1 Introduction

Omicron variants of the SARS-CoV-2 virus have spread worldwide despite vaccination, and are highly transmissible and associated with escape immune responses [1]. During the pre-Omicron era, the high mortality of critically ill patients was attributed primarily to age and comorbidities [2]. With the emergence of Omicron variants, vaccination status contributed to the mortality of patients admitted to critical care units (CCUs). The mortality was observed in patients who received three doses of COVID-19 vaccination compared with unvaccinated patients and those who received one or two doses of vaccination after adjustment for age, comorbidities and disease severity [3, 4]. Although the overall hospitalization and mortality have been lower for the Omicron than the Delta variant, immunocompromised patients infected with Omicron often require CCU admission [5]. The COVID-19 pandemic and the emergence of SARS-CoV-2 variants placed a tremendous burden on CCU bed utilization [6, 7]. The duration of SARS-CoV-2 viral shedding is variable, and the development of de-isolation protocols in CCUs has been challenging, given the heterogeneity among patients [8]. A recent study has indicated that 56% of mechanically ventilated patients have infectious viral shedding for more than 20 days [9]. Rapid antigen detection tests (RADTs) have been used to determine the infectiousness of patients with COVID-19 in outpatient settings because they are less expensive and more practical than RT-PCR. RADTs have 94.7% sensitivity with respect to positive viral cultures [10]. In another large cohort study comparing RADT, viral cultures and RT-PCR in patients with mild COVID-19, all patients with negative viral cultures had negative RADTs [11]. The role of RADTs in de-isolation decision-making for CCU patients with COVID-19 was not previously studied. Therefore, we describe the serial results of RADT, RT-PCR and viral culturing in three critically ill patients infected with the Omicron variant.

2 Methods

SARS-CoV-2 RT-PCR (Abbott Real-time SARS-CoV-2 EUA, Abbott Park, IL, USA), RADTs (Panbio™ COVID-19 Antigen rapid test, Abbott, USA), viral culture and whole viral genome sequencing were performed as previously described [12, 13]. Serial RADTs were performed every 1–3 days in critically ill patients to determine the time at which COVID-19 isolation could be ended, according to our Department of Infection Control and Hospital Epidemiology protocol for COVID-19 de-isolation. COVID-19 viral cultures and RT-PCR were performed after two negative consecutive RADT findings.

2.1 RNA Extraction and SARS-CoV-2 Detection

Anterior nasal swabs were collected from all included patients. Abbott Real-time SARS-CoV-2 EUA tests were used for the diagnosis of COVID-19 (Abbott Park, IL, USA). The test is authorized by the US FDA. RNA extraction was performed on the Abbott m2000sp platform, and DNA amplification was performed on the Abbott m2000rt platform. The assay detects dual targets of RdRp and N-genes, with a detection limit of one copy per milliliter.

2.2 RADTs

Panbio™ COVID-19 Antigen rapid tests (Abbott, USA) were performed according to the manufacturer’s instructions, within 15 min from the time of nasal swab collection. Briefly, the presence of the test line and the control line within the result window, regardless of which line appeared first, indicated a positive result. The presence of only the control line and no test line within the result window indicated a negative result. If the control line was not visible within the result window after the test was performed, the result was considered invalid.

2.3 Cell Line and SARS-CoV-2 Cultures

All experiments involving live SARS-CoV-2 virus were performed at the biosafety level 3 facility of the Special Infectious Agents Unit–BSL-3 at King Fahd Medical Research Center, King Abdulaziz University, Jeddah, Saudi Arabia. Vero E6 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, 0.25 µg/ml fungizone and 10 mmol/L HEPES (pH 7.2). A human SARS-CoV-2 clinical patient isolate (SARS-CoV-2/human/SAU/85791C/2020, GenBank accession number: MT630432) was used in all experiments as the positive control.

2.4 Detection of Replicating SARS-CoV-2

Collected samples were diluted 1:10 in DMEM with 10% FBS, inoculated on Vero E6 cells in six-well plates in duplicate, and incubated for 1 h at 37 °C. The inoculum was then removed and replaced with 2 ml DMEM with 2% FBS, and cells were incubated at 37 °C for 3 days or until cytopathic effects were observed in 85–90% of cells in the positive control samples.

