Perspectives on WHO’s SOLIDARITY Trial: Disappointing Findings Limit Remdesivir’s Use in COVID-19 Patients

by Sam Sun

Last week, the World Health Organization (WHO) published data from its SOLIDARITY trial on a preprint server. The SOLIDARITY trial is a global, pragmatic, adaptive, master protocol trial investigating 4 therapies for hospitalized COVID-19 patients, including remdesivir. Disappointing findings for remdesivir in the SOLIDARITY trial elicited a statement from Gilead Sciences, questioning the validity of SOLIDARITY, and emphasizing its ACTT-1 trial as the “gold standard” regarding remdesivir’s value for COVID-19.

Here, we look at remdesivir in two phases.

· First, we take a close look at the raw data from the randomized clinical trials of remdesivir in COVID-19 patients

· Second, we discuss the potential impact of the SOLIDARITY trial on the clinical use of remdesivir with critical care physicians and inDemic.org advisors, Dr. Bob Dickson and Dr. Hrishikesh Kulkarni

Before looking at the raw clinical trial data, it’s important to discuss the design and methodology of the SOLIDARITY trial. Gilead Sciences criticized the SOLIDARITY trial’s:

· Open-label design

· Prioritization of broad access, with heterogeneity in trial adoption, implementation, controls, and patient populations

Gilead’s ACTT-1 trial was double-blind, although its SIMPLE-Severe trial was open-label. The strength of an open-label trial is its simplicity and lower costs, practical advantages in a global pandemic. A blinded trial may be at lower risk of bias than an open-label trial, because it minimizes impact of investigators’ opinions on post-randomization treatment decisions and on reporting of subjective outcomes [1].

Still, we believe that SOLIDARITY’s design (and hence the validity of its findings) is at least complementary to that of ACTT-1, if not superior. SOLIDARITY, despite being open-label, is nonetheless randomized and controlled. Further, SOLIDARITY’s endpoint is overall mortality, whereas ACTT-1’s endpoint is time-to-hospital-discharge. No matter what biases SOLIDARITY investigators may’ve had, overall mortality is not subjective or open to interpretation, so an open-label design does not undermine SOLIDARITY’s overall findings. In addition, the UK RECOVERY trial also uses an open-label design, and it is arguably the most influential trial of COVID-19 therapies to date. Finally, perhaps partly due to the simplicity of SOLIDARITY’s enrollment process and open-label design, SOLIDARITY enrolled several times more patients than ACTT-1 onto its remdesivir vs. placebo arms.

As for the “heterogeneity” of SOLIDARITY, we view this as a strength of the trial. It more closely reflects real-world use of remdesivir, and given that the FDA granted an Emergency Use Authorization (EUA) to remdesivir for all hospitalized COVID-19 patients, it is appropriate to evaluate remdesivir in a heterogeneous patient population.

Looking at the patient population in SOLIDARITY by level of oxygen support, it’s striking that patients on mechanical ventilation may have been harmed by remdesivir. Similarly, in the ACTT-1 trial [2], patients on mechanical ventilation also did not benefit from remdesivir (Figure 1; excerpted from [2]).

Figure 1: Estimates of cumulative recoveries in patients receiving remdesivir vs. placebo, with separate plots (B-E) for patients with different levels of COVID-19 disease severity. 

Broadly speaking, the early stages of COVID-19 disease are dominated by virus-induced pathophysiology, whereas the late stages of COVID-19 disease are dominated by host-related pathophysiology (e.g. complement overactivation, etc.). Dr. Leon Henderson-MacLennan, an internist and advisor to inDemic.org, notes that most effective antiviral therapies (e.g. for hepatitis C, HIV, etc.) are used in the ambulatory care or outpatient setting. Hence, since remdesivir is also an antiviral, perhaps it would be more effective in the early stages of COVID-19, prior to or just after hospitalization.  

Unfortunately, the limited data on remdesivir suggests it is a poor antiviral against SARS-CoV-2. Neither ACTT-1 nor SOLIDARITY’s preprint study looked at immunologic correlates of remdesivir’s purported activity against SARS-CoV-2. However, an earlier study from Wuhan, China looked at the viral load after treatment with remdesivir vs. placebo [3]. Viral load in the upper and lower respiratory tract decreased at similar rates for patients treated with remdesivir vs. placebo (Figure 2; excerpted from [3]).    

