The preliminary report of the Adaptive COVID-19 Treatment Trial (ACTT-1), a multi-center, placebo-controlled RCT on remdesivir is here!1 The one we’ve all been waiting for!
Patients were recruited from 60 sites in several countries. Inclusion criteria were patients who were hospitalized due to COVID-19 with evidence of lower respiratory tract infection, defined in terms of meeting at least one of the following criteria:
- Radiographic pulmonary infiltrates.
- Oxygen saturation <94% or requiring supplemental oxygen.
- Mechanical ventilation or ECMO.
Important exclusion criteria are as follows. These are stealthily hidden in the supplemental appendix, so they may be widely overlooked:
- AST or ALT over 5 times the upper limit of normal.
- “Impaired renal function as determined by calculating an estimated glomerular filtration rate” (no actual cutoff value for the acceptable glomerular filtration rate is stated).
- Need for hemodialysis or hemofiltration.
- Pregnancy or breast-feeding.
Baseline characteristics are shown below. Groups were generally well matched, although intubation was a bit more common in the placebo group:
Primary endpoint: Time to clinical recovery
The primary endpoint was the time to clinical recovery (defined as either discharge from the hospital or remaining in the hospital solely for the purpose of isolation). Median time to recovery was 11 days in the remdesivir group versus 15 days in the placebo group (p<0.001).
So it looks like remdesivir accelerates recovery, which is nice. Getting out of hospital earlier is good. However, a much more important endpoint is whether or not patients recover (regardless of timing). The latest numeric data point provided within the study is absolute recovery by day 14. More patients in the remdesivir group did recover within that 14 day period (334/538 (62%) in the remdesivir group versus 273/521 (52%) in the placebo group; p= 0.002).
The key question is then whether this benefit in recovery rates is durable over time, or whether it is a short-lived phenomenon. This is important to sort out:
- If recovery differences are maintained over time, that suggests that remdesivir causes some patients to recover who wouldn’t have recovered otherwise. That’s huge.
- If recovery differences disappear over time, this suggests that remdesivir accelerates the recovery among some patients who would have recovered anyway (without ultimately changing any patient’s fate). That’s much less exciting.
Without additional data it’s impossible to distinguish between these possibilities. The curves shown above seem to start converging, suggesting that benefit may be lost over time. However, some data is still missing (because many patients had not been followed up for 28 days), so it’s impossible to tell.
Secondary endpoint: Mortality
The study reports the mortality numbers at 14 days, a rather early timepoint. Depending on how you crunch these numbers, it may look like remdesivir caused a mortality benefit. However, the authors were appropriately careful not to make any claims about mortality benefit.
The reason not to get excited about these mortality numbers at 14 days is that this looks like a time-limited anomaly. Examining the mortality curve from the supplemental data suggests that mortality curves diverge around 14 days, but then converge at around 25 days. So, when the 28-day mortality numbers are finalized, it’s likely that there won’t be any sustained mortality benefit. Unfortunately, it’s too soon to be sure.
Of course, absence of mortality benefit doesn’t indicate that a medication isn’t beneficial. It’s extremely difficult to achieve proven mortality benefit in critical care trials (discussed here and here). Mortality may not be a terrific outcome to use in COVID-19 patients, because it lumps together patients who truly recover with those who require tracheostomy and chronic ventilation.
Secondary endpoint: Ordinal outcomes at day #14
The original primary endpoint was ordinal endpoint after two weeks. This was changed to a time-to-recovery endpoint above. This change was made based on the realization that COVID-19 could be a slower disease process than the investigators initially thought (prior to seeing any of the data).
The investigators got a lot of flak about changing the endpoint, but I think switching the endpoints was a sensible decision (the mistake might have been to originally use an ordinal endpoint at all). The road to hell is paved with ordinal endpoints, so any de-emphasis of these monstrosities is appreciated.
The ordinal endpoint does reveal benefit from remdesivir at day #14 (p=0.001). So regardless of whether or not the primary endpoint was switched or not, it still would have been positive.
Remdesivir hastened recovery in mildly ill patients. There was no signal for benefit among patients on high-flow oxygen, noninvasive ventilation, or invasive ventilation. This is consistent with the concept that later-phase COVID-19 is largely an immunopathological phenomenon (so anti-viral therapy may not help patients with advanced illness very much).
These subgroup analyses are under-powered and ultimately inconclusive. Thus, this data shouldn’t be interpreted to mean that intubated patients shouldn’t be treated with remdesivir. However, this does cast shade on the popularized strategy of allocating limited stores of remdesivir to the sickest patients.
There are no signals of harm due to remdesivir. This is consistent with the prior trial in the Lancet.
The trial design at clinicaltrials.gov lists 28 secondary endpoints. The supplemental appendix describes collection of blood tests and nasopharyngeal swabs for viral load:
Much of this data is conspicuously absent from the manuscript, including:
- All laboratory tests seem to be missing in action. The entire manuscript and supplemental appendix lacks a single actual blood test value. This is very strange – hopefully this data will be forthcoming in a subsequent publication. Surely some baseline lab values were collected – what happened to them??
- Viral load data is absent.
