Over the course of the last several years, there have been major trends in the reconsideration of the structure, intent, and research site locations for registrational clinical trials. This has been driven by three forces: (1) FDA mandates that trials provide meaningful information for the outcomes of the populations who will ultimately receive approved medicines; (2) a practical and moral imperative to have trials available to all patients, especially in the community setting, which represents the location where 75%+ of patients received their care; and (3) meeting goals for greater diversity in the trial populations, assuring a statistically meaningful representation of ethnic, racial, and economic subpopulations that may be most adversely impacted by a specific disease. This is a significant change for the industry that places new requirements on sponsors, research sites, and supporting service intermediaries (e.g., contract research organizations and clinical trial technologies companies).
Historically, most sponsors used past trial designs and those of recently approved therapeutics in the same drug class or the same disease. Most large biopharma organizations have protocol systems of records and other tools assembled over the years that support this process. These solutions were complemented by site and investigator databases that emphasized the research entities and specific investigators they were successful with in the past. This process of ‘mutual affinity’, site-to-sponsor and sponsor-to-site, sustained a regular flow of studies to leading academic centers that had the scale to accrue patients in even rare indications, and the staffing levels to commit to significant levels of studies. Academic enterprises were further supported by NCI and NIH funding that assured research personnel well above the level of any other care setting. As an example, local, next generation sequencing infrastructure might have started as NCI funded and then transitioned to being supported by more standard reimbursement approaches. The imperatives of having neutral to positive margins are also lower here given the multi-part funding model. To be clear, these centers performed an exceptionally valuable role in bringing the deepest level of disease understanding and access to large patient populations. Still, the patients they saw tended to be located in more prosperous urban areas, were less diverse, and had broader private payer or even private pay coverage.
With the imperatives to have trials reflect the ultimate populations receiving newly approved medicines, and for trial diversity to better represent subgroups that may be disproportionately negative impacted, this will need to change. Legacy trial designs will inform, but not dictate, future designs. Instead, large-scale, highly representative standard-of-care datasets will have more influence on the key endpoints of interest, intermediate measure of safety and response, and the range of clinical activities being called for. These rich and specially prepared datasets can reflect standard of care clinical activities, such as number and timing of serial biopsies. This allows for several complementary goals to be achieved, such as assuring that the study is no more complex or burdensome than the current standard of care, and that the trial design is executable in a community versus an academic setting, to name two critical objectives.
The addition of Social Determinants of Health data to the clinical data being used to model trial designs allows an understanding of narrow populations that may be most impacted by a specific cancer. It can then inform a specific trial design to assure that the protocol explicitly includes considerations for these populations and avoids unwittingly excluding these populations. At ConcertAI, we are now including medical claims and Social Determinants of Health as standard in 100% of the data sets and SaaS solutions that we use in clinical development and other applications. Doing so enables consultative conversations with regulators in satisfaction of their mandates for greater meaningful diversity in trials, supports iterative and AI optimized trial designs, and bolsters selections of sites that may contribute to achieving the diversity goals.
All this comes together at the community research site level. We believe it is important to reframe our expectations for what is termed ‘community oncology research sites’. Community site capabilities have evolved and matured a good deal over the last five years. As compared to academic centers, they navigate corridors in clinical decision making and assignment to therapy that are bounded by what is reimbursed and what patients can afford to pay. Yet with these very practical and responsible considerations, they have advanced their models of care to include robust clinical pathways aligned to the latest clinical evidence, greater use of next generation sequencing testing, advanced imaging, etc. In short, the differences across settings are less about the infrastructure, capabilities, and training, but rather economics of reimbursement, capacity for self-pay, and clinical trials.
Clinical trial access is not entirely under the control of the healthcare provider or research site. Trial sponsors have to want these sites to have access to their trials. Since sponsors tend to go where they have deployed studies in the past, it has been slower than anticipated for trials accessibility to grow in the community. Today there are four categories of community research centers: (a) high volume research centers with dedicated staff often having 100 to 150 active trials underway at any one point in time; (b) research capable sites with active programs, 10 to 25 trials underway, and a staff with previous experience at academic or large regional health systems as investigators; (c) research aspirational with limited research underway but staff with prior investigator and trial experience at scale; and (d) research naïve sites with limited physician experience as investigators. Based on the practice data we see, we estimate that 15 to 20% of community sites are in the first category, 30% in the second and growing, 25% in the third, and 25% in the fourth.
So, to meet the imperative of broadening trial access in the community so that the participating patients more closely resemble the ultimately treated patients post-approval, it will take sponsors’ policies and focused efforts to realign their trial sites to those in the (a) and (b) categories, while taking a longer-term view on the (c) category. This can be done as follows:
It is impractical to assume that community research sites will move towards academic level staffing or organizational models – that is just not their role in the healthcare system. In the community, margins have been significantly pressured over the last five years, the number of newly diagnosed patients per medical oncologist has grown, and ability to attract and retain clinical staff and research staff challenged. Still, it is an imperative and requirement to broaden trial access and diversity. Therefore, new tools, infrastructure, contracting models, and support models will be needed. ConcertAI’s approach is to advance AI SaaS solutions for research sites and sponsors that bring the expertise of highly trained research personnel into software intelligence that is integral to clinical workflows and that automates the most laborious of activities, such as matching patients to trials and the population of trial data into electronic case report forms. While we are early in this important journey, the results thus far are tremendous and promising. Together, with our community provider partners and biopharma sponsors, we are committed to meeting the requirements of the three forces of trial population generalizability, trial diversity, and great community trial access.