NEW YORK – In newly released draft guidelines, the US Food and Drug Administration is encouraging industry players to submit plans for improving the inclusion of racial and ethnic minorities who have been historically underrepresented in clinical trials.
Last month, the FDA released a draft document that expands on an earlier guidance from 2016, in which the agency outlined how sponsors should collect and present race and ethnicity data in product submissions. In the latest document, the agency is inviting drug and test makers to develop and submit how they plan to "address inclusion of clinically relevant populations" and discuss them with regulators early in the development cycle of their products.
The longstanding challenge of increasing clinical trial diversity has been exacerbated in the era of precision oncology. Although the development programs for all types of medical products suffer from exclusion of racial and ethnic minorities, the challenge to achieve better representation is more acute in precision oncology drug trials where the pool of participants is defined by rare biomarkers.
Even when social and cultural factors such as trial site location, transportation, and trust issues are addressed, studies may still struggle to enroll underrepresented groups in the biomarker-positive cohort because they are using biomarkers derived from genomic studies, which are themselves heavily biased toward European participants. Before precision oncology drug and diagnostic clinical trials can be truly inclusive, there need to be more studies characterizing prognostic and predictive biomarkers prevalent in populations with non-European genetic ancestry.
An analysis published last year found that minority racial groups in the US were significantly underrepresented in precision oncology drug trials when compared to the incidence of cancer in these groups. Meanwhile, non-Hispanic white participants were overrepresented in these studies, comprising more than 82 percent of the participants in 93 studies. In comparison, 10 percent were Black participants. Other underrepresented groups in this analysis included Hispanic, American Indian, and Alaskan Native participants.
Relatively little data exists on racial and ethnic variations in the incidence of targetable cancer biomarkers, though more research is being done in non-European populations in recent years. The best known and oft-cited are EGFR mutations, which are linked to increased benefit with a variety of EGFR inhibitors, and are most common among East Asians, occurring in around 30 percent of lung tumors compared to just 8 percent of patients from other backgrounds.
A study from Latin America revealed that Native American ancestry was associated with targetable mutations in EGFR and KRAS, and that tumor mutational burden and STK11 mutations may be prognostic biomarkers.
However, few studies such as these are available to inform clinical development of precision oncology products. The distribution of individuals included in genome-wide association studies is heavily skewed to those of European ancestry, and although GWAS have growing representation from Asian patients, people of African, Latin American, Arab and Middle Eastern, Pacific Island, and Native American descent still comprise a small percentage of participants.
This has long raised concerns that cancer drugs the FDA approves based on data from clinical trials may not be applicable to all cancer patients. "Clinical trials provide a crucial base of evidence for evaluating whether a medical product is safe and effective," the agency said in releasing the latest guidance. "Therefore, enrollment in clinical trials should reflect the diversity of the population that is ultimately going to use the treatment."
The agency is hoping to make concrete gains on this front by asking drugmakers to come up with strategies to address the lack of diversity in the trials they conduct for regulated products.
In the draft guidance, the agency advises drug and device makers to submit a clinical trial diversity plan early in the development cycle of products, for which they expect to submit an application for regulatory approval. In these diversity plans, sponsors should specify their goals for enrolling underrepresented racial and ethnic minorities and assess "any data that may indicate the potential for a medical product to have differential safety or effectiveness associated with race or ethnicity."
The FDA's draft guidance aligns with legislation recently introduced into the US House of Representatives, which also aims to improve diversity in clinical trials. The Diverse and Equitable Participation in Clinical Trials (DEPICT) Act, sponsored by US Reps. Anna Eshoo, D-Calif., Brian Fitzpatrick, R-Pa., and Robin Kelly, D-Ill., would require investigational new drug and investigational device exemptions applicants to report clinical trial enrollment targets by demographic subgroup, as well as provide a rationale for those targets and a diversity action plan for reaching those goals.
While recommendations the FDA makes within a guidance are not legally binding on industry players, this bill has teeth the agency can use to ensure sponsors are sticking to their trial diversity plans. For example, the bill gives the FDA authority to mandate post-market studies when sponsors miss diversity targets without justification. And it requires the FDA to publish an annual report on diversity and contains provisions and funding for education and outreach efforts.
The genomics gap
"The FDA is focusing on strategies to achieve diversity in clinical trials and believes that the recommendations provided in the guidance will help achieve diversity across all types of trials by enabling all populations who possess the biomarker under study in a precision oncology trial to participate," Lola Fashoyin-Aje, associate director of the FDA's Science and Policy Program to Address Disparities, said in an email.