- Duration of oxygen use, non-invasive ventilation, or high-flow oxygen is absent.
- Incidence of intubation and escalation of oxygen therapy is absent.
The absence of any data on creatinine or renal function is particularly bothersome. Patients with renal insufficiency were excluded from the trial (apparently without any well-defined GFR cutoff?). Lack of a GFR cutoff or information about patients’ GFR values makes it difficult to know which patients can safely receive remdesivir. Enthusiasm about remdesivir will inevitably lead to its administration among patients with renal insufficiency who would have been excluded from the trial, among whom the safety of remdesivir is unknown.
The presented data to date suggests that remdesivir hastens recovery by a few days. It remains unclear whether remdesivir has a durable effect to increase the number of patients who recover. Available evidence suggests that remdesivir probably has no impact on mortality.
This study is compatible with a prior study in Lancet by Wang et al, which found a 1.2-day reduction in the median time to clinical improvement in patients treated with remdesivir.2 Importantly, Wang found no impact of remdesivir on either mortality or viral load. So, overall remdesivir faired similarly in both studies, causing a small acceleration in clinical recovery without exerting major effects on outcome.
The value of an intravenous drug which accelerates recovery by a few days is unclear:
- The obvious part: Some patients may benefit from the drug, with reduced hospital length of stay.
- The less obvious part: Some patients who are on the borderline between being admitted or discharged from the emergency department may end up getting admitted for the purpose of receiving IV remdesivir! Thus, the availability of remdesivir will increase the hospital length of stay for these patients!
For now, the benefits of remdesivir may likely outweigh risks among hospitalized COVID-19 patients with adequate renal function. However, further investigation is also urgently needed. Given the persistence of many basic questions about remdesivir, it seems premature to declare that remdesivir should be the standard of care for COVID-19.
Remdesivir: The opiate of the people?
Regardless of whether remdesivir works, the perception that remdesivir works could have enormous public health implications. Up until now, there has been no disease-specific treatment for mild COVID-19. As such, minimally ill people were often encouraged not to seek evaluation at all. Why bother? Even if they had COVID-19, the treatment would be the same (go home, have some chicken soup, etc.).
The availability of a disease-specific therapy could change all of that. If we have a therapy, that would encourage people to seek early testing and possibly hospital admission for IV remdesivir. This could have very broad-reaching consequences, including both threats and opportunities:
- Improved disease tracking: Encouraging people to seek COVID-19 testing early could help track the disease and control spread (especially if paired with contact tracing and other public health interventions).
- Isolation of infected people: Early admission for infected people could help limit disease spread, by sequestering them from the community during the period of maximal contagiousness.
- Early intervention for infected patients: Early admission of at-risk patients could promote early stabilization (e.g., awake proning, volume optimization, DVT prophylaxis, and discontinuation of potentially harmful medications such as some anti-hypertensives and NSAIDs) – potentially preventing patients from progressing to more severe illness.
- Improved understanding of early illness: studies of patients admitted earlier in the disease course could help identify factors involved in illness progression, potentially improving treatment.
- Paralysis of emergency and inpatient services: An increased volume of patients seeking evaluation and inpatient admission could rapidly congest hospitals in hot zones. For example, this could exacerbate shortages in PPE.
If hospitals were capable of meeting increased demand for evaluation and therapy, this might have a positive effect on COVID-19. However, it’s unclear whether the benefits outweigh the costs (especially compared to simpler public health interventions which could have more favorable cost/benefit ratios).
- The ACTT-1 preliminary report suggests that remdesivir accelerates recovery by a few days. It remains unclear whether remdesivir might affect long-term outcomes (available evidence suggests that it might not, but long-term follow-up data is limited). Additionally, lots of information is missing from the manuscript, impeding a full understanding of the study (e.g., there are no laboratory values).
- Patients with renal failure were excluded from the ACTT trial, due to concern for possible nephrotoxicity. This restriction should not be forgotten as remdesivir use expands virally (undergoing the inevitable indication creep).
- Available evidence suggests that benefits likely outweigh risks for the use of remdesivir in hospitalized COVID-19 patients without renal insufficiency. However, more complete data on the ACTT trial and additional RCTs are needed before concluding that remdesivir has robust benefit and thus deserves to be considered the standard of care.
- The availability of early, disease-specific therapy for COVID-19 could have profound implications for how COVID-19 patients interact with the healthcare system. Such effects (both positive and negative) could easily overshadow the marginal efficacy of remdesivir.
- Wang RCT on Remdesivir
- Two more trials just published on Remdesivir (Salim Rezaie, RebelEM)
- Compassionate use case series on Remdesivir in NEJM (PulmCrit)
- 1.Beigel JH, Tomashek KM, Dodd LE, et al. Remdesivir for the Treatment of Covid-19 — Preliminary Report. N Engl J Med. Published online May 22, 2020. doi:10.1056/nejmoa2007764
- 2.Wang Y, Zhang D, Du G, et al. Remdesivir in adults with severe COVID-19: a randomised, double-blind, placebo-controlled, multicentre trial. The Lancet. Published online May 2020:1569-1578. doi:10.1016/s0140-6736(20)31022-9