Mene Pangalos, head of biopharmaceuticals research for AstraZeneca Plc, went to bed on Monday, March 22, feeling good for the first time in a while. After working around the clock through the weekend, he'd just announced better-than-expected interim results from the company's large US vaccine trial: The shot was safe and 79% effective at preventing cases of symptomatic Covid-19. Positive news, at last, after months of questions about everything including safety and supply shortfalls.
But at around 5 the next morning, Pangalos was jolted out of bed in the UK by a call from AstraZeneca Chief Executive Officer Pascal Soriot, ringing from Australia to ask what on earth was happening. The US National Institute of Allergy and Infectious Diseases (Niaid) had just issued a late-night statement announcing that the safety board overseeing the trial was concerned AstraZeneca "may have included outdated information" in its results, which "may have provided an incomplete view of the efficacy data." Niaid, which is part of the National Institutes of Health, urged the company to release up-to-date figures as soon as possible.
While they were asleep in Europe, Pangalos and his team had received a harsh email from the Data and Safety Monitoring Board, the independent committee of experts appointed by the NIH to oversee the trial. The DSMB accused the company of releasing "potentially misleading" figures, saying the data were "the most favorable for the study as opposed to the most recent and most complete." The experts added that "decisions like this are what erode public trust in the scientific process." Astra executives were stunned. No one had reached out to them to discuss the concerns first, and Niaid's decision to publicize the independent committee's critique of the interim results was unprecedented.
The contents of the DSMB's letter quickly leaked to the Washington Post and the New York Times. Within hours, Anthony Fauci, director of Niaid, was on Good Morning America, talking about the flap. "It really is unfortunate that this happened," he said. "This is really what you call an unforced error, because the fact is this is very likely a very good vaccine, and this kind of thing does nothing but really cast some doubt."
Astra had received approval the Friday before from the safety board to conduct its interim analysis, based on 141 verified Covid cases dating to Feb. 17. The board had advised the Astra team to be mindful of the unverified cases that had come in since then, to be sure they wouldn't produce a significantly different efficacy figure, according to people familiar with the discussions. The team had looked at the cases, but, believing that their potential effect on the efficacy number was negligible, they hadn't mentioned in their statement that the final percentage might change.
Now, with the DSMB's letter public knowledge, the trial team raced to review the additional 49 cases, a process that normally takes weeks. Two days later, AstraZeneca announced the updated results. The differences were statistically insignificant: Overall efficacy dropped 3 percentage points, to 76%, though it actually rose 5 percentage points in seniors, to 85%. The vaccine was 100% effective at preventing hospitalization and severe disease. This time, AstraZeneca was careful to note that there were still 14 more possible or probable Covid cases to be assessed, which could cause the numbers to fluctuate. The company's executives were deeply upset by the events of the week. "It's just hugely frustrating," one says. "Honestly, it breaks our hearts."
The week after the updated results came out, Fauci downplayed the drama. "I didn't question their data at all," he said when asked about the incident by Bloomberg Businessweek at a White House briefing on March 31. "This is a good vaccine that is going to have a very important role in the global response to this outbreak."
The story of the vaccine that AstraZeneca developed with the University of Oxford is marked by noble intentions, communication blunders, messy trials, manufacturing nightmares, and political and economic rivalry. Most seriously, the shot is facing a spate of reports that a small number of people who received it, most of them younger than 60, developed a rare form of cerebral blood clotting. The World Health Organization and British and European regulators said that they'd found a possible link between the vaccine and the clots but that the benefits of getting the shot outweigh the risks. The EU regulator has recommended that clotting be listed as a rare side effect, while the UK is advising that the vaccine not be used on people under 30. Some governments have already halted the shot's use in younger people. Despite all the issues, many European Union leaders have been clamoring for doses, complaining that AstraZeneca has failed to meet its supply promises and even threatening to block vaccines made or bottled in the EU from being exported.
In the span of a year, Astra has gone from favored child to problem child in the family of coronavirus vaccines. "There's this little cloud that's followed it around everywhere it goes, which I think is slightly unfair," says John Bell, the Canadian-born Oxford professor of medicine overseeing relations between the university and AstraZeneca. "If people keep beating this vaccine up, nobody's going to have the confidence to use the darn thing, and then we've got a massive problem—because it's deployable, it's cheap, you can use it globally, and it's clearly highly effective."
