Clinical Researcher—June 2024 (Volume 38, Issue 3)
SCIENCE & SOCIETY
Rohit Nambisan, MS, MA
Precision medicine brings about transformational advances in how we treat and prevent disease. As our understanding of biology progresses, we advance our ability to precisely target therapies to individuals with specific variants of broad diseases. The continued evolution of biomarker-driven therapeutics has given rise to unique, novel clinical trial designs that are largely focused on complex rare conditions and indications. This is a level of precision and complexity that is helping us take important steps forward in our ongoing battle against the most vexing clinical needs.
While this development is overwhelmingly positive, especially for patients, it creates unintended challenges. As clinical sciences rapidly progress, the processes and technologies that are used to run trials have not kept pace. As a result, we endanger critical advancements in patient care.
Evolving Dynamics of Growing Complexity
The complexity of clinical trials has dramatically increased in recent years. Today, more than 3.5 million datapoints are collected for each late-phase trial, which is three times the number that were collected in similar trials just a decade ago. Moreover, these datasets include a growing number of primary and secondary endpoints—averaging 26 for late-phase trials—as well as the specific biomarkers, which are required in half of all oncology studies and in about one in six of other trials.
Additionally, new studies are increasingly focused on niche sub-indications, driven by the need for sponsors to show improved efficacy over existing treatments. As a result, nuances that are often found in rare disease trials are becoming more prevalent across a broader range of trials, increasing operational complexity, especially in patient-finding. For example, in a recent oncology cell therapy study, eligibility was based in part on expression of human leukocyte antigen (HLA)-A*02:01. A Precision for Medicine analysis found that prevalence of this antigen varied in different parts of the world, ranging from 38.5% to 53.8% in Europe to 16.8% to 47.5% in North America.
Regulatory guidelines add yet another layer of complexity. For example, the Consolidated Appropriations Act 2023 (H.R. 2617) recently signed into law now requires sponsors of many trials to submit a diversity action plan to the U.S. Department of Health and Human Services to address the representation disparity that has plagued clinical trials for decades. Patient demographics vary across each disease, making it especially critical for sponsors to effectively represent those disease-specific demographics in their studies. Misrepresentation increases the risk of commercial failure due to low awareness from minority populations, or worse, adverse reactions since the therapies were not rigorously tested across the appropriate patient demographics. Yet, while it is crucial to ensure population representation in trials, it is also difficult to achieve, as it requires sponsors to recruit from a narrower demographic pool.
Heightened complexity also means that trials take longer and are more expensive. Between 2008 and 2019, Phase III participant recruitment timelines jumped by 39%, requiring, on average, a total of 18 months to complete. Recruitment challenges also commonly lead to delays and can often derail trials entirely.
More Data, More Challenges
The industry’s need for more precise data adds to today’s clinical research challenges. Sponsors, contract research organizations, and sites struggle to integrate the multitude of disparate datasets from a variety of institutions and systems. A 2023 survey of trial sites punctuates how the proliferation of systems increasingly burdens trial operations. More than half the respondents indicated that setup and training on sponsor technology is their biggest challenge during a trial, with 70% of respondents reporting that each of their studies requires six (or more) unique system logins to have access to all the required data.
Inevitably, the industry’s collective challenges are increasing costs without commensurate gains in efficiency. Between 2012 and 2022, inflation-adjusted research and development spending increased by 44%, from about $170 billion to $247 billion. Yet, during this same period, the number of U.S. novel drug approvals remained flat.
Out With the Old: New AI-Optimized Approaches
Today’s obstacles represent an existential threat to our continued ability to engage in effective clinical research. The pharmaceutical industry is naturally driven to follow tightly controlled processes—which, on the one hand, yield repeatable, reproducible performance. Moreover, these ingrained processes are difficult to change without significant upheaval in change management. However, science has progressed and is demanding new approaches from trial operations. What worked in 1990 will not yield the same performance today.
It’s time to reconceptualize operations to tackle modern clinical science’s complexities, otherwise novel therapies for specialized diseases will remain out of reach for patients in need.
In the past two years, artificial intelligence (AI) has upstaged all technology applications in its ability to transform multiple industries, regardless of domain. Now, AI-optimized approaches can address the substantial challenges facing global clinical research as well.
For example, sponsors can leverage AI-based feasibility analysis that incorporates real-world data, amongst other critical data sources, to make informed and accurate enrollment predictions for their trials. An analysis by consulting firm McKinsey notes that such tools could result in trial recruitment that is accelerated by as much as 20%. AI-optimized approaches can also help to predict optimal trial sites, troubleshoot problems that might arise before becoming too late to mitigate, and even evaluate latent risks before a trial is launched.
While it impacts change management, utilizing real-world, patient-level data to drive site identification and study forecasts can drive better outcomes. Given the increasing complexity and costs associated with clinical trials, we should leverage the best possible information on where highly niche patient populations are receiving care, informing recruitment projections and opportunities to access and enroll participant populations. In much the same way that biotech leaders extol the virtue of “failing fast” when it comes to experimental therapies, assessing trial sites for their likelihood of recruiting appropriate participants beyond the scope of investigator feasibility surveys, which typically measure investigator motivation instead of enrollment potential, can save critical time and resources.
Further, consider conducting feasibility analyses regularly during a trial rather than just one time, before the study starts. Automated data tracking solutions can surface accurate insights throughout study conduct so teams can dynamically reassess the trial plan and course-correct at any time. For instance, each time a new country opens, a new site starts recruiting, or there is a market change (i.e., a competitor suddenly goes to market), continuous feasibility analysis identifies how these events impact the trial timelines and budgets. Prescriptive or causal AI models can empower trial managers by recommending scenarios to ameliorate trial performance and explain why the recommendations will produce desired outcomes.
Conclusion
We should no longer force yesterday’s one-size-fits-all approach onto biotech’s third wave of advanced drug development. As precision medicine effectively makes every trial today a rare disease trial, the operations of these trials must keep pace or else we squander today’s most promising clinical advancements. By embracing AI-optimized approaches, we can improve operations and ensure that decades of scientific innovation achieve the ultimate goal: improving the human condition.
Rohit Nambisan, MS, MA, (rohit.nambisan@lokavant.com) CEO/Co-Founder of Lokavant, is trained as a neuroscientist and is an executive with experience leading product development organizations in pharma, medical devices, personalized medicine, health IT, healthcare data and analytics, and AI. Prior to co-founding and leading Lokavant, he was most recently the Head of Digital Product at Roivant Sciences and the Head of Product at Prognos Health.
