 27 October 2020. The average SARS-CoV-2 prevalence at the CHU of Liège during the sample selection period was 43.7%. Following the RT-PCR analysis, 15 negative samples and 48 positive samples were selected based on convenience. Positive samples were chosen on the basis of Ct calculated for the N gene by Qiagen RT-PCR in triplicate for each Ct value ranging from 20 to 35.

Phase 1 bis: Evaluation of samples based on the predefined Ct value after methodology improvement. In order to improve the HeRDT sensitivity, 30 RT-PCR-positive samples were tested according to an adapted protocol between 28 October and 30 October 2020.

Phase 2: Evaluation of randomized samples. The validation was performed from 3 November to 11 November 2020. At the CHU Liège, the prevalence of SARS-CoV-2 positivity during this period averaged 29%. Following the Qiagen RT-PCR analysis, 50 negative and 50 positive samples were randomly selected.

#### *2.3. Diagnostic Procedures*

RT-PCR: The reference method used was the QIAprep&amp Viral RNA UM Kit (Qiagen, Hilden, Germany), combining liquid-based sample preparation with the one-step RT-qPCR detection of SARS-CoV-2 targets from human respiratory samples. Three differentcolored channels (FAM™, HEX™, and Cy5) were used for the multiplex detection of SARS-CoV-2 (N1/N2 genes), human genetic material integrity (B2M and Rnase P), and inhibition control (synthetic transcript), respectively. RT-PCR detection was performed on a LC480 thermocycler (Roche, Basel, Switzerland) according to the following cycling conditions: 50 ◦C for 10 min, 95 ◦C for 2 min, 40 cycles at 95 ◦C for 5 s, 58 ◦C for 30 s. The quantification of the viral load was carried out using the calibrated standards provided by the Belgian National Reference Centre for Respiratory Viruses so that the virus concentrations could be estimated from the Qiagen Ct values (Table 1). Samples were considered negative when N1/N2 genes were not detected or they presented a Ct value >40.

**Table 1.** Correlation between Qiagen PCR Ct and standard viral quantification.


Ct, cycle threshold.

RDTs: We evaluated the COVID-19 Ag Respi-Strip kit from Coris Bioconcept (CoRDT) and the coronavirus antigen rapid test cassette from Healgen Scientific, LLC (HeRDT). Both assays are qualitative membrane-based immunochromatographic tests that use monoclonal antibodies to detect nucleocapsid proteins from SARS-CoV-2 in direct nasopharyngeal swabs. For both tests, a nasopharyngeal specimen was taken using a sterile swab and the test was performed directly.

HeRDT manufacturer test procedure: The swab was dipped for 1 min in a dropper bottle containing 10 drops of buffer. After removing the swab from the bottle, 4 drops of the solution were placed in the well of the immunochromatographic test. The result was read 15 min later.

CoRDT manufacturer test procedure: The swab was dipped into a tube containing 8 drops of buffer and stirred thoroughly. After removing the swab, the immunochromatographic strip was placed in the tube. The result was read 30 min later.

Sample processing methods were modified from the manufacturer's instructions in order to perform the RDT from the same transport media as were used for the RT-PCR. We analyzed the transport medium containing the swab rather than the swab directly, as recommended by the manufacturer. As the tests were carried out in our laboratory, samples arrived in a virus stabilization tube containing phosphate-buffered saline solution. For both tests, 100 μL of transport fluid was added to the buffer.

In addition to the adapted sample use, we evaluated a second modification to the operating protocol to improve the sensitivity (by halving the lysis buffer) on thirty RT-PCRpositive samples. Instead of adding 10 drops of buffer, we added 5 drops for HeRDT and 4 for CoRDT instead of 8.

The results were read double blind and the technicians performing the RDT were also blinded to the RT-PCR results.

#### *2.4. Clinical Data*

We reviewed medical records to find clinical data regarding the presence of COVID-19 symptoms and, when present, the time from the onset of symptoms.

#### *2.5. Statistics*

The performance of both CoRDT and HeRDT was evaluated according to the criteria of sensitivity and specificity with a corresponding 95% confidence interval (95% CI). Qiagen RT-PCR was considered as the gold standard for this evaluation. Sensitivity was calculated for two groups: (1) on positive NP swabs including all Ct values and (2) in NP swabs with RT-PCR cycle threshold (Ct) values ≤31. This cut-off was used in the host laboratory to assist with the interpretation of Ct: a positive sample with a Ct value >31, equivalent to a viral load ≤1000 copies/mL, was considered to be a weakly positive result [8,9,15]. The agreement between the two techniques was evaluated using Cohen's kappa coefficient, κ, with a corresponding 95% CI. Analyses were performed using the R software [19].

