On March 30, the Brazilian drug regulator, Anvisa, published a report listing a litany of quality problems with Bharat Biotech’s manufacturing plant for its inactivated COVID vaccine, Covaxin. These problems included insufficient measures to ensure that the SARS-COV-2 virus was completely killed and that the vaccine was free of microbial contamination. The report also talked about possible variations in Covaxin’s potency from one dose to another.
On the face of it, these problems sound deadly serious. Bharat Biotech makes Covaxin by inactivating the SARS-COV-2 virus, so that the virus does not make people sick even as its antigens – the proteins on the surface of the virus – provoke an immune response. But if the company wasn’t inactivating the virus properly, it would mean that the vaccine itself could give patients COVID-19. Brazil’s response was appropriately cautious; the country won’t go ahead with its plan to buy 20 million Covaxin doses until Bharat Biotech fixes these problems.
Yet, confusingly, several commentators downplayed the seriousness of Anvisa’s findings. The Hindu quoted “diplomatic sources” as saying the problem was merely “about bureaucratic process and not pertaining to the quality of the vaccine.” Oommen Kurien, a public health researcher with the Delhi-based Observer Research Foundation, tweeted that without further information, it was hard to say if Anvisa’s report was driven by quality issues – and not by political considerations.
To understand how serious Anvisa’s report was, The Quint spoke to several pharmaceutical manufacturing experts. Their verdict: ensuring the inactivation of a virus is a critical quality control measure, required by both Indian and Brazilian laws under a set of standards called Good Manufacturing Practices (GMP). These quality-control measures are a direct result of historical vaccine-manufacturing accidents, especially the 1955 Cutter Incident, where partial inactivation led to a public-health disaster.
So, if these GMP have indeed been violated, the Brazilian regulator’s observations are anything but a bureaucratic issue. “Based on the deficiencies identified, including the lack of specifics, I probably would not volunteer to receive the vaccine,” said Barbara Unger, president of the California-based Unger Consulting that helps pharma companies comply with GMP requirements. Other experts said that given the vaccine had been given to over 9 million people in India, Bharat Biotech ought to issue a public clarification about the shortcomings identified by Anvisa.
When The Quint reached out to Bharat Biotech, a spokesperson characterised Anvisa’s report as “a misunderstanding”, and said the Brazilian agency would be clarifying soon. But the spokesperson didn’t answer questions about whether the company had indeed skipped the GMP requirements pertaining to viral inactivation.
Read on to understand why Indian and Brazilian laws require GMP, and what the violations mean.
Typically, the first step in making an inactivated vaccine is to grow the virus in cells. In Covaxin’s case, Bharat Biotech used Vero-cells, which come from the kidney of the African green monkey. The next step is to filter the liquid, and separate the virus from the cell debris. After this, manufacturers use heat or a chemical to inactivate the virus. Bharat Biotech uses a chemical called beta-propiolactone, which jumbles the viral genome so that the virus cannot replicate anymore. A few steps later, the vaccine is prepared by combining the inactivated virus with an adjuvant (a chemical which enhances the body’s immune response), and excipients (chemicals which add volume to the product and keep it stable).
Along the way, however, manufacturers must take multiple measures to ensure the process goes as planned. And Anvisa says Bharat Biotech skipped some of these measures.
One measure is to prove that the inactivation process worked. Manufactures can do this in many ways. For the rabies vaccine, the vaccine-maker can add the killed virus to a cell line (such as Vero cells), and check for rabies virus antigens using a method called immunofluorescence. Alternatively, the vaccine-maker can inject the fluid from the virus-cell mixture into the brains of mice. If the mice develop rabies symptoms within two weeks, it means the inactivation method has failed.
Another measure is to show the so-called kinetics of inactivation, which maps the rate at which the virus gets killed during the inactivation process. Broadly, this means the manufacturer will plot a curve with the number of surviving viral particles on the vertical axis and the time for which the virus has been exposed to the inactivating heat or chemical on the horizontal axis.
