scientists wearing masks working in labsShare on Pinterest
New research questions the study design of clinical trials that test the benefits of antivirals for COVID-19. ER Productions Limited/Getty Images
  • Observational and randomized clinical trials evaluating the efficacy of antiviral drugs for COVID-19 treatment have reported negative or inconsistent findings.
  • A recent modeling study suggests that observational studies may show inconsistent results due to variation in viral dynamics among COVID-19 patients.
  • The model also suggested that the efficacy of antiviral drugs was optimal when starting treatment within a day of developing COVID-19 symptoms and diminished after this period.
  • Negative results from antiviral clinical trials may be due to a lack of consideration for the time of onset of COVID-19 symptoms while designing the study or late timing of treatment initiation.

While wealthy nations have been able to immunize a sizeable proportion of their populations with COVID-19 vaccines, many countries have been unable to access them.

Moreover, while vaccines greatly reduce the risk of getting ill from SARS-CoV-2, no vaccine is 100% effective, and breakthrough infections do occur. So, we still need antiviral treatment options for anyone who contracts the virus and gets ill.

While some drugs have shown promise in combating SARS-CoV-2 in cell lines and animal models, human studies testing the efficacy of these antivirals have reported inconsistent or negative findings.

For more advice on COVID-19 prevention and treatment, visit our coronavirus hub.

These human studies included data from either compassionate use programs or clinical trials.

Compassionate use programs involve using an experimental drug to treat a patient with a life threatening condition — such as severe COVID-19 — when doctors have exhausted alternatives.

In contrast to this approach, randomized controlled trials are ideal for reducing biases and evaluating the effectiveness of a drug.

Some scientists have attributed the lack of consistent results from these studies to their imperfect design.

A recent study investigated the specific reasons for the failure of these studies using mathematical modeling.

The research, which appears in PLOS Medicine, found that individual differences in response to SARS-CoV-2 infection are likely to bias the results of compassionate use programs.

The study also found that antiviral drugs may have limited efficacy against SARS-CoV-2 unless treatment was initiated immediately after developing COVID-19 symptoms. Failure to adequately account for the effect of the timing of antiviral treatment initiation may therefore be responsible for unsuccessful antiviral clinical trials.

To define the reasons behind the lack of success of antiviral drugs in human studies, the researchers initially investigated individual differences in response to the virus.

The team used published clinical data from 30 individuals with a SARS-CoV-2 infection to assess changes in the viral load or amount of virus in the upper respiratory tract over time. Since the viral load data from these patients were not complete, the researchers used a mathematical model to reconstruct changes in viral load over time.

By reconstructing the viral dynamics, the team was able to estimate parameters, including the replication rate of the virus and the rate of decline of viral load or decay rate for each patient.

They found that the decay rate differed significantly among these patients. Based on these differences between individuals, the researchers categorized participants into three groups: rapid, medium, and slow decay groups.

The researchers also modeled the duration of viral shedding — the release of viral particles — in each patient.

The team observed that the duration of viral shedding mirrored the trend in viral decay rates among patients. In other words, the slow decay group shed the virus for the longest time, while the rapid decay group shed the virus for the shortest period.

The duration of viral shedding has links with the severity of COVID-19 and can be useful in assessing the effectiveness of treatments in improving clinical outcomes.

Therefore, heterogeneity in viral shedding duration among COVID-19 patients suggests individual differences in response to the virus and the severity of the disease.

Healthcare practitioners initiate treatment depending upon their assessment of illness severity in compassionate use programs. The severity of the disease also determines the eventual clinical outcome.

Therefore, the inconsistent results that experts observed in compassionate use programs may be due to the variability in the severity of COVID-19 and the heterogeneous response of individuals to the disease.

The timing of antiviral treatment initiation after symptom onset is known to influence clinical outcomes in other viral illnesses such as influenza.

Therefore, the researchers used their model to understand how the three groups with different viral load decay rates would respond to different treatment initiation times.

They considered treatment initiation times between 0.5 and 5 days after COVID-19 symptom onset. This is because the virus levels in the upper respiratory tract peak during this timeframe.

The researchers found that initiation of a potent antiviral after 0.5 days but before 5 days following symptom onset was effective in reducing viral load and viral shedding duration in all groups.

The researchers then used a mathematical model to mimic a randomized control trial, where patients would receive either an antiviral drug or a placebo.

The team found that the sample size required to produce a statistically meaningful difference in viral shedding duration between the antiviral and placebo groups was dependent on the time of treatment initiation after symptom onset.

While the researchers did not specify the time for treatment initiation after symptom onset, the clinical trial would require at least 11,670 participants in both the treatment and placebo groups to detect a statistically significant antiviral effect.

However, if the model assumed that patients would enroll in the clinical trial within the first day of developing COVID-19 symptoms, a sample size of 458 patients in each group would be sufficient to detect the impact of a highly effective drug.

The researchers obtained the sample size data by simulating the effects of an antiviral drug that would reduce viral replication by 99%. They caution that experts need larger sample sizes to obtain statistically significant results in studies involving antiviral drugs with a lower inhibition rate.

Therefore, achieving a statistically significant antiviral effect without considering the timing of treatment initiation would require an unreasonably large sample size.

The researchers recommend that early enrolment in clinical trials after COVID-19 symptom onset or only including patients in clinical trials who develop symptoms in a specific time window would be more likely to result in studies that could detect a genuine antiviral effect.

The authors conclude that their model emphasizes that, besides the dose and mechanism of action, the time of treatment initiation and individual differences in response to the virus can influence the efficacy of an antiviral drug.

Considering the strengths of the study, the researchers note, “We used clinical data from [patients with a SARS-CoV-2 infection] for the simulation. Thus, our numerical results are realistic and directly interpretable for drug development for SARS-CoV-2. In other words, our approach is flexible and can be applied to other antiviral drugs for other diseases by replacing the dataset.”

However, the authors also state that their model was minimalist and based only on viral load data. They note:

“At such time that further data and appropriate scientific information about infection dynamics becomes available, more complex models may be able to capture additional details of within-host viral dynamics.”

“We did not use viral load data under treatment to evaluate our model because such data were not sufficiently available. Estimating antiviral effects from such data and using that in the sample size calculation would strengthen our approach, but we need to wait until such data are accumulated,” elaborated the researchers.