#### **3. Results**

For the Phase 1 validation, 63 samples were selected from patients presenting at the COVID testing center of CHU Liège. This cohort included 15 negative (23.8%) and 48 positive NP swabs (76.2%). Positive samples presented RT-PCR Ct values between 20.2 and 35.9 Ct (median Ct = 28.3), corresponding to a SARS-CoV-2 viral load of between 1000 and 10,000,000 copies/mL. Of the 48 positive samples, 29 samples were detected as positive with HeRDT; the test was positive for each of the three specimens in triplicate, reaching a Ct value of 28. However, the results were increasingly inconsistent for those above this value. The sample with the lowest viral load detected with HeRDT presented a Ct value of 31.5 (corresponding to a weakly positive result ≤1000 copies/mL). All RT-PCR-negative samples were negative with both HeRDT and CoRDT, resulting in a specificity of 100%. As shown in Table 1, the sensitivity for HeRDT was 60.4% while CoRDT—which detected 18 positive samples out of the 48 RT-PCR-positive NP swabs—presented a sensitivity of 37.5%. Since the samples with positive RT-PCR presenting a Ct value >31 were usually considered to be weakly positive [15], the performance of both HeRDT and CoRDT was calculated for a subgroup of specimens. This subgroup included all negative samples (*n* = 15) and positive samples with a Ct value ≤31 in RT-PCR (*n* = 39). The improved sensitivity (84.9% for HeRDT and 54.5% for CoRDT) observed in this subgroup is also presented in Table 2.

By halving the volume of the lysis buffer (5 drops instead of 10, with 150 μL of sample) for HeRDT, we observed an improved sensitivity (73.3%). Of the 30 Qiagen RT-PCRpositive samples selected for this validation phase (Ct range: 24.1–33.8; median Ct 29.1), 22 samples were positive with HeRDT. The same specimens were also analyzed with CoRDT, which identified only four positive results out of the 30 RT-PCR-positive samples (leading to sensitivity of 13.3%). The results are shown in Table 3.


**Table 2.** Phase 1 performance of HeRDT and CoRDT compared to RT-PCR (P = positive, N = negative). The results for all Ct values ranging from 20.2 to 35.9 are presented in (**A**), and only Ct values ≤31.00 are presented in (**B**).

CI, confidence interval; CoRDT, COVID-19 Ag Respi-Strip from Coris Bioconcept; HeRDT, coronavirus antigen rapid test cassette from Healgen Scientific.

**Table 3.** Sensitivity of HeRDT and CoRDT compared to RT-PCR after the improvement of operating protocol (P = positive, N = negative). The results for all Ct values ranging from 24.1 to 33.8 are presented in (**A**).


Phase 2 validation was performed on 100 randomly selected routine NP swabs containing 50 negative and 50 positive Qiagen RT-PCR samples, with Ct values in the range of 16.7 to 37.3 (median Ct = 23.6); the viral load ranged from 1000 to >10,000,000 copies/mL) (Figure 1). For NP swabs taken from patients admitted to the emergency department and from the testing center of the CHU Liège, 76% presented COVID-19 symptoms with a median duration of 3 days (from 0 to 11 days), and 86.8% of symptomatic patients had symptoms ≤5 days. As shown in Table 4, HeRDT identified 44/50 positive samples (88% sensitivity). The limit of detection was a sample with a Ct value of 33.8. The κ expressing the agreement between Qiagen RT-PCR and HeRDT was 0.880 (95% CI: 0.788–0.972), which indicated a strong agreement between the two techniques [20]. Thirty-one positive results out of the 50 RT-PCR-positive samples were detected with CoRDT (indicating 62% sensitivity), and the lowest viral load detected had a Ct value of 26.5 (corresponding to a positive result between 1000 and 100,000 RNA copies/mL). The agreement κ index between CoRDT and Qiagen RT-PCR was 0.620 (95% CI: 0.477–0.763), indicating a moderate agreement. All RT-PCR-negative samples were also negative with both HeRDT and CoRDT, leading to a 100% specificity. The performance of both assays was calculated for the subgroup, counting only negative (*n* = 50) and positive samples with a Ct value ≤31 in RT-PCR (*n* = 45). HeRDT showed a sensitivity of 91.1%. Cohen's kappa coefficient for this subgroup was 0.915 (CI: 0.833–0.997), indicating almost perfect agreement between HeRDT and RT-PCR. In this subgroup, the sensitivity of CoRDT was also improved (68.9%), but the κ expressing agreement between RT-PCR and CoRDT continued to indicate a moderate agreement (κ = 0.699 (CI: 0.560–0.837)).

**Figure 1.** HeRDT and CoRDT results according to the viral load (Ct).

**Table 4.** Phase 2 performance of HeRDT and CoRDT compared to RT-PCR (P = positive, N = negative). The results for all Ct values ranging from 16.7 to 37.3 are presented in (**A**), and only Ct values ≤31.00 are presented in (**B**).


#### **4. Discussion**

Effective testing strategies combined with the good performance of diagnostic methods are essential for the control and management of COVID-19 patients, as well as for asymptomatic carriers [21]. Multiple manufacturers have proposed RDTs for the detection of SARS-CoV-2, with reported performances that should be confirmed by clinical validation to reach minimum performance requirements (sensitivity ≥80% and specificity ≥97%) [8]. In this study, we assessed the performance of two RDT kits (CoRDT and HeRDT) in comparison with RT-PCR, which is considered as the gold standard method. The present validation revealed a 100% specificity for both kits, while the sensitivity was, respectively, 88% and 62% for HeRDT and CoRDT. These results were applicable when considering all positive samples from the validation Phase 2. Moreover, we decided to determine the diagnostic accuracy of both RDTs in a subgroup of samples with RT-PCR-positive samples not exceeding 31 Ct. In accordance with the literature and the quantification of the viral load by calibrated standards, a SARS-CoV-2-positive individual with a value >31 Ct (corresponding to a viral load <1000 copies/mL) is considered to be weakly positive [22]. In this studied subgroup, the sensitivity was higher, with 68.9% for CoRDT and 91.1% for HeRDT. In October 2020, Sciensano (National Institute of Public Health in Belgium) updated the COVID-19 testing strategies: RDT can be used as a diagnostic method for

