Researchers at the Memorial Sloan Kettering (MSK) Cancer Center have shown that cancer patients from different ancestral backgrounds have not benefited equally from FDA precision oncology drug approvals over time, with lowest biomarker-based clinical actionability observed among people of African ancestry.
Writing in JAMA Oncology, MSK molecular geneticist Debyani Chakravarty, PhD, and co-authors say that the “difference in eligibility for precision oncology treatments is a concerning trend that should be monitored and may partially contribute to the larger, well-documented disparities of care access that face Black individuals in the U.S. and other historically marginalized populations.”
Chakravarty and team retrospectively analyzed samples from 59,433 patients with solid cancers who underwent clinical sequencing using the MSK-IMPACT assay, which looks for cancer-causing mutations in 505 cancer-associated genes, between 2014 and 2022.
Of these, 60.1% of patients were classed as having non–Ashkenazi Jewish European ancestry, 15.6% as Ashkenazi Jewish European, 5.0% as African, 5.7% as East Asian, 1.9% as South Asian, 0.4% as Native American, and 11.3% as having admixed ancestry.
Chakravarty, who was senior author of the study, explained that the classification was based on a previously published ancestry inference method developed by her team.
“Briefly, patient DNA sequencing data can identify subsets of heritable single nucleotide polymorphisms (SNPs) that vary across ‘continental’ populations in part due to population migration patterns,” she said. “The proportion of a patient’s genetic ancestry is determined by quantification of how much or how little of a specific subset of SNPs are present in the patient’s genome. Patients were assigned to the ancestry group that comprised at least 80% of their genetic makeup.”
She added: “Patients of European ancestry (i.e. where the European ancestry SNPs comprised ≥80%) were further subgrouped based on the presence or absence of a different set of SNP markers common in Ashkenazi Jews. If the Ashkenazi Jewish-associated SNP markers were present, the patients were considered of inferred Ashkenazi Jewish European ancestry, if absent, the patients were of inferred non-Ashkenazi Jewish European ancestry.”
After determining the participants’ ancestry, Chakravarty and team used OncoKB, another MSK-developed database that contains hundreds of molecular biomarkers for FDA-approved precision oncology drugs across different cancer types, to calculate the proportion of patients with clinically actionable biomarkers between 1998 and 2023.
They found that, across all cancers, the two largest increases in clinical actionability occurred in 2013 and 2020.
In 2013 this coincided with the approval of erlotinib for EGFR-mutant lung cancer, which contributed to a 9.6% increase in actionability in patients of inferred East Asian ancestry and a 3.9% increase in patients of inferred South Asian ancestry. By comparison, the increases were 2.0% in the patients of inferred non–Ashkenazi Jewish European and 1.6% in the patients of inferred African ancestries.
In 2020, approval of the anti-PD1 immunotherapy agent pembrolizumab for high mutational burden (TMB-H) solid tumors disproportionately benefited the patients of inferred European ancestry.
Altogether, compared with a 9.1-fold (3.6% to 32.3%) increase in clinical actionability in the non–Ashkenazi Jewish European group between 2012 and 2023, the increase in the African group was only 6.0-fold (4.7% to 27.8%).
Overall, the odds of receiving a personalized drug increased approximately 9.1-fold (3.6% to 32.3%) for patients of European ancestry between 2012 and 2023, whereas the increase was just 6.0-fold for patients of African ancestry (4.7% to 27.8%). Patients of East Asian and South Asian ancestries benefitted from 8.5- and 6.8-fold increases in clinical actionability, respectively.
Furthermore, from 2019 onwards, patients of inferred African ancestry had significantly lower clinical actionability than those of European ancestry.
Since 1998, precision oncology therapies that require tumor genomic biomarker profiling to guide patient selection have comprised 43% of FDA–approved oncology drugs.
“In the early years of precision oncology therapy, there were no significant differences in the proportion of clinically actionable mutations among the ancestry groups,” said MSK computational biologist Kanika Arora, lead author of the paper. “But as more drugs have received approval, the differences have grown wider.”
Chakravarty said this finding was unsurprising. “Cancer disparity research has long known that first, different subtypes of cancer, are more prevalent in patients of one ancestry compared to patients of another ancestry (such as triple-negative breast cancer or right-sided colorectal cancers that are more prevalent in patients of African ancestry and associated with worse disease prognoses).”
“Second, it has also long been known that patients of different ancestry or self-reported race have different prevalence of specific mutation events.”
The researchers say that one of the main reasons underpinning the differences in the biomarker-based clinical actionability by patient ancestry is the underrepresentation of non–European ancestries in clinical research databases and biobanks.
“Over 80% of the samples included in these pooled datasets are from self-reported white patients, predominantly of European ancestry, because these are the patients who historically have been most likely, and often most able, to participate in clinical trials,” Chakravarty said. “This means that biomarker discovery and subsequent efforts to develop molecularly matched drugs are based on data from patients who are overwhelmingly of European ancestry.”
The study authors therefore concluded: “To prevent growing disparities in biomarker-based clinical actionability, those not of non–Ashkenazi Jewish European descent must receive equitable access to biomarker testing and clinical trial enrollment to ensure that future biomarker discovery and drug development efforts are based on a foundation of genomic data that equally represents patients across all ancestries.”
