How an AI driven platform can help sponsors save failing trials, hone protocols, and find patients most likely to benefit from drugs in development.

Clinical development has a high attrition rate. Only 14 percent of drugs in development make it to approval, which means 86 percent of assets in development fail to achieve marketing authorization. The latter number drops further to 97 percent for oncology drugs. ⁱ

But even in trials where the majority of patients don’t respond as expected, there may be sub-populations who show benefit from taking the drug. In the past, those trials would be shut down and the drugs shelved, leaving those patients, and the sponsors, back at square one.

But it doesn’t have to be that way.

The high rate of failure in drug development can now be mitigated through the use of artificial intelligence (AI) platforms that are able to identify sub-populations who are more likely to show positive response to a treatment, allowing sponsors to adapt the trial design accordingly.

This innovative application of AI and machine learning is the next evolution of precision medicine, bringing predictive analytics to the forefront of drug development.

The best news is that this isn’t just theory but is now reality. We’ve worked with many clients who have seen measurable results using our Sub-Population Optimization and Modeling Solution (SOMS). This statistically rigorous machine learning analytical platform can identify predictive biomarkers or population variables in treatment populations with stronger outcomes than the general population of study participants. When leveraged throughout the drug development lifecycle, this platform can find subgroups with particular strong response to a therapeutic, reduce the rate of trial failure, rescue promising molecules for further development, better capture primary and secondary endpoint data to support approval and payer valuations, and find populations with elevated adverse event risk that should be avoided.

Here are a few examples of how sponsors have used SOMS to improve clinical trial decision making, reduce risk, and generate more value from their investments.

Biomarkers support a move to phase 3

Sponsors often first leverage the SOMS platform to review phase 2 trial results to proactively identify sub-populations most likely to respond to treatment. Early analyses are used to uncover any genetic, biologic, and/or environmental characteristics that define these promising populations. This information can be used to adapt inclusion/exclusion criteria, reassess endpoint selection, and potentially reduce the trial sample size by recruiting patients with strong treatment effect signals.

In one example, a global pharma company wanted to evaluate the safety and efficacy of a new drug compared to the standard of care for treatment of a specific cancer. Using an analysis method called Subgroup Identification based on Differential Effect Search (SIDES) within the SOMS platform, they identified six biomarkers in their phase 2 data that were predictive of treatment effect. The sponsor used those biomarkers to identify subgroups where the treatment effect was substantially different from the overall population.

That analysis played a key role in the development of a tailoring strategy and the selection of the patient population for phase 3 trials.

Bacterial Infection drug wins FDA approval

In studies where trial results are inconsistent, SOMS analyses can support adaptive trial designs, where sponsors modify key components of the trial in response to the data collected. This ensures they are focusing on patients most likely to experience positive outcomes, which improves patient safety, reduces liability risks, and increases the likelihood of gathering positive outcomes data to support regulatory approval.

In this example, an emerging biopharma company was conducting two global phase 3 trials to evaluate the safety and efficacy of a new drug versus standard of care for treatment of bacterial infections. The original analysis of the clinical trial data demonstrated no overall treatment effect. However, a retrospective analysis conducted using the SOMS platform identified 26 biomarkers that were predictive of positive treatment effect. The analysis simultaneously identified characteristics indicative of reduced treatment effect in the complement subgroup due to a safety concern that could be mitigated.

Thanks in part to this analysis, FDA approved the new drug with a black box label for challenging cases when alternative treatments are not suitable.

Remove bias from decision making

Many sponsors have in-house biostatisticians who vet trial data to determine whether a trial is delivering promising enough results to continue. While these analyses are a vital part of the drug development process, they can be unintentionally biased, particularly when a company is heavily invested in a trial’s success. An impartial AI-driven analysis using the SOMS platform can provide a fresh perspective on their recommendations, improving objectivity and potentially yielding more insightful results.

In one case, a global pharma company had phase 2b study results of a treatment in oncology showing superior effect on overall survival for a specific subgroup. However, the phase 3 study yielded no meaningful results. Using SIDES within the SOMS platform, they conducted a retrospective analysis on the phase 2b results with a more detailed subgroup assessment and found the data did not show consistent meaningful outcomes for any sub-population. As a result, the company abandoned further investment in the development of that asset.

These are just a few of the many examples of how sponsors are using this technology today to improve trial results, lower their rate of failure, and make development more efficient. To see examples, learn more about the technology, watch our video or request a demo at our website.