COVID-19 in symptomatic patients with symptoms ≤5 days consulting at the emergency department [21]. In fact, antigen detection methods are more effective in patients presenting high SARS-CoV-2 viral loads, most often corresponding to the recent onset of symptoms. As reported in other studies [14,23], a higher sensitivity of RDT was observed when testing patients at an early stage of the disease. Similar results were observed in our study: a predominant subgroup of patients presented COVID-19 symptoms with a symptom onset delay ≤5 days. Clinical data were missing in 24% of samples included in this validation step, while only 8% were taken from asymptomatic patients. Due to the high sensitivity of HeRDT and the updated recommendations from the Belgian authorities for the use of RDTs, we decided to implement HeRDT as a screening test for patients admitted to the emergency department at the CHU Liège. However, all negative antigenic test results had to be confirmed by RT-PCR.

In addition to the viral load and the number of days of post-symptoms, other factors may influence the performance of the RDT. Among these, our modified sample processing method may impact the results [14,23]. We chose to process RDTs directly from the transport medium, which was also used for RT-PCR, in order to compare results from the same sample. The adaptation led to a dilution of the NP swab that may partly explain the differences in the performance compared to those announced by the kit's manufacturers or described in other studies [13,24]. Carrying out the RDT directly using the NP swab transport medium—without diluting it in the kit buffer—may improve the sensitivity, as already demonstrated in other studies [14].

Rapid diagnostic tests (RDTs) can provide multiple advantages compared to PCR methods. They cost less and are easier to use (simpler equipment, requirement of basic skills compared to molecular tests). They present a shorter turnaround-time compared to RT-PCR and provide results at any hour in the day. However, RDTs may present a series of disadvantages, such as the immunochromatographic tests needing to be read by humans, the interpretation of the results possibly being subjective, and the manual encoding of data being a possible source of error in the Laboratory Information Management System (LIMS), especially in the case of decentralized point-of-care testing configurations. It is essential for laboratories to easily retrieve the results of these RDTs in the LIMS, which can be used for submission to authorities for pandemic monitoring or other purposes. However, the main disadvantage of RDTs is associated with their lower sensitivity compared to RT-PCR, often leading to the verification of negative results [9,25].

As previously described [23,24,26], the performance of an RDT depends on the viral load of SARS-CoV-2 in samples, the delay from the onset of symptoms, any adaptation of the testing protocol, and how the results are interpreted by the reader. Moreover, the performance of an RDT is also influenced by the SARS-CoV-2 infection rate; a higher prevalence of infection in the population corresponds to a higher positive predictive value (PPV). Conversely, as the disease prevalence decreases, PPV decreases and the risk of obtaining false positive results increases. During the beginning of the study period (22 October to 27 October 2020), the prevalence of SARS-CoV-2 at the host hospital was 43.7%, leading to a PPV of 100% and a negative predictive value (NPV) of 91.5% for HeRDT. For CoRDT, the calculated predictive values for the same period were 100% for PPV and 77.2% for NPV. For instance, in another context where the prevalence of COVID-19 is 5.0% (prevalence at CHU Liège in January 2021), HeRDT would show a PPV of 100% and an NPV of 99.4%. For CoRDT in the same situation, PPV and NPV would be 100% and 98%, respectively. Contrary to other studies [24], our specificity reached 100%; therefore, it was not possible to predict a PPV. It would be interesting to re-evaluate the specificity during a period with a lower prevalence of COVID-19. It should also be noted that the Phase 2 validation performed in our laboratory was carried out on only 100 samples (50 RT-PCRpositive and 50 RT-PCR-negative NP swabs). However, the European Centre for Disease Prevention and Control (ECDC) recommends testing at least 100 positive and 100 negative samples [25]. At the beginning of this validation study, these recommendations were not yet available. However, a routine daily verification was performed for several weeks and

confirmed the sensitivity observed during the second phase of the validation of HeRDT. From 11 November to 31 December, 984 samples (971 negative and 13 positive) were analyzed in parallel using a HeRDT test and QIAGEN RT-PCR assay, revealing a 97% concordance between the two methods.

Increasing the number of sample tests and the examination of different prevalence conditions remain relevant for validating RDTs and precisely confirming the manufacturer's diagnostic accuracy in real-life conditions.

In conclusion, RDTs have a real role in the COVID-19 testing strategy thanks to their ease of use and capacity for mass screening and the rapid detection of SARS-CoV-2 24/7. This validation study showed that it is necessary to confirm the test performances announced by manufacturers before implementing RDTs, especially in the case of protocol adaptation. Although it is crucial to realize that RDT sensitivity is lower than RT-PCR, their use can be of benefit in cases of limited access to molecular methods or when the RT-PCR testing capacity is overburdened.