As Bell suggests, Astra's vaccine is more easily transported and stored than the mRNA-based vaccines from Moderna and Pfizer-BioNTech. The UK company has promised to deliver as many as 3 billion shots in 2021, selling them on a not-for-profit basis, at a few dollars a dose. Only the single-shot vaccine from Johnson & Johnson, which has also agreed to sell doses for no profit, is expected to make a comparable contribution to ending the pandemic, but it's far behind Astra's on production. By contrast the Pfizer-BioNTech vaccine is expected to bring in $15 billion in revenue this year, while Moderna Inc. is forecasting $18 billion for its shot. More than 135 million Astra doses have thus far been distributed around the world, and the shot has become the workhorse of Covax, the WHO-backed program that provides vaccines to low- and middle-income countries.
AstraZeneca and Oxford have had their share of bad luck, but they also made a series of missteps that derailed their front-runner status. "It's shocking that they made so many errors," says Ezekiel Emanuel, a medical ethics and health policy professor at the University of Pennsylvania who advised President Biden's Covid transition team. For all the blunders, the world desperately needs Astra's shot to end the pandemic. This is how it's gone wrong.
"We must be able to trust the vaccines. ... These are findings we cannot ignore"
Last spring, scientists at Oxford were advancing quickly through lab and animal testing of their vaccine candidate, which they were planning to deliver in a single dose. They were under pressure to team up with a big pharmaceutical company, given that manufacturing at scale would require months of intensive planning and vast logistical capacity. Early talks with US-based Merck & Co. ground to a halt after UK officials, wary that American nationalism could leave them without access to doses, said they wanted to keep development and manufacturing in British hands. GlaxoSmithKline Plc, a major UK vaccine developer, passed on the opportunity to join forces. That left the only other British player of any size, AstraZeneca.
Astra was on a roll under Soriot, who'd fended off a $117 billion hostile bid from Pfizer Inc. in 2014 and developed a reputation as a miracle worker in getting new drugs, especially treatments for cancer, to market. When Bell spoke with Soriot last April, he said Oxford had two conditions: the vaccine had to be sold on a not-for-profit basis at least for the first year, and Astra had to help ensure the developing world got doses. Bell says Soriot didn't hesitate to agree, replying, "I'm completely with you." Astra would get a massive public relations coup and, down the road, lots more business if booster or annual shots proved necessary.
Other than a nasal spray for flu, though, Astra had almost no experience in vaccines. "They're in this game for the first time, so they're making their way," Bell says. "They don't have five authoritative big guns in the vaccine world to stand up and say, 'Hey guys, pay attention to this. This is really good stuff.' "
Some potential for dysfunction was baked into the process from the start. By the end of April, when it reached the final partnership agreement with Astra, Oxford had already begun human trials, whose protocols would become quite complex. It had large Phase III trials set to start in the UK and Brazil, along with a smaller Phase I-II in South Africa. For the US Phase III trial, Astra took charge, designing a straightforward two-arm protocol.
With the backing of the UK government, the partnership moved swiftly on manufacturing, building on the university's early work with Oxford Biomedica, a small nearby biotech company, to get domestic production off the ground. In mid-May, the British government announced it had preordered 100 million doses, and Operation Warp Speed, former President Donald Trump's vaccine program, pledged as much as $1.2 billion to accelerate the shot's development and secure 300 million doses for the US Oxford soon started its Phase III trial in the UK Everything seemed to be coming along.
Out of the public eye, though, the scientists at Oxford realized they'd miscalculated the concentration of the vaccine, which had led some Phase III trial participants to receive a half-dose. Around the same time, the team decided to move from a one-shot to a two-shot regimen after seeing signs it would produce better protection, settling on a dosing interval of about four weeks. The UK regulator authorized them to continue with the half-dose group and follow up with a full dose, but the miscalculation and the decision to add a second shot meant the team would need to make more vaccine. That in turn delayed the second jab for a large number of volunteers.
Continuing with the half-dose group seemed justifiable to many of those in the loop, given that the pandemic was accelerating, a vaccine was urgently needed, and it's normal to try different dosage regimens, if not usually on the fly. But some US officials heard about the changes to the protocol and thought, "What are these guys doing?" according to a former senior Trump administration figure. The moves undermined American confidence in the British developers—just as the NIH was working with AstraZeneca on setting up the US Phase III trial.
Astra's team had been hoping to start the US trial in July, which ended up being about the time Moderna and Pfizer began their Phase IIIs. But US officials wanted to make sure manufacturing of the vaccine could be scaled up and tested before an expensive trial began, and that took longer than expected, according to three former senior officials. Also, the Food and Drug Administration and some Warp Speed officials were asking for detailed immune-response data in over 65s, people familiar with the situation say; Oxford had been slow to recruit older adults to its large UK trial. With the delays, AstraZeneca wasn't able to begin dosing volunteers in the US until the end of August.
