Pharmacological treatment for patients with coronavirus disease 2019: systematic review of randomized controlled trials

Bahaa Abu-Raya (  baburaya@bcchr.ubc.ca ) Vaccine Evaluation Center, BC Children's Hospital Research Institute Hana Mijovic Vaccine Evaluation Center, BC Children's Hospital Research Institute Robine Donken Vaccine Evaluation Center, BC Children's Hospital Research Institute Adriana Cabrera Vaccine Evaluation Center, BC Children's Hospital Research Institute Vanessa Kitchin University of British Columbia Rachael Chow McGill University Rose Leishman McMaster University Christina Pan University of British Columbia Vanessa Paquette Department of Pharmacy, British Columbia Women's Hospital and Health Centre; University of British Columbia Ashley Roberts Division of Infectious Diseases, Department of Pediatrics, University of British Columbia Manish Sadarangani Vaccine Evaluation Center, BC Children's Hospital Research Institute


Results
Overall, 6346 references were screened for eligibility. Of these, 39 full-text articles were assessed against inclusion and exclusion criteria. Nineteen articles met the inclusion criteria and were included in the systematic review and qualitative synthesis (Supplementary Figure 1). Thirteen out of 19 studies included patients with moderate-to-severe disease, and 16 studies enrolled hospitalized patients. Most of the studies were deemed to have low risk of bias (Supplementary Tables 1-3).
Two RCTs assessed the effect of remdesivir on hospitalized patients with laboratory-con rmed moderate-to-severe COVID- 19. While a small study from China (n=237) did not nd difference in time to clinical improvement between remdesivir and placebo during 28 days follow up, a larger multi-country study (n=1059) reported that 10 days of remdesivir was associated with shortened time to recovery (11 vs. 15 days) during 28 days follow up. Neither study found reduction in mortality at day 14 after enrollment (Table 1). ROB was low/unclear in some domains (Supplementary Table 1).
A study investigating lopinavir/ritonavir in patients with laboratory-con rmed COVID-19 did not meet the primary outcome of the study of time to clinical improvement, evaluated at day 28, and also did not show reduction in mortality (Table 1). ROB was low (Supplementary Table 1). Studies of combination regimens of lopinavir/ritonavir with other antivirals and other immunomodulators were also included. In patients receiving lopinavir/ritonavir combined with novaferon, compared with novaferon alone, the percentage of patients with negative SARS-CoV-2 PCR was higher at day 3 and 6 after randomization, and the median time until negative SARS-CoV-2 PCR was shorter (Table 1). In another study, the combination of lopinavir/ritonavir with IFN b-1b and ribavirin (RBV) was associated with shorter time until negative SARS-CoV-2 PCR, shorter time to resolution of symptoms and decreased hospital length of stay (LOS), compared with lopinavir/ritonavir alone. ROB was low in both studies (Supplementary Table 1).
Four RCTs investigated HCQ use in hospitalized and clinic patients. Two of these RCTs reported that early treatment of outpatients with mild disease (4-5 days after symptoms onset) did not reduce disease severity, hospitalization rate, death or SARS-CoV-2 viral load (Table 1). ROB was variable in these studies (Supplementary Table 1). Two other RCTs found no difference in virological clearance during 28 days follow up after HCQ administration in hospitalized patients with mild-to-moderate disease. HCQ administration was also not associated with difference in symptoms, need for respiratory support or death (Table   1). ROB was low for most domains in these two studies (Supplementary Table 1).
Two published RCTs that assessed the effect of administration of steroids on 28-day mortality were identi ed. The rst was an interim analysis of the RECOVERY trial that found that administration of dexamethasone resulted in a lower case-fatality rate compared with standard of care. Subgroup analyses indicated that the nding was prominent among patients receiving invasive mechanical ventilation, but not observed in patients not receiving respiratory support at randomization. A study in Brazil with a lower number of patients randomized, reported that methylprednisolone did not result in a reduction in 7day, 14-day or 28-day case-fatality rate compared with placebo ( Table 2). The two studies had low ROB (Supplementary Table 2).
Four RCTs that evaluated immunomodulators (other than steroids) as single agents in treating COVID-19 patients were identi ed ( Table 3). The rate of hospitalization was not different in outpatients treated with febuxostat compared with HCQ. Receipt of IFN b-1a did not result in improvement in time to clinical response, although mortality bene t was noted, especially if administered within 10 days of symptoms onset. In a small study, colchicine was associated with improved time to clinical deterioration. Ruxolitinib receipt did not result in reduction in time to clinical improvement or mortality bene t.
Overall, the studies were found to have low ROB (Supplementary Table 3).