**Author Contributions:** Conceptualization: Y.S., C.F., M.-P.H., and J.D.; Data curation: Y.S., A.N.D., and A.-F.D.; Formal analysis: A.N.D. and A.-F.D.; Investigation: Y.S. and J.H.; Resources: C.M.; Supervision: M.-P.H.; Writing—original draft: Y.S. and J.D.; Writing—review and editing: Y.S., F.M., A.N.D., A.-F.D., M.-P.H., and J.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical concerns.

**Acknowledgments:** The authors wish to thank Sebastien Bontems for his review of the manuscript and all microbiology laboratory staff for their assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Impact of Kidney Failure on the Severity of COVID-19**

**Dorota Zar ˛ebska-Michaluk 1,\*, Jerzy Jaroszewicz 2, Magdalena Rogalska 3, Beata Lorenc 4, Marta Rorat 5,6, Anna Szymanek-Pasternak 5, Anna Piekarska 7, Aleksandra Berkan-Kawi ´nska 7, Katarzyna Sikorska 8, Magdalena Tudrujek-Zdunek 9, Barbara Oczko-Grzesik 2, Beata Bolewska 10, Piotr Czupryna 11, Dorota Kozielewicz 12, Justyna Kowalska 13, Regina Podlasin 14, Krzysztof Kłos 15, Włodzimierz Mazur 16, Piotr Leszczy ´nski 17,18, Bartosz Szetela 19, Katarzyna Reczko <sup>1</sup> and Robert Flisiak <sup>3</sup>**


**Abstract:** Background: Patients with kidney failure are at an increased risk of progression to a severe form of coronavirus disease 2019 (COVID-19) with high mortality. The current analysis was aimed to assess the impact of renal failure on the severity of COVID-19 and identify the risk factors of the fatal outcome in this population. Methods: The analysis included patients from the SARSTer database, a national real-world study evaluating treatment for COVID-19 in 30 Polish centers. Data were completed retrospectively and submitted online. Results: A total of 2322 patients were included in the analysis. Kidney failure was diagnosed in 455 individuals (19.65%), of whom 373 presented moderate stage and 82 patients, including 14 dialysis individuals, presented severe renal failure. Patients with kidney failure were significantly older and demonstrated a more severe course of COVID-19. The age, baseline SpO2, the ordinal scale of 4 and 5, neutrophil and platelet count,

**Citation:** Zar˛ebska-Michaluk, D.; Jaroszewicz, J.; Rogalska, M.; Lorenc, B.; Rorat, M.; Szymanek-Pasternak, A.; Piekarska, A.; Berkan-Kawi ´nska, A.; Sikorska, K.; Tudrujek-Zdunek, M.; et al. Impact of Kidney Failure on the Severity of COVID-19. *J. Clin. Med.* **2021**, *10*, 2042. https://doi.org/10.3390/ jcm10092042

Academic Editor: Lionel Rostaing

Received: 29 March 2021 Accepted: 6 May 2021 Published: 10 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

estimated glomerular filtration rate, and C-reactive protein concentration as well as malignancy and arterial hypertension were the independent predictors of 28-day mortality in logistic regression analysis. Conclusions: Underlying kidney disease in patients with COVID-19 is among the leading factors associated with a higher risk of severe clinical presentation and increased mortality rate.

**Keywords:** kidney failure; SARS-CoV-2; COVID-19; mortality

#### **1. Introduction**

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has rapidly spread worldwide since it was first identified in December 2019 in Wuhan. Despite an unprecedented global public health effort, the outbreak became pandemic on 11 March 2020. After one year, more than 120 million affected people with nearly 3 million deaths globally due to coronavirus disease 2019 (COVID-19) were documented [1]. The clinical spectrum of SARS-CoV-2 infection ranges from asymptomatic through mild and moderate respiratory illness to critical life-threatening viral pneumonia with respiratory failure, septic shock, and multiple organ dysfunction. The higher risk of the severe clinical presentation of COVID-19 is associated with older age, immunosuppressive therapy, and underlying comorbidities including cardiovascular and chronic pulmonary illnesses, diabetes, cancers, and chronic kidney diseases (CKD) [2–4].

The progressive loss of renal function in CKD results in alterations of the innate and adaptive immune system, including decreased leukocyte phagocytic activity, dwindling dendritic cells responsible for presenting antigens, depletion and dysfunction of B lymphocytes, and impaired cell-mediated immunity through an accelerated T cell turnover and increased apoptosis of cluster of differentiation (CD) 4+ and CD8+ lymphocytes [5]. The impaired immune response is associated with higher incidence and more severe course of infections which appear to be responsible for a large part of the mortality, especially in patients with end-stage renal disease (ESRD). Alongside secondary immunodeficiency, the immune activation in patients with chronic kidney disease is observed [6]. The increased production and decreased clearance of pro-inflammatory cytokines lead to systemic inflammation and oxidative stress, which contribute to atherosclerotic cardiovascular disease and other conditions worsening the prognosis of patients with SARS-CoV-2 infection.

The current analysis was aimed to assess the impact of kidney failure on the severity of COVID-19 and to identify the risk factors of the fatal outcome of the disease in this population in the real-world setting.