Only a week later, the health news service Stat reported that the US trial had been paused after a volunteer in the UK experienced an unspecified illness. It soon emerged that vaccination had been suspended in all the partnership's global trials as the situation was investigated. It isn't uncommon for clinical trials to be halted as a precaution, especially in the large-scale final phase—it's a sign that safety is being taken seriously. But this wasn't a normal trial; it was one of the most closely watched scientific projects in history.
The first reported pause to a coronavirus vaccine trial set off a media storm. Under huge pressure to divulge further details, the developers refused, citing participant confidentiality—standard practice with such events and one that usually goes unnoticed. A throwaway comment by Matt Hancock, the UK health secretary, then revealed that Oxford's Phase III trial had been paused over a safety concern once before.
During a subsequent private investor call, Soriot revealed some of the details journalists and scientists had been demanding. The participant was a woman who'd begun suffering from neurological symptoms consistent with transverse myelitis, an inflammation of the spinal cord that can be caused by a viral infection. Soriot said it was unclear whether this specific condition had been diagnosed and whether the vaccine was to blame. Again, the information leaked, and again, a storm ensued.
Soriot's selective disclosure renewed the demands for details, which the company once more refused to divulge. The woman who suffered from the adverse reaction was complaining that her health was being dissected in the media around the world, according to people familiar with the situation. Releasing additional information about her condition before a full investigation was conducted could also have compromised the integrity of the trial and potentially encouraged other participants to report similar issues where none were occurring.
Regulators in Brazil and the UK cleared Oxford to resume its trial roughly a week after the pause began, but the US kept the study there on hold for almost seven weeks as Pfizer and Moderna zipped ahead with theirs. American officials were asking Astra for detailed information on every neurological event in any participant who'd ever received the vaccine, according to people familiar with the request. Three former senior US officials say they grew frustrated at how long Astra was taking to provide them the data they requested. The FDA had no desire to slow things down given the urgency of the pandemic, one former official says. Other companies were able to respond quickly to FDA requests, and several US officials began to assume AstraZeneca and Oxford were having problems communicating.
Some working on the UK government's vaccine effort were baffled by the holdup. On Oct. 12, Stat reported that Johnson & Johnson had paused its trial because of an unexplained illness. Like AstraZeneca, the American company didn't announce the pause beforehand and declined to release details, citing patient privacy; it later revealed in a report to the FDA that a 25-year-old male volunteer had suffered blood clotting in his brain, resulting in cerebral hemorrhage. After concluding that the event was unrelated to the vaccine, J&J resumed its US trial, saying it would start back up less than two weeks after the pause began.
By then some American officials had soured on the AstraZeneca vaccine, safe in the assumption they had Pfizer and Moderna in the bag, with Johnson & Johnson not far behind. "AstraZeneca looked like the gang that couldn't shoot straight," says the former senior Trump administration figure. "It was almost an attitude of: If they make it, they make it; if they don't, they don't. But we're not going to go out of our way to make AstraZeneca successful. That's for sure. They had fumbled the ball so many times."
By November, Pfizer and Moderna had produced a set of clear results for their vaccines in US trials, showing efficacy of about 95%—better than anyone had dared hope for. When Oxford reported interim results from its Phase III UK and Brazil trials on Nov. 23, they were also good, but they looked messy by comparison.
Whereas Pfizer had enrolled more than 40,000 participants, and Moderna had more than 30,000, Oxford reported interim efficacy data on fewer than 12,000. Oxford also didn't produce a single efficacy figure, instead coming out with two that seemed counterintuitive—an issue that traced to the now-public fact that one group had been given a half-dose for their first shot. The vaccine was 62% effective in one cohort who got two full doses spaced about a month apart. Surprisingly, in the smaller group of about 2,700 people who'd received the half-dose for their first shot, the number rose to 90%.
Initially, Oxford and AstraZeneca offered conflicting stories about what had happened with the dosing. Sarah Gilbert, the Oxford scientist who led the vaccine's development, explained it away as a function of experimentation. "We don't want to stick to a very low dose and discover we have an immune response that is too low," she told Bloomberg News. "On the other hand, we would like a vaccination regimen that's well-tolerated." The next day, Pangalos told Reuters the Oxford scientists had actually "underpredicted the dose by half." He cast the lower dose's relative success as "serendipity."
The muddled trial, coupled with Oxford and Astra's inability to coherently explain what had happened, overshadowed the strong efficacy number for the smaller group in many people's minds. As one former senior UK official says, it "wasn't their finest hour."
