Representative photo: Markus Spiske/Unsplash
On August 7, scientists of the Indian Council of Medical Research (ICMR) made an interesting claim via a preprint paper: that mixing Covaxin and Covishield, the two major vaccines in India’s COVID-19 vaccine drive, may not only be safe but may also induce a greater immune response.
This may seem at first glance to be a heartening statement, but as we have seen so often during the course of this pandemic, there is a lot here to unpack.
The study was led by Pragya Yadav, a scientist at the maximum containment facility of ICMR-National Institute of Virology, Pune. She and her colleagues have described the results of their analysis of antibodies to the novel coronavirus among 18 people who received a first dose of Covishield and, accidentally, a second dose of Covaxin at a vaccination centre in Uttar Pradesh, in May 2021.
The researchers studied the immunological effects of this administrative fiasco. That they hadn’t planned to examine the effects of a mixed-vaccine schedule makes theirs a serendipitous study, instead of a deliberate experimental trial. The title of their paper is ‘Serendipitous COVID-19 vaccine-mix in Uttar Pradesh, India: Safety and immunogenicity assessment of a heterologous regime’.
Serendipity is to find something “interesting or valuable by chance”. One of the more famous examples of a serendipitous finding from the history of medicine is the discovery of penicillin. On September 3, 1928, when Alexander Fleming returned to his laboratory after holiday, he observed that a Petri dish he had ‘infected’ with staphylococci bacteria contained a small patch of fungus, and the bacteria immediately surrounding it had been destroyed. The bacterial colonies further away seemed okay. The fungus belonged to the genus Penicillium, and produced an antibacterial substance that later came to be refined into the world’s first antibiotic.
But for being the product of accidents, serendipitous discoveries are perfectly honourable as long as they germinate in the right conditions. Many experts have even formally documented and studied them.
The bigger problems with the present study, about mixing Covaxin and Covishield in a single vaccination regimen, emerge from other parts.
A good idea to mix?
Using a combination of vaccines as a strategy against COVID-19 is not new. Technically called heterologous prime-boost vaccination, the idea is that, in the short-term, mixed schedules could help overcome shortages of particular vaccines that could in turn affect the rollout of a given country’s vaccination programme. And if a combination boosts the immune response and, later, the effectiveness, that would be a major bonus.
In early April 2021, French public health authorities recommended that people younger than 55 years and who had received a primary dose of the Oxford-AstraZeneca vaccine should receive a second dose of one of the mRNA vaccines – either Pfizer-BioNTech or Moderna. They also called for clinical trials on such heterologous, or mixed, schedules and for the state health apparatus to continue its pharmacovigilance.
In July, authorities in Ontario, Canada, backed a similar proposal for a mixed vaccine schedule, and also suggested a first dose of the AstraZeneca vaccine and second dose of an mRNA vaccine for better protection.
In a preprint paper published by The Lancet in May 2021, scientists at the University of Oxford had reported a study of such a schedule in the Com-COV clinical trial. This was a phase 2 study with a cohort of 830 participants. Their goal was to find that a heterologous COVID-19 vaccination schedule didn’t induce lower antibody levels than a homologous schedule. They succeeded. They also reported more frequent adverse reactions following the second dose of the heterologous schedule – but the reactions were short-lived and minor.
Studies like this are complicated, but the headline results thus far have been encouraging. That is, there is sense in pursuing a heterologous prime-boost vaccination schedule.
This said, unlike the Oxford Com-COV study – a planned, well-designed phase 2 trial – the ICMR study was opportunistic. So before we further unpack the latter, let’s remember that the most it is capable of showing us is that the heterologous vaccine schedule didn’t seriously compromise the participants’ antibody response.
The ICMR study
Pragya Yadav and her colleagues identified the 18 people who had been inadvertently given Covaxin as their second dose after having received Covishield as the primary dose.
Covishield uses a fragment of a virus protein to induce an immune response in the human body. Covaxin uses inactivated whole-virus particles.
At the time, in May, there had been some anxiety about the effects of the mix-up, especially on vaccine uptake and hesitancy, so the researchers decided to formally check the after-effects.
