key: cord-0806173-n39hdybs
authors: Sher, David J.
title: An Ounce of Prevention and a Pound of Cure: Randomized Clinical Trials of Therapeutics Against COVID-19 and an Assessment of Personal Protective Equipment and Distancing
date: 2020-10-01
journal: Int J Radiat Oncol Biol Phys
DOI: 10.1016/j.ijrobp.2020.08.001
sha: 4354368cbde9da3edf3c9d691b3f0c3750ee84ee
doc_id: 806173
cord_uid: n39hdybs

nan

Given the multitudinous clinical characteristics of SARS-CoV-2 infection (the cause of the COVID-19 pandemic), the medical profession's early response to managing afflicted patients effectively amounted to throwing everything against the wall and hoping something would stick. Clinical trials were seemingly written and initiated overnight, but in the absence of evidence of a successful therapy, clinicians were trying anything. Yet despite the mayhem and challenges in emergently developing randomized clinical trials, several groups successfully completed high-quality studies that have defined evidencebased treatment options for moderately to severely ill patients with SARS-CoV-2. This review discusses the core of what has been learned to date from this initial frenzy of investigations.

Almost any discussion of drugs in the COVID-19 era must start with hydroxychloroquine. The international excitement at the potential benefits of the medicine were catalyzed by the experience of the French virologist Didier Raoult, who publicized his purported success in treating hospitalized patients with hydroxychloroquine and azithromycin. The results were initially published as a preprint, accepted by the International Journal of Antimicrobial Agents and subsequently impugned for methodologic flaws. 1 There is a scientific rationale for the use of this drug against COVID-19: It has diverse antiviral activity in vitro, including against coronaviruses, and the drug has known immunomodulatory effects. 2 Because hydroxychloroquine has been used for decades in both malaria and autoimmune disease with an established safety profile, it was routinely given off-label to patients who received a diagnosis of SARS-CoV-2 infection, 3 and dozens of trials were initiated to study its efficacy.

The most influential randomized study published thus far evaluated the use of hydroxychloroquine as postexposure prophylaxis. Investigators at the University of Minnesota were able to quickly initiate a binational doubleblind, randomized, placebo-controlled study of hydroxycholoroquine for individuals exposed to laboratory-proven SARS-CoV-2 infection as a health care worker, first responder, or a member of the household of someone with a diagnosis of the disease. 4 Patients were enrolled within 4 days of exposure, defined as proximity to a positive individual of within 6 feet for more than 10 minutes. Individuals were considered high risk if they wore neither a facemask nor eye shield or moderate risk if they used a facemask but no eye shield. Most participants reported a high-risk exposure. The active arm involved a total of 19 tablets (200 mg/tablet), taken over 5 days using a tapering-down regimen. The primary endpoint was symptomatic illness confirmed by polymerase chain reaction, or simply COVID-19eassociated symptoms if testing was not available. Nonconfirmed cases, which comprised the vast majority of events in the study, were categorized as probable or possible based on the nature and number of symptoms. The study had 3 interim analyses, and it was stopped after the third analysis (and 821 patients enrolled) due to futility.

The study population was young (median age of 40 years) and predominantly composed of health care workers with no coexisting conditions. In brief, the study was negative. There was no statistical difference in the number of confirmed or probable COVID-19 cases between the arms, and there was no difference in symptom severity between the arms. The drug clearly increased the risk of side effects (most commonly gastrointestinal), although they were minor in severity.

