A representative photo of various drugs littered on a table. Photo: Anna Shvets/Pexels.
Bengaluru: Medical experts have accused the Mumbai-based Glenmark Pharmaceuticals of cherrypicking the results of its phase 3 clinical trials to claim that its antiviral drug, favipiravir, is effective against COVID-19. While the Indian drug regulator approved the drug based on the trial in June, the company only published the trial’s full results on November 16. And experts pointed out that the full results don’t bear out the efficacy claims that Glenmark has made about the drug.
Since the drug’s approval, the company has headlined two press releases (here and here) about the trial with the claim that favipiravir recipients were cured of COVID-19’s signs faster than those who received standard of care, and that the difference was statistically significant. These press releases also said patients in the favipiravir arm cleared the virus faster (as measured by how quickly patients turned negative on an RT-PCR test) than controls, although this difference was not statistically significant.
According to Glenmark and the trial’s principal investigator, Mumbai based pulmonologist Zarir Udwadia, these findings together showed that favipiravir can treat mild and moderate COVID-19.
But critics have countered that the faster viral clearance in the favipiravir arm was not statistically significant, which means, strictly speaking, that the drug failed to show efficacy in the trial. In other words, favipiravir recipients could have turned RT-PCR negative faster by chance alone, and not due to the drug. Critics also said the lack of statistical significance in viral clearance was the key result of the trial, rather than the difference in cure times. This is because viral clearance was the Glenmark trial’s primary endpoint while the cure times difference was the secondary one.
Drawing conclusions based on secondary endpoints when the trial has failed on the primary is a bad idea because it can lead to false positives.
The way Glenmark framed its press releases could mislead laypersons into believing that the trial’s results were promising when they are actually not, C.S. Pramesh, the director of Tata Memorial Hospital, Mumbai, told The Wire Science.
Following the press release, at least two newspapers reported that the results were strongly in favour of favipiravir. On November 21, the Mumbai Mirror ran an article with the headline that Udwadia had found “proof that favipiravir works for the moderately ill”. The Times of India reported on the same day that favipiravir cuts treatment time.
Neither newspaper mentioned favipiravir’s failure to make a statistically significant difference on the trial’s primary endpoint.
Even the claim that favipiravir cures patients faster is questionable, said Sahaj Rathi, a faculty member in the department of medicine at the Mahatma Gandhi Institute of Medical Sciences, Sevagram. Rathi pointed out that the trial had an ‘open-label’ design – which means doctors overseeing the patients knew who received favipiravir and who didn’t. And these doctors also took a call on which patient was considered to have been cured by checking whether the patient’s body temperature had dropped below 98.8º F, respiratory rate had fallen to less than 20 times a minute, blood oxygen levels (SPO2) had returned to over 95%, and cough was either mild or absent.
But the knowledge of which patients received the drug could easily influence these doctors’ decisions, given that even temperature, respiratory rate and cough severity can be subjective, Rathi said. “Open-label trials are heavily prone to biases… body temperature varies with time and use of antipyretics, respiratory rate varies, and mild cough is a very loose term.”
Put together, Rathi and others have argued, none of the findings support the company’s claims that favipiravir is effective. “Faced with a patient with mild or no symptoms – should a doctor recommend favipiravir?” Jammi Nagaraj Rao, an epidemiologist in the UK, asked. “Based on this study, no.”
But the weak results haven’t deterred widespread use of the drug. Since June, favipiravir has gained traction among doctors treating COVID-19 patients across India. In a recent editorial published in the Indian Journal of Medical Ethics, Rathi and his coauthor wrote that favipiravir is now in routine use across the country. At least three states – Maharashtra, Kerala and Karnataka – have recommended the drug in their COVID-19 treatment guidelines as well.
While a Glenmark spokesperson said the company didn’t have a number for favipiravir sales since June 2020, Mumbai Mirror reported that the Municipal Corporation of Greater Mumbai alone has amassed a stockpile of five lakh tablets.
Primary endpoints matter
One reason why experts warn against shifting focus to secondary endpoints when a trial has failed on its primary ones is that trials are designed around the primary, and not the secondary. In other words, a trial should be considered a ‘win’ only if the primary endpoint is met. Secondary endpoints help gather further ‘good to know’ information on the drug’s efficacy and can help generate hypotheses for future trials – but can’t be the basis of efficacy claims.
