A patient, wearing an oxygen mask sits outside Lok Nayak Jai Prakash Narayan Hospital, New Delhi, April 22, 2021. Photo: Reuters/Adnan Abidi
Mumbai: On April 23, 2021, the Ahmedabad-based pharmaceutical giant Zydus Cadila announced that its drug Virafin had received restricted, emergency use approval from the Drug Controller General of India (DCGI).
‘Virafin’ is pegylated interferon alpha-2b. Interferons are signalling proteins that help the body’s immune system defend against viral infections. Before it was repurposed for COVID-19, pegylated interferon alpha 2b was used to help treat Hepatitis C.
According to the company’s registration on the Clinical Trial Registry of India, Virafin’s phase 2 clinical trial was conducted with 40 participants, and phase 3, with 250 patients. The company’s press release also suggests that the approval was granted based on data from both the trials.
However, The Wire Science could not find any preprint or peer reviewed papers corresponding to the phase 3 trial.
The paper describing the phase 2 trial results was published in the International Journal of Infectious Diseases, in its April 2021 issue. It was conducted with 40 patients who had moderate COVID-19 – 20 of them were assigned to the control arm and 20 to the treatment arm.
According to the paper, the researchers ascertained the ‘presence’ of moderate COVID-19 thus: “RT-PCR confirmed SARS-CoV-2 infection, pneumonia with no signs of severe disease, respiratory rate 15-30 breaths/min, oxygen saturation 90%–94%.”
The 20 in the control arm received standard of care and the 20 in the treatment arm received Virafin along with standard of care.
The primary endpoint – the principal outcome the researchers were checking for – was whether a patient had a score lower than 2, on a seven-point scale designed to categorise clinical improvement, after 15 days.
(Note: The paper states that the researchers used the WHO’s seven-point scale – but the WHO’s scale has eight points. Some other large trials, including one involving a combination of lopinavir and ritonavir, have used a modified seven-point scale, similar to the one here, in which WHO’s points 6 and 7 are merged into one.)
This trial was ‘open label’. That is, the clinician who decided the patient’s score also knew if the patient belonged to the treatment arm or the control arm. “Open-label trials are heavily prone to biases,” Sahaj Rathi, a faculty member at the Mahatma Gandhi Institute of Medical Sciences, Sevagram, had told The Wire Science earlier. These biases stand to be amplified if the clinician has a favourable view of the drug or treatment being tested – in this case Virafin.
Primary endpoint results
After 15 days, 19 out of 20 patients in the treatment arm and 13 out of 19 patients in the control arm reportedly showed improvement – with scores of lower than 2 on the scale. (According to the paper, the 20th participant in the control arm withdrew consent midway and dropped out.) The p-value for this endpoint was 0.0436.
The p-value is a measure of how reliable a result is against chance. A value lower than 0.05 is considered to be statistically significant – and the difference between the drug and the standard of care is considered unlikely to be a coincidence.
But there is a catch: if 14, instead of 13, participants out of 19 in the control arm had shown improvement, the p-value suddenly jumps to 0.0915. This appreciably more than 0.05 and renders the result statistically insignificant.
Put another way, the primary endpoint’s statistical significance hinges on a clinical decision as to whether a single patient scored 1 or 2 on a subjective scale. Together with the trial being open-label – i.e. the clinician knowing which group each patient belongs to – a single instance of bias can dramatically swing the result. And when such a possibility exists, the clinical trial is not considered to be robust.
In technical terms, when a single patient can reverse the results of a study, the study is said to have a value of 1 – considered bad – on the fragility index.
Out of 19 mild adverse events, the researchers reported 11 were from the treatment arm and eight from the control arm.
Along with safety, the trial used four other secondary endpoints:
1. Qualitative RT PCR
2. Duration of supplemental oxygen
3. Requirement of Mechanical Ventilation
4. Duration of hospitalisation
There was no statistically significant difference between the treatment and the control arms for three out of these four endpoints; the exception was qualitative RT PCR on days 7 and 14 after receiving standard of care or Virafin + standard of care.
(At least according to one study, qualitative RT PCR refers to whether a test has identified the presence of the viral genome in the patient sample – ‘yes’ or ‘no’. Unlike quantitative RT PCR, the qualitative variety doesn’t indicate whether the viral particles are viable, what the viral load could be, etc.)
