This systemic review of medical literature and meta-analysis in patients with COVID-19 did not reveal any clinic benefit of using HCQ/CQ, even if these medications were initiated early after diagnosis. Despite large heterogeneity of the individual publications, our results suggested that the use of HCQ/CQ was associated with an increased risk of in-hospital mortality in these patients. Moreover, we showed no consistent rapid viral clearance of SARS-CoV-2 by using HCQ/CQ. The heterogeneity of the drug efficacy on mortality might be attributed to the comorbidities of enrolled subjects, such as hypertension, diabetes and chronic lung disease.
In theory, HCQ and CQ may be ideal drugs to treat SARS-CoV-2 infection as they can inhibit the virus via their antiviral effects and help mediate the cytokine storm via immunomodulation. HCQ was found to be more potent than CQ to inhibit SARS-CoV-2 in vitro , and exerted less side-effects in the daily clinical practice for the treatment of systemic lupus erythematosus and rheumatoid arthritis. Although the anti-viral effects of HCQ/CQ on SARS-CoV-2 were confirmed by several independent in vitro studies, their clinical efficacy on COVID-19 was controversial in clinical studies. Early investigations and clinical trials reported different outcomes of HCQ/CQ treatment on clinical recovery, virus-negative conversion, symptomatic progression and death. In this study, we opted to study viral clearance rate by RT-PCR and all-cause in-hospital mortality - two clinically relevant objectives - in our meta-analysis to evaluate the clinical efficacy of HCQ/CQ in patients with COVID-19.
The earliest reported study on the use of HCQ in COVID-19 was from Marseilles, France, and showed 50% viral clearance, as tested by RT-PCR, at Day 3 compared to 6.3% in the control group. Addition of azithromycin in this cohort seemed to augment the viral clearance . The study had major limitations because it included 26 patients on HCQ from one center and 16 patients as controls from another hospital, and 6 patients from the treatment group were lost to follow-up during the survey because of early cessation of treatment. Later on, four other observational studies reported conflicting results on viral clearance at different time-points during the disease. Mallat J et al.  reported that viral negativity was significantly lower with HCQ compared to control group, whereas studies from Chen X et al  and Huang M et al  did not observe rapid viral clearance by CQ. A large multi-center cohort containing 373 patients (mainly mild to moderate COVID-19) by Huang MX et al  reported a significantly higher rate of viral clearance at Day 10 (91.0% vs 57.0%) and Day 14 (96.0% vs 80.0%) in the CQ treatment group. These authors suggested that early treatment initiation (e.g. <3 days or 3-7 days) contributed to shorter time to undetectable viral RNA by RT-PCR. In contrast, two recently reported RCTs on HCQ from Chen J et al and Tang W et al did not reveal a higher proportion of viral clearance in patients admitted to hospital with mainly mild to moderate COVID-19 than standard of care alone [8, 22]. Our primary meta-analysis revealed a positive effect of HCQ/CQ on viral clearance by RT-PCR testing. However, this benefit was mainly attributed to the study reported by Huang MX et al, and the odds ratio was no longer significant when this study was omitted by sensitivity analysis. More remarkablly, our meta-regression analysis did not observe the coefficient of viral clearance with dose or early administration of HCQ/CQ. Thus, along with the strong evidence gathered from the above-mentioned RCTs, we can conclude that HCQ/CQ has no beneficial effect on viral clearance in COVID-19 patients.
Mortality outcomes were addressed in 7 observational studies and 2 RCTs. The studies from Yu B et al from China  and Membrillo FJ et al from Spain  reported reduced in-hospital mortality with the use of HCQ compared to standard care, but there was non-negligible bias in these studies. The first study reported clinical efficacy of low dose (400mg/day) of HCQ on critically ill patients, whereas the later study reported that high dose (800mg/day) of HCQ was associated with improved survival, which was statistically significant only among patients mild COVID-19. Both of these studies had unmatched number of patients in the treatment and control groups (48 vs 502 and 123 vs 43, respectively) as a statistical limitation. On the other hand, two observational studies from Rosenberg E et al  and Geleris J et al , showed significantly higher mortality in the HCQ group compared to the non-treatment group, but the treated patients were obviously more severe than those in the control group. Thus, both studies performed multi-level statistical adjustments and concluded that HCQ administration was not associated with either a greatly improved or increased risk of the mortality endpoint. In our meta-analysis, we extracted adjusted events rate from these two studies to reduce the study bias. The strong evidence from the RCT by Mahevas M et al  also demonstrated that HCQ treatment did not have any effect on ICU admissions or death at Day 21 after hospital admission. In combination with these RCTs and real-world observational studies, our meta-analysis did not reveal a mortality benefit with administration of HCQ/CQ. On the contrary, it seems to show that HCQ/CQ treatment may even increase mortality to some extent (OR=1.26, 95%CI 1.06-1.51).
A novel finding of our meta-regression analysis was that the large heterogeneity of the findings of the pooled publications might have been due to the characteristics of study cohorts, and in particular, due to their comorbidities such as hypertension, diabetes and chronic lung disease. In the presence of higher proportion of these comorbidities, there was an increased risk of mortality (OR>1) with the use of HCQ/CQ. Previous studied had clearly demonstrated that patients with cardiopulmonary comorbidities had adverse outcomes, including progression into cardiac injury and death [26-28]. The potential mechanisms were also proposed based on the observations of several molecular and pathology studies [29-31]. Despite the fact that HCQ/CQ were commonly used for rheumatoid arthritis and systemic lupus erythematosus in the general population, there has been major concern with widespread use of these medications for COVID-19 due to possible QTc prolongation and induction of life-threatening arrhythmias. There is general consensus that QTc prolongation after HCQ/CQ use may occur in a dose-dependent way [11, 32-35]. Previous studies demonstrated that HCQ/CQ block the hERG potassium channels, extending ventricular repolarization and action potentials, and hence, may trigger ventricular tachyarrhythmias [36, 37]. Whether patients with underlying cardiopulmonary disease or those with de-novo cardiac injury have a greater predilection to cardiac toxicity and increased mortality after exposure to HCQ/CQ remains uncertain but plausible.
An observational study, based on patients with systemic lupus erythematosus receiving long-term treatment of HCQ, having blood concentrations of the drug within therapeutic range, concluded that HCQ does not prevent COVID-19 in these patients . Likewise, our meta-analysis did not identify any clinic benefit of using HCQ/CQ. In addition, it did not reveal any association between clinical efficacy and HCQ/CQ dose as well as disease severity. Moreover, whether these medications were administered early during the course of disease or not had no effect on clinical efficacy. Altogether, these results suggest a lack of clinical benefit with the use of HCQ/CQ in patients with COVID-19. In any case, the results of large scale, randomized clinical trials are needed to be waited before making any definitive conclusions on this controversial topic.