If trials are designed around primary endpoints, it means their sample sizes are also calculated based on how effective the drug is expected to be on these endpoints. For example, in Glenmark’s favipiravir trial, investigators calculated that for the primary endpoint of viral clearance, they would need to recruit 150 patients.
But when it came to measuring the secondary endpoints, the sample sizes were far smaller. To calculate time to clinical cure, for instance, the investigators analysed only 102 patients because only this subset of patients had symptoms like fever and cough.
Yet another secondary endpoint was how long patients took to deteriorate enough to need oxygen, ventilation or extracorporeal membrane oxygenation (ECMO, in which an external machine does the job of the heart and lungs. This machine is only deployed when patients worsen even on ventilation). The investigators measured this endpoint on a mere 14 patients – because only these patients ended up needing oxygen, ventilation or ECMO.
Yet Glenmark cited its findings among these small groups to bolster its argument that favipiravir helped patients. Such tiny sample sizes make such comparisons meaningless, sad Rao. “The time to needing oxygen is based on so few patients that it is best not to draw any conclusions.”
In their replies to emails from The Wire Science, the trial investigators defended their decision to rely on secondary endpoints, however. Monika Tandon, the head of clinical development at Glenmark and an investigator on the trial, said the scientific understanding of viral clearance, as measured with RT-PCR tests, had evolved during the course of the pandemic.
According to her, when the trial protocol was drafted in March, RT-PCR tests were thought to reflect the duration of COVID-19 infections accurately. “RT-PCR was the primary endpoint in most trials then.” But later studies showed that patients could remain RT-PCR-positive long after they had been clinically cured. This explained why the primary endpoint did not correlate well with clinical cure, she said.
Udwadia added that analysing secondary endpoints was not that unusual. “If the primary endpoints were all that were looked at, there would be no successful COVID-19 trial except for dexamethasone in all of COVID-19 literature.” He added that compared to drugs like hydroxychloroquine, ivermectin and tocilizumab, which have been used with little evidence, the small improvements in clinical cure that favipiravir showed were “meaningful”.
While it’s true that pharmaceutical companies have been known to cherry-pick from secondary endpoints to make claims about drugs’ efficacies, experts say this practice is a big no-no.
A 2017 guidance for the industry from the US Food and Drug Administration reiterates this in the following words: “Positive results on the secondary endpoints can be interpreted only if there is first a demonstration of a treatment effect on the primary endpoint family.”
Further, the Indian drug regulator requires all trials to be registered prospectively on the Clinical Trial Registry of India today, making it harder than before for investigators to shift goalposts surreptitiously at the end of the trial.
Cost of favipiravir
The widespread use of favipiravir would have been of little concern if the drug was completely safe. But in Glenmark’s trial, favipiravir recipients experienced adverse effects like elevated blood uric acid and abnormal liver function tests around 36% of the time, while controls did so only 8% of the time.
Tandon and her colleagues have argued that these adverse effects were mild and reversible, and worth the trouble given the drug’s benefits. But this might not be true. One of the adverse effects experienced by favipiravir recipients more often, for example, was viral pneumonitis – the inflammation of lung tissue, captured on X-ray scans. Pneumonitis can result in future complications, like scarring of the lung tissue, leaving patients incapacitated for a long time.
Then there is the matter of the drug’s cost. When favipiravir was first launched, Glenmark priced it at Rs 103 per tablet, which means a full course of 122 tablets would have cost over Rs 12,500. Today, with several other manufacturers making the same product, Glenmark’s price has dropped to Rs 37-38 per tablet, according to a company spokesperson. But this, too, adds up to at least Rs 4,514 per full course – not an insignificant amount for many.
Should a patient then take the drug, given its benefits remain unproven? Critics of the favipiravir approval say there isn’t enough justification to do this. “The best I can say is that this trial improves on existing evidence for favipiravir,” Pramesh said. “But by no means can this trial be considered a definitive phase 3 trial.”
Glenmark officials say the company isn’t planning any larger controlled studies of favipiravir alone, although they pointed out that a few randomised studies are already happening in other parts of the world. Favipiravir may yet turn out to be a drug worth prescribing to COVID-19 patients – but we will have to wait for these ongoing trials to finish to know if this is the case.
As of now, experts say, it’s best to hold off.
The reporting for this story was funded by a public health journalism grant to Priyanka Pulla from The Thakur Family Foundation. The Wire Science retained full editorial control.
Priyanka Pulla is a science writer.