Global Disparities in Ancestral Representation in Genetic Studies of Psychiatric Disorders

A disproportionate majority of participants in published genetic studies of psychiatric disorders are of European ancestry (+72%). ^, As depicted in Fig. 1 , using schizophrenia (SCZ), bipolar disorder (BIP), and major depressive disorder (MDD) as examples, the proportion of ancestral populations in large-scale psychiatric genetic studies ( left boxes ) are not reflective of the relative proportion of global ancestral populations ( right box ). All ancestral populations are significantly underrepresented except European ancestry. To-date, the genome wide association studies (GWAS) of psychiatric disorders in non-European ancestry populations are typically many orders of magnitude smaller than European ancestry cohorts, and if unaddressed, this will have serious scientific and ethical consequences. Furthermore, the inclusion of non-European populations in large-scale genetic studies has consistently enhanced the discovery of genetic associations across ancestral groups, improved the fine-mapping of causal loci, and increased the transferability of polygenic scores (PGS) across populations. This underrepresentation has implications for future developments in precision medicine and clinical care. Furthermore, the limited representation of individuals from non-European ancestry in genomic databases results in a higher proportion of variants of uncertain significance for non-Europeans, which limits the diagnostic yield and variant interpretation of multi-gene panels and exome sequencing data. ^, Achieving sample size parity across global populations is critically needed to advance the understanding of similarities and differences across global populations in etiology, clinical presentation, phenotypic heterogeneity, and genetic and environmental risk architecture.

Ancestral representation in published psychiatric genetic studies compared with global populations.

(Roseann E. Peterson et al., Genome-wide Association Studies in Ancestrally Diverse Populations: Opportunities, Methods, Pitfalls, and Recommendations, Cell, 179 (3), 2019, 589-603, https://doi.org/10.1016/j.cell.2019.08.051 .)

A clarification regarding race, ethnicity, and ancestry terminology – The terms ‘ race ’, ‘ ethnicity ’, and ‘ ancestry ’ are often conflated, largely due to a lack of understanding of the differences in these constructs. ^, The terms race and ethnicity are social constructs that offer insights into societal influences on disease risk, including experiences of social injustice. Ancestry specifically refers to genetic lineage, indicating the geographic clines of an individual’s recent biological ancestors as reflected by DNA inherited from them. Shared genomic signatures among individuals with similar ancestral backgrounds result from common ancestral migrations, genetic mutations, recombination, drift, and natural selection. These processes lead to variations in allele frequencies of genetic variants and linkage disequilibrium (LD) patterns across populations, which must be appropriately considered to prevent false positive findings arising from statistical artifacts due to population stratification. It is imperative for clinicians and researchers to accurately distinguish between these terms to disentangle biological, psychological, and social factors impacting health outcomes.

Ethical Importance of Global Diversity in Psychiatric Genetics

The human population has grown exponentially over the past millennium. Much of this has been made possible by innovation in science and technology, which transformed agriculture and industry, and also allowed modern medicine to reduce death rates. ^, However, for historical reasons, ideas and methods of science have spread unevenly across the world. This is due to several factors, including systemic inequality, which is an integral part of colonial and racial divisions that, in many ways, provided the resources for the development of certain regions. This has a particular relevance to medicine in general, and psychiatry in particular. Inequalities in access to health care are widely known—a fact which has ramifications for equitable access to genomics —but asymmetries in the knowledge and skills that can be used to address these inequalities are also quite apparent.

The burden of global mental health conditions is characterized not only by a treatment gap but also a quality and prevention gap, particularly in low and middle income countries (LMICs). LMICs have lower expenditures in research and development, hindering their participation in large-scale genetic studies, limiting the applicability of emerging genetic discoveries, and potentially exacerbating existing global mental health disparities. ^{, , ,}

The ethical imperative of ensuring diversity in psychiatric genetics research centers around concepts such as equity and (distributive) justice with respect to participants, researchers, research institutions, and societies. This is an area of inquiry that is relatively lacking with respect to global psychiatric genomics.

Genetic Studies Across Populations Advance Insight into Psychiatric Disorder Etiology

Including a more diverse sample has repeatedly increased the yield of psychiatric GWAS, and in some cases, identified ancestry-specific loci. One salient example comes from Alzheimer’s disease (AD), where the apolipoprotein E (APOE) allele exhibits different frequencies and associated risk across populations. First identified by Pericak-Vance and colleagues in a case-control study, the ε3 and ε4 alleles show increased risk, while the APOE ε2 allele has been found to be protective. The APOE ε4-associated AD risk is weaker in individuals of African ancestry but stronger in individuals of Japanese descent when compared with Europeans. Below, we highlight several studies that identified ancestry-specific loci for schizophrenia, bipolar disorder, and major depressive disorder (MDD). We are bound to advance our understanding of both shared and unique loci as sample sizes grow and become more diverse.