#### **2. Materials and Methods**

The study population consisted of patients included in the national database SARSTer, which is an ongoing project supported by the Polish Association of Epidemiologists and Infectiologists and covers 2784 adult individuals treated for COVID-19 between 1 March and 31 December 2020 in 30 Polish centers. All the patients were diagnosed with COVID-19 based on positive results of the real-time reverse transcriptase-polymerase chain reaction (RT-PCR) from the nasopharyngeal swab specimen [7].

The therapeutic management decisions were taken at the discretion of the treating physician following the current medical knowledge and in line with the national recommendations [8–10]. The SARSTer study had the approval of the Ethical Committee of the Medical University of Białystok with a granted waiver of informed consent from study participants due to its retrospective design, and the local bioethics committees in case of the off-label use of medication in patients with COVID-19.

Patients' data were retrieved retrospectively from hospital files and completed online by a platform operated by "Tiba" sp. z o.o. The parameters gathered on admission included age, gender, body mass index (BMI), comorbidities and concomitant medications, clinical symptoms of SARS-CoV-2 infection, lung computed tomography scan, and selected lab

values. The baseline laboratory data consisted of complete blood count, inflammatory indicators (C-reactive protein (CRP), procalcitonin (PCT), ferritin, and interleukin 6 (IL-6) concentration if tocilizumab (TCZ) prescription was considered), coagulation parameters such as D-dimer, international normalized ratio (INR), and fibrinogen, the activity of liver enzymes (aspartate and alanine aminotransferases, gamma-glutamyl transpeptidase, lactate dehydrogenase), and renal function tests. Estimated glomerular filtration rate (eGFR) was calculated with the MDRD Study equation and, using this measure, CKD was defined as eGFR < 60 mL/min/1.73 m<sup>2</sup> along with a history of kidney disease from medical records [11]. According to renal function on admission, patients were stratified into three groups: eGFR < 30 mL/min/1.73 m2, eGFR 30–60 mL/min/1.73 m2, and eGFR > 60 mL/min/1.73 m2.

The COVID-19 severity on hospital admission was determined based on blood oxygen saturation (SpO2) and clinical status was defined as symptomatic stable with SpO2 > 95%, symptomatic unstable with two levels of baseline saturation SpO2 91–95% or SpO2 ≤ 90%, and critical with acute respiratory distress syndrome (ARDS). The information on the medications applied for the treatment of COVID-19, including remdesivir (RDV), tocilizumab (TCZ), dexamethasone, convalescent plasma, low weight molecular heparin, and antibiotics, as well as drug-related adverse events, were collected during the hospitalization.

The patients were scored at baseline and then every 7 days during the following 28 days after admission on an ordinal scale, which includes eight categories: 1. unhospitalized, no activity restrictions; 2. unhospitalized, no activity restrictions and/or requiring oxygen supplementation at home; 3. hospitalized, does not require oxygen supplementation and does not require medical care; 4. hospitalized, requiring no oxygen supplementation, but requiring medical care; 5. hospitalized, requiring normal oxygen supplementation, low-flow by mask or nasal prongs; 6. hospitalized, on non-invasive ventilation with high-flow oxygen equipment; 7. hospitalized, for invasive mechanical ventilation or ECMO; 8. death.

The study outcomes included death, need for mechanical ventilation, and clinical improvement defined as at least a 2-point decrease in an ordinal scale classification from baseline to 14, 21, and 28 days of hospitalization.

To evaluate the impact of chronic kidney disease on the outcome of COVID-19, the analysis was performed concerning the eGFR at baseline.

#### *Statistical Analysis*

The results are expressed as mean ± standard deviation (SD) or *n* (%) and odds ratios with 95% confidence intervals. *p* values of <0.05 were considered to be statistically significant. The significance of difference was calculated by Fisher's exact test for nominal variables and by Mann–Whitney U and Kruskal–Wallis ANOVA for continuous and ordinal variables. Due to the highly variable group size, the Fisher's *p*-values were accompanied by OR as the sample size independent effect size measures. The association between variables was measured by Spearman's rank correlation coefficient and its significance test *p*-values. Survival analyses between patients with different eGFR ranges (Kaplan–Meier curves) were performed by Log-rank (Mantel–Cox) Test. Forward stepwise logistic regression models with Bayesian Information Criterion (BIC) as a model selection criterion were performed with death within 28-days after the start of hospitalization as the dependent variable. Among independent variables tested for the best model were age, sex, BMI, arterial hypertension, diabetes, coronary artery disease, chronic obstructive pulmonary disease, malignancy, GFR range, baseline levels of SpO2, CRP, procalcitonin, WBC, lymphocyte and neutrophil counts, platelets, D-dimer, ALT as well as therapy with dexamethasone, remdesivir, tocilizumab, and heparins. Logistic regression models were calculated by use of Statistica 13.0 (TIBCO Software Inc., Palo Alto, CA, USA).

