NEW YORK – Amidst calls for greater diversity in clinical trials and better inclusion in the life sciences in general, a smaller group of voices is calling for greater diversity in the cell lines used in primary research, often the first living things in which potential drug candidates are tested.
In 2009, Sean Morrison and his colleagues at the University of Michigan published a study analyzing the genetic diversity of the then most widely used human embryonic stem cell lines. Of those 47 cell lines, the researchers determined that most came from individuals of northern and western European ancestry.
Other studies have gone on to both confirm and expand upon these results, making similar findings with immortalized cell lines and even finding several cases of cell line ancestry having been miscategorized.
Because candidate therapeutic compounds are generally tested first in vitro in cells, the lack of genetic diversity so early in the drug development process can have many knock-on effects down the line.
Furthermore, the pan-cancer genomic and proteomic atlases used to dissect disease pathways and find potential biomarkers are built from cell line data, meaning that their built-in genetic biases will necessarily be reflected in any research based on them.
But to what degree does this lack of genealogical diversity among cell lines impact research and drug development?
"The simple thing is, we don't know, and we probably should have more diversity so that we don't have any unforeseen issues," said Yaw Bediako, founder and CEO of Ghanaian cancer research company Yemaachi.
As an example of issues potentially arising from a lack of diversity in research, Bediako points toward efavirenz, an antiretroviral used to treat and prevent AIDS/HIV, which was found to have significantly more adverse side effects among Black Zimbabweans than among the largely Caucasian patients that it was first tested in, within the US.
Compared to the US population, Zimbabweans experienced higher rates of treatment failure and neuropsychological toxicity. These outcomes were traced to a single nucleotide polymorphism that is considerably more prevalent among Zimbabweans — and among African populations more generally — than among people of European ancestry.
This SNP, called CYP2B6*6, causes a glycine-to-histamine amino acid change that impacts efavirenz metabolism.
While it's hard to draw a straight line from in vitro assays to efavirenz's effects among Zimbabweans, Bediako points out that at some point, a toxicity study had to have been done in an in vitro cell model before determining the dosing to be used in human studies.
"If we had better in vitro model[s]," he said, "it may allow us to design better studies, and maybe safer trials as well, because the in vitro studies would allow us to define the parameters that could then go into the trial."
Peter Fekkes, VP of drug discovery at 54gene, a Nigerian genomics company, echoed Bediako's sentiment.
"In the end, the ideal initial model to validate an observation in the field is a model with a similar genomic makeup," he said. "If you can then show that the association between a disease and a genomic variation holds up in other models as well," he added, "the chance that the therapy that comes from this work can be applied globally increases."
If a potentially actionable insight into a biological process — such as identifying new drug targets — is prevented because of having a genetically limited set of models to interrogate, "that's a missed opportunity to gain more insight in the molecular mechanisms of a disease," Fekkes said.
Lila Collins and Uta Grieshammer, both science officers at the California Institute for Regenerative Medicine (CIRM), which both funds stem cell research and banks stem cell lines, pointed out another case where genetic background altered a medication's effects.
"A relevant example," they wrote via email, "is the discovery of PCSK9 as a target for lowering LDL cholesterol, which was enabled by the discovery of a mutation, prevalent in African Americans, that causes low plasma levels of LDL."
Collins and Grieshammer stressed that it is important not to forget that environmental and social determinants of health contribute significantly to health disparities. The role that genetic background does play in disease outcomes, however, makes testing in an ancestrally diverse collection of cell lines vital, as it maximizes the chance of discovering relevant genetic differences and new drug targets that might only appear in specific populations.
Yemaachi and 54gene are both working to address diversity issues in biomedical research on multiple fronts.
Yemaachi is actively involved in cataloguing genomes representative of Africa's highly diverse population, and in generating an African liver tissue bank, with which to explore drug metabolism and enzyme activity. Eventually, Bediako expects that these initiatives will contribute to the creation of more genealogically diverse stem cell and immortalized cell lines.
On a similar note, 54gene has been actively trying to expand our knowledge of African genomes through an effort to sequence 100,000 genomes across the continent, as well as through other African genomics-related collaborations.
"As we find out more and more about how genealogical differences affect clinical outcome, [and] the pathology of disease, I think it will be more and more important to have cell lines that are a bit more reflective of global diversity," Bediako said.
Yemaachi and 54gene join a growing effort to create more diverse genomic datasets.
The Human Pangenome Reference Consortium, for example, recently completed a draft of a more diverse reference genome. In contrast to the current reference genome, GRCh38, of which a single person accounts for 70 percent, the updated genome will feature contributions from 47 individuals from the 1,000 Genomes Project, which represents 26 global populations.
Impact on prostate cancer
Prostate cancer is one of the more heterogeneously expressed cancers and high-throughput in vitro screening assays have been instrumental in identifying lead compounds for the development of therapies.