Lack of Ancestral Representation Hinders Polygenic Risk Prediction and Precision Psychiatry

Psychiatric disorders are highly polygenic. Each genetic variant identified in a GWAS explains a small proportion of the heritability but exhibits an additive effect. PGSs are the aggregation of genetic variants across the genome identified in a GWAS weighted by their effect size estimates. It is generated in a training set and then evaluated in an independent testing dataset. PGS can be used to predict an individual’s genetic propensity to a wide range of phenotypes, with some caveats as discussed below.

PGS has been proposed as a tool for precision medicine efforts. To date, thousands of PGSs have been developed, many of them with potential clinical utility in predicting disease risk, even more accurately than clinical models like in breast cancer, type 1 diabetes, and coronary artery disease. For complex traits like psychiatric disorders, PGS performance is modest, as demonstrated for schizophrenia and depression. Considering that PGS account for only a small fraction of the phenotypic variance in psychiatric disorders, their clinical utility remains far from being a practical reality.

Most PGS to date have been developed from GWAS using primarily European ancestry populations and show limited transferability to other ancestral populations. ^, Studies evaluating the performance of PGS across ancestries have found that the predictive power of PGS is significantly attenuated in non-European ancestry populations, particularly among African ancestries. This has been observed across a range of psychiatric disorders including schizophrenia, bipolar disorder, and major depression. ^{, , ,}

The accuracy, transferability, and potential clinical utility of PGS are limited among non-European ancestry populations for several reasons. First, most data used to generate PGS are derived from GWAS with strong ancestry biases toward European ancestry. ^, Second, ancestral populations differ in LD patterns and allele frequencies. Third, there may be distinct heritability estimates for the trait of interest across different populations. Fourth, individuals from different populations experience unique environmental exposures and gene-environment interactions. ^, Additionally, PGS can be affected by uncorrected population stratification. Finally, there is a lack of methods specifically designed to address recent ancestry admixture.

The limited generalizability of PGS can be trait-specific. ^, For example, while schizophrenia exhibits a very similar genetic risk (PGS are highly correlated) among European and East Asians, the different prevalence of alcohol use disorder across populations is influenced by differences in availability of alcohol as well as genetic variants involved in alcohol metabolism (eg, ALDH1 variants). Further, poor transferability can be observed within the same ancestral group. In Africans, a recent study showed that PGS performance of lipid traits remained limited when comparing African Americans with African individuals from sub-Saharan Africa, specifically South African Zulu and Ugandan populations. African ancestry is highly admixed and heterogeneous, even more so when comparing Africans in the American and African continents. The poor transferability emphasizes the impact of geographic, historic, and environmental differences, as well as admixture composition in the PGS performance.

While the potential of PGS as a clinical tool for predicting disease risk is highly promising, its widespread implementation in the clinic remains premature due to evident limitations in transferability across populations. Inaccurate risk stratification in already underserved and vulnerable populations is concerning and can be harmful for multiple reasons, including: (1) Increased health disparities by widening the gap to health care access and worsening health outcomes for those underserved, (2) Misallocation of resources , which may lead to unnecessary interventions and resulting economic burden or overlooking those at higher risk but do not receive adequate health care, (3) Misleading information , which can cause a false sense of security or anxiety for those who receive inaccurate PGS leading to inappropriate health behaviors or avoidance of necessary health care services, (4) Mistrust from underserved communities in health care systems and providers which can discourage them from seeking health care services, leading to delayed diagnosis and treatment, or even death. The early and premature implementation of PGS is also ethically wrong and may result in the exploitation of underserved and vulnerable populations, perpetuating systemic inequalities in health care access and services. This is particularly concerning in countries like the United States, where these communities are already being affected by significant health disparities embedded in our health care system.

Data—Fully Utilize Existing Data While Large-Scale Data Collection Is Expanded Globally

Advancing health equity in psychiatric genetics will require a concerted effort to fully utilize existing datasets while also significantly expanding global large-scale data collections. These efforts are crucial for addressing the persistent disparities in genetic research that have limited our understanding of psychiatric conditions across populations. Current psychiatric genetic studies are heavily skewed toward populations of European descent, with only modest inclusion of individuals from other groups (eg, African, Indigenous, Latin American, East Asian ancestries). This underrepresentation not only hampers our ability to detect genetic associations across populations but also limits the scope of research that can be conducted on gene-environment interactions, which are crucial for understanding the full complexity of psychiatric disorders.

The first important step is to fully utilize existing datasets that include diverse populations. Some underutilized resources, such as ancestral groups within the UK Biobank, offer a wealth of data that could be leveraged to address gaps in our understanding. By aggregating and integrating these existing resources with new data from underrepresented regions, researchers can build a more comprehensive and inclusive picture of the genetic and environmental factors that contribute to psychiatric disorders. ^,

One of the primary challenges in psychiatric genetics is the underrepresentation of non-European populations in existing datasets. While there has been some inclusion of African, Latin American, and East Asian individuals in European and North American datasets, their representation remains disproportionately small. ^, This underrepresentation limits the ability to draw meaningful conclusions about genetic risk factors across populations and reduces the generalizability of findings. To truly advance health equity, it is essential to move beyond the current focus on European populations and extend data collection efforts globally to include participants from regions such as Africa, Latin America, Asia, and the Middle East.