2.5 Whole Genome Sequencing

The samples were subjected to nucleic acid extraction with a MagMAX™ Viral/Pathogen Nucleic Acid Isolation Kit (Cat No. A42352, Thermo Fisher Scientific; MA, USA). All samples positive for SARS-CoV-2 were converted to cDNA with SuperScript™ IV VILO™ Master Mix (Thermo Fisher Scientific, USA). The cDNA was amplified with the Ion AmpliSeq™ SARS-CoV-2 Insight Research Assay, according to the manufacturer’s instructions. RNA (5 μl) was used for cDNA synthesis. Amplified products were ligated with unique barcode adaptors with the Ion Xpress Barcode Adaptors 1–16 kit (Thermo Fisher Scientific, USA) and purified with a 1.5 × volume of Agencourt AMPure XP Reagent (Beckman Coulter, USA). The libraries were constructed and normalized to 33 pM with nuclease-free water, and as many as 16 libraries were equally pooled for further processing. Pooled libraries were used as the template input for emulsion PCR, and enrichment of template-positive particles was performed with an Ion Chef automated system and Ion 510 Kit-Chef kit (Thermo Fisher Scientific, USA), according to the manufacturer’s instructions. The obtained data were processed (base calling, base quality recalibration, alignment, assembly and variant calling) primarily with Torrent Suite Server, version 5.12 (Thermo Fisher Scientific, USA). De novo assembly of the contigs was performed with the assembly Trinity plugin (v1.2.1), and consensus sequences of each sample were generated with the IRMA plugin (v1.2.1). Variant call files were analyzed with the COVID19AnnotateSnpEff plugin to identify and annotate variants with public and private databases.

2.6 Statistical Analysis

Serial results of both RADTs and viral cultures in the three patients are presented. Categorical data are presented as numbers and percentages. The agreement between the results of both tests was examined with kappa statistics. p-Values were two-tailed, and values < 0.05 were considered significant. Statistical Package for Social Sciences (IBM, SPSS, version 25) was used for statistical analysis.

3 Results

This study included three patients with confirmed Omicron infection and severe COVID-19 pneumonia requiring intubation, ventilation and CCU admission.

Whole genome sequencing confirmed the presence of Omicron B.1.1.529 in the three isolates of cases 1, 2 and 3.

3.1 Case 1

A 33-year-old man with advanced HIV infection and Kaposi sarcoma was admitted to the CCU 7 days post symptom onset (PSO) with severe COVID-19 pneumonia. He had received two doses of COVID-19 vaccination. His COVID-19 therapy included remdesivir, tocilizumab and dexamethasone. His hospital course was complicated by extensive drug resistant Pseudomonas bacteremia, progressive Kaposi sarcoma and refractory septic shock. Serial RADTs were positive on day (D) 10, D13, D16, D23 and D26 PSO. Serial RADTs were negative on D28, D29 and D32 PSO. Viral cultures were negative on D32 PSO. Concordant negative results between RADTs and viral cultures were found on D32 PSO. He was hospitalized in the CCU for 31 days and died on D38 PSO.

3.2 Case 2

A 83-year-old man with diabetes mellitus, hypertension, chronic kidney disease, coronary artery disease and heart failure was admitted to the CCU on D4 PSO with severe COVID-19 pneumonia. He had not received any COVID-19 vaccination. His COVID-19 therapy included dexamethasone and tocilizumab. His RADT was positive on D6 PSO. His serial RADTs were negative on D10, D11, D13, D14 and D15 PSO. His serial viral cultures were negative on D13 and D15 PSO. Concordant negative results between RADTs and viral cultures were found on D13 and D15 PSO. He was hospitalized in the CCU for 16 days and died on D20 PSO.

3.3 Case 3

A 58-year-old woman with chronic autoimmune liver disease and Behçet disease, who was taking immunosuppressive medication and had received two doses of COVID-19 vaccination, had been isolated in her home for 21 days after her COVID-19 diagnosis. Subsequently, she required non-CCU admission for COVID-19 pneumonia. Her condition deteriorated, and she was admitted to the CCU at D36 PSO with severe COVID-19 pneumonia. Her COVID-19 therapy was dexamethasone. Her hospital course was complicated by Salmonella bacteremia, Escherichia coli pyelonephritis, Stenotrophomonas maltophilia pneumonia and probable positive COVID-19-associated pulmonary aspergillosis, on the basis of serum galactomannan 4 positivity. Serial RADTs were positive on D35 and D39 PSO. Serial RADTs were negative on D46 and D48 PSO. Serial viral cultures were positive D35 PSO, and negative on D46 and D48 PSO. Concordant negative results were observed between RADTs and viral cultures on D46 and D48 PSO, and concordant positive results were observed between RADTs and viral cultures on D35 PSO. She was hospitalized in the CCU for 18 days and died on D54 PSO.

The RT-PCR findings remained positive in all cases at time points in which viral cultures and RADTs were negative (Table 1).

