Joining Novartis, Janssen and Eli Lilly, Pfizer and Roche’s membership in the consortium, known as the ADDPLAN DF consortium, seems rooted in their desire to improve dose-selection in late-phase clinical work, which remains a major barrier to resolving high failure rates in Phase III trials.
Andy Grieve, SVP of clinical trial methodology at Aptiv, told Outsourcing-Pharma.com, “The goal of the consortium is to advance statistical dose-finding methodology and translate these new methods, plus already existing methods, into user friendly dose-finding software that will be available to statisticians across industry. Additionally the consortium aims to develop any necessary execution solutions for these designs in order to make sure that implementation is not a barrier to their uptake.”
The ADDPLAN DF Consortium statisticians are currently collaborating to expand the utility of the Multiple Comparison Procedure and Modeling (MCP-Mod) approach, specifically by developing robust methodology for incorporating adaptive functionality.
“Whilst there are other software packages that provide tools for dose-response studies this is the first package, validated to industry standards, that provides the ability to design, simulate and analyse studies utilising the MCP-Mod approach recently qualified by the EMA [European Medicines Agency],” Grieve said.
The current MCP-Mod procedure enables Proof of Concept (PoC) and dose selection to be established in a single trial.
“The capabilities in the software to support the design, simulation and analysis of early-phase dose-response studies have been enhanced in the most recent release by the provision of Bayesian dose-escalation designs for Phase I and Phase IIa,” Grieve explained. “In the subsequent releases the family of early-phase Basyesian dose-escalation designs will be extended and an adaptive version of MCP-Mod will be developed by the Consortium members to allow changes to the allocation rates to be made at an interim analysis based on accumulating data. The consortium members will be critical to the success of these developments as their experience in designing and running such complex designs will ensure that the implementation meets the needs of pharmaceutical sponsors.”
He also explained to us the way in which the MCP-Mod approach may be more beneficial than pairwise comparisons.
“MCP-Mod approach gets round the use of pairwise comparisons for proving a dose-response effect. Optimal contrasts based on assumptions of likely dose-response shapes provide weighted scores for the different dose-levels,” Grieve explained. “This allows the determination of true dose-response effects with a greater power, compared to a standard pairwise comparison. Additionally, model-based optimal contrasts do not suffer from multiplicity issues and so dose-levels can be added to a study to test PoC [Proof of Concept] without an additional penalty for multiplicity. Such additional dose-levels provide increased sensitivity for determining a reliable working model for the estimation of an appropriate dose for Phase III.”