The agency's guidance has been welcomed by industry and patient advocates who hope the move will finally make clinical trials more accessible to patients from underrepresented groups. However, a gap remains in the genomics data needed to understand the distribution of widely used biomarkers among different ancestry groups and to identify new biomarkers that have been missed in Eurocentric genomic studies.
"This is a broad topic of discussion now," said Nicole Richie, Genentech's global head of health equity, science, and strategy. She noted that there have been numerous reports of incorrect outcomes based on studies using a homogenous population. "When we think about how we address this as a pharmaceutical company, we need to have a focus on what are the subsets, how do they vary, and are there differences in biomarker prevalence … but right now, there's a really big gap in what's known."
In a roundtable discussion on clinical trial diversity hosted by the FDA's Oncology Center of Excellence, Brian Rivers, director of the Cancer Health Equity Institute at Morehouse School of Medicine, emphasized that in cancer research there is a need to focus on genetic heterogeneity in minority racial and ethnic groups in addition to social determinants of health. "African Americans and Blacks are disproportionately impacted by cancer," said Rivers. "The question begs itself in terms of why? What are we not doing? What strategies can we put forth? What changes can we make? The first step is recognizing that there is this biological heterogeneity that we have not uncovered yet."
More granular real-world data are needed to equip drugmakers to come up with the type of diversity plan the FDA is asking for, identify which groups may have differential responses to their treatments, and refine enrollment strategies accordingly. Toward that end, Genentech conducted a study based on real-world data on 60,000 advanced non-small cell lung cancer patients and found that patients with African ancestry had significantly higher TMB scores, while patients of Asian ancestry had lower TMB scores.
High TMB is associated with better outcomes in advanced NSCLC patients taking immunotherapy compared to patients with high TMB but on other treatments. These kinds of results have implications for clinical trial design. For example, if a clinical trial is using TMB as a predictive biomarker to enroll NSCLC patients into an immunotherapy trial, and it does not have sufficient representation of African American participants, the study could underestimate the treatment's benefit. "If we are thinking about a patient population that will benefit the most from a treatment, we should make sure that we're thinking about that holistically," Richie said. "And that includes differences that could occur across patient subsets."
Richie noted that in some cases, unknown differences in biomarker prevalence could become a hurdle to diversifying a clinical trial. She pointed to EGFR, which is more prevalent in East Asian populations, and KRAS, which is more common in European populations, as examples.
"Eighty percent of all genomic data currently available in the world is of patients of European ancestry," said Richie. "That makes it even more important that we have an explicit focus on patients that have historically been understudied in these settings so that we don't actually undermine the intention of precision oncology and precision medicine to deliver the best possible care at the right time for patients by not understanding how medicines might work differently. We might need to have different types of recommendations for diagnostics based on population science and population-related differences."
Although genomically informed cancer care has taken off in the past decade, lack of diversity remains deeply problematic in this space. Polygenic risk scores, for example, are at the cutting edge of genetic testing to assess cancer risk, though these tests are rarely developed on diverse datasets.
Laura Raffield, a professor of genetics at the University of North Carolina, Chapel Hill, who is researching genetics of hematological traits, said that many early GWAS studies have focused on subjects with European ancestry because the analysis was simpler. Disease correlations with genetic variations are more complex in non-homogenous populations or populations that have admixtures from different continental regions.
"It's sometimes the easier choice to just focus on folks of European ancestry, since that's where we have the biggest sample size in almost all cases in human genetics," Raffield said. "But I think it's not the right choice. You could be missing important findings, and the findings that you have are less applicable across global populations if you make that analytical choice."
Raffield took on the task of doing that more complicated analysis looking for genetic variants prevalent in non-European populations within a notoriously European-biased dataset, UK Biobank. She said that although the vast majority of the 400,000 individuals represented by UK Biobank are of white European ancestry, it includes more than 20,000 participants with non-European ancestry.
"We focused on the African, South Asian, East Asian ancestry participants in UK Biobank, and we actually did have a number of findings that were not identified in white British individuals," Raffield said. For example, researchers discovered two variants associated with HbA1c in South Asian people not previously reported in the GWAS catalog. Diabetes diagnoses and lab values tracking glycemic control could be inaccurate in people with those variants because their baseline HbA1c values are lower than the norm.
According to Raffield, there's a fundamental problem with using white European populations as a standard for genomic studies. She's working with a consortium established by the NIH, Polygenic Risk Methods in Diverse Populations (PRIMED), which aims to improve methods and applications of polygenic risk scores in diverse populations. Because all human populations arose from Africa, and genetic diversity is greatest in African populations, the sections of genes that tend to be inherited together, or haplotypes, are shorter. Non-African populations tend to have longer stretches of DNA that haven't been broken up by historical recombination events.