Because such a high percentage of SARS-CoV-2 infections can be asymptomatic, especially in the younger population, it is disappointing that molecular testing was not mandatory during the study, but understandable given the limited testing capability at the time. It is theoretically possible that hydroxychloroquine did prevent asymptomatic cases of the infection, but at a minimum, it is clear that the drug as postexposure prophylaxis does not change the risk of symptomatic SARS-CoV-2 infection. This particular study does not provide any information on the efficacy of this drug for known-positive patients (with any degree of symptomology), although the result may have presaged the negative findings from the British RECOV-ERY trial patients, which has thus far been press-released and preprinted but not published in the peer-reviewed literature. 5 In contrast to hydroxychloroquine, a relatively cheap drug that was repurposed as a COVID-19 therapy, remdesivir is an expensive RNA polymerase inhibitor initially developed (unsuccessfully) against hepatitis C, then repurposed (mostly unsuccessfully) against the Ebola virus, and then rerepurposed against SARS-CoV-2 because it had been shown to be tolerable and had in vitro efficacy against coronaviruses. Although there was an initially negative trial from China, 6 a larger study, ACTT-1, sponsored and primarily paid for by the National Institute of Allergy and Infectious Disease, changed the landscape of managing severe COVID-19 cases. 7 This study is a platform trial of novel therapeutics, in which the first stage was this randomized, double-blind, placebo-controlled trial of remdesivir, given over 10 days or until discharge or death, whichever came first. The published study in the New England Journal of Medicine does not flesh out most of the details of the design, which were relegated to the appendix. The eligibility criteria were broad and included laboratory-confirmed SARS-CoV-2 infection plus either radiologic pulmonary infiltrates, clinical pulmonary involvement (physical examination or oxygen saturation), or oxygen/mechanical ventilation. Patients were stratified by site and disease severity, but severity was simply dichotomized as severe versus mild/moderate, with the latter requiring O2 saturation greater than 94% without tachypnea and supplemental oxygen. The primary outcome in the paper was time to recovery, defined as hospitalization without oxygen and no longer receiving medical care (ie, inpatient for infection control reasons) or better. It is notable that the initial primary outcome was the difference in clinical status at day 15, but this was changed by statisticians with no knowledge of any trial-related data; the final chosen endpoint is significantly more clinically meaningful, and its use strengthened the study.

A total of 1063 patients were randomized over a 2month period, which is a testament to both the explosive nature of the epidemic and to the investigators who believed in the importance of the trial. The interim analysis was performed after all patients were enrolled but before follow-up was completed; because the study was positive, it was released for public consumption. The mean age of the patients was approximately 59 years, and the majority were male (relevant because male patients tend to do worse with COVID-19). Importantly, over 75% of patients had at least 1 coexisting condition and over half had 2 coexisting conditions, so this cohort was high risk at baseline for disease-related mortality. Nearly 90% of patients were on at least supplemental oxygen, and one-quarter were intubated or on extracorporeal membrane oxygenation. The primary endpoint was met: The time to recovery was significantly shorter among patients given remdesivir (11 days vs 15 days; P < .0001). The formal test of interaction between baseline severity and recovery was not significant, although there was a suggestion that the patients with the most severe disease (ie, ventilated or on extracorporeal membrane oxygenation) were least likely to benefit (recovery rate ratio [RR], 0.95; 95% confidence interval [CI], 0.64-1.42, where higher is better). Similarly, patients with mild-moderate symptoms seemed to recover too quickly with placebo to benefit from the intervention (median of 5 days to recovery for both; RR, 1.09; 95% CI, 0.73-1.62). The original primary outcomedodds of improvement in the severity scale at day 15dwas also significantly improved in the remdesivir arm (odds ratio, 1.50; 95% CI, 0.47-1.04), but mortality was not significantly different, albeit favoring remdesivir (hazard ratio for death, 0.70; 95% CI, 0.47-1.04). Importantly, there were no deaths due to the treatment itself, and there were no significant differences in severe adverse events.

Thus, this remdesivir trial was a statistical success and established the drug as the first evidence-based therapeutic for COVID-19. The clinical gains for patients with protocol-defined severe disease appear real, with a decrease of 6 days in median time to recoverydmeaningful to both the patients and hospitalization rates. Yet the devil of its clinical application is in the details, especially given its limited supply and cost. Patients at either edge of the severity spectrum appeared to gain only modest, if any, clinical benefit. In addition, hospitals were allowed to deliver other treatments if specified in a policy or guideline, so it is possible that concurrent steroids (discussed later) or other therapies could have interacted with remdesivir. Over time, we will presumably learn whether a survival benefit becomes statistically significant, and I hope data on longterm disease-related morbidity become available. Because it is clear that infection can lead to significant chronic symptomatology in survivors, both pulmonary and systemic, medications that lessen this burden are also needed.

