New logic to fix a flawed system
Dr Phil Birch discusses adaptive design in clinical trials and how the clinical development landscape is evolving.
The interplay between innovative statistical methodology, smart trial execution platforms, and clear guidance from regulators, is dramatically changing the clinical development landscape and heralding a new dawn for the efficient development of medicines to treat serious disease.
The solution to fixing the high failure rate of clinical development is not to employ temporary fixes to pare down the cost of failure, but to create a new, smarter system that redefines the fundamentals of clinical trial design and execution, and thus the odds for success. In 2013, three milestones were achieved along the path toward redefining the traditional approach to testing clinical hypotheses and should compel sponsors to rethink their current trial design and execution infrastructure.
1. EMA Qualification Opinion on MCP-Mod
It is well known that the selection of the wrong dose in Phase II contributes to a fifty percent failure rate in Phase III trials, making dose selection one of the most difficult challenges in the drug development process.
There are two flaws in most dose-finding designs.
The first is that current approaches for estimating the target dose are prone to uncertainty, which can result in the use of an inappropriate statistical model that over- or under-estimates the true effective dose.
In a significant shift for the industry, the EMA released in October 2013 a draft qualification opinion that challenges the way most dose finding studies are performed, essentially saying that the number of doses and the dose range selected are sub-optimal.
The opinion focused on MCP-Mod, a statistical methodology for dose-finding studies that the EMA said has the potential to “enable more informative Phase II study designs” and “provide a more solid basis for all subsequent dose selection strategies and decisions.”
The MCP-Mod approach is unique in that it defines several plausible candidate dose-response models, tests them for significance, and then identifies the most appropriate statistical approach to model the dose-response and estimate the target dose. According to the EMA, MCP-Mod uses available data better than traditional pairwise comparisons, such as an ANOVA approach. MCP-Mod is an example of an innovative approach that will help fix some of the fundamental problems inherent in dose finding studies.
The second flaw is that sponsors also tend to focus on a very narrow range of doses. A recent analysis of Phase II trials registered on ClinicalTrials.gov between 2002 and 2011 showed that the majority of trials evaluated just two or three doses that covered, on average, a four-fold dose range.
Focusing on such a narrow region of the dose-response curve is risky and often means that finding the minimum effective dose requires sponsors to repeat studies, which increases trial cost, duration, and unnecessary patient risk. Innovative approaches such as adaptive dose finding, which allow inclusion of more doses without significantly increasing cost, are also gaining traction. These more sophisticated designs require a different approach to trial execution but are now achievable through the availability of integrated technology platforms specifically designed for running adaptive trials.
2. Increased Support from Regulatory Agencies for Adaptive Designs
By allowing predefined adjustments to the trial in reaction to real-time data, adaptive designs increase the utility of the information that these studies produce for patients, physicians, regulators, and the sponsor.
While simple adaptive designs such as futility stopping have obvious ethical and monetary benefits, the true benefit of adaptive design is in more sophisticated exploratory trials where a number of parameters, such as dose, endpoint and patient subpopulation, have to be assessed.
Adopting these approaches in exploratory development dramatically improves development decision-making, which when applied consistently at the product portfolio level, will drive significant commercial value.
Regulatory agencies increasingly support adaptive trials, and, in 2013, the FDA included potential priority review status for Investigational New Drug (IND) applications whose clinical studies employ adaptive design.
3. Success for I-SPY2
The industry is closely watching the progress of I-SPY2, an innovative Phase II adaptive trial already in progress for patients with cancer confined to the breast. The trial uses genetic profiles to highlight biomarker differences among patients and match them with drugs that predict a benefit based on those biomarkers. The adaptive design allows interim data to inform the drug assignments for patients who enter the trial later.
As part of the design, the I-SPY2 team can screen a number of investigational drug candidates from several companies in parallel. Drugs shown to be efficacious in targeted patients on the basis of interim data graduate to Phase III; those that show poor efficacy are dropped, and new candidate molecules are added in their place without having to stop the trial.
