Associations of viral ribonucleic acid (RNA) shedding patterns with clinical illness and immune responses in Severe Acute Respiratory Syndrome Coronavirus 2 (SARS‐CoV‐2) infection

Abstract Objectives A wide range of duration of viral RNA shedding in patients infected with Severe Acute Respiratory Syndrome Coronavirus 2 (SARS‐CoV‐2) has been observed. We aimed to investigate factors associated with prolonged and intermittent viral RNA shedding in a retrospective cohort of symptomatic COVID‐19 patients. Methods Demographic, clinical and laboratory data from hospitalised COVID‐19 patients from a single centre with two consecutive negative respiratory reverse transcription‐polymerase chain reaction (RT‐PCR) results were extracted from electronic medical records. Kaplan–Meier survival curve analysis was used to assess the effect of clinical characteristics on the duration and pattern of shedding. Plasma levels of immune mediators were measured using Luminex multiplex microbead‐based immunoassay. Results There were 201 symptomatic patients included. Median age was 49 years (interquartile range 16–61), and 52.2% were male. Median RNA shedding was 14 days (IQR 9–18). Intermittent shedding was observed in 77 (38.3%). We did not identify any factor associated with prolonged or intermittent viral RNA shedding. Duration of shedding was inversely correlated with plasma levels of T‐cell cytokines IL‐1β and IL‐17A at the initial phase of infection, and patients had lower levels of pro‐inflammatory cytokines during intermittent shedding. Conclusions Less active T‐cell responses at the initial phase of infection were associated with prolonged viral RNA shedding, suggesting that early immune responses are beneficial to control viral load and prevent viral RNA shedding. Intermittent shedding is common and may explain re‐detection of viral RNA in recovered patients.


INTRODUCTION
The global tally for coronavirus disease 2019 (COVID-19) had crossed the 4.5-millon mark and accounted for more than 300 000 deaths by 17 May 2000. 1 The impact of the pandemic has placed an extra burden on healthcare systems, causing capacity limitations. 2 In Singapore, all individuals with confirmed SARS-CoV-2 infection were initially admitted into airborne infection isolation rooms and attending staff wore personal protective equipment in accordance with the United States Centers for Disease Control and Prevention guidelines. 3 More than 25 000 COVID-19 cases have been reported in Singapore as of 16 May 2020. 4 Of these, approximately 1095 (4.3%) were still hospitalised while 17 881 (71%) who were clinically well but still tested positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus causing COVID-19, had been discharged to community isolation facilities.
A large proportion of patients with COVID-19 experience only mild disease and do not require inpatient hospital care. 5 However, several countries such as Singapore have a policy to hospitalise or isolate all patients with confirmed infection to mitigate the risk of secondary community transmission. However, the infectious period of COVID-19 infection is at present still unclear, and the recommended duration of isolation required has varied across regulatory authorities. [6][7][8] In a cohort of 191 patients in China, the median duration of viral ribonucleic acid (RNA) shedding by RT-PCR was 20 days. 9 Therefore, the length of stay in healthcare or isolation facilities can be much longer than medically required.
In a study of 113 patients in China, risk factors associated with prolonged SARS-CoV-2 RNA shedding included male gender, delayed admission to hospital after illness onset and requirement for invasive mechanical ventilation. 10 At our centre, the National Centre for Infectious Diseases (NCID), we observed patients with prolonged and intermittent viral RNA shedding of SARS-CoV-2 RNA by RT-PCR which precluded discharge from our healthcare or isolation facilities. In this report, we describe the prevalence of prolonged and intermittent viral RNA shedding and investigated associated factors as well as correlation with host immune responses.

Baseline characteristics and clinical outcomes
We identified 205 patients who were hospitalised with at least one positive RT-PCR for SAR-CoV2-2 and discharge followed by two consecutive negative results. As four patients were asymptomatic, they were excluded from analysis. Of the remaining 201 patients, 105 (52.2%) were male. The median age was 49 years (Interquartile  range  [IQR] 16-61 years). The common comorbidities were hypertension (23.4%), hyperlipidaemia (23.4%) and diabetes mellitus (13.9%). Median duration from illness onset to hospital admission was 5 days (IQR, 3-8 days). One hundred and six patients (52.7%) had pneumonia on chest radiograph, 43 (21.4%) required supplemental oxygen for hypoxia (oxygen saturation <94%), and 10 (5.0%) needed invasive mechanical ventilation. Death occurred in two patients. The patient characteristics are summarised in Table 1.