Finally, in a departure from typical protocol for this section, I will review a preprinted publication, the "Effect of Dexamethasone in Hospitalized Patients with COVID-19 e Preliminary Report." 8 Despite the absence of peer review, its impact in changing the standard of care was tremendous, and the results are highly relevant for management of hospitalized patients. The RECOVERY trial is a platform study constructed in the United Kingdom to rapidly assess potential therapeutics, including dexamethasone, hydroxychloroquine (subsequently stopped), azithromycin, tocilizumab, and convalescent plasma. The dexamethasone component included approximately 6400 patients randomized to dexamethasone (6 mg/day for 10 days) versus usual care at a 1:2 ratio. Patients were required to have suspected or confirmed SARS-CoV-2 infection for enrollment. The study was powered for overall survival at day 28 from randomization, obviously an impactful endpoint. The mean age of the population was 66.1 years (markedly older than the prior 2 studies), and 56% of the population had at least 1 comorbidity. Importantly, 24% of patients did not need oxygen at baseline, 61% did require supplemental oxygen, and 15% were mechanically ventilated.

The use of dexamethasone was a clear success, reducing the risk of overall mortality from 24.6% to 21.6% (RR, 0.83; 95% CI, 0.74-0.92), but a prespecified analysis tells the important part of the story. The greatest absolute and relative improvement was in patients who were ventilated at diagnosis (RR, 0.65; 95%, CI 0.51-0.82; absolute improvement approximately 15%), followed by those requiring oxygen (RR, 0.80; 95% CI, 0.70-0.92; absolute improvement approximately 5%). For patients who did not meet these criteria, dexamethasone was borderline harmful (RR, 1.22; 95% CI, 0.93-1.61; absolute decrement approximately 5%). Patients who had a longer duration of symptoms (ie, more than 7 days) appeared exclusively to receive the benefit, although this population was typically also in the sickest subgroup. Secondary endpoints were also met with dexamethasone, including a higher rate of discharge by 28 days and a reduced progression to mechanical ventilation.

This study essentially changed practice overnight for patients requiring some type of respiratory support, which makes sense given the therapeutic ratio of this relatively low dose of dexamethasone. The pathophysiology of SARS-CoV-2 is complex, and a simplistic conceptualization is that the infectious/viral component is the driver of the initial clinical condition, but the downstream second phase is primarily due to the cytokine storm and inflammatory response. That dexamethasone appears solely to benefit patients later in their disease course may speak to its action on the inflammatory cascade, a hypothesis supported by the apparent negative impact of steroids earlier in the infection, when the immune host response is required to fight the infection.

Both of these released randomized trials highlight some of the critical considerations in future COVID-19 trial design and interpretation, including identifying the optimal timing of a given therapy and the required degree of disease severity to gain a meaningful benefit. Moreover, the pragmatic nature of these trials, with generous, nonrestrictive enrollment criteria and clinically meaningful endpoints, were crucial for ensuring rapid accrual and reasonably generalizable, practice-changing results. These are important lessons that can apply to prospective studies across disciplines.

In conclusion, I hope the public and lay government now understand the absolutely irreplaceable nature of the randomized, controlled trial on which we can base medical decisions. Without the perseverance of these investigators and the willingness of both doctors and patients to accept placebo, we would not understand the potential benefits or harms of these drugs, and we would have made no tangible progress in our treatment paradigms. As funding mechanisms for prospective studies presumably dry up with the worldwide recession, these studies should serve as a constant reminder of the seminal importance of randomized trials. It also should not escape notice that the 2 positive results arose from adaptive platform trials. Every discipline, obviously including oncology, faces challenges in trial efficiency and yield, issues that will become more pressing over time. As shown in ACTT-1 and RECOVERY, the adaptive platform design provides unique flexibility to address the limitations of the conventional phase 3 randomized study. Platform trials are already underway in oncology, including breast cancer and glioblastoma, and for those readers interested in novel trial design, I would highly recommend a review article on this topic. 9 Although the race is on to identify other active therapeutics, these 2 studies have established remdesivir and dexamethasone as the only 2 evidence-based drug treatments against COVID-19, with hydroxychloroquine destined to be remembered more for its sociologic than scientific impact. Although these studies still leave many questions unanswered, treating physicians finally have proven therapies that may save lives. Given the vast numbers of afflicted patients and the modest benefits of these 2 medications, more therapies are clearly necessary, not just for treating severely ill individuals but also for those with mild symptoms early in the disease process. An efficacious postexposure prophylaxis would, of course, be an important advancement. There are multiple promising targets, ranging from novel antivirals to immune modulators (which may either augment the immune response such as with convalescent plasma or tamp it down to treat the cytokine storm), and prospective studies of these treatments are ongoing. Biomarker research in SARS-CoV-2 infection is in its infancy, but exploring predictors of serious infection may help stratify patients for different interventions and generate hypotheses for innovative therapeutics.

Of course, preventing infection in the first place is the most important public health directive, and Chu et al performed a systematic review and meta-analysis to estimate the efficacy of physical distance, facemasks, and eye shields to prevent coronavirus transmission to healthy individuals. 10 This systematic review returned 172 studies of severe acute respiratory syndrome, Middle East respiratory syndrome, or COVID-19, with 44 comparative papers included in the analysis. The bottom line from this analysis is that physical distancing greater than 1 m, the use of facemasks, and the use of eye protection appear to provide significant protection against virus transmission, with estimated relative improvements ranging from 0.15 (facemask) to 0.34 (eye protection) and an absolute reduction in risk of around 10%. The authors tried to distill out differences between N95-type protection versus a surgical mask, and the N95 masks do appear superior, especially in cases of exposure to aerosol-generating procedures, but the overall interaction between mask type and efficacy was only 0.09. A brachytherapist would clearly agree with the distance and barrier conclusions but would also ask whether duration of exposure to the infected patient is important, and this metaanalysis could not assess this important variable.

On the bright side, this Lancet study has no relationship to SurgiSphere, 11 but the limitations of this study must be recognized to appreciate the degree to which these results can be generalized. The types of studies reviewed in this meta-analysis are typically compromised by recall bias, and presumably there was a fair amount of heterogeneity in the nature of the exposure itself. Although nearly all of the included studies required confirmed cases, the timing of the exposure and its duration (especially in sudden acute respiratory syndrome, when patients were most infectious when symptomatic and ill) may have influenced the result. Moreover, out of the 44 included studies, only 7 studied SARS-CoV-2 specifically. For the analysis of proximity, COVID-19-affected patients comprised a total of 31 transmission events out of the 808 reported in these papers, and the respective numbers were 133 events out of 709 for the analysis of mask use. Whether these interventions have the same impact across these coronaviruses remains to be seen; therefore, although they are provocative, I would still argue these effect estimates need to be taken with a grain of salt. Especially because SARS-CoV-2 infection can be asymptomatic, it is also possible that transmission could have been higher than reported but never established.

Another key point in this study is that 34 of the papers exclusively studied transmission in the health care environment, and so it is very difficult to draw any inferences on the benefits of these social maneuvers in a community setting. Finally, 4 out of the 7 COVID-19 papers are preprints and not peer reviewed, including the study that contributed 5442 out of the 6674 individuals studied. Given the challenge in assessing physical distance and compliance in mask use retrospectively, a careful analysis of the methodology of these studies is particularly important to ensure the result is valid; thus, it is questionable whether nonepeer-reviewed studies should have been included in this meta-analysis.

In summary, these data do suggest that particularly in the health care setting, distance and physical barriersdincluding eye protectiondappear to confer protection against coronavirus transmission. These are relatively simple interventionsdprovided personal protective equipment is availabledthat can potentially save lives. Yet the limitations of this study are nontrivial, and whether the reported degree of protection is adequate to prevent SARS-CoV-2 transmission specifically cannot confidently be inferred from these data.

Hydroxychloroquine and azithromycin as a treatment of COVID-19: Results of an open-label nonrandomized clinical trial

Effects of chloroquine on viral infections: An old drug against today's diseases?

Observational study of hydroxychloroquine in hospitalized patients with Covid-19

A randomized trial of hydroxychloroquine as postexposure prophylaxis for COVID-19

No clinical benefit from use of hydroxychloroquine in hospitalised patients with covid-19

Remdesivir in adults with severe COVID-19: A randomised, double-blind, placebo-controlled, multicentre trial

Remdesivir for the treatment of COVID-19 -preliminary report

Dexamethasone in hospitalized patients with COVID-19 e Preliminary report

Author correction: Adaptive platform trials: Definition, design, conduct and reporting considerations

Physical distancing, face masks, and eye protection to prevent person-to-person transmission of SARS-COV-2 and COVID-19: A systematic review and meta-analysis

Retraction-hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: A multinational registry analysis