#### **3. Results**

Among 2784 adult patients included in the SARSTer project, the data on kidney function were provided for 2322 individuals with a mean age of 60.4 ± 17.1 years and male predominance (53%). Among them, 455 individuals presented kidney impairment, a moderate stage of renal insufficiency was diagnosed in 373 patients with eGFR 30–60 mL/min/1.73 m2, of which six underwent kidney transplantation, and 82 patients with eGFR < 30 mL/min/1.73 m2 were diagnosed with severe renal failure (68 patients with non-dialysis dependent CKD) and ESRD (14 dialysis patients). Among patients with renal failure, 328 with moderate and 74 with severe stage had the diagnosis of chronic kidney disease based on the medical file, whereas in the remaining 53 individuals we were not able to confirm CKD due to incomplete records or disturbed communication with patients. Despite the lack of a previous diagnosis of CKD and no follow-up during three months after discharge from the hospital, we included these patients based on the depth-analysis of the available data concerning comorbidities and taking into account the age of patients as a risk factor of CKD and no improvement in the renal function after hydration. The detailed baseline characteristics of the patients according to kidney function on admission to the hospital are summarized in Table 1.

**Table 1.** Baseline characteristics of patients according to kidney function.


eGRF, estimated glomerular filtration rate; SD, standard deviation; ECM, extracorporeal membrane oxygenation.

Patients with renal failure were significantly older and demonstrated a more severe course of COVID-19 on admission, defined by the higher rate of patients with an oxygen saturation ≤ 90% and a greater percentage of the more advanced categories on the ordinal scale. Patients with CKD more frequently suffered from diabetes and cardiovascular diseases including arterial hypertension, coronary artery disease, heart failure, and atrial fibrillation, and were more likely to be treated with insulin, oral antidiabetics, and antihypertensive drugs, compared to non-CKD individuals. Among medications reducing blood pressure, beta-blockers and diuretics were used predominantly in patients with renal failure (Supplementary Table S1).

Significantly higher values of inflammatory parameters including the concentration of CRP, PCT, and IL-6, as well as white blood cell and neutrophil counts, and lower platelet counts were documented in patients with CKD on admission (Table 2).


**Table 2.** Baseline laboratory indicators according to the baseline kidney function.

CRP, C-reactive protein; ALT, alanine aminotransferase.

Patients with CKD were more often treated for COVID-19 with IL-6 inhibitor tocilizumab (TCZ), dexamethasone, and low molecular weight heparin compared with patients without kidney abnormalities. The application of remdesivir (RDV) was significantly lower in CKD patients, and five individuals with severe renal failure received off-label RDV, four of them were concurrently treated with TCZ, and two with dexamethasone—all were scored on admission in category 5 on the ordinal scale. Antibiotics were administered more frequently in those with severe renal failure (Table 3).

**Table 3.** In-hospital treatment for COVID-19 according to the baseline kidney function.


\* 1208 patients received prophylactic dose and 98 therapeutic dose. \*\* 251 patients were on prophylactic dose only, 17 received prophylactic dose on admission and then therapeutic dose during hospitalization, and 31 patients were on therapeutic dose from admission. \*\*\* 55 patients were on prophylactic dose, 14 patients received prophylactic then therapeutic and remaining 8 were on therapeutic dose.

Continuous renal replacement therapy was continued only in 14 hemodialyzed patients, this therapy was not initiated by anyone else, and no patient was treated with hemoperfusion to remove cytokines.

As shown in Table 4, 28-day in-hospital mortality and the need for mechanical ventilation significantly increased in direct proportion to the degree of renal impairment. Moreover, clinical improvement was significantly slower in patients with advanced renal impairment.


**Table 4.** Outcome according to baseline kidney function.

The analysis of the outcome according to baseline kidney function that also takes into account the selected parameters at the admission is presented in the supplementary tables (Supplementary Tables S3–S5).

The analysis performed depending on the survival revealed that patients who died were significantly older, with a higher proportion of males, a greater percentage of baseline oxygen saturation ≤ 90%, more severe clinical presentation on admission in terms of oxygen demand, more frequent comorbidities and treatment with concomitant medications, with significantly higher values of inflammatory parameters and D-dimer level, and lower platelet count. Those patients were more likely to be treated with TCZ, dexamethasone, convalescent plasma, low molecular weight heparin, and antibiotics. Furthermore, higher rates of individuals with moderate and severe kidney failure were reported in this group of patients (Table 5).

Among patients with eGFR < 30 mL/min/1.73 m2, no statistically significant differences were noticed between dialyzed and not dialyzed patients regarding baseline demographic and clinical measures as well as effectiveness outcomes (Supplementary Table S2). However, those who died were significantly older, with a higher proportion of baseline oxygen saturation ≤ 90%, were more likely to have coexisting conditions, with higher AST activity on admission, and were more frequently treated with antibiotics during hospitalizations (Table 6).


**Table 5.** Comparison of patients who died or survived regardless of kidney function.


**Table 6.** Comparison of patients with eGFR < 30 mL/min who died or survived.

One of the five patients with eGFR < 30 mL/min/1.73 m<sup>2</sup> who received RDV died due to sepsis. In the remaining four, no safety issues were observed and no deterioration in renal function was documented, and in two of them clinical improvement—in one after 21 and in another after 28 days of hospitalization—was reported.

The independent predictors of 28-day mortality in logistic regression analyses were age, baseline SpO2, the ordinal scale of 4 and 5, neutrophil and platelet count, eGFR, and CRP concentration (Figure 1, Table 7).

