Even before the era of COVID-19, significant attention was channeled to the overwhelming potential of adaptive MAMS designs. Short for multi-arm multi-stage designs, these trials enable numerous therapies to be tested on a single platform with a single comparator arm. When patients are too few or there are several therapies in competition with each other to enroll, adaptive MAMS designs expedite the discovery of new drugs.
This feature of adaptive MAMS designs is undoubtedly why it has been so attractive for organizations conducting COVID-19 research. The World Health Organization’s SOLIDARITY trial, the RECOVERY trial conducted in the UK, and the ACTT-3 Trial sponsored by the National Institute of Allergy and Infectious Diseases in the USA, have all used an adaptive MAMS platform. Together they have managed to enroll thousands of patients within months, reaching conclusive findings in highly accelerated timelines.
As adaptive MAMS designs gain normalcy in clinical development and medical discovery, statisticians are confronted with the question of which methods for adaptive MAMS are most likely to deliver gains in power rapidly. There are traditional adaptive MAMS methods (Stage-Wise methods) and newer Cumulative MAMS methods.
According to new research by Cyrus Mehta, which compares Stage-Wise MAMS methods to Cumulative MAMS methods, the Cumulative method displays remarkable strengths over Stage-Wise MAMS.
Mehta’s research demonstrates conclusive evidence that Cumulative MAMS generates significant gains in power over the Stage-Wise methods. A recent case study of the SOCRATES trial shows a 5% power gain when using a Cumulative MAMS methods, corresponding with 60 fewer patients needing to enroll to obtain the same power. Mehta and his team also generalize these results using both analytical methods and simulations, displaying clear dominance of Cumulative MAMS.
Additional to the gains in efficiency produced by adaptive MAMS methods, there are also regulatory implications for their more widespread use. Currently adaptive MAMS designs must demonstrate strong control of the Family Wise Error Rate, a higher hurdle to overcome than a pairwise comparison of one new therapy to a control. This essentially disincentivizes larger companies from participating in collective platforms that utilize adaptive MAMS designs. Cyrus Mehta ends the webinar by discussing whether strong control of FWER should be a necessary regulatory guide, given the efficiency gains obtained with such designs.
To learn more about Cyrus Mehta's new research and hear the discussion about regulatory options, listen to a replay of the webinar by clicking the button below:
