Since the start of the COVID19 pandemic, extensive efforts have been made by pharmaceutical companies to create monoclonal antibodies that can be administered to people during the viremic phase of illness. The goal of these drugs is to give patients antibodies prior to their body mounting a response in an effort to prevent progression of disease. We have previously reviewed the EUA drug bamlanivimab (LY-CoV555) – a monoclonal antibody. In that review, we note the absence of any difference in patient centered outcome as well as serious methodological flaws. Here we review a monoclonal antibody cocktail – REGN-COV2 (casirivimab/imdevimab).
Paper: Weinreich DM et al. REGN-COV2, a Neutralizing Antibody Cocktail, in Outpatients with COVID-19. NEJM 2020. PMID: 33332778 Free Link Clinical Trials Listing
Clinical Question: Does REGN-COV2 reduce viral loads, lead to improvements and reduce medical attendance in outpatients with COVID19?
Population: Adults > 18 years of age with a confirmed SARS-CoV2 infection within 72 hours of randomization and symptom onset no more than 7 days prior to randomization who were not hospitalized
Intervention: High (8.0 gm) or low (2.5 gm) of REGN-COV2 (casirivimab/imdevimab)
Control: Placebo
Outcome (primary): No primary endpoint was prespecified. Instead, the research team selected multiple endpoints for investigation including time-weighted average change in viral load, cumulative incidence of COVID-19 related medical attended visits (telemedicine, office visit, emergency department visit, hospitalization etc) and adverse events
Outcomes (secondary): There were 40 pre-specified secondary outcome measurements
Design: Randomized, double-blind, placebo-control, phase 1-3 trial – interim analysis
Excluded:
- Patient requires admission to the hospital
- Patient was previously involved in a research study evaluating convalescent plasma, monoclonal antibodies or IVIG in the last 3 months
- Prior current or planned future use of convalescent plasma, monoclonal antibodies, IVIG, systemic corticosteroids or other COVID19 treatments
Results:
Primary Results
- 275 patients underwent randomization
- 269 patients received either REGN-COV2 or placebo
- REGN-COV2 Low-dose: n = 92
- REGN-COV2 High-dose: n = 90
- Placebo: n = 93
- 30/275 (11%): tested negative for SARS-CoV2 by RT-PCR
- 17/275 (6%) did not have baseline viral load data
- Ultimately, only 228/275 randomized patients were analyzed
- 269 patients received either REGN-COV2 or placebo
- Antibody status prior to treatment
- 123/275 (45%) had antibodies to SARS-CoV2 spike protein
- 113/275 (41%) did not have antibodies to SARS-CoV2 spike protein
- 39/275 (14%) of patients had unknown antibody status
Critical Results
- Antibody negative at baseline: viral load was statistically significantly lower in low-dose and high-dose REGN-COV2 groups in comparison to placebo at day 7
- All patients analyzed (including those with antibodies prior to treatment): viral load was statistically significantly lower in the high-dose and in patients treated with any dose of REGN-COV2 but not with low-dose
- Clinical Efficacy (Change in medically attended visits)
- Overall: REGN-COV2 – 3% vs Placebo – 6% (-3% CI -16 to 9) – NOT STATISTICALLY SIGNIFICANT
- Antibody negative: REGN-COV2 – 6% vs Placebo – 15% (-9% CI -29 to 11) – NOT STATISTICALLY SIGNIFICANT
Strengths:
- Randomized, double-blind, placebo-controlled trial
- Broad inclusion criteria with minimal exclusions
- Study enrolled people early in disease process which should be where a med attacking the virus should be effective
Limitations:
- No primary endpoint was selected a priori. The authors clearly state, “no formal hypothesis testing was performed.”
- Patients were not consecutively enrolled
- Only 83% of enrolled patients were included in the full analysis
- The authors do not include a detailed account of medical comorbidities
- Regeneron “designed the trial; gathered the data, together with the trial investigators; and analyzed the data.” Additionally, “certain employees of Regeneron had access to unblinded early data from the trial.”
- Viral load is a non-patient centered outcome
- Medically attended visits included Telehealth, clinic or office visit, ED visit or hospitalization. These outcomes are not the same.
- The study is too small to make any determinations about safety
Discussion:
- The authors state, “because of the lack of a priori information that would allow us to correctly select end points, and because certain employees of Regeneron Pharmaceuticals (who had no role in the conduct of the trial) had access to unblinded early data from the trial as described in the protocol, no formal hypothesis testing was performed.”
- In order for research to be scientific, a hypothesis must be established and then tested.
- Essentially, this tells us that this study was designed as a fishing expedition – give the drug, track lots of different things looking for some benefit
- This is not how clinical studies should be done. This data can be used to generate hypothesis but cannot be used to direct treatment
- The authors (read Regeneron) play up the random positive outcomes they found while attempting to downplay the large number of non-statistically significant findings
- Similar to the BLAZE-1 study on bamlamivinab, the researchers combine all medically attended visits ignoring the very obvious difference between a Telehealth visit and an admission to the hospital
- Both BLAZE-1 and this study highlight the fundamental flaw in the role of monoclonal antibody therapy
- The idea is to take a patient early on in the viremic phase of disease who has not mounted an immune response and give the patient antibodies to block the virus infecting cells
- This approach has a number of problems
- Identifying this group is challenging as patients can be asymptomatic or minimally symptomatic for days while still creating antibodies (evidenced by the fact that 45% of patients had antibodies on enrollment)
- The majority of patients who become infected will recover regardless of treatment. Since we can’t identify on day 1 which group of patients will decompensate, the drug would have to be given indiscriminately. Even if it works, an enormous study would be needed to find a statistically significant benefit
- The NEJM publishing this study further signals the blatant lack of scruples the journal maintains. This is nothing more than an advertisement for Regeneron and is far short of a high-quality publication. In the past, a pharmaceutical sponsored study with no primary endpoint and no prespecified hypothesis would be relegated to the trash heap. Seeing this published in the once venerable NEJM is a clear indication that the publication’s loyalties lie with big pharma, not with clinicians or patients.
Author Conclusion: “In this interim analysis, the REGN-COV2 antibody cocktail reduced viral load, with a greater effect in patients whose immune response had not yet been initiated or who had a high viral load at baseline. Safety outcomes were similar in the combined REGN-COV2 dose groups and the placebo group.”
Our Conclusions: No conclusions can be drawn from a study published with no hypothesis. Our stance is that this cocktail, as with other monoclonal antibodies, should ONLY be given to patients in the setting of a clinical trial. On a related note, the NEJM is no longer a reliable, trusted source of high-quality clinical information.
Clinical Take Home Point: There is no role for the use of REGN-COV2 outside of a clinical trial. A large, well-done, preferably non-industry sponsored trial is necessary prior to this drug being used outside of this setting.
Thanks to Josh Farkas for sharing his thoughts on this trial.
For More Thoughts on This Topic Checkout:
- Brief19: Regeneron’s Antibody Cocktail Shows Promise in the Lab but, Little Difference for Patients
- REBEL EM: BLAZE-1: COVID-19 Neutralizing Antibody (Bamlanivimab)
- EMCrit: I’m So Confused About Bamlanivimab
Post Peer Reviewed By: Salim R. Rezaie, MD (Twitter: @srrezaie)