**Figure 1.** Kaplan–Meier survival curves of the CKD groups dependent on the eGFR.



*p*-value < 0.001.

Among comorbidities, most notably malignancy as well as arterial hypertension (HA) and ischemic heart diseases were associated with mortality, while diabetes mellitus (DM) and chronic obstructive pulmonary disease (COPD) were not. Interestingly, in this cohort of 2322 COVID-19 patients, we were not able to show an independent effect of therapies, BMI, and baseline D-dimers, ALT, or procalcitonin on overall 28-day mortality.

#### **4. Discussion**

Chronic kidney disease is an increasing public health issue affecting 8–16% of the population worldwide [12]. Patients with CKD are highly susceptible to COVID-19 and are at an increased risk of progression to a severe or critical form of the disease because of impaired immunity; additionally, they are at enhanced risk of SARS-CoV-2 infection due to frequent hospital attendance [2,13–18]. The prevalence of CKD in patients with COVID-19

has been reported in a wide range of approximately 1–47%; however, it is suggested that the lowest values result from underestimation [16,19–21]. Among patients with COVID-19 hospitalized in 30 Polish centers included in the current analysis, nearly 20% were diagnosed with CKD, of which 18% presented severe renal failure. In addition to the worse patient's status regarding the severity of COVID-19 when admitted to the hospital, we found the pre-existing renal disease to be independently associated with higher in-hospital mortality, especially in those with severe kidney failure, and our findings are in line with the results of previous studies [13,14,22–26]. The Global Burden of Disease (GBD) collaboration recently estimated the risk factors for severe COVID-19 worldwide using results from international databases and large multimorbidity studies from different countries, and determined that CKD is a condition conveying the highest risk for the severe presentation of the disease and COVID-19-related death [27]. In the current study, no difference in mortality was demonstrated between dialysis and non-dialysis dependent patients with severe renal failure, but it must be emphasized that the group of dialysis patients in the analyzed cohort was relatively small—only 14 patients. However, our findings of the comparable death risk regardless of the dialysis are consistent with observations from the study conducted by Flythe et al.

The reported death rates of 50% in 143 dialysis and 521 non-dialysis dependent individuals with CKD and 35% in 3600 non-CKD patients are higher than those noted in our analysis, but it is noteworthy that the abovementioned study included only critically ill patients with COVID-19 treated in intensive care units (ICU) [23]. The investigation performed by Yang et al. in 836 patients revealed in-hospital mortality rates of 9%, 50%, and 66.7% in non-CKD, non-dialysis dependent CKD, and dialysis patients, respectively. Of note, the proportions of individuals with the moderate presentation of COVID-19 were similar among those without and with non-dialysis dependent CKD (73.7% and 75%, respectively), and much lower in dialysis patients (40%), who were more frequently scored as severe cases on admission, which has had an impact on the fatal outcome [25].

The negative impact on the outcome was demonstrated for the baseline oxygen saturation corresponding to the severity of COVID-19 on admission—35.5% of patients with moderate and almost 56% with severe renal failure classified at baseline as SpO2 ≤ 90% died compared to a 16.5% mortality rate among non-CKD individuals. The category 5 in an ordinal scale on admission associated with the need for oxygen supplementation was an independent factor related to higher mortality, and our findings are consistent with the results of the other studies [28,29].

We confirmed older age as an independent strong predictor for in-hospital mortality, which is in line with previous reports and calculations performed by GBD collaboration in patients with CKD [2,3,15,22,23,27,30]. On the contrary, Cai et al., in a meta-analysis of 12 studies including CKD patients, documented a higher mortality rate in those below 70 years compared to older patients, explaining this finding by the more frequent rate of the other comorbidities with stronger than CKD association with increased risk of death among the elderly [14].

We did not demonstrate gender as a factor influencing the clinical status on admission and in-hospital mortality in CKD patients, and among those with severe renal failure, the death rate was nearly equal between females and males. Thus, the results of our study differ from other reports documenting a higher risk of severe course and COVID-19-related death rate in males [2,4,22,27,30,31].

In the current analysis, patients with CKD presented significantly increased baseline leukocyte and neutrophil counts, a higher level of inflammatory markers, including CRP, PCT, and IL-6 concentration, as well as D-dimer level compared to patients without CKD. Those parameters were increased in direct proportion to the degree of renal impairment and probably corresponded to a proinflammatory state. The high CRP concentration, the increased neutrophils, and the decreased platelet counts on admission were independently associated with a significantly higher in-hospital death rate in patients with CKD, which supported results from other studies [22,23,25,30,32,33]. Among comorbidities, arterial

hypertension, ischemic heart disease, and malignancy were found to be independent negative predictors of 28-day survival and these findings, whereas diabetes and COPD were not. The impact of coexisting diseases was investigated in many studies and results are divergent depending on the analyzed population, sample size, and the nature of the study. Meta-analysis performed among Iranian patients by Merjalili et al. revealed that diabetes was associated with higher mortality, while arterial hypertension was not [26]. Park et al. analyzed the Korean population and found both diabetes and hypertension to be risk factors for death in the course of COVID-19 [34]. The same results were achieved by Gupta et al. among Indian patients [35]. Chen et al. demonstrated only ischemic heart disease and cerebrovascular disease to be independently associated with high mortality in Chinese patients [36]. According to the results of the meta-analysis conducted by Chaoqun et al., the presence of cerebrovascular disease, DM, COPD, malignancy, and hypertension was related to higher mortality in the course of COVID-19 [15]. Factors associated with an increased risk of death documented by Wiliamson et al. included ischemic heart disease, DM, and malignancy, especially hematological [2]. It should be pointed out that the percentage of diabetes among patients with severe renal failure was significantly higher compared to those with moderate kidney impairment and non-CKD individuals, which allows us to suppose that diabetic nephropathy is responsible for some cases of ESKD.

