LONDON – Astrazeneca plc promised to publish the primary analysis of the phase III U.S. trial of its COVID-19 vaccine by the morning of March 25, after chastisement from the data safety monitoring board (DSMB) for using “outdated” information in the interim results published on March 22.

That followed a statement from the U.S. National Institute of Allergy and Infectious Diseases (NIAID), put out late on Monday night, saying the DSMB had expressed concern the interim data showing the vaccine is 79% effective “may have provided an incomplete view.”

NIAID said, “We urge the company to work with the DSMB to review the efficacy data and ensure the most accurate up-to-date efficacy data be made public as quickly as possible.”

In its statement, issued shortly before noon U.K. time on Tuesday – and just as Anthony Fauci, director of NIAID was being interviewed about the matter on “Good Morning America” – Astrazeneca defended its position, noting the data cutoff of Feb. 17 was prespecified for the interim analysis.

The Cambridge, U.K.-based company said it has now reviewed the preliminary assessment of the primary analysis and has almost completed the statistical validation. Those results, which are “consistent with the interim analysis” will be immediately shared with the DSMB and published within 48 hours, the company said.

That may make the NIAID’s extraordinary intervention look like stirring up a storm in the teacup. But letting the presentation of data showing its COVID-19 vaccine is highly effective and safe blow up in this way is another misstep by Astrazeneca and its academic collaborators in the way trial data have been reported.

What makes it more unfortunate is that the U.S. phase III trial was expected to allay concerns about apparent inconsistencies and protocol changes in the U.K., Brazil and South Africa phase III trial run by Oxford University, on which conditional or emergency use approvals granted in 70 countries worldwide are based.

Amid distractions

Presenting the data on March 22, Mene Pangalos, executive vice president of biopharmaceuticals at Astrazeneca and Ruud Dobber, president of the company’s U.S. biopharmaceuticals business unit, described getting the data from the DSMB over the weekend and discussing it with U.S. government officials on Sunday morning.

That was happening as they were dealing with the news that their colleague, José Baselga, executive vice president of oncology R&D, had died at the age of 61.

It was also against the backdrop of a diplomatic row in Europe over supplies of the vaccine, with the EU threatening to introduce export controls.

But it seems it was case reports from Europe of thrombotic events that derailed activities. Those reports prompted the DSMB to call in a neurological specialist to look for any similar problems in U.S. phase III participants, delaying publication of the interim analysis. With the full primary analysis now promised for Thursday, Astrazeneca apparently has the full dataset, as the DSMB knows, and hence its notification to NIAID that the company “may have included outdated information” from the phase III trial.

Pangalos was completely open about the fact there were more data in the bag, saying the interim analysis was based on 141 events and there was “more to come for the full primary analysis.” The DSMB apparently expected the company to present all of the available data on March 22.

Another factor may be that both Astrazeneca and its academic partners at Oxford University were distracted by focusing on preparing to answer questions about the differing efficacy results from the U.S. phase III, and the Oxford-led U.K., Brazil and South Africa phase III.

Oxford University’s press release made a pre-emptive strike, saying absolute efficacy in the U.S. phase III was higher than the Oxford-led trials because of the protocol case definitions and the populations in which the vaccine was studied.

Andrew Pollard, professor of pediatric infection and immunity, who was lead investigator for the Oxford trials of the vaccine, told attendees of the briefing held to discuss the interim data, “Every trial protocol is different in the way data is collected and [in] case definitions, and also health care systems may differ.”

Given Astrazeneca now says the primary analysis is in line with interim data, the DSMB’s complaint may be a mere technicality.

It was NIAID, not the DSMB, that went public about the issue. NIAID had put out its own press release repeating the data in Astrazeneca’s release, and Fauci spoke to the media about the results in glowing terms, which may explain NIAID’s intervention, once it had seen the DSMB’s complaint.

As Stephen Evans, professor of pharmacoepidemiology at the London School of Hygiene & Tropical Medicine, commented, “It is not unknown for a DSMB to disagree with investigators over interpretation of trial results. It is usually done in private, so this is unprecedented in my opinion.”