What's Causing So Many Phase III Trials to Head for Avoidable Failure?

Gen Li, president and founder of Phesi, spoke with Pharmaceutical Executive about issues pharma companies are having during Phase III trials and how new technologies can solve these issues.

Pharmaceutical Executive: What's causing so many Phase three trials to head for avoidable failure?
Gen Li: The challenges can be put into two different categories. The first is: how good are we designing trial protocols? The problems that we’re seeing can be caused by a mismatch in the design and the patient population being targeted. In those cases, the trial protocol must be amended. The consequence of that is that this causes a delay and prolongs to trial time, which brings increased financial costs.

The other category is investigating the right sites and recruiting the right patients. Oftentimes, we don’t do the best ion getting the sites to recruit the patients.

These two things conjugate with each other. When you have poor design, that in and of itself can cause poor performance with enrollment, which compounds with the issues the site may already have with enrollment. The main problem is the mismatch between the experience and the knowledge of the investigator and the patients being targeted by the protocol itself. These are the issues that cause problems in Phase III.

Another issue that’s overlooked is the competition. This can come in many different levels: site, country, and even disease. Using dermatitis as an example, it’s an area that suddenly saw a lot of innovation. Everyone wanted to be the first to reach approval, hence the competition became fierce, which further impacted poor enrollment performance and prolonged the cycle times.

PE: How Can Real World Data and AI Help Solve Issues with Clinical Trials?
Li: AI has many different ways of helping the industry. From our perspective, we have a large patient database that we can use to construct digital patient profiles. These can be made in alignment with a particular product design and the patient population being targeted. By constructing a digital patient profile, we can then find the issues associated with the design. Oftentimes, it’s found some misalignment between what a particular clinical trial is targeting in terms of patient population and what the actual patient population may actually look like.

By solving that misalignment, this leads to better trial design. We can potentially perfect the design by looking at the outcome measures.

PE: Where has AI been successfully implemented already?
Li: We’re current working with three of the top 10 pharmaceutical companies to create better clinical trial designs, better patient recruitment, and improve site performance. We’re also working with a number of medium to small size companies as well.

The exciting part part of AI application in clinical development is the use of the digital patient profile, which can lead to the creation of a digital twin. These can be used to either partially or totally replace control arms. We’re not there yet, though.

We’re working with our clients to create a path forward. We’ve realized that we need a lot of stakeholders in different categories to move us in the right direction. We’re working with experts from Harvard and Oxford to build a solid medical and data foundation. We’re exploring ways of working with FDA and other regulatory agencies.

It takes a lot of players to be in the same place and looking at the technology to get us to the place where we can use the digital tools to either partially or totally replace control arms. We’re not there yet, but it’s only a matter of time.

PE: Do you have a timeline for those results?
Li: It could be next year, it could be two or three years. But there is no doubt, in my view, that we will get there.