Although relations between AstraZeneca and Oxford didn't appear strained through all this, a common refrain emerged. When AstraZeneca executives talked about the UK trial, they would say, "Well, it's Oxford's trial. We had nothing to do with it." When Oxford scientists were asked about delays to the US trial, they would say, "Well, that's AstraZeneca's trial. We're not involved." Bell says the relationship is good. "We're still in there swinging with AstraZeneca, trying to help. We've been fused at the hip from the beginning."
The UK regulator authorized the Oxford vaccine on Dec. 30, recommending going as long as 12 weeks between shots. Its experts concluded that the higher efficacy number for participants who'd received the half dose for their initial shot likely owed to the increased interval before the second dose, rather than to the amount of the first one. Production in the UK was by then up and running at several sites. The country was preparing for a major rollout starting on Jan. 4, about a month into a national vaccination program that had begun with the shot from Pfizer-BioNTech. Oxford Biomedica had worked for months to iron out the kinks in the manufacturing process for the AstraZeneca shot, overcoming yield issues and learning to grow cell banks that would prove capable of producing 2 million doses a week. By early spring, Astra would account for a little more than half of all doses administered in the UK, with data showing that it was driving down deaths and hospitalizations.
The rollout was much slower elsewhere in Europe. In late January, a few days before the European Medicines Agency granted its approval for Astra's shot, the company announced that it would be able to deliver only 31 million of the 120 million doses it had originally planned to supply the EU in the first quarter. The bloc hadn't signed its contract, for as many as 400 million doses, until the end of August, leaving European manufacturing months behind the UK's. Output at Astra's Belgian subcontractor was proving lower than expected, and the EMA wouldn't grant authorization for another facility, in the Netherlands, to begin production until the end of March.
News of the shortfall enraged officials in Brussels. Thierry Breton, a former businessman who serves as the EU's internal market commissioner, tried to persuade his fellow Frenchman, Soriot, to boost deliveries, to no avail. Making the optics worse, Soriot wasn't physically present in Europe. He'd been working from Sydney, where his family lives, since before Christmas. Breton told a French radio station he awoke every day at 4:30 a.m. to call Soriot. "I'm not criticizing him. Each of us has to manage this situation as best they can," Breton said on Europe1 radio on March 14. "But you see, nearly every week I go to two or three factories. I'm not saying I know their factories better than they do, but I am on site." (AstraZeneca says Soriot is working around the clock from Australia.)
In mid-March, European Commission President Ursula von der Leyen renewed an earlier threat to ban the export of all vaccines manufactured in the EU. Paranoia that Astra doses were being siphoned off from Europe to the UK crested a few days later, when Brussels ordered Italian leaders to have police raid a vial-filling factory in the town of Anagni that had 29 million doses on-site. The stock turned out to be destined for Europe and Covax.
The supply pressure came even as European leaders were raising efficacy and safety concerns about the vaccine. In early February, at least 10 European countries had restricted use of the shot in over 65s, given the lack of data for that group from Oxford's Phase III trial. More bad news would follow from South Africa, where the Oxford trial showed the vaccine provided minimal protection against the dominant variant there. Then, in mid-March, Denmark, Norway, and Iceland announced they were temporarily suspending the Astra shot to investigate a rare form of blood clotting that had been detected in several vaccine recipients, at least one of whom had died. A few days later, Germany said it would suspend use of the shot, with several European countries following suit even as the EMA recommended continuing. Most resumed use after the agency reviewed the data again and concluded that the benefits of the shot far outweighed the risks, but it wasn't long before more cases appeared. At the end of March, Canada halted AstraZeneca shots for people 55 and younger, and Germany reimposed its suspension for under 60s. "We must be able to trust the vaccines," German Chancellor Angela Merkel said, looking exhausted. "These are findings we cannot ignore."
She was referring to 31 German cases of a form of clotting, known as cerebral venous sinus thrombosis (CVST), that had been reported out of about 2.7 million doses administered—an incidence rate higher than normal. Some of the cases were accompanied by low platelet levels; most involved younger and middle-aged women. Nine patients died.
The EMA investigated 86 cases of blood clots, 18 of them fatal, in Europe and elsewhere that had been reported as of March 22, out of 25 million people who'd received the vaccine. On April 7 the agency said that it had received a total of 222 reported cases of brain and abdominal clots and that the data didn't indicate a cause or specific patient profile. The situation "clearly demonstrates one of the challenges posed by large-scale vaccination campaigns," said Emer Cooke, the EMA's executive director. "When millions of people receive these vaccines, very rare events can occur that were not identified during the clinical trials."
The UK regulator said the overall incidence of cerebral clots was about 4 cases for every 1 million people who'd received the vaccine. The reported risk in Europe is 1 in 100,000, according to the EMA.
(The much lower number in the UK could be because the country has yet to start vaccinating under 50s, though the EMA has said there's no evidence the clots are connected to age or gender.) The Johns Hopkins University School of Medicine estimates that normally 1 in every 200,000 people of all ages develop CVST in a given year.
In an April 7 statement responding to the latest regulatory findings, AstraZeneca said it would be making recommended changes to the vaccine's label, and it was "already working to understand the individual cases, epidemiology and possible mechanisms that could explain these extremely rare events."
Any hope AstraZeneca executives may have harbored that their larger and more robust US trial would reassure the world that their vaccine was safe and effective were dealt a major setback when the safety board's letter leaked. The runup to the interim data release had been fraught—even by the standards of the pandemic. Astra was under pressure from European regulators, who wanted more insight into the vaccine's efficacy in seniors, and then into the threat of clotting. Executives were also acutely aware of the heat the company had received in the past for not being forthcoming, and most importantly, Astra was under strict obligations, as a publicly traded company, to report material data quickly. On March 21, the day before the interim results were set to come out, executives got an emotional curveball when they learned that Astra's head of oncology research, José Baselga, had suddenly died.
That same Sunday, AstraZeneca informed some US officials about their assessment that the later cases wouldn't appreciably change the results. In a video call attended by Fauci, NIH Director Francis Collins, and other US health officials (but no one from the DSMB), the company presented its results and stepped everyone through the unvalidated additional cases, noting that the final efficacy number might drop by a couple of percentage points and would actually go up for the elderly. There was no pushback, according to people familiar with the discussion.
In announcing the interim results on Monday, Pangalos did mention that additional cases had come in, but Astra's decision not to acknowledge in its statement how this might affect its efficacy figure stood in stark contrast to what Pfizer and Moderna had done months before. In Pfizer's case, the initial estimate of 90% efficacy was upgraded two weeks later to 95%—a win on both the efficacy and public-relations fronts. The DSMB's letter, by contrast, said that Astra's efficacy could drop to between 69% and 74% once all the cases were validated—a calculation it didn't share with the company until after it had issued its release. In itself, that wasn't unusual; what was unusual was calling out the company publicly for its number. The DSMB is primarily responsible for reviewing safety data and deciding when a trial has met a threshold for efficacy. Final figures are normally decided by independent statisticians working with the company on pre-agreed terms, and disagreements with the DSMB are usually worked out behind closed doors.
Polly Roy, a professor of virology at the London School of Hygiene and Tropical Medicine, says this was a case where precision should have been paramount for AstraZeneca. "They should not have said a particular number if they weren't sure," she says. "There's nothing wrong with the vaccine itself. It's the way they reported on the clinical trial. That was the wrong way to do it. Communication is the problem."
Other experts say Niaid's public intervention was unnecessary and extremely damaging. "It was grossly irresponsible for the National Institutes of Health to publish statements that there was something improper about the way the data had been collected and therefore you couldn't trust it, which is how it will have been read no matter what it actually said," says Peter English, former chair of the British Medical Association's public health medicine committee. The inference, he adds, is "that things might have been terrible. This brings the whole vaccine into disrepute. It makes people think there's something dodgy about it. It will affect confidence in the AstraZeneca vaccine but also vaccines in general."
Astra's next major hurdle will be to secure an emergency use authorization from the FDA. The company plans to submit data for review in the first half of April, but a final ruling could take weeks. Astra executives are under no illusion that the process will be smooth. They must continue to work with the DSMB, and they have one of the most complex datasets of any Covid vaccine developer, with thrombotic events and real-world evidence from millions of people vaccinated. They also got a new challenge at the end of March when one of their main subcontractors in the US, Emergent BioSolutions Inc., mixed up ingredients from Johnson & Johnson and AstraZeneca, a major manufacturing error. Emergent will now exclusively produce J&J doses while US officials work to find AstraZeneca new manufacturing capacity.
Many people inside Astra are deeply distressed by the extent of the criticism they've taken while trying to lead the way to a nonprofit vaccine that could help end the pandemic. As one executive puts it when asked if they'd do it again: "Not in a million years. All we've had is grief."
Stephanie Baker is a Senior Writer at Bloomberg News/Businessweek for projects and investigations.
Suzi Ring is a Healthcare Reporter for Bloomberg. Ex-legal reporter and all things Brexit.
Disclaimer: This opinion first appeared on Bloomberg, and is published by special syndication arrangement