Of COVID-19-related medications, dexamethasone and tocilizumab were used more frequently in patients with CKD, probably as a result of more severe clinical presentation of the disease, so we were not able to show an independent effect of those therapies. According to the summary of product characteristics, the application of RDV was significantly lower in CKD patients; however, five individuals with severe renal failure scored at baseline in category 5 on an ordinal scale and received off-label RDV, and one of them died; however, it should be pointed out that in the remaining four patients no deterioration in renal function was documented [37]. Similar observations in a small series of 20 patients with ESKD treated with RDV were published by Pettit et al., where therapy appeared to be relatively safe and the potential benefit outweighed the theoretical risk of renal toxicity [38].

We are aware of several limitations of our study associated with its retrospective design—the analyses were based on the clinically available electronic captured data with possible entry errors, a lack of information about the causes of CKD, and in some cases a lack of confirmation of CKD due to incomplete records, and also missing laboratory tests in some patients, which did not allow us to assess the impact on the mortality of the selected parameters. As treatment with RDV and TCZ is not indicated in patients with eGFR <30 mL/min/1.73 m<sup>2</sup> and, in turn, the absence of such treatment may influence the investigated outcomes, we may overestimate the causal effect of CKD on study endpoints. Lastly, due to the lack of control laboratory tests during hospitalization in some patients, we performed analysis taking into consideration the baseline renal status of patients with CKD, not looking at the development of acute kidney injury (AKI) in the course of COVID-19 in those individuals, although the experience from previous reports showed that AKI is associated with higher mortality [24,39–41].

However, the strengths of our study include a large number of patients from the heterogeneous population, covering different parts of the country, which increases the generalizability of the findings—all of them had a laboratory-confirmed SARS-CoV-2 infection and the patients were followed up for 28 days.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/jcm10092042/s1, Table S1: Medications for the treatment of comorbidities, Table S2: Patients with eGFR < 30 mL/min according to dialysis, Table S3: Patients with baseline CRP ≥ 100 mg/L outcome according to kidney function, Table S4: Patients with baseline SpO2 ≤ 90%—outcome according to kidney function, Table S5: Patients with baseline D-dimers ≥ 1000 ng/mL—outcome according to kidney function.

**Author Contributions:** Conceptualization—D.Z.-M., R.F., methodology—D.Z.-M., formal analysis— D.Z.-M., J.J., investigation—D.Z.-M., validation—R.F., writing—original draft preparation—D.Z.-M., writing—review and editing—R.F., supervision—R.F., project administration—R.F., funding acquisition— R.F., data collection—D.Z.-M., J.J., M.R. (Magdalena Rogalska), B.L., M.R. (Marta Rorat), A.S.-P., A.P., A.B.-K., K.S., M.T.-Z., B.O.-G., B.B., P.C., D.K., J.K., R.P., K.K., W.M., P.L., B.S., K.R., R.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Medical Research Agency, grant number 2020/ABM/ COVID19/PTEILCHZ, and the Polish Association of Epidemiologists and Infectiologists.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki. The SARSTer study had the approval of the Ethical Committee of the Medical University of Białystok (APK.002.303.2020).

**Informed Consent Statement:** Patient consent was waived due to the retrospective design of the study.

**Data Availability Statement:** Data supporting reported results can be provided upon request from the corresponding author.

**Conflicts of Interest:** D.Z.-M. reports personal fees from Gilead and Abbvie, outside the submitted work. R.F. reports grants from Abbvie, Gilead, Merck, personal fees from Gilead, Abbvie, Merck, Roche, and non-financial support from Abbvie, Gilead, and Merck outside the submitted work. J.J. reports personal fees from Gilead, Abbvie, Bausch Health, Merck, Promed, Roche, and nonfinancial support from Abbvie, Gilead, and Merck outside the submitted work. D.K. reports personal fees from Gilead and Abbvie outside the submitted work. J.K. reports personal fees from Gilead, Merck, ViiV, Janssen outside the submitted work. R.P. reports personal fees from Gilead outside the submitted work. W.M. reports grants and personal fees from Gilead, Abbvie, Abbott, Roche, Janssen outside the submitted work. P.L. reports grants and personal fees from Abbvie, Roche, UCB, Lilly, Novartis, BMS, Amgen, Janssen, Abivax, Viela-Bio outside the submitted work. BS reports grants from Abbvie, Gilead, Janssen, personal fees from Gilead, Abbvie, Janssen outside the submitted work. MRog, B.L., MRor, A.S.-P, A.P., A.B.-K., B.B., M.T.-Z., D.K., K.K., K.R., P.C., B.O.-G., K.S. declare no competing interests.

#### **References**