Viral RNA shedding patterns and association with clinical characteristics
The median duration of viral RNA shedding in this study was 14 days (IQR 9-18) (Figure 1a). Prolonged viral shedding was thus defined as the duration of SARS-CoV-2 RNA shedding being longer than 14 days. The median duration of hospital stay was 13 days (IQR 9-17), and each patient received a median of 7 SARS-CoV-2 PCR tests. Duration of viral shedding was not significantly associated with sex, age, presence of comorbidities, baseline investigations, prolonged fever (≥ 7 days), pneumonia, need for supplemental oxygen and use of experimental antiviral agents. Median duration of viral RNA shedding was significantly longer in patients requiring invasive mechanical ventilation (19 days, 95% confidence interval [CI] 17.5-20.2) compared to 14 days (95% CI, 13.1-14.9) in patients not requiring mechanical ventilation, P-value = 0.01 ( Figure 1b). Duration of viral shedding was also significantly associated with lopinavir/ritonavir treatment (median 13 days, 95% CI 12.1-13.9) vs. 16 days (95% CI 14.3-17.8), P-value = 0.026. However, in the multivariate logistic regression analysis, none of the factors analysed were statistically significant (Table 3).
Seventy-seven patients (38.3%) had intermittent viral RNA shedding, which we defined as alternating between positive and negative PCR tests on serial testing, before cessation of viral shedding with two consecutive negative PCR tests. The median duration from symptom onset to intermittent viral RNA shedding was 13 days (IQR 10-16.5), and the median duration of intermittent viral RNA shedding was 3 days (IQR 2-5)

