How Pfizer’s ‘Science Will Win’ Mantra Played Out Behind The Scenes
March 8, 2021 | One year after COVID-19 was declared a pandemic by the World Health Organization, a handful of Pfizer clinical operations leaders talked openly about the “one-team mindset” that helped usher a safe and highly efficacious vaccine to market in record time. Over 2 billion doses of the newly authorized mRNA vaccine, co-developed with BioNTech, are expected to be in arms by the end of 2021—a manufacturing feat that is itself headed for the history books.
Early on, Pfizer launched a “Science Will Win” campaign coloring dialogue around the vaccine both inside the company and in the court of public opinion. The copyrighted phrase has been used in ads run on television and social media and has been emblazoned on t-shirts worn by the study team to daily morning meetings.
Vaccine themselves “used to be niche but not anymore,” says Nicholas Kitchin, M.D., senior director in Pfizer's vaccine clinical research and development group. The first three authorized for emergency use against COVID-19 by the U.S. Food and Drug Administration (FDA) could help end the pandemic.
Kitchin was among a team of eight presenting on Pfizer’s vaccine trial execution strategies at the recent Summit for Clinical Ops Executives (SCOPE). He was joined by Helen Smith, team lead for clinical development and operations; Beth Paulukonis, study management group lead for clinical development and operations; Darren Cowan, head of site management and monitoring for North America; Brett Wilson, vaccines data monitoring and management lead; Ralph Russo, global head of clinical database management; Liping Zhang, vaccines statistical programming and analysis lead; and Demetris Zambas, global head of data monitoring and management for biometrics and data management.
Pfizer’s vaccine story began in 2018 when it joined forces with BioNTech on a mRNA-based flu vaccine, says Kitchin. The companies pivoted to co-development of a COVID-19 vaccine, using the same mRNA platform, after reports emerged from China about the novel and rapidly spreading coronavirus.
The development team initially relied on what was known about severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) which, like SARS-CoV-2, are caused by coronaviruses. Flexibility was crucial, as was collaboration with regulatory authorities, Kitchin says. “We had no idea where this path would take us. We had to be ready to respond to changes quickly.”
Early development was conducted largely in the U.S., and phase 1, 2 and 3 trials were combined in a single study that has “served us extremely well these past 12 months,” Kitchin continues. Phase 1 trials utilized small, specialty sites where multiple vaccine candidates were tested in different dosages and age groups to select the best product and dose.
COVID-19 being brand new, he adds, there was no threshold-of-protection bar to be met and “that remains the case. So, we had to do placebo-controlled trials and count cases.”
Phase 2 and 3 trials broadened to over 46,000 participants and launched in locations around the globe expected to have high rates of infection in the coming weeks based on “calculated guesses,” says Kitchin. “We came to accept that protocol amendments would be a part of our life.”
Study sites committed to submitting data daily, so information amassed quickly, Kitchin says. Since the vaccine was untested in humans until May, daily review of safety data at the individual and aggregate levels was paramount. Only a separate, external data monitoring committee (DMC) had an unblinded view of safety signals.
The CEO “put everything” behind efforts to develop a vaccine in an extremely short period of time without shortchanging quality, he says. “We’ve had more than 50 emergency use marketing approvals around the world.” As of late February, 100 million doses of the Pfizer/BioNTech had already shipped out globally.
All this required cross-functional collaboration by many team members, Kitchin notes. The effort involved over 1,000 colleagues representing 18 Pfizer functional lines and 20 vendors partners.
All Hands On Deck
The leadership of Pfizer and BioNTech gave the internal team permission to move at pace and remove the roadblocks—and the sites, vendors, and regulators they worked with were similarly motivated, says Smith. Their one-team mentality fostered open communication where everyone was free to express their opinions and frustrations and have their feelings validated.
Progress toward goals was reviewed during daily calls, she says. Challenging timelines were met because of the focus on the “removal of white space” from standard processes, so what might normally take a week would get done in two days. Even so, “almost excessively” long hours were put in by many people across several continents.
“If there was a meeting for the COVID study, you attended,” Smith notes. Reporting dashboards provided critical, almost-real-time operational metrics, including diversity rates, helping to proactively identify processes that needed to be modified. “Senior leaders were involved every day.”
Upwards of 20 study managers, versus the typical one or two, were assigned to the vaccine trial, says Paulukonis. Sites were entering data and responding to queries on a same-day basis and a commensurate level of commitment came from CROs and vendor partners—even as those individuals were being personally impacted by the pandemic as they quarantined at home.
Importantly, people were placed where they would excel and trusted to deliver while remining interconnected with one another, Paulukonis says. “We needed each other to succeed… and pivot when we needed to,” she adds, referencing the one-team mindset and dashboard data that would change on a nearly hourly basis.
Team members, most of whom had never worked together previously, became fast friends, she says. Emotions ran high the day the vaccine’s high efficacy (nearly 95%) was announced because of what it meant to their parents, siblings, and friends.
It was past midnight in the U.S., Zambas says, and the team broke out in “songs and poetry.”
Starting At Speed
The trusted partnership model is what enabled the fast pivots and real-time dialogue with investigators and sites, and it had to be meaningful since trial participation involved showing up at sites, says Cowan. In some instances, he adds, recruitment by investigators was paused when the number of COVID-19 cases in the area was low and picked back up again when it was rising.
An internal team provided dedicated support for every detail in the protocol and external vendors were used to relieve burdens on sites as much as possible to provide a “flawless customer service experience,” he says. The pre-study work of the dedicated internal team focused on getting sites up and activated at the right time, and what issues needed to be addressed with ethics committees.