In December 2013, I-SPY2 graduated its first two successful candidates, AbbVie’s veliparib and Puma Biotechnology’s neratinib. Continued success of these investigational drugs in Phase III and beyond would be a major win for adaptive design.
In the same month, the Innovative Medicines Initiative (IMI) announced the launch of a €53 million adaptive design project for early stage clinical trials aimed at evaluating novel drugs in Alzheimer’s disease. This fund, open to private and public applicants alike, is nearly double the entire budget the NIH approved in September 2013 to fund Alzheimer’s research.
The IMI project model is similar to that of the I-SPY2 trial and is the first time a multi-product adaptive design approach will be used for Alzheimer’s disease.
IMI-funded trials will simultaneously compare several drugs to a placebo and place only 20% of patients into the placebo group, breaking away from the traditional design of one drug per trial and half of patients taking placebo. The adaptive design will enable decisions to be made based on the analysis of interim data, allowing drugs that are ineffective to be withdrawn, and new candidate drugs to be added in their place.
Traditional trial designs do not adjust to patient response and can in certain cases become a “one-shot educated gamble.” Rethinking the logic of current approaches to trial design and execution means investing in innovative methodologies such as MCP-Mod, adding flexibility using adaptive design, and deploying novel execution platforms that enable these sophisticated designs to be efficiently run.
The strategic application of these approaches will not only reduce costs, but also fundamentally increase the likelihood of success in later clinical development. This is essential for both large pharmaceutical companies struggling to drive productivity across their portfolio as well as biotech companies that grow as a consequence of success.
Looking forward into 2014, we expect to see similar reform and innovation in the area of clinical site monitoring as sponsors seek to trim costs. Currently, 100% source data verification (SDV) is the most common method of monitoring on-site data, and one of the largest contributors to trial cost.
The 100% SDV approach distributes quality control resources across all clinical trial sites equally, even though the quality of sites can vary greatly with some sites producing high-quality data, and others requiring greater scrutiny.
Rather than use brute force, smarter approaches exist that allow resources to be applied to sites that need it most. The cost savings generated through this can then be reinvested in quality assurance approaches that improve data quality by ensuring clinical protocol adherence, rather than simply monitoring for transcription errors in the trial database.
These approaches, called risk-based monitoring, are no less vigilant in the oversight of the trial, but rather provide a mechanism to focus on mitigation of important and probable risks. Regulatory authorities have released specific guidance on the acceptability of risk-based monitoring approaches1-3 and these strategies are beginning to be adopted across industry.
The key to success will be the use of approaches that are based on robust statistical methodologies and the deployment of trial execution platforms that enable data quality at individual sites to be assessed in real-time.
The solutions for redefining the trial design and execution landscape are clear and implementable. The interplay between innovative statistical methodology, smart technology platforms for trial execution, and clear guidance and support from regulators, heralds a new dawn for the successful development of medicines to treat serous disease.
1. EMA. Quality Risk Management (ICH Q9). 31 January 2011.
2. MDC/DH/MHRA Joint Project. Risk-adapted Approaches to the Management of Clinical Trials of Investigational Medicinal Products. 10 October 2011.
3. FDA. Guidance for Industry: Oversight of Clinical Investigations – A Risk-Based Approach to Monitoring. August 2013
About the author:
Dr Phil Birch, Senior Vice-President, Global Strategic Marketing, Aptiv Solutions
Aptiv Solutions is a visionary company that creates and delivers practical solutions and services to meet today’s product development challenges. Dr Birch has worked in the pharmaceutical industry for over 28 years and has held senior positions in corporate development, business development and R&D across a number of consulting, biotechnology and top 10 pharmaceutical companies. At Aptiv Solutions, he is responsible for the industry-wide adoption of innovative approaches to product development. Dr Birch has a D.Phil. and first class honours degree from Oxford University, where he studied Neuroscience.
Closing thought: Is it time to rethink the logic of current approaches in trial design?