DISCUSSION
The duration of viral RNA shedding in COVID-19 ranged from 1 to 35 days. The factors that correlated significantly with the duration of viral RNA shedding on univariate analysis were the requirement for invasive mechanical ventilation and lopinavir/ritonavir treatment. Age, gender, comorbidities and disease severity were not found to be significantly associated with duration of viral RNA shedding. In the multivariate analysis, none of the factors analysed was statistically significant ( Table 3). The duration of viral RNA shedding is similar in SARS and MERS coronavirus infections. In SARS, patients with severe disease had higher viral load and 47% of patients with SARS-CoV had detectable virus RNA at day 21 of illness. 11 In Middle East respiratory syndrome coronavirus (MERS-CoV), patients with more severe disease had higher and more prolonged levels of viral RNA tested by RT-PCR. 12 The median time to negative test was 17 days in MERS. 13 As the period of infectivity of COVID-19 was unknown, the demonstration of 2 consecutive negative RT-PCR tests was recommended to deisolate patients with COVID-19, similar to MERS. 14 Xu et al. 10 conducted a retrospective study of 113 patients with COVID-19 and found that male gender, older age, severe disease, late   presentation, use of corticosteroids and invasive mechanical ventilation were associated with prolonged viral RNA shedding. Age, gender distribution, comorbidities and days to hospital presentation were similar to our cohort. However, more patients in this study required invasive mechanical ventilation (15.9% vs 5.0%), which could partially explain the longer median viral RNA shedding duration of 17 days. Additionally, 56.6% patients in the cohort received systemic glucocorticoid therapy compared with 0.5% (1 of 201) in our cohort, which has been associated with longer viral RNA shedding in patients with SARS and MERS. 15,16 In another study of 147 patients with COVID-19, the use of systemic corticosteroids, but not supplemental oxygen requirement, was associated with prolonged viral RNA shedding. 17 In contrast, we found on univariate analysis that patients who required invasive mechanical ventilation and patients who received lopinavir/ritonavir treatment to be significantly associated with longer viral RNA shedding. The association of prolonged viral RNA shedding with invasive mechanical ventilation could be confounded by the testing of endotracheal samples which are known to have higher viral load and sensitivity in patients with COVID-19 pneumonia. 18 Further, invasive mechanical ventilation was not found to be an independent risk factor of prolonged viral RNA shedding in our multivariable model.
We observed that prolonged viral RNA shedding was associated with elevated levels of cytokines actively involved in pulmonary inflammatory response, [19][20][21][22] particularly in patients who required mechanical ventilation. Interestingly, patients with prolonged viral RNA shedding demonstrated lower levels of IL-1b and IL-17A during the acute phase of infection. Both IL-1b and IL-17A are pivotal cytokines involved in activation of anti-viral T-cell responses. 23,24 Exhaustion of T-cell activation was reported to be associated with increased duration of viral RNA shedding in SARS-CoV infections. 25 Hence, we postulate that more effective T-cell activation at the early phase of infection promotes virus clearance and subsequently shortens viral RNA shedding duration in COVID-19. Further studies are warranted to comprehensively assess the roles of effector T cells in mediating virus clearance during SARS-CoV-2 infection.
Recovered COVID-19 patients who were diagnosed with re-infection by subsequent positive RT-PCR have been reported. 26 As a result of RT-PCR test characteristic, very low levels of RNA at the threshold of detection may cause false-negative or false-positive results. 27 This can account for intermittent viral RNA shedding observed in our study. In our cohort, intermittent viral RNA shedding was detected in 77 (38.3%). While most patients had short duration of intermittent viral RNA shedding, 12 (6.0%) had intermittent viral RNA shedding for more than 7 days. In another report, 4 patients who demonstrated 2 negative RT-PCR tests re-tested positive again after 5-13 days despite being asymptomatic with no new radiographic changes. 28 While re-infection is a concern, prolonged intermittent viral RNA shedding may account for re-detection of viral RNA.
Viral RNA shedding may be intermittent and is affected by the immune status of the patients. Previous studies have reported that virus reactivation may be provoked by anti-inflammatory therapies. 29,30 We noted that the patients had significantly lower pro-inflammatory cytokine levels during intermittent viral RNA shedding period, compared with those without intermittent viral RNA shedding after complete virus clearance. A weaker inflammatory response or suppression of inflammatory responses could be triggering virus re-activation, resulting in intermittent detection in the patients. Whether milder inflammation and virus re-activation are causally related remains to be explored.
While current de-isolation strategies generally rely on demonstration of non-detectable viral RNA, detection by RT-PCR is only a surrogate marker for infectivity. Positive RT-PCR test may represent non-viable viruses or remnant nucleic acid products. A study of 9 patients showed that SARS-CoV-2 was readily isolated from respiratory samples within the first week of symptoms, with greater success from sputum than from upper respiratory swabs, and from samples with higher viral loads. Additionally, viral isolation was unsuccessful when patients were beyond day 8 of illness. The authors suggested that patients beyond 10 days of illness with < 100 000 viral RNA copies per mL of sputum could be safely isolated at home. 31 If the suggested de-isolation strategy can be applied, it can significantly shorten duration of isolation and demand for scarce hospital beds during periods of peak COVID-19 activity.
There are limitations to our study. As viral culture was not done, we were unable to determine whether the detection of viral RNA by RT-PCR was related to viable virus or shedding of remnant non-viable genetic material. This would have public health implications on the period of infectivity and duration of isolation of COVID-19 cases. The duration of viral RNA shedding relied on negative tests by RT-PCR which may be influenced by collection technique or the respiratory site from which the sample was obtained. Patients with predominantly lower respiratory infection may have a negative nasopharyngeal test result. The failure to find any clinical correlates of prolonged viral shedding may have been due to inadequate power to detect small differences, including the small number of patients who required invasive mechanical ventilation.
In conclusion, our study observed no independent risk factor associated with prolonged viral RNA shedding. While the duration of viral RNA shedding varied widely, there were no identified demographic or clinical determinants in the duration of viral RNA shedding. COVID-19 patients with less active T-cell responses during the initial phase of infection shed viral RNA longer, and these patients also presented lower levels of pro-inflammatory cytokines during intermittent viral RNA shedding. Intermittent viral RNA shedding is a common phenomenon, and patients with prolonged intermittent viral RNA shedding may explain reports of re-detection of viral RNA in recovered COVID-19 patients.