"Because of this genetic diversity, if you build the [polygenic risk] score just in individuals of European or East Asian ancestry and then try to apply it in African populations, it almost always works significantly worse," said Raffield.
Some organizations and biotech companies including the Latin American Genomics Breast Cancer Consortium (LAGENO-BC), Myriad Genetics, and Invitae are using more diverse data to develop polygenic risk scores.
LAGENO-BC is pooling resources among researchers from seven countries to identify breast cancer risk variants unique to Hispanic and Latina populations. Myriad has recalibrated its riskScore for breast cancer to provide more accurate estimates for women of all ancestries. Meanwhile, Invitae recently partnered with Allelica to develop a polygenic risk score for breast cancer incorporating multiple ancestries.
Rakesh Patel, chief medical officer of digital health at Invitae, said the lack of diversity in genomic databases limits their ability to inform risk prediction for a broad swath of the population. "Diversity in these databases is important to capture both differences in population genetics, as well as to account for the gene-environment interactions which are essential for an informed understanding of disease risk," said Patel. "We would certainly hope that this important issue is addressed in the FDA's draft guidance, as a database that lacks representativeness has significant methodological and ethical implications."
Raffield said trial sponsors should be aware that biomarkers may have a genetic component that is not completely understood, especially in diverse populations, and that diagnostic test results can be misleading. For example, she has found genetic polymorphisms that interfere with antibody binding that are rare in European populations.
Immunoassays that rely on such biomarkers could produce inaccurate results when broadly marketed. "The currently used commercial and clinical-use assays are not robust to human genetics and people are not always measuring what they think," said Raffield.
Strategies to increase diversity
Although it's important to be aware of disease biology and genetic biomarker differences across racial and ethnic groups, these aren't the main drivers of healthcare access and outcomes disparities. Social determinants of health, systemic bias, and racism are to blame for a lot of inequities in healthcare and carry over into clinical trials.
While the challenges may be greater for diversifying precision oncology trials that rely on targeting rare tumor markers, some of the strategies drugmakers are investing in to improve diversification, including more outreach and establishing trial sites in underserved communities, will likely be the same for biomarker-driven and non-biomarker trials.
As a drug industry leader, Richie said Genentech must set expectations for improving clinical trial diversity, and a big part of that is making sure diverse populations have access to research sites. "We know that 80 percent of minorities in the US live in 20 percent of the zip codes," said Richie. "We have to be really purposeful about where we put our [trial] sites and what kind of investigators we partner with, so that we really have better access and capabilities to reach minority patients."
Even though communities that have traditionally had limited access to clinical trials may trust scientists less, Richie noted that a survey her group conducted showed that patients from these underserved communities were just as likely as people in other places to partake in a trial if asked. It's just that patients from underserved groups are less likely to be asked. "What we found is it's much less likely, for example, for a Black patient to be asked to participate in a clinical trial," said Richie.
These sentiments were echoed in the FDA's roundtable discussion. Representatives from Pfizer and AstraZeneca, as well as researchers and patient advocates, agreed that putting clinical trials in communities typically excluded in research and including investigators that look like and understand the experiences of the people living there would go a long way toward encouraging study participation.
This is a more resource-intensive proposition than how globally positioned drugmakers currently conduct clinical trials. Richard Pazdur, director of the FDA's Oncology Center of Excellence, noted that almost all pivotal trials for cancer drugs take place across multiple countries. "Twenty percent [of the data] come from the US, which is not a lot considering the size of our market," said Pazdur. "We sometimes see more people coming from selected Eastern European countries than from the entire US."
"Should we be taking a more careful look at how companies, especially global companies, select sites?" he wondered during the panel discussion.
Sandra Amaro, head of clinical trial diversity at Pfizer, agreed that these conversations need to take place and that even the term "diversity" may need to be reconsidered for the purposes of the guidance. "When you start to think globally, the term diversity is different," said Amaro. "Even the way we collect race and ethnicity is not translatable outside the US."
Some drugmakers seem to be taking the agency's message seriously, though. Serban Ghiorghiu, VP of clinical development oncology R&D at AstraZeneca, said that his company is considering introducing a diversity plan even before a drug is tested in humans and making trials cost-neutral to patients and caregivers.
"We will use a data-driven approach to identify new sites and investigators. If needed, we'll be prepared to support and train investigators. Patients coming from racial and ethnic minority disenfranchised populations will connect and trust much better investigators of the same background," Ghiorghiu said, noting that the company wants to strike "the right mix between diversity and quality of the data."