This curve is important because in some processes, the virus dies quicker in the beginning, while in others, the inactivation is delayed. And because it is virtually impossible for a vaccine-maker to detect every last live virus in every last vial, this curve can predict how long it will take for the virus to become completely harmless. This calculation helps companies produce vaccine vials in which the chances of a live virus being present are vanishingly small.
A third measure the manufacturer must take is to “validate” the proof of inactivation. In other words, they must show that the proof of inactivation is specific and sensitive, says Sumant Baukhandi, the managing director of GMP Pharma Private Limited, a Dehradun-based GMP consulting firm and training institute.
In the example of rabies virus vaccine described earlier, high specificity would mean that only the live rabies virus, and no other, is detected in the proof of inactivation, while high sensitivity would mean that even small amounts of rabies virus are detected.
After the vaccine is formulated, manufacturers must again do a number of tests. One test measures the potency of each dose, typically by measuring the viral antigen which provokes an immune response in humans, eventually protecting against disease.
For instance, in the case of the inactivated influenza vaccine, potency is measured in terms of the hemagglutinin antigen, a key protein on the flu virus’ surface.
According to one Anvisa document, Bharat Biotech did not demonstrate the kinetics of inactivation or validate its proof of inactivation. Further, the document says the company was using a potency method which did not ensure that every dose was equivalent. “Thus, the patient may receive non-uniform and non-representative doses of the product,” the report notes. Finally, Anvisa was dissatisfied with the measures to ensure the sterility and purity of the vaccine (See question: Anvisa also said Bharat Biotech didn’t ensure the sterility and purity of Covaxin? What does this mean?)
In a separate document, Anvisa listed out dozens of provisions under Brazilian Good Manufacturing Practices, which it says were violated by Bharat Biotech. These provisions require the company to ensure that treated products aren’t contaminated with untreated ones, and pertain to the filtration and sterilization during manufacture.
Unger said that while the report lacked specifics, an inspection with 12 major findings was “quite problematic”.
GMP for vaccines have evolved tragedy by tragedy, shaped by past instances where inadequate quality-control led to deaths and disability.
The most infamous of these tragedies is the 1955 Cutter Incident, involving the inactivated polio vaccine. That year, five American companies starting manufacturing a polio vaccine developed by virologist Jonas Salk. But the companies – especially the California-based Cutter Laboratories - failed to fully inactivate the vaccine, leading to one of the worst manmade polio outbreaks in history. Around 40,000 children developed a mild version of polio called abortive polio, 164 people were paralysed, and ten died as a result.
The Cutter Incident, a book by American paediatrician Paul Offit, describes the sequence of manufacturing mistakes which led to the tragedy. Among them, when Cutter Labs began manufacturing its vaccine on a large scale, the company changed the filtration method it was using to separate the virus from the cells it was grown in. When Salk developed the inactivation process originally, he had used asbestos filters to separate the virus from the monkey-kidney cells it was grown in. But because asbestos filters were too slow for large-scale manufacturing, Cutter Labs switched to glass filters, which were faster.
This had an unintended consequence; some of the glass filters were less effective than others, and ended up leaving cell debris behind.
Another mistake that precipitated the tragedy, Offit explains in his book, was that Cutter Labs did not map the kinetics of inactivation. In his inactivation experiments, Salk had found that the graph with the number of poliovirus units per dose on the X axis, and length of exposure to formaldehyde on the Y axis, was a straight line; in other words, the drop in the number of surviving viral particles was proportional to the time it was exposed to the chemical.
But when Cutter Labs and other manufacturers tried to repeat the process on a larger scale, their experience was different. They found that with time, the rate of inactivation slowed down, likely due to interference with the cell debris. And even though Salk advised Cutter Labs to confirm that its own inactivation curve was straight, the company did not do so. This left the vaccine-maker with a blind spot – they didn’t know how long it would take to fully inactivate the virus, and to make a completely safe vaccine.
These problems could have been caught had the company’s proof of inactivation been reliable. However, the method it used wasn’t sensitive enough. The test which the United States government had recommended at the time, and which the firm used, involved inoculating monkey-kidney cells with the virus, and checking through a microscope to see if the virus was destroying the cells. In a separate test, manufacturers also inoculated monkeys with the vaccine, and waited for thirty days to check if they were paralysed. But investigators later found that these monkeys weren’t that sensitive to live virus, and sometimes escaped without falling sick.
