Novadiscovery’s trial simulation predicts AstraZeneca phase 3 results
The French-U.S. company Novadiscovery has hailed a “watershed moment for clinical trial design” as its trial simulation tool successfully predicted the outcome of a phase 3 oncology trial run by AstraZeneca.
“This further proves that in silico trials offer a critical, reproducible and customizable tool to power pharmaceutical clinical trial designs in the future,” said François-Henri Boissel, co-founder and CEO of Novadiscovery, in a public statement.
The trial followed by Novadiscovery was the phase 3 FLAURA2 trial run by AstraZeneca. Since cancer patients often develop resistance to AstraZeneca’s non-small-cell lung carcinoma treatent osimertinib, the study investigated whether combining osimertinib (Tagrisso) with chemotherapy and maintenance therapy in patients with non-small cell lung cancer would extend their progression free survival compared with osimertinib alone.
A project led by Michaël Duruisseaux at the Université de Lyon, France, used Novadiscovery’s clinical trial simulation tool jinkō to predict the results of the FLAURA2 trial. The team used publicly available data from AstraZeneca’s phase 1 and 2 populations and the published FLAURA2 trial protocol. They then designed 5,000 virtual patients based on demographic data from 30 lung cancer patients in the safety run-in trial of FLAURA2, and Novadiscovery’s own non-small cell lung cancer disease model.
Days before the release of the FLAURA2 trial results, Duruisseaux posted the group’s predictions on the social media site X, formerly known as Twitter. They predicted that the risk of disease progression would be around 40% lower in the combination therapy arm than in the osimertinib arm. They also predicted a median time to progression (a surrogate of median progression free survival) at 25.9 months in the experimental arm, more than 17.3 months in the osimertinib arm.
Results resemble prediction closely
The actual results released by AstraZeneca last week resembled the prediction closely. According to the findings, the risk of disease progression or death in patients given the combination therapy was 38% lower than patients given osimertinib alone. The median progression free survival was also higher in the combination therapy group at 25.5 months compared with 16.7 months in the osimertinib arm.
The collaboration ran for three weeks and the simulation process itself took one hour, far less than the 33 months it took to complete the trial itself. Neither entity worked with AstraZeneca or obtained proprietary information to complete the simulation.
“Results like these, if leveraged before human trials begin, will enable the recruitment of the most relevant patient population, optimize trial design, and ultimately accelerate all therapeutic development in our shared hope that cancer becomes an eminently curable disease,” said Duruisseaux in a public statement. He added that in silico trials could be used to build the statistical hypothesis of the next generation of future trials.
Novadiscovery was founded in 2010 and launched its jinkō platform in 2022. The technology allows customers on many scales to plan out clinical trials, including biotech and pharma companies in addition to academic research centers and university hospitals.
The in silico clinical trials market was valued at around $3 billion in 2023 and is expected to grow by 8% per year up to 2033, according to a report from Visiongain. The closest competitors to Novadisovery include Certara and SimulationsPlus, Boissel told Outsourcing Pharma, adding that Novadiscovery stands out with its capacity to industrialize the practice of quantitative systems pharmacology (QSP).
“It takes time and considerable expertise to build a disease model, as human biology is exceptionally complex,” said Boissel. “Our jinkō platform is built with the objective to make it much more efficient by reusing pre-built models, and utilizing parallel computing to run trial simulations on thousands of patients in seconds or minutes.”
While no computational model will be a perfect replacement for human trials, Boissel noted that using trial simulations “can help identify the best responding patients and therefore reduce the human trial sample size.”