Figure 2: Viral load by quantitative PCR on the upper respiratory tract specimens (A) and lower respiratory tract specimens (B). 

In contrast, another class of “antivirals”, the antibody cocktails against SARS-CoV-2, clearly reduce viral load compared to placebo. For example, Lilly’s antibody cocktail (LY-CoV555 and LY-CoV016) reduced viral load compared to placebo in an interim analysis of the BLAZE-1 trial of outpatients with COVID-19 (Figure 3).  

Figure 3: Lilly’s antibody cocktail reduces viral load compared to placebo in outpatients with COVID-19. 

Regeneron has released similar data for its antibody cocktail, REGN-COV2, showing that it reduces viral load compared to placebo in outpatients with COVID-19 (Figure 4).  

Figure 4: Regeneron’s antibody cocktail reduces viral load compared to placebo, particularly in patients with a high baseline level of SARS-CoV-2 virus. 

It is difficult to claim that remdesivir acts as an “antiviral” against SARS-CoV-2 without data that it reduces viral load more effectively that placebo.

Next, we will discuss remdesivir and the SOLIDARITY trial with critical care physicians and inDemic.org advisors, Dr. Bob Dickson and Dr. Hrishikesh Kulkarni. 


Previously, we discussed the SOLIDARITY trial, which showed that for hospitalized adults with COVID-19, remdesivir plus local standard of care (SOC) did not improve mortality or hospitalization compared to local SOC alone. On October 21, 2020, the FDA granted regulatory approval to remdesivir for use in adult and pediatric patients 12 years or older with COVID-19 requiring hospitalization. In its Q&A on remdesivir, the FDA said that “the SOLIDARITY results do not refute these findings of…shorter time to recovery and better odds of clinical improvement…with remdesivir.”

Not all physicians and health agencies agree with the FDA’s synthesis of Gilead’s trials and the SOLIDARITY trial. In fact, we predict that the next catalyst for remdesivir is likely to be guidelines from the WHO, NIH, and professional societies (e.g. Society for Critical Care Medicine) regarding its use in hospitalized patients. Here, as a preview, we discuss the potential impact of the SOLIDARITY trial on the clinical use of remdesivir with critical care physicians and inDemic.org advisors, Dr. Bob Dickson and Dr. Hrishikesh Kulkarni.


What do you think of the SOLIDARITY trial? 

Dr. Dickson: I think it’s a spectacular effort, the kind of innovative, international, adequately-powered study that one would hope we’d be swimming in by this point with COVID-19 given the number of cases & the urgency. 
I think it (again) closes the door on hydroxychloroquine, though at this point, the door has been closed on it over and over. By this point, I’m not sure any question in pulmonary and critical care research has been more definitively answered than: “does hydroxychloroquine improve outcomes in patients with SARS-CoV-2 infection?” I really don’t see any justification for continuing hydroxychloroquine trials at this point, especially given the opportunity costs! Patients are still being enrolled on trials for an antimalarial therapy with effectively 0% chance of showing benefit, when we still don’t have good data to guide anticoagulation practices for patients with COVID-19 (i.e. prophylactic dose vs. more aggressive therapeutic dosing).
My expectations were already low for lopinavir and interferon based on earlier (smaller) studies, but the SOLIDARITY results do provide some clarity and reassurance.

Great. Let’s move on to remdesivir. With SOLIDARITY being a negative study with overall mortality as the primary endpoint, and ACTT-1 being a positive study with time-to-recovery as the primary endpoint, how do you reconcile their findings?