That’s not all. Manufacturers must also take precautions against pathogens from sources other than the vaccine virus. Even the cells in which the virus is grown, or the air in the manufacturing facility can be such a source. In the 1960s, for example, yellow-fever vaccines in England and the United Kingdom were found to be contaminated with the avian leukosis virus. The virus, which can cause cancer in chickens, likely came from the chicken eggs in which the yellow-fever virus was grown. More recently, in 2010, a rotavirus vaccine made by GlaxoSmithkline was found to be contaminated with porcine circovirus, which causes disease in pigs. This likely came from an enzyme extracted from pigs and used to grow the cells in which the virus was cultured
Anvisa’s report doesn’t share any details on this issue.
Sterility refers to the absence of any microbes in a product. Because vaccines are injected into the body, bypassing the protections of the gastrointestinal system, any microbes in a vaccine pose a far higher health risk than microbes in an oral drug. So, sterility is crucial.
Purity, on the other hand, refers to the lack of impurities other than microbial ones– a vaccine shouldn’t have anything extra apart from the inactivated virus, excipients that keep it stable, and the adjuvant.
“It is not sufficient to simply test a portion of the lot for sterility and assume all is well,” said Unger. Manufacturers have to comply with dozens of best practices, like keeping tabs on the quality and flow of air in the room where vials are filled, donning sterile gowns the right way, and ensuring there is no contamination when the product is filled into vials.
A 2019 inspection report of pharmaceutical company Cipla’s Goa plant gives an idea of the kind of violations that can compromise sterile drugs. The United States Food and Drug Administration sent a warning letter to the company after discovering dozens of problems at its plant: among them, air ducts weren’t cleaned properly, and there was bacterial growth in parts of the plant which possibly came from soiled socks worn by workers. And last week, KHN reported about repeated quality problems at Pfizer’s largest sterile injectable plant in the USA. The Kansas based plant, which is slated to manufacture the company’s COVID vaccine soon, has faced persistent problems with mould in the area where sterile drugs are filled into vials.
Unger said she agreed with this observation. According to her, the antigen on the surface of SARS-COV2-virus, which is the protein responsible for the potency of the vaccine, can degrade with time. This could render the vaccine less potent, but would still not impact the measurement of total protein, because a total-protein test would measure both intact and degraded protein. “Not all protein is the same,” Unger says.
Hans Meerburg, a GMP consultant based in Netherlands, agreed with this assessment. “Only using total-protein is not enough,” he said.
Baukhandi added that a manufacturer could use total protein as a measure of potency too, but that they would first have to demonstrate that this number was correlated with potency.
Whether Bharat Biotech did so is not clear.
A spokesperson representing Bharat Biotech did not respond to The Quint’s questions, except to say that the Anvisa report was the result of “a misunderstanding”.
But these issues must be clarified because they could imply that CDSCO overlooked quality problems with Covaxin in an effort to get the vaccine out quickly. Over the last year, the CDSCO, the Indian Council of Medical Research (Bharat Biotech’s co-developer for Covaxin), and Bharat Biotech have shown a tendency to skip due diligence in the interest of speed.
These incidents have been worrying. In July last year, the director general of ICMR wrote to all the investigators in the first-in-human trials of Covaxin, asking them to begin enrolling participants within six days. The decision was heavily criticised by clinical research experts, because it could have hurt the quality of the trial. Then, in December, Bharat Biotech controversially applied for accelerated approval of its vaccine in India, even though it had not yet shown the efficacy of its product. A few weeks later, in January 2021, participants at the Bhopal site for Covaxin’s Phase 3 trial complained of serious protocol breaches, claims that the company, ICMR and the CDSCO dismissed without addressing.
Could the hurry to push out an indigenously developed vaccine have led the CDSCO to waive GMP requirements for Covaxin? Given the events of the last one year, the possibility certainly exists.
(Priyanka Pulla is freelance science journalist based in Bangalore. The reporting for this story was funded by The Thakur Family Foundation. The Foundation exercised no editorial control on the contents.)