There were also no differences between the two arms on seven biomarkers (although the supplementary data mentions that the researchers assessed 14 biomarkers). A biomarker is a measurable parameter that doctors and researchers use to understand, say, the effects of a disease on the body. The biomarkers in the Virafin phase 2 trial included D-dimer, Interleukin-6 and C-reactive protein levels.
As it happens, this trial is not the first one to study interferons vis-à-vis COVID-19. In the WHO-sponsored SOLIDARITY trial conducted last year, researchers studied the effects of interferon beta with 4,100 COVID-19 patients. Interferon beta is similar to Virafin. The trial found that the drug didn’t improve the mortality rate, postpone the initiation of ventilation or reduce duration of hospitalisation.
In another instance, researchers have been studying the effects of interferon alpha 2b with about a thousand COVID-19 patients, in a trial sponsored by the US National Institute of Allergy and Infectious Diseases.
In October 2020, the trial’s data safety and monitoring board suspended the recruitment of patients with severe COVID-19 after it found more serious adverse events in the treatment arm relative to the control arm. The study continued to recruit patients with moderate infections, and concluded on December 21, 2020.
While the results may not be directly comparable to Virafin, since severe adverse events were restricted to people with severe COVID-19, it offers a cautionary tale – that Virafin’s efficacy itself may be restricted to patients with moderate disease. Far from being ‘harmless’, administering it to people with severe COVID-19 could prove counterproductive. Of course, we need well-designed clinical trials to ascertain this.
(Virafin however is pegylated interferon alpha 2b. Pegylation is simply the process by which polyethylene glycol – or PEG – is attached to a drug or therapeutic molecule to enhance its properties, like ability to circulate for longer in the body or not draw undue attention from the immune system.)
As it happens, the Virafin study has several methodological and reporting flaws that substantially undermine its results.
First, the clinical trial entry for the phase 2 trial includes quantitative RT PCR as a secondary endpoint – but the paper doesn’t mention it at all nor does it explain why the researchers left it out.
Second, the trial relies on a scale, which is a soft endpoint, instead of an objective measure. This creates room for bias, as mentioned earlier.
Third, it uses a modified intention to treat (mITT) analysis instead of the intention to treat (ITT) analysis. In the latter, participants who don’t complete the trial are also included in the analysis. One cross-sectional study found in 2011 that mITTs are often associated with post-randomisation exclusion and misleading outcomes.
Fourth, the size of the trial is unreasonably small. In fact, the paper admits that “There was no formal calculation of sample size for this study.”
Dr Jammi Nagaraj Rao, a public health physician in the UK, told The Wire Science that this study was in fact very poorly designed.
According to him, a phase 2 trial would ideally test different dosage regimes and check the feasibility of measuring outcomes in a small sample, ahead of a larger phase 3 trial. In the context of COVID-19, the only outcomes that matter to patients is mortality reduction. But in this study, even if the results are claimed to be statistically significant, they are of marginal clinical significance.
Next, according to the paper, its corresponding author Kevinkumar Kansagra is employed at the Zydus Research Center, Ahmedabad – a firm that stands to profit financially from the results of the study. But at the end of the paper, all the authors declare no conflict of interest. This is a violation of the journal’s rules and ethical research practices.
The DCGI’s approval of Virafin is thus based on dubious or non-existent data – in keeping with a trend it has exhibited during the COVID-19 pandemic. It bulldozed the restricted-use approvals of two other drugs, favipiravir and itolizumab, despite several similar methodological flaws in their respective trials.
Virafin has been indicated for use in patients with moderate COVID-19, whose oxygen saturation is in the range 94-90%. Given the number of cases currently active in India, and the fraction of them that have only moderate disease, Virafin’s potential consumer base is very large. This does not bode well. Each dose of Virafin for a patient with a body mass of 50 kg costs Rs 9,000 per injection.
“I suspect the manufacturer will deploy all its marketing resources to push this drug based on this questionable study,” public health activist Dinesh Thakur said. “And given the general atmosphere of panic in the country, I am afraid they will capitalise on it and push it out to unsuspecting patients across the country.”
According to Thakur, the law requires a drug or therapy to have exceptional therapeutic efficacy before it receives even limited approval – and said that Virafin is not at all close.
Taken together with the presence of adverse effects, the unreliable study design and reporting, and the absence of phase 3 trial data in the public domain, the DCGI seems once again to have acted against patients’ best interests.
Requests to the DCGI’s office and Zydus Cadila for comments were unanswered at the time of publishing. The article will be updated as and when they respond.
Ronak Borana is a science communicator based in Mumbai. He tweets at @ronaklmno.