Although prostate cancer rates have declined in the US and Europe, men of mainly European ancestry have benefited most from this trend, responding more readily to current chemotherapies than men of African ancestry.
Although hypothetical, a better cell line model may plausibly have uncovered disease mechanisms that could have led to treatments more likely to benefit men of African ancestry, if not more men in general. One study, for example, found that tumors obtained from African American men were more likely to carry gene splicing variants associated with tumor aggressiveness and poor outcomes than tumors from European Americans.
Despite the increased mortality among men of African ancestry, one study found that approximately 97 percent of the 32 listed human prostate cell lines available from major suppliers ATCC, Sigma Aldrich, and ECACC are European in origin where the cell line's ethnicity is known.
"The goal of cancer research is to ensure that laboratory studies are reflective of patient populations," Jesse Boehm, a principal investigator at the Broad Institute who focuses on functional genomics as applied to cancer, said, "because you want to translate the work from those laboratory studies into the clinic."
When models fail to reflect real-world populations, he explained, "there might be missed opportunities to discover things that require new types of cell models. And there might be false positives that work in one population and not in others."
In a presentation earlier this year at the American Association for Cancer Research annual meeting, Sean Misek, a postdoctoral researcher in Boehm's lab, made an example of how ancestry-based differences could cause experimental bias in CRISPR screening.
Cuts made by the Cas9 enzyme in cells, he explained, such as cancer cells, cause double-strand breaks that make those cells more likely to die. A SNP in the target sequence, however, can cause a mismatch between that and the guide targeting sequence, preventing Cas9 from binding and cutting. These cells then won't get any genome editing and will survive.
In a study presented at AACR in June, whose abstract is published in Cancer Research, Misek tested how differing ancestries might cause artifacts within CRISPR screening data. Misek and his colleagues tested four CRISPR guides targeting the Claspin gene in cell lines from diverse ethnic backgrounds.
One of those four guides showed a mismatch in approximately 60 percent of East Asian cell lines, compared to about 10 percent of European cell lines. Extending these results to larger datasets, Misek and his colleagues computationally mapped SNPs from approximately 75,000 genomes in the gnomAD database to guide targeting sequences for six commonly used CRISPR libraries.
"What we found is that there's a range of between 2 and 5 percent," he said. "What this means is that depending on how you design your guides, between 2 and 5 percent of the guides are going to be affected by this artifact."
While this may sound small relative to the overall dataset, significantly more mismatch occurred among individuals of African descent.
"What we saw across every single CRISPR library is that although every individual regardless of ancestry group is affected, individuals of African ancestry are more affected," Misek said, adding that "the absolute magnitude of the signal is quite low, but the relative magnitude of the signal between individuals of [diverse] ancestry is quite high."
Misek and his colleagues at the Broad are now designing a tool, currently available as a beta version, to screen CRISPR guides for their susceptibility to ancestry bias.
Other cellular model systems
Immortalized cell lines aren't the only type of cell line available for early biomedical research. Nor, in the opinions of some, do they even make ideal systems for the cancers they're meant to model.
"They may have extra chromosomes, or they may have lost certain chromosomes," said William Skarnes, director of cellular engineering at Jackson Labs Genomic Medicine. "They have all sorts of rearrangements in their genome. They're not even human, in my opinion."
The Jackson Laboratory (JAX), a nonprofit biomedical research institution, aims to improve the state of cell line diversity through a panel of induced pluripotent human cell lines made from people representing a wide variety of ethnic backgrounds.
Skarnes recently submitted a grant proposal seeking to establish the provisionally named JAX Diversity Panel through the institution's Director's Innovation Fund.
In its initial phase, the panel would consist of approximately 10 iPS cell lines made from people of non-European ancestries.
The idea for the Diversity Panel grew out of a large project involving the development of Alzheimer's disease cell models. The group behind that project, led by Skarnes, has completed one cell line of northern European descent and is now looking to engineer the same mutations in cell lines of more diverse backgrounds, which has proven to be challenging.
"Either they've been reported in the literature, but you can't find where to get them," Skarnes said, "or if you try to get them, there are all kinds of restrictions on their use, which makes them impossible for us to work with because we want to be able to hand these cell lines out to the entire community. So JAX made the decision to say, OK, we're just going to make our own."
The JAX Diversity Panel will join several other iPS cell resources, such as the European HipSci resource and the CIRM iPSC Initiative. HipSci contains over 700 human iPS cell lines derived from fibroblasts of normal healthy individuals of mostly British descent, while CIRM contains approximately 1,500 patient-derived iPSCs across multiple disease areas.
While these represent some of the largest-scale resources available, neither presents a case of equitable diversity. Of the 1,554 cell lines in the CIRM collection at the time of writing, 80 were annotated as coming from Black or African American donors, 111 were annotated as Asian or Asian/Other, and only two were annotated as American Indian.
