EPIC-HR Evaluates Nirmatrelvir-Ritonavir for COVID-19
“When she smiles without her teeth in place she looks like a witch. But with them in her mouth she’s very pretty.”
The SARS-CoV-2 genome encodes structural proteins [i.e., spike, envelope, membrane, nucleocapsid] and polyproteins [i.e., pp1a and pp1ab]. When the guts of SARS-CoV-2 spew into a cell, viral mRNA acts as a template for polyprotein inception. During and after translation of the polyproteins, their existence is internally modified – cleaved and processed into the non-structural proteins by an embedded main viral protease [Mpro, also known as ‘non-structural protein 5’, 3CL]. In addition to MPro, the other non-structural proteins are existential features of the virus as they become replication centres. Given that Mpro is vital to viral multiplication, that Mpro is unaffected by spike protein mutations and that protease inhibition has been a successful approach for both HIV and hepatitis C, opposing Mpro is an attractive pharmacological approach.
PF-07321332, also known as Nirmatrelvir, is a SARS-CoV-2 main protease [Mpro] inhibitor. Nirmatrelvir potently inhibits Mpro from all coronavirus types known to infect humans, including the particularly vexing beta-coronaviruses [e.g., SARS-CoV-1, SARS-CoV-2, MERS-CoV, etc.]. Further, Nirmatrelvir has no known inhibitory effects on mammalian proteases and its plasma concentration is ‘boosted’ by blocking cytochrome P450 3A4. While there are many medications that alter CYP 3A4 activity, the inhibitor ritonavir is used to enhance the antiviral medications lopinavir and darunavir with success. Accordingly, the combination of the Mpro inhibitor nirmatrelvir with its pharmacological booster ritonavir, is Paxlovid [NIR-RIT].
The first large, now completed, clinical evaluation of NIR-RIT is the EPIC-HR study, or the Evaluation of Protease Inhibition for COVID-19 in High-Risk patients. The EPIC-HR results are currently available in press release form, though this study forms the basis of the emergency use authorization in the United States granted in December 2021.
What they did
2246 adult participants were randomized to either NIR-RIT or placebo. The study was double-blinded and randomization was 1:1.
Patients were included if they had: confirmed SARS-CoV-2 infection within 5 days prior to randomization, signs or symptoms of COVID-19 within 5 days of randomization and had at least 1 sign or symptom present on the day of randomization, agreed to use highly effective contraception if fertile, at least 1 underlying medical condition associated with increased risk of severe COVID-19.
Patients were excluded if they had: received, or were expected to receive, SARS-CoV2 vaccine prior to Day 34 follow up visit, previously had SARS-CoV-2 infection, a history or, or needed, hospitalization for COVID-19 treatment, oxygen saturation of less than 92%, active liver disease, dialysis or moderate-to-severe renal insufficiency, HIV with a viral load more than 400 copies/mL, current use of medications dependent upon or inducers of CYP 3A4, pregnancy, received convalescent plasma, participation in another COVID-19 interventional trial.
The primary outcome measure was the proportion of patients with COVID-19-related hospitalization or death from any cause through 28 days. The secondary outcomes evaluated were many, including: adverse events, treatment-emergent adverse events, duration and severity of symptoms, all-cause death, nasal swab viral titres, COVID-19-related medical visits other than hospitalization, number of days in the hospital and ICU for COVID-19 therapy if hospitalized.
What is reported
For patients randomized within 3 days of symptom onset, 0.7% of patients who received NIR-RIT were hospitalized or died due to COVID-19 through day 28 following randomization [i.e., 5/697 hospitalized, no deaths]. In those who received placebo, 6.5% were hospitalized [i.e., 44/682 hospitalized with 9 subsequent deaths]; this was clinically and statistically-significant.
In a secondary endpoint, NIR-RIT reduced the risk of hospitalization or death for any cause compared to placebo in patients treated within five days of symptom onset – 0.8% [i.e., 8/1039 hospitalized with no deaths], compared to 6.3% of patients who received placebo [i.e., 66/1046 hospitalized with 12 subsequent deaths], this was also statistically-significant. In the overall study population through Day 28, no deaths from any cause were reported in patients who received NIR-RIT as compared to 12 [1.2%] deaths in patients who received placebo.
Treatment-emergent adverse events were comparable between NIR-RIT and placebo, most of which were mild in intensity. Fewer serious adverse events and discontinuation of study drug due to adverse events were observed in patients received NIR-RIT, compared to placebo.
These thoughts are based upon the press release for NIR-RIT, so it is best to await the full, peer-reviewed publication before an in-depth critique and final conclusions. Nevertheless, the headlines of ‘90% reduction in hospitalization and mortality’ have been widely disseminated, so it is important to consider the source of these statements.
As in many studies, what is reported is the relative risk reduction [RRR]; that is, the reduction in events in the study arm relative to the events observed in the placebo arm. RRR is different from the absolute risk reduction [ARR] which is simply the mathematical difference in outcome between the two arms. The ARR essentially answers the question: if I treat 100 patients with placebo and 100 patients with active drug, how many receiving the active drug will derive benefit? The difference between the RRR and ARR can be notable. For example, if the observed outcome for placebo is 1% and 0.1% for therapy X, this is a 90% relative risk reduction but an absolute risk reduction of only 0.9%. In other words, treating 100 patients with the drug X prevents the outcome in 0.9 patients. This is the genesis of the number needed to treat [i.e., NNT = 100/ARR]; in this hypothetical example, the NNT is 111.
For the primary outcome reported above, the ARR in hospitalization or death due to COVID-19 [for those randomized within 3 days, more below] afforded by NIR-RIT is 6.5% minus 0.7% or 5.8% or a NNT of approximately 17. This is, nonetheless, an impressive number. Stated another way, if you treat 17 patients meeting the inclusion and exclusion criteria for EPIC-HR, an instance of hospitalization or death due to COVID-19 is prevented.
Yet, what caught my eye in the reported results were the denominators. Only the final peer-reviewed manuscript will shed light here. The denominators of the reported primary outcome were 697 and 692 for NIR-RIT and placebo, respectively. Yet 2246 patients were randomized in EPIC-HR. It is suggested by the wording that the unexpectedly small denominators for the primary outcome are because they only reported those who were randomized within 3 days of symptom onset while the inclusion criterion was anyone within 5 days of symptom onset. Ostensibly, the trial was powered based on the 5-day inclusion criterion, so this reporting is curious. Was the primary outcome not statistically different when all patients randomized within 5 days were considered? This is all-the-more interesting because in the secondary outcome [i.e., death or hospitalization by any cause at day 28], they report a significant difference between the groups for anyone with symptoms within 5 days. In these groups, the denominators sum to 2085, which is still less than the 2246 planned.
One final thought, regardless of any of the above, is that the inclusion and exclusion criteria are kept in mind when translating EPIC-HR clinically. The enrolled patients were outpatients, SARS-CoV-2 immune naïve [i.e., unvaccinated, without history of infection], at low risk for ritonavir complications [e.g., CYP 3A4 inhibition], had at least one medical comorbidity that increased their COVID-19 risk and were with oxygen saturations above 92%.
As Judy Blume reminds us, a smile can look quite different depending on perspective. How the results of EPIC-HR truly appear remain until the final manuscript is published.
Best wishes, warmest regards,