A high-trust relationship was built up front and leveraged throughout the study with a “one-stop person” at Pfizer available for problem-solving, Cowan says. Equity and access were calls to action for sites since people of color were disproportionately impacted by COVID-19, and Pfizer provided study materials to help reach into those minority communities in real time.
“Not a lot of time was spent second guessing things or making a plan B,” he says. “Plan A worked,” in part because sites were also highly engaged. “Investigators felt like they have a true partnership with us, and they are still fully committed.”
Within a day of announcing the partnership with BioNTech to develop the vaccine, Pfizer began pulling together the internal team, says Wilson. The focus was on identifying a core set of individuals who could “start at speed” and would subsequently be putting in the most hours. Within four weeks, the study protocol spanning the three trial phases was written.
Dealing With Data
The vaccine trial veered significantly from pre-COVID norms when studies involved many sequential activities and multitasking by team members, says Russo. New standards also had to be developed for database builds involving mRNA technology.
The clinical database was ready five days after the approved protocol, Russo says. Pfizer took a staged testing approach that started with the patient report form, in addition to robust database unit testing and peer review.
Accommodating the cadence of the study’s adaptive design also consumed a lot of overtime hours, he adds. Data cleanup tools were built in parallel with database components.
The conduct phase rollout was a “massive undertaking” involving over one million forms, more than 150,000 queries, and 34 million discrete data points, according to Wilson. Matching the inflow of data with how it was validated, monitored, and reviewed required close coordination across multiple roles.
Alignment was achieved through a variety of “sprints,” or short meetings, focused on a single deliverable such as data validation review and central monitoring frequency outputs, Wilson continues. One notable output was “participant visit to clean data in four days,” which was easily accomplished with commitment to the goal across every function, including sites.
Leading into the FDA’s Emergency Use Authorization of the vaccine, says Wilson, the team committed to additional batch cleaning targets and ensuring investigator attestation to the data. Throughout the trial, a clean data rate of 90% to 95% was maintained with the help of the customized dashboards.
To fit the enrollment pace of the study, central monitoring of safety signals happened four days a week versus the more typical once-a-month reviews, Wilson says. Cross-functional review happened on data surveillance calls convened one day each week. Quality tolerance limits were also established within the first month and reviewed regularly.
Every system that had data flowing through it was tested during key events, such as database changes and database locks, he adds. Most problems were prevented, and the rest were quickly fixed. “Anyone working on COVID received enhanced support as a technical user from our help desks,” says Wilson. Additional laptops were rolled out as needed.
Data refresh rates allowed visibility to insights “multiple times a day,” Wilson says. Transfers of trial data, included patient-reported outcomes, happened daily and the electronic data capture system used by stakeholders was refreshed every six hours.
Over the course of four months, the machine learning tool Smart Data Query (SDQ) adjusted and reconciled over 103 million combinations of clinical data points to detect discrepancies and prompt queries. SDQ is credited with reducing the cycle time to identify and call out a query by 50% within the COVID-19 vaccine trial and by more than 95% relative to pre-pandemic norms.
‘No Magic Fairy Dust’
The development timeline ran from March 17 through Dec. 2 (when the vaccine received the world’s first EUA by the Medicines & Healthcare Products Regulatory Agency in the U.K.) and beyond, says Zhang. The phase 1 study started in the U.S. on May 4, and her team’s job was to summarize the data for safety and immunogenicity to help the internal review committee determine the best vaccine candidate and dose to bring to subsequent phases.
Three days after the first patient was enrolled, the team was needed to provide support and deliver the output, she says. Immunogenicity data had to be delivered overnight for next-day review once a week for several months.
In less than three months, a single vaccine candidate was selected to take into phase 2/3 trials, says Zhang. The team produced numerous safety and DMC reports and, starting in early September, ran daily case counts that triggered the first interim analysis for efficacy on Nov. 8. The programming team delivered the study positive interim analysis report to the DMC within 24 hours, she notes.
The study met efficacy targets and FDA-required safety endpoints on Nov. 18 and, within the space of three days, Pfizer and BioNTech submitted its EUA request to the FDA. The rapid turnaround was the result of an around-the-clock team effort, she says.
Between May 4 and Nov. 20 her group made 70 data cuts, each with its own reporting event and delivered in a matter of hours, she says. Multiple EUA submissions were made globally and responses to queries from regulators were made within two days and, often, within 24 hours.
From programming’s perspective, Zhang cites seven key factors that contributed to accelerated development and EAU submissions for the vaccine: teams with a dedicated focus to their deliverable, use of programming standards, limiting output to key messages, leveraging technology (internal CDISC system and Pinnacle 21 software), regulatory engagement, a rigorous quality control process, and planning and cross-functional alignment.
“There was no magic fairy dust,” says Zambas, just a “team of chefs demonstrating their skills.” Redefining R&D has been a 10-year journey for Pfizer and responsible for the existence of his department, he adds.
In planning for the mRNA vaccine trial, “we were given very clear instructions from leadership,” Zambas says. “Science matters, quality matters, and the rest is a blank check.”
Other than a cohort of 12-to-15-year-olds, the study has been unblinded for most trial participants in the interest of getting as many people vaccinated as possible, says Kitchin. Ongoing safety monitoring is being handled through routine adverse event reporting processes as well as active database safety studies.
The upside of that is that huge numbers of people around the world have already received the vaccine and the efficacy profile demonstrated in the study appears to be reflected in real-world use, based on published data coming out of Israel and the U.K. So far, the observed safety profile is also “very similar” to what was characterized in clinical trials, he adds.