Dr. Dickson: We should always be wary of “proxy” or “surrogate” endpoints in these studies. In ACTT-1, the primary endpoint was “time to recovery” using an ordinal scale driven mainly by whether patients were hospitalized or not, and how much oxygen support they need. They reported mortality as a secondary endpoint, which suggested separation at day 15 that was less clear at day 29, as the confidence interval for the hazard ratio crossed 1.0.
While I understand the rationale, I’m extremely skeptical of “soft” surrogate endpoints, especially when “hard” endpoints like 28-day mortality can be assessed. There is a graveyard of therapies that improve surrogate endpoints in critical care (like oxygenation or time-to-discontinuation-of-pressors) that either don’t improve mortality or actually result in worse mortality. We wish that the diseases of critical illness fell into neat, linear causal pathways and we could use easily measurable variables in the causal pathway as surrogate endpoints, but that’s never worked. And our therapies always have off-target consequences that may cause harm. 
I like that the SOLIDARITY authors did us the favor of baking a meta-analysis into their manuscript (Figure 5).
Figure 5: A meta-analysis of mortality in trials of random allocation of hospitalized COVID-19 patients to remdesivir vs. control (from SOLIDARITY preprint)
There were simply WAY more patients in SOLIDARITY than ACTT-1. The low/hi-flow O2 arm in SOLIDARITY was 5-6x the size of the equivalent arms in ACTT-1. I’m always wary of over-interpreting subgroups, but it does help to see how aberrant the “low-flow O2” arm of ACTT-1 was compared to all other subgroups, as it’s the only subgroup that suggests a mortality benefit to remdesivir. For all three “invasive ventilation” subgroups, the point estimate suggests a mortality detriment to remdesivir.
Again, not to over-interpret, but it really cuts against ANY argument that remdesivir helps patients sick enough to need substantial oxygen support. I can’t speak to the FDA’s approval of remdesivir, but I see no indication for its use in anyone sick enough to be in my ICU.

As you know, the FDA used the ACTT-1 trial extensively in its review of remdesivir. This is quite a loaded question, but which is a “better” or “more trustworthy” trial, ACTT-1 or SOLIDARITY? 

Dr. Dickson: I appreciate all RCT data in the COVID-19 era, and don’t mean to belittle the work that went into ACTT-1, but I will say that I will almost always give more weight to hard endpoints like 28-day mortality than soft endpoints like “time to clinical recovery.” I’ll also say that larger sample sizes are important for minimizing risk of both Type II AND Type I errors (false positives).

Will the SOLIDARITY results change your use of remdesivir for hospitalized COVID-19 patients?

Dr. Dickson: I will not be using remdesivir in COVID-19 patients requiring high-flow O2 or mechanical ventilation. I was already leaning this way based on the ACTT-1 subgroup analysis, and SOLIDARITY has made me more confident.

Let’s turn to Dr. Kulkarni.  Any thoughts on how the SOLIDARITY trial will affect remdesivir’s use in hospitals, or in a clinical research setting?

Dr. Kulkarni: Dr. Dickson has already covered most of what I would say (plus a lot more). 
It also concerns me to some extent that remdesivir is already being considered standard of care in some hospitals for many patients, and as a part of most phase 3 clinical trials for COVID-19 therapies. And SOLIDARITY’s results should make us take a step back and re-evaluate this approach of making it the “standard of care”. Given some of the points that Dr. Dickson brought up, including “the point estimate suggests a mortality detriment to remdesivir for patients on invasive ventilation” – it makes me wonder, will making remdesivir the standard of care in ongoing phase 3 trials mask a potential benefit of the drugs being tested. Moreover, remdesivir is not used for every critically ill patient, for example (e.g. it is not used in those with a creatinine clearance under 30). Thus, this could result in the interpretation of clinical trial results becoming harder.

Great, thank you so much, Dr. Dickson and Dr. Kulkarni!  We appreciate your clinical perspectives.

inDemic.org will be revisiting remdesivir again after SOLIDARITY results are formally published in NEJM, and after the WHO and other organizations publish guidelines on use of remdesivir in hospitalized COVID-19 patients.

References

[1] Beigel, John H., et al. “Remdesivir for the treatment of Covid-19.” New England Journal of Medicine (2020).

[2] Beyer‐Westendorf, J., and H. Büller. “External and internal validity of open label or double‐blind trials in oral anticoagulation: better, worse or just different?.” Journal of Thrombosis and Haemostasis 9.11 (2011): 2153-2158.

[3] Wang, Yeming, et al. “Remdesivir in adults with severe COVID-19: a randomised, double-blind, placebo-controlled, multicentre trial.” The Lancet (2020).

© 2020 inDemic. All Rights Reserved.