What 30 Years of Oncology Trial Data Reveal About Racial Representation: Rahman Olusoji, MBBS

Rahman Olusoji, MBBS, discusses a study that used artificial intelligence (AI) to conduct a 30-year analysis of racial and ethnic representation in oncology clinical trials. In this interview with The American Journal of Managed Care^® (AJMC^®), Olusoji, a hematology and medical oncology fellow at The Ohio State University, covers the study's key findings, including explosive enrollment growth in Asia that masks stagnant or declining participation among Black and Hispanic patients, the socioeconomic and regulatory barriers driving these disparities, and what interventions could meaningfully improve diversity in clinical trial enrollment.

This interview has been lightly edited.

AJMC: Your study uses AI to conduct a 30-year systematic analysis of racial and ethnic representation in oncology trials—a scope that would be impossible through manual review. Could you talk about the background and methodology of the study?

Olusoji: The study was born out of curiosity about what has been going on in the field of oncology clinical trials—specifically, how far we have come in terms of diversity and globalization. There has been a lot of advocacy about making sure underrepresented populations are represented in clinical trials, and it seems everyone is doing something to try to make that happen. The study was designed to find out whether we are truly making progress, and if so, what the impact has been on enrollment.

To have a retrospective view going back 30 years, reviewing all the clinical trials published in that period would probably take years for humans to extract the data and get a true picture. That was the main reason for using AI to speed-read these publications and extract the variables we needed.

We looked at trials from 1995 to 2025—a 30-year span—and analyzed temporal trends across 3 decades: 1995-2005, 2006-2015, and 2016-2025. We searched PubMed for oncology clinical trials published in The Lancet and New England Journal of Medicine over that period, identified close to 1500 publications, and used Gemini 2.5 to extract variables including race, ethnicity, trial type, trial phase, country, and cancer type.

AJMC: What were some of the most notable findings? And what did the scale reveal that a shorter or narrower analysis might have missed?

Olusoji: In terms of volume, there was growth across the 3 decades, from around 200-something publications in the first decade, to 300-something in the second, to 600-something in the third. Most were phase 3 trials, rising from 68% to about 70% in the last 2 decades, though that may partly reflect the journals we selected.

The most striking finding was the explosive growth in enrollment from Asia. About 14% in the first decade, to about 20-something percent in the second, to 45% in the last decade. This shows increased globalization, but the growth is concentrated in Asia. Continents like South America and Africa had minimal to no participation. Africa had almost nothing.

Once we removed the Asian population from the analysis, the true picture of what the rest of the world is experiencing became clear. There was not necessarily any improvement in the inclusion of underrepresented populations like Hispanics and Blacks. Hispanic participation was essentially flat: about 6.8% in the first decade, declining to about 4% in the second, and recovering only slightly to about 4.6% in the last decade. For Africa, there was a significant decline—from about 5% to 8% in the first decade, to about 2% in the second, to just 1.5% in the last decade.

What makes this especially concerning is that we are now in the age of targeted therapy and genomic profiling. Africa has the largest and most genetically diverse population in the world. If we conduct trials in China using genomics and apply those findings as the global standard of care, patients in Africa, who may have entirely different genomic profiles, are receiving treatments based on data that do not represent them, because they were never included in those trials. That is something we really need to work on.

AJMC: Taking a step back— from a managed care perspective, for healthcare providers and people making decisions about what treatments to cover, why is this disparity in clinical trial enrollment a problem?

Olusoji: Part of what we identified from the literature review is that it seems easier to conduct clinical trials in Asia because there are fewer regulatory bottlenecks compared with the US. There are also sociodemographic barriers in continents like Africa that affect participation.

Even within the US, Black and Hispanic enrollment in clinical trials is significantly lower compared to other racial groups, and I think that also comes down to socioeconomic factors. These populations tend to have more comorbidities and often lack the insurance coverage needed to access trials. We also have stringent inclusion and exclusion criteria that further limit eligibility.

These patients are generally not in a position to travel from, say, Columbus, Ohio, to MD Anderson for a second opinion before starting treatment. But I have seen patients come to clinic, we tell them what trials are available, and they say, "I hear you, but I want to go to MD Anderson, I want to go to MSK [Memorial Sloan Kettering], to see what other options are available." For someone to do that, they have to be well-informed and at a certain socioeconomic level. Unfortunately, many Black and Hispanic patients do not fall into that category, and that affects their trial enrollment as well.

AJMC: Based on your research, if you could make one specific change—whether in how trials are designed, where they're conducted, enrollment requirements, or how they're reported—what would it be?

Olusoji: What I would do is apply the same approach that drove the explosive growth in Asia to underrepresented populations. Specifically, through an aggressive campaign to start enrolling patients in Africa and South America. For example, if we carry out a trial in Nigeria, where I am from, and we come out with a practice-changing outcome, I am confident that finding would also be applicable to the African American population here in the US. So, I would actively recruit and encourage participation from specific racial groups, because it seems that while we are trying our best in the US, the unique populations we most need to reach are also outside the country. We have to look beyond here to get more enrollment and participation.