Discussion
There is an urgent need to establish the optimal pharmacological treatment of patients with COVID-19. In this systematic review, we aimed to assess the effect of different pharmacological treatments on outcomes in patients with COVID-19. Remdesivir was found to be associated with shortened time to clinical improvement in patients hospitalized with moderate-severe disease. In addition, an improved survival rate was noted for steroids administered to patients, especially those who required mechanical ventilation and/or supplemental oxygen at diagnosis. Lopinavir/ritonavir was not associated with clinical improvement and did not demonstrate mortality bene t. HCQ administered to outpatients or inpatients did not reduce the severity of disease or hospitalization rates, enhance virological clearance, improve symptoms, reduce need for respiratory support or prevent death.
Remdesivir is an inhibitor of the viral RNA polymerase and has been shown to have inhibitory activity against SARS-CoV-2, SARS-CoV-1 and MERS-CoV in vitro (8)(9)(10)(11)(12). While a small study did not nd that remdesivir administration was associated with shortening of time to clinical improvement (13), a multi-country study found that a 10 days course of Remdesivir was associated with shortening of time to recovery (14). However, current evidence from both studies did not support that there is reduction in mortality at day 14 after randomization. It should be noted that the former study did not complete full enrollment (due to the end of outbreak in China) of the target number of patients, and thus had a lower sample size that might have precluded any de nite conclusion. In addition, the modest clinical bene t observed (14) with a highly expensive drug might challenge its use in some settings (e.g. low-middle income countries). Experience with other human coronaviruses is limited. In non-human primate studies, remdesivir reduced MERS-CoV lung virus levels and lung damage when initiated 12 hours after inoculation with the virus (15,16).
Lopinavir is a protease inhibitor, and is combined with ritonavir to increase lopinavir's plasma half-life through the inhibition of cytochrome P450. This combination is an established agent in the treatment of HIV. Lopinavir has in vitro inhibitory activity against SARS-CoV-1 (17,18), the causative agent of SARS disease. Lopinavir also has activity against MERS-CoV observed in vitro (19) and in an animal model (20). The addition of RBV to lopinavir/ritonavir reduced the risk of ARDS or death, as well SARS-CoV-1 viral load among patients with SARS (21). The combination of lopinavir/ritonavir, RBV and IFN α has been associated with survival in case reports of patients with MERS (22)(23)(24). In this systematic review, one RCT did not nd lopinavir/ritonavir to be associated with shorter time to clinical improvement, or have mortality bene t in patients with SARS-CoV-2 (25). One RCT found that the combination of IFN b-1b, lopinavir/ritonavir, and RBV was associated with shorter time to negative SARS-CoV-2 PCR, shorter time to resolution of symptoms and decreased hospital LOS, compared with lopinavir/ritonavir alone (26). This might have practical implications related to isolation precautions for patients. In addition, reduction of hospital LOS might enable health service to cope with higher load of patients, provide better care in severe disease, and might also reduce costs of hospitalization.
Chloroquine and its hydroxyl analogue HCQ are well known as antimalarial drugs. Both drugs have been shown to block the viral replication of SARS-CoV-2 in cell cultures, suggesting that they might have potent antiviral activity against SARS-CoV-2 in vivo (12,27). In our systematic review, two RCTs performed on outpatients found no reduced disease severity or hospitalization rates after treatment with HCQ early in the course of mild disease (28,29). Two RCTS on the use of HCQ in hospitalized patients also failed to demonstrate that HCQ administration for patients with mild-moderate disease enhanced virological clearance, improved symptoms, reduced need for respiratory support or prevented death (30,31).
One of two published studies showed that dexamethasone treatment was associated with reduction in 28-day mortality, especially among patients who required invasive mechanical ventilation or supplemental oxygen (32). While this nding was not supported by a recent study from Brazil, which also recruited patients with moderate to severe COVID-19, differences in the results might stem from the lower number of patients, the later presentation after symptoms onset and the higher baseline mortality rate without steroids in the latter compared with the former study (33). A recent prospective meta-analysis of RCTs that evaluated the e cacy of corticosteroids in critically ill patients found that corticosteroid administration was associated with a reduction in 28-day mortality when compared with SOC or placebo (odds ratio 0.66, 95% CI: 0.53-0.82) (34). Thus, The World Health Organization (WHO) is currently considering amending their COVID treatment guidelines to recommend the use of steroids in the treatment of critically ill patients. Limited data from earlier SARS outbreaks were not conclusive regarding the bene t of steroids (35). Systematic review of the use of corticosteroids in patients with MERS did not suggest a reduction in mortality, and was associated with delayed MERS-CoV RNA clearance (36).
Investigation of other immunomodulatory agents as a single agent has shown less promising results, and published articles were limited by small number of patients. Among the 4 articles identi ed, receipt of IFN -1a might have mortality bene t especially if administered early in the course of the disease (37). However, the high costs of these drugs might preclude their use, especially in low-middle income countries settings. IFN b was found to be the most potent inhibitor of SARS-CoV among other IFNs (38). However, a retrospective study showed that RBV plus recombinant IFN (rIFN-α2a, rIFN-α2b, or rIFN-β1a) did not result in a reduction in 90-day mortality amongst patients with severe MERS-CoV infection (38).

Conclusions
Remdesivir is associated with shortening time to clinical improvement in patients hospitalized with moderate-severe disease, and its use should be supported by future research. Currently, there is no evidence to support the use of HCQ in either outpatients with mild disease or inpatients with mild-moderate disease, nor the use of lopinavir/ritonavir in hospitalized patients. Our systematic review supports the use of steroids in critically ill patients.
There is a need to investigate the role of pharmacological treatment in adults with some underlying comorbidities (e.g. human immunode ciency virus infection) or other speci c populations (e.g. pediatric population, pregnant women) who have up to now largely been excluded from RCTs. In addition, there is a need to further explore the role of steroids in the treatment of moderate-severe COVID-19 as well as the optimal dose and duration of corticosteroid therapy in critically ill cases. Meta-analysis of effects of different pharmacological treatments may be possible as additional RCTs are published.   * Number out of total and %; ** 88% con rmed in dexamethasone and 89% in SOC group; *** 83% of MP and 79% of Pla were laboratory con rmed by SARS-CoV-2 RT-PCR