Importance of Designing Clinical Trials from a Program Perspective

September 9, 2020

Cytel’s co-founder, Nitin Patel, conducted a webinar on designing clinical trials from a program-level perspective. His presentation helps us understand the value of designing clinical trials considering downstream consequences. Watch the on demand webinar to get insights on the role of simulation in optimizing clinical trials' performance from a program perspective and understanding the hybrid Bayesian-frequentist approach to clinical trial design.

Access Replay

We also had the opportunity to interview Nitin about his journey since he co-founded Cytel and got his views on implementing a program-wide strategy for pharma and biotech companies. Read the blog here.

Continue reading this blog for key highlights from the webinar.

Need for program perspective

There is a vast opportunity for using a Program Perspective in designing clinical trials as recent estimates show low Probability of Success (PoS) rates at all phases and high, as well as rising costs. Studies show that confirmatory trials cost tens of millions of dollars and fail on average 40% of the time. When conducting clinical trials, we are often faced with the difficult question of whether we should continue the development of a drug based on the data in hand. This is referred to as the Go/No-go decision. A program-wide strategy helps to reduce drug development cost by improving Go/No-go decision making between phases.

Phase 2 trials aim to establish proof-of-concept and to select the best dose for Phase 3. Studies by several researchers suggest a major factor for high failure rate in Phase 3 is that the right dose is not selected in Phase 2. Program-level simulation for Phase 2 designs can substantially improve the program. It enables evaluation and optimization of several important design choices, for example, sample size, dose selection methods, number of doses, dose response model selection and others.

Adaptive Program work stream

The DIA set up a working group of professionals from industry, academia and research centers with Dr. Carl-Fredrik Burman as chair to research program level modelling of clinical trials in 2010. The focus of the team was on late stage trials, mainly Phase 2 and Phase 3 trials. Several decision problems were considered such as:

  • Go / No Go?
  • Number of doses?
  • Dose choice?
  • Trial adaptation?
  • Sample size?
  • Population?
  • Treatment duration?
  • Recruitment Rates? Etc.

General approach to model drug development program

There is generally a tendency to go too quickly into Phase 3 from Phase 2. We need to be mindful of the fact that Phase 3 studies are extremely expensive. The DIA working group focused primarily on how to model Phase 2 and Phase 3 trials as components in a program to gather enough information in Phase 2 to make better decisions about launching and designing confirmatory Phase 3 trials. The group considered the commercial value which is often not taken into account in modeling for clinical trials. They made this exception and aligned designs quantitatively to commercial objectives like Expected Net Present Value (ENPV). ENPV is a method to value future cash flows and adjusts for uncertainty by calculating net present values under different scenarios.

A hybrid Bayesian-frequentist framework was used in their case study, and a parametric model was designed to study dose-response.

Performance criteria of a good design

The difficulty with using just the PoS is that it states, the larger the sample size the higher is the PoS. The challenge with increasing the sample size is that it increases trial cost and delays time to market. There is a big difference in revenue when you are marketing a drug within its patented period as opposed to outside the patent period when there is competition. ENPV is a measure of commercial success that balances these opposing effects of sample size increases. It considers trial costs, regulatory hurdles, proving the efficacy and safety, and the market value if NDA is approved. Then we optimize using different sample sizes in simulations with respect to dose selection, Phase 2 and Phase 3 designs, Go/NoGo rules, and other decision variables.

Commercial Model

Revenue flow is a complex concept that includes many variables such as, the characteristics of the market, other available drugs, the target population etc. A financial model is created to estimate what the revenues will look like post the release of the drug in the market. The cash revenue steeply falls after the patent expiration. Hence, time and cost estimates are made for the different phases of a trial. If the duration of the overall trial is stretched or too much time is spent in between phases, we begin to lose money as the patent life is eroded.

We need an approach that explicitly addresses the value proposition during the clinical development process by incorporating research and development, marketing, commercial, and medical practice perspectives. A clinical trial simulation tool is necessary to compare multiple development options with respect to the expected value of the product, which clearly depends not only on the quality of the product itself but also on the quality of the development program.

Neuropathic Pain drug development program

As a case study, Nitin describes a phase II trial design based on program-level optimization. A hybrid Bayesian-frequentist framework was utilized to evaluate the impact of Phase 2 design choices on the probability of Phase 3 success, clinical utility, time to market, trial costs and the ENPV of the product. These factors include Phase 2 sample size, go-no go and dose selection decision rules for Phase 3, and Phase 3 sample size. Using neuropathic pain as an example, we use simulations to illustrate the framework and show the benefit of including these factors in the overall decision process.

Watch the on demand webinar by clicking the button.

Access Replay


About Nitin Patel

nitinOver 30 years in the US, Nitin Patel has led development of several innovative statistical software products and worked on collaborative research projects with major pharmaceutical companies. He has served as a member of PhRMA and DIA working groups on adaptive trials contributing in the areas of dose finding trial design, drug supply, adaptive programs and portfolio optimization. In India he worked at Tata Consultancy Services where he was a member of the founding management team in 1966. He joined IIM Ahmedabad as professor in 1975.

He co-founded Cytel with Cyrus Mehta in 1987 where he is a member of the board and a lead researcher at the Innovation Center. He has Masters degrees in Management and in Computer Science-Electrical Engineering, and a Ph.D. in Operations Research from MIT. He has been a visiting professor at MIT and at Harvard University for over a decade. He is a Fellow of the American Statistical Association and a Fellow of the Computer Society of India. He has served as president of the Operational Research Society of India and as vice president of the International Federation of Operations Research Societies. He has published over 70 professional peer-reviewed papers and book chapters and co-authored a book on data mining that is now in its 3rd edition.