The researchers assembled two groups of 40 people each, who had received both doses of either Covishield or Covaxin, to serve as the control groups. In effect, the study involved three groups of people:
- Those who had received two doses of Covishield,
- Those who had received two doses of Covaxin, and
- Those who had received one dose of Covishield followed by one dose of Covaxin
As Shreya Dasgupta has written for The Wire Science, in a typical randomised controlled trial (which this study was not):
… investigators randomly allocate participants to either the group that receives the intervention or to a control group that doesn’t (which instead receives a placebo or standard care), but is nearly identical in all other ways. The investigators then administer the treatment being studied and observe the two groups over a period of time. At the end, they compare the groups in terms of outcomes – or endpoints – that they had set out to study with the trial.
The ICMR preprint paper, however, does not tell us how the scientists selected the two groups of people. Ideally, they should have sampled the people from the community that had been vaccinated at around the same time as the group set to receive the intervention, and at the same vaccination centre. But this was a serendipitous experiment and there could be no random allocation. So we need other assurances that the groups were comparable in every respect other than the vaccination schedule. These assurances are missing.
Next, the scientists observed the endpoints – immunological response and adverse events – nine weeks after the first dose. It’s possible that this was too soon. The Covaxin recipients had been vaccinated with two doses given four weeks apart, and thus had completed five weeks after the second dose at the nine-week point. On the other hand, Covishield recipients and the recipients of the heterologous schedule received their two doses six weeks apart, and thus were only three weeks past the second dose at the nine-week point.
Also read: CDSCO Expert Panel Gives Nod to Study on Mixing Doses of Covaxin, Covishield
The findings
There were no major differences between the reactions and side-effects among participants in the three groups. This said, 18 participants in the heterologous group and 40 each in the other two ‘control’ groups is hardly a big enough cohort from which we can draw major conclusions about adverse effects following immunisation. All we can say is that the experience of these 18 people raises no major concerns vis-à-vis safety.
The scientists studied the immunological, or antibody, response by measuring the levels of antibodies – or antibody titres – against different proteins in the virus (e.g. the receptor binding domain and the N-protein) among all participants. They also measured the titres of neutralising antibodies against four strains of the virus: B.1, B.1.1.7 (alpha), B.1.351 (beta) and B.1.617.2 (delta).
With so many measurements, the scientists could perform a large number of statistical comparisons.
However, the preprint paper presents the data in a dense and confusing manner. Some of the charts appear to show a near-total overlap in the spread of values for the antibody titres across groups – but the tests are reported to show statistically significant differences. This shouldn’t be possible.
Instead, the scientists could have used one table to show the differences and another to show the results of the significance tests. It also seems that in the rush to upload the paper into the public domain, the authors have made a few silly mistakes. For example, the 95% confidence interval (CI) for a mean value of 2260 is listed as “95% CI: 2716-1881” – instead of ‘1881-2716’.
But the final conclusion, we are told, is that the levels of the immunoglobulin G antibodies were higher in the heterologous group than in the Covaxin or the Covishield groups. However, a chart with a logarithmic scale (i.e. showing not the absolute numbers but the orders of magnitude) indicates that the levels in the heterologous group varied considerably.
Next, the data on the antibody titres is also very confusing – so much so that it’s hard for even the keenest reader to make sense of what the researchers are saying. There is no comfortable way to connect the mass of numbers on page 11 and those presented in figure 4, to which it relates. The p-values (numbers denoting the significance of a result; p < 0.05 is widely considered, but of late controversially, to be ‘good enough’) quoted in the text also bear no resemblance to those in the figure.
One p-value is listed as p > 0.999. If this is not a misprint, it would mean the datapoint it applies to is so insignificant as to undermine the study’s conclusions.
It would also have been better if the scientists had specified the study’s primary outcome at the outset. Doing so is important to reassure the rest of us that the scientists knew what they wanted to test for before the study began – instead of fixing it after obtaining the data (akin to drawing the target and firing arrows at it versus firing the arrow and then drawing the target around it).
In sum, the ICMR study is a preliminary study. Its strength is limited, the authors acknowledge, by its small numbers and the opportunistic nature of its design. It is good that the scientists conducted the study at all, to allay fears about the side-effects of a heterologous schedule – but we can’t read much into the findings. We can only say the claim that “a heterologous schedule with Covishield followed by Covaxin may offer advantages over two doses of either vaccine” merits further examination.
Such examination will necessarily have to be in the form of a properly resourced, planned and designed clinical trial. But given India’s track record thus far on conducting and reporting trials for COVID-19 vaccines, it’s hard to say if it will happen.
Dr Jammi Nagaraj Rao is a public health physician, independent researcher and epidemiologist in the UK.
Vasudevan Mukunth edits The Wire Science.