Pharmaceutical companies are introducing business planning at an earlier stage of clinical development and exploring statistical models and other predictive tools in an effort to boost the strike rate of new drug candidates, discussions among R&D leaders suggest.
A roundtable recently convened by the US-based Tufts Center for the Study of Drug Development (CSDD) found that drug developers and their partners are increasingly concerned with measuring the probability of clinical success and accelerating time to market.
Indicative of this trend, for example, is a shift in decision-making toward data-driven models rather than intuition and prior experience in the development process, the Tufts Center noted.
Net present value
It also pointed to more rigorous use of risk-adjusted net present-value calculations earlier in clinical development as a means to improve decision-making on how to structure trials.
Another theme that emerged from the roundtable discussions was the significant potential for transforming clinical studies through advances in the development of personal genomic information, increased voluntary sharing of patient data, and an explosion in data on cell metabolism.
At the same time, effective mining of the data generated by these advances presents a major challenge to research centres, the assembled R&D leaders felt.
Tufts CSDD director Kenneth Kaitin highlighted a simple algorithmic model created by his team with a pharmaceutical company to predict whether oncology products emerging from Phase II trials are likely to secure marketing approval.
This Approved New Drug Index (ANDI) was modelled on the five-factor Apgar score, widely used in delivery rooms to evaluate the health of newborn babies.
The Tufts team concluded that, compared with the prevailing industry metric, data drawn from this model supported assigning a much higher probability of success to oncology drugs with top ANDI scores of 7 and 8, and much lower probabilities of success to candidates with scores of 0 to 4.