Patients, clinical data and management
During the study period, all patients with confirmed COVID-19 infection in Singapore were mandated to be admitted to hospital regardless of illness severity. A total of 201 patients with confirmed COVID-19 infection admitted in National Centre for Infectious Diseases (NCID) from 22 January 2020 to 5 April 2020 and discharged after obtaining two consecutive negative RT-PCR results at least 24 h apart were included. The discharge date was censored on 21 April 2020.
Some of the patients received supportive therapy which included supplemental oxygen, empirical antibiotics and/or oral oseltamivir if there was clinical concern of communityacquired pneumonia (Table 1). Complete blood cell count, renal and liver function, C-reactive protein (CRP) and lactate dehydrogenase (LDH), and chest radiograph were performed for all patients as standard of care.
Co-formulated lopinavir-ritonavir, interferon beta-1b and hydroxychloroquine were prescribed to selected patients at the treating physicians' discretion after shared decisionmaking and provision of oral informed consent. Remdesivir was prescribed for patients enrolled in clinical trials in accordance with trial protocols (ClinicalTrials.gov: NCT04280705, NCT04292899, NCT04292730). Corticosteroids were only given for septic shock in the intensive care unit. Demographic, clinical and laboratory data were obtained from the electronic medical records. Data collection was approved under the Infectious Diseases Act with waiver of informed consent. 32 Respiratory samples for SARS-CoV-2 testing included nasopharyngeal swab, throat swab, sputum and endotracheal aspirate. The diagnosis of COVID-19 infection was confirmed through reverse transcription-polymerase chain reaction (RT-PCR) testing for SARS-CoV-2. Repeat samples were performed on average, every other day after the patients were afebrile and clinically improving. A cycle threshold value (Ct-value) above 35 indicates low viral load in the sample; this guided the physicians to repeat viral RNA sampling by PCR 24 h later. National Public Health Laboratory developed test targeting the N and ORF1ab genes was used at the start of the outbreak in Singapore in late January 2020. From 6 February 2020, a commercial assay was used. 33 Specific consent for SARS-CoV-2 testing was not obtained as all testing was part of routine clinical care. Written informed consent was obtained (approval from the National Healthcare Group Domain Specific Review Board, Study Reference 2012/ 00917) for measurement of immune mediator serum samples.
Duration of viral RNA shedding was defined as the number of days from symptom onset to the last day of positive RT-PCR. Duration of persistent viral RNA shedding was defined as the number of days from symptom onset to the first negative RT-PCR. Duration of intermittent viral RNA shedding, if present, is defined as the number of days from the first negative RT-PCR test to the last positive RT-PCR.

Statistical analysis
The Mann-Whitney U-test was used for comparison of continuous variables, and chi-square (v 2 ) or Fisher's exact tests for categorical variables as appropriate. Kaplan-Meier survival curve analysis was used to assess the effect of gender, age, severity of illness and experimental anti-viral agents on the duration of viral RNA shedding. Significant risk factors identified on univariate analysis were further analysed by adjusted multivariate logistic regression analysis to identify independent risk factors associated with the prolonged duration of viral RNA shedding. Nonparametric Mann-Whitney U-tests were conducted on the logarithmically transformed concentration of immune mediators between patients with prolonged or intermittent viral RNA shedding and those without. Correlation analysis was carried out using Spearman's rank correlation. P-values < 0.05 were considered statistically significant. Data analysis was performed using SPSS, version 26.0 (IBM Corp., Armonk, NY, USA). Plots were generated using GraphPad Prism version 8 (GraphPad Software, San Diego, CA, USA).

ACKNOWLEDGMENTS
We thank Professor Olaf R€ otzschke, Dr Bernett Lee, Wilson How and Norman Leo Fernandez from the Singapore Immunology Network (SIgN), for their help in running multiplex microbead immunoassay. We are also grateful to Dr Danielle Anderson and her team at Duke-NUS, for their technical assistance in virus inactivation procedures with Triton TM X-100. This work was supported by a National Medical Research Council COVID19 Research Fund (Ref: COVID19RF-001) to Barnaby Young. The multiplex immunoassay studies were supported by the A*STAR COVID-19 Research funding (H/20/04/g1/006) provided to Singapore Immunology Network (SIgN) and core fund allocated to SIgN by the Biomedical Research Council (BMRC), A*STAR, and by a National Research Foundation grant (#NRF2017_SISFP09) to the SIgN Immunomonitoring platform.