Table 1 Results of serial RADT, viral cultures, RT-PCR and RT-PCR Ct, and concordance between RADT and viral culture findings
Full size table

Significant agreement was observed between the RADT and viral culture findings among the three cases (kappa statistic = 1.0 and p = 0.014; Table 2).

Table 2 Agreement between RADT and viral culture findings for de-isolation of three CCU cases infected with Omicron variants (kappa statistic = 1.0 and p = 0.014)
Full size table

4 Discussion

In our case series, we found a significant concordance between RADT and viral culture findings during viral clearance. The correlation between viral culture and RADT findings for de-isolation decision-making has been described in a recent study in healthy patients with mild Omicron infection. The specificity of RADTs with respect to viral cultures was 90% on day 6, thus indicating that RADTs can be useful in de-isolation decision-making for patients with mild SARS-CoV-2 infection [14]. In addition, RADTs have been used in decision-making to stop the isolation of athletes in colleges and health care workers in hospitals, who had mild COVID-19 in the outpatient setting during the Omicron surge, to avoid unnecessary sick days and the risk of ongoing infectiousness [15, 16].

In our case series, the duration of infectious viral shedding in CCU ranged from D13 to D46 PSO. The main reasons for these differences were associated with the heterogeneity among critically ill patients infected with Omicron variants; a rapid fatal course was observed in the unvaccinated 83-year-old man with multiple comorbidities who had viral shedding for 13 days, whereas a more prolonged course was seen in the 53-year-old immunocompromised woman who received two COVID-19 vaccine doses, had multiple hospital acquired infections and had viral shedding for 46 days. The variation in viral clearance in critically ill patients has also been described by another study demonstrating positive SARS-CoV-2 viral cultures from 13 to 29 days. In that study, the cycle threshold (Ct) value in COVID-19 RT-PCR was a good predictor of infectious viral shedding: Ct values < 25 were associated with positive viral cultures. RADTs were not performed in that study [17]. In our study, a limited number of samples had serial RT-PCR data available, and a correlation of RT-PCR Ct values with either viral culture or RADT findings could not be determined. In addition, all patients had positive RT-PCR results at time points with negative RADT and viral culture findings. In case 1, the viral cultures were negative, and the RT-PCR Ct value was 13, whereas in case 3, the viral cultures were negative, and the RT-PCR Ct value was 24. A recent study has indicated that the correlation between RT-PCR Ct values and SARS-CoV-2 viral cultures lacks utility, because it varies by upper and lower respiratory tract sample type in critically ill patients with COVID-19 [9]. Furthermore, the role of RT-PCR Ct values in predicting viral infectiousness has been questioned in a study evaluating their correlation with viral culture findings during the pre-Omicron and Omicron periods. During the Omicron period, compared with the pre-Omicron period, viral isolates had higher Ct values and were more likely to have positive viral cultures, thus indicating that the RT-PCR Ct value is a poor indicator of infectiousness in patients infected with Omicron variants [18].

The mortality rate in our case series was 100%. The patients had several risk factors for severe COVID-19 and admission to the CCU, including age, multiple comorbidities and immunosuppression. In addition, all patients had incomplete vaccination status: one patient was unvaccinated, and the other two patients had received only two COVID-19 vaccine doses. Several studies and a systematic review have indicated that three COVID-19 vaccine doses is significantly more effective than two COVID-19 vaccine doses against Omicron variants [19]. A real-world study in the UK has confirmed the effectiveness of three COVID-19 vaccine doses in decreasing hospitalization, respiratory failure and CCU admission in patients infected with Omicron variants, including patients older than 75 years, compared with controls [20]. Another study in Canada has confirmed that three COVID-19 vaccine doses is 61% effective in preventing symptomatic infection and 95% effective in preventing hospitalization or death, as compared with the values in controls [21].

The main limitation of our study is the inclusion of a small number of cases. Larger studies are therefore required to confirm our results. The main strength of our study was the serial COVID-19 RADT, RT-PCR and viral culture testing during the hospitalization of critically ill patients infected with Omicron variants. Viral cultures remain the standard tests for evaluating infectious viral shedding, rather than RT-PCR and RADTs.

5 Conclusion

In our case series, on the basis of significant concordance between viral culture and RADT findings, RADT is a potentially reliable test to determine the necessary duration of isolation in patients with COVID-19 in CCUs This test may help overcome the challenges of the variable duration of viral shedding in the era of SARS-CoV-2 variants, and the burden of CCU bed utilization during surges in upper respiratory tract infections, including influenza, respiratory syncytial virus and COVID-19. Future studies are required to confirm our findings.