Discussion
This study is the first to investigate the relationship between viral shedding duration in COVID-19 and patient prognosis and disease progression to critical illness. In this study we determined that prolonged viral RNA shedding duration in COVID-19 is associated with improved patient prognosis. Furthermore, patients with long viral shedding duration as compared to short viral shedding duration had a lower rate of progression to critical illness. Finally, we determined the mortality of COVID-19 in patients with long viral RNA shedding duration (≥26 days) was nearly 50% less as compared to patients with short viral RNA shedding duration (<26 days). In this study, we have excluded 5 patients whose viral RNA shedding of SARS-CoV-2 was also detectable until death. So, the mortality is not a confounder of viral RNA shedding duration.
This finding is inconsistent with previous findings in different viruses including avian influenza A (H5N1) and respiratory syncytial virus, where longer viral shedding duration has been shown to be associated with increased risk of complications2,4,5,7,8. Furthermore, longer duration of viral shedding in H1N1 has been correlated with worse disease severity9. There are many differences in the pathology of SARS-CoV-2 that could potentially account for the different observations in the correlation between viral shedding duration and patient prognosis in COVID-19 as compared to other viruses (such as influenza). One possibility is that a destabilizing mutation in SARS-CoV-2 in the nsp3 protein (that is not observed in other viruses, such as SARS) could explain the unique correlation between decreased viral shedding duration and worse disease progression observed with COVID-19, although this remains to be tested10.
Secondly, differences in virus subtypes of SARS-CoV-2 as compared to other viruses might explain the contrasting finding that increased viral shedding duration is associated with better patient prognosis in COVID-19. A recent study investigated the molecular divergence of SARS-CoV-2, in which it was determined the virus has two major types, L-type and S-type, each that has unique characteristics11. This study determined that L-type SARS-CoV-2, as compared to S-type, is detected in approximately 70% of all patients and is more prevalent and may be more transmissible than the L-type. It has been suggested that the diversity of COVID-19 mortality rates in different regions of China may be explained by the different virus subtypes, among other characteristics12,13. Currently, few studies have focused on defining the clinical features of these two sub-types of SARS-CoV-2. As we do not know the SARS-CoV-2 subtypes of the patients in our study, further analysis of the SARS-CoV-2 mRNA sequence of the virus type could help further stratify the findings in our study.
Previous studies have shown that correlations between viral load and immune system response are important determining factors in disease progression. It has also known that SARS-CoV-2 infection can activate innate and adaptive immune responses14. Therefore, it is possible an immune system response may influence the differences in viral load observed in the patients in our study. Consistent with prior studies that have shown lymphopenia is a common feature in severe COVID-1914,15, we also frequently observed lymphocytopenia in the patients in our study. Patients with long viral RNA shedding duration as compared to short viral RNA shedding duration had a significantly higher absolute lymphocyte count, as well as higher levels of inflammatory markers (such as hs-CRP and IL-6). This suggests that patients with prolonged viral RNA shedding durations may also have increased innate and adaptive immune responses, although this remains to be tested.
Previous studies have observed that the median duration of viral shedding in COVID-19 survivors ranges from 17 days (IQR 13-22 days) to 20 days (IQR 17-24 days)3,16. The longest reported duration of viral shedding in COVID survivors was 37 days3,16. In our study, the median SARS-CoV-2 shedding duration was 26 days, and the longest duration was 62 days (reported in a 72-year-old male patient who survived COVID-19). Approximately 15% of in our study patients had a viral shedding duration of more than 6 weeks. This duration is significantly longer than has been previously reported3,16. This difference may be attributed to the fact that the patients enrolled in our study were from Wuhan, China where more severe cases of COVID-19 have been reported. Delayed hospital admission may also be associated with longer viral shedding duration as patients would not have received any medical care that could have minimized COVID-19 symptoms. This could also potentially contribute to the longer viral RNA shedding times detected in our studies.
There are limitations to this retrospective study. First of all, although we were able to detect viral RNA of SARS-CoV2, viral RNA shedding may not perfectly correlate with total viral load as quantified by virus isolation. However, viral RNA can be used as a surrogate for estimating the total viral load in a patient. Secondly, the viral RNA specimens were obtained from pharyngeal swabs and it is possible there may be instances of false negative viral RNA detections. Respiratory tract specimens such as viral RNA detection from sputum, endotracheal aspirate, or bronchoalveolar lavage fluid would provide a more accurate level of detection for viral RNA, but obtaining these specimens from non-critical patients is not practical as it requires invasive mechanical ventilation. Finally, there were variations in the antiviral treatments patients received in this study. Lopinavir/ritonavir and interferon-α were used for many patients in our study, and because of the lack of standardized antiviral therapies for the patients in our study, we do not have statistical power to determine if these antiviral treatments had an effect on viral RNA shedding.