To provide something of a counterbalance, Skarnes said that JAX is currently aiming to collect samples only from non-Europeans. JAX also wants its panel to be more comprehensive in the cellular data available for each line. The CIRM biobank, for instance, currently lacks genomic and transcriptomic data for each cell line, and HipSci contains whole-genome sequence information for only a fraction of its lines.
CIRM acknowledges that the lack of cell line diversity is an issue and is working to increase that of its own library.
"There are certainly many more hESC and hiPSC lines available for use today, in part thanks to CIRM's efforts," wrote CIRM's Collins and Grieshammer.
Although most of their cell lines are still derived from people of European ancestry, they noted that their collection is growing more diverse, as are those of other institutions, such as the New York Stem Cell Foundation, which offers its own Ethnic Diversity iPSC Panel.
"We are seeing more and more of our grantees taking advantage of this diversity and using these cell lines as tools against which to test their therapeutic approaches," Collins and Grieshammer wrote.
CIRM, JAX, and others expect that funding for such diversity panels will increase as demand increases.
Tenneille Ludwig, senior scientist of the Wisconsin International Stem Cell Bank, a highly regarded stem cell repository affiliated with the Wisconsin Alumni Research Foundation, commented that "ensuring inclusion necessitates larger studies, which brings larger costs. Accordingly, this needs to be a priority not only for scientists, but funders as well."
Push to fund diversity
Some funders are taking notice and acting. The Human Cancer Models Initiative (HCMI), for instance, is an international consortium working to develop cancer cell models that better represent the hallmarks and diversity of human cancer. Consortium funders include the National Cancer Institute, Cancer Research UK, and the Wellcome Sanger Institute, while institutions developing cell and organoid models for the consortium include the Broad Institute, Hubrecht Organoid Technology (HUB), and Cold Spring Harbor.
"Cancer Research UK are actively encouraging researchers to diversify their research materials, to better understand how cancer affects everyone," said Sam Godfrey, research information lead at CRUK.
"We are working with our scientists, including the experts reviewing applications for new studies, to understand where diversifying biological materials adds the most valuable insights about cancer," he added.
Godfrey noted that the "long route" from research done in the lab to improvements in patient care seen in the clinic, as well as the time it takes to generate more diverse datasets, present challenges to this effort. CRUK, he said, actively seeks to fund research projects aimed at diagnosing and treating cancer among people of different ethnicities.
Cell line distributor ATCC banks and distributes the cell lines and organoids developed by HCMI participating organizations, and a spokesperson for the nonprofit commented that the vendor "recognizes the need for an industry-wide evaluation of genetic diversity to support the scientific community."
"The generating institutions," the spokesperson said, "deposit the models into ATCC, where they are authenticated, expanded, preserved, and made available for global distribution."
Although ATCC didn't provide specific information on measures aimed at narrowing the cell line diversity gap, the company noted that the material it receives from HCMI generating institutions such as CRUK and HUB include cell lines and organoids derived from understudied populations.
Rob Vries, HUB's CEO, suggested that organoids might one day supplant cell lines as early drug discovery and screening models.
"When it comes to direct relevance for patients and therefore for drug development," he said, "[cell lines] have very limited potential."
In part, he explained, echoing JAX's Skarnes, their limits stem from the adaptations to make them viable in vitro.
"In contrast," he said, "organoids can be established for most patients in a relatively short time frame, while maintaining all patient characteristics including heterogeneity."
The Broad's Boehm sees this confluence of funders and researchers as a step in the right direction toward better, more diverse, cellular models, but pointed out that more is needed.
"It's a more complicated problem than funding alone can solve," he said. "We need patients everywhere to be engaged and to donate bits of their tissue so that we can make cell models from them, especially if they come from communities that have historically distrusted medical research."
Such distrust can run deep in underrepresented communities.
The infamous Tuskegee Study, for instance, ran from 1932 to 1972 and remains one of the most highly referenced studies in explaining the high levels of mistrust that African Americans report feeling toward medical research in general.
Originally called the "Tuskegee Study of Untreated Syphilis in the Negro Male," it sought to characterize the natural course of the disease. The scientists conducting the study did not obtain informed consent and did not treat participants with penicillin, even though it had become widely available as the treatment of choice for syphilis in 1943.
"It was common practice to do research on humans without their consent and typically, the marginalized, primarily Black people were more accessible targets," wrote Simone Badal, a researcher with the University of the West Indies, in a comment on the need for ethnically diverse cell lines for drug testing in Nature Reviews Cancer earlier this year.
"We have to find ways of addressing that [distrust] and inviting people from those communities to fully participate," Boehm said. "We have to incentivize researchers in the lab to look and mitigate source of bias and we have to be prepared to provide funding for those types of projects. I think everyone has a role to play, but I'm excited that there's now a lot of attention on these issues."