Collecting data from participants residing in these regions, rather than relying solely on diaspora populations in Europe and North America, is vital for studying gene-environment interactions in their environmental and cultural contexts. While diaspora studies provide valuable insights, they cannot fully capture gene-environment interactions that are influenced by the unique environmental factors present in the regions where these populations originate. ^, For example, factors such as diet, climate, socio-economic conditions, and local health care practices can significantly influence the prevalence and presentation of health and wellness. By expanding data collection efforts to these underrepresented regions, researchers can develop an understanding of how these factors interact with genetic predispositions to influence psychiatric outcomes.

Equitable Collaboration

Establishing large-scale discovery cohorts from diverse ancestries is a critical need. However, such efforts will require extensive collaboration across international borders and disciplines, bringing together researchers, health care providers, and communities to create more representative and comprehensive datasets. Collaborative efforts can leverage the expertise and resources of multiple stakeholders, ensuring that data collection is conducted in an ethical and culturally sensitive manner. These collaborations can facilitate the sharing of data and knowledge, accelerating the pace of discovery and translating research findings into clinical practice, but must be done in an equitable manner.

Such data collection at the required scale relies on equitable global collaboration and capacity building. As highlighted by Emmanuel and colleagues, research in less resourced countries can only be considered ethical if collaborative; this means that such research must be co-developed with researchers, policy makers, and host communities, with shared responsibility for planning, implementation, evaluation, and integration of the research findings into the health care system. Importantly, Emmanuel and colleagues also benchmarked capacity building, that is, developing the capacity of researchers, policy makers, and the community to become “full and equal partners in the research enterprise”. In the highly technical terrain of genomics research, this is an ethical as well as a scientific imperative, as for such research to fit comfortably in the larger global domain, it must be of the highest quality, comparable to similar work in other parts of the world. Comprehensive approaches to capacity building will recognize that the enterprise of genomic research involves a cycle of project planning and management, clinical/field work, laboratory processing, and bioinformatic analysis, with different training needs at each step. In order for capacity building efforts to be done right, it must be scalable and compatible with the regular work and study commitments of recipients. At best, such efforts will occur in tandem with data generation efforts so that trainees have the opportunity for direct hands-on application, as this is the best way to consolidate learning. Good models of collaboration such as Neurogap and H3Africa/H3ABioNet ^, have successfully implemented this model of embedding training and career development opportunities within data and sample collection efforts. As more funding projects spread across LMICs, it is crucial to replicate this model and to utilize mechanisms such as embargo periods and data management/access committees to ensure that researchers from LMICs have first access to the data that they generate.

Importance of Global/Ancestral Representation Among Research Teams

These efforts have been crucial in diversifying psychiatric genomics research and, to some extent, in broadening the workforce. However, additional efforts are needed to further improve the representation of diverse researchers in the workforce. It has been well established that diverse research teams advance innovation by bringing together a wide range of perspectives, experiences, and expertise. It can foster creativity, help identify and mitigate biases in research design, implementation, and interpretation, and be driven by more inclusive research questions that likely reflect the needs and concerns of various populations. Further, researchers from diverse backgrounds offer a nuanced understanding of environmental factors impacting genetic studies, such as cultural practices and societal structures. They also comprehend the genetic diversity within their regions and the sources of the genetic variation, including geographic isolation, migration patterns, and historical contexts. This comprehensive perspective is essential for ensuring ethical research practices and producing findings and tools that are robust, informative, and pertinent to the populations involved. Promoting diversity in genomics research not only advances scientific knowledge, but also facilitates the development of more equitable and impactful health care solutions that can benefit all populations, regardless of their backgrounds.

Advanced Methods

Analyzing genetic data across diverse ancestral populations has often been seen as challenging, but it is both scientifically and ethically essential. The growing availability of analytical tools is making this work increasingly feasible. Traditional GWAS have relied on ancestral homogeneity to minimize population stratification, which has led to the exclusion of admixed and non-European ancestry individuals. However, emerging trans-ancestry methods now provide the means to capture ancestral diversity and promote inclusivity. Expanding the ancestral diversity of study populations will enhance the effectiveness of genomic medicine by broadening our understanding of genetic risk and resilience factors, ultimately improving our insights into disease etiology. Increased representation of diverse populations brings many benefits for locus discovery, fine-mapping, PGS, and addressing health disparities.

Despite this progress, further methodological developments are needed to fully realize health equity in psychiatric genetics. Current population reference panels do not adequately represent global populations, leaving gaps in understanding genetic variation in underrepresented groups. Moreover, many post-GWAS methods still require single-ancestry input, limiting their use across diverse populations. To address these challenges, cross-ancestry methods for heritability estimation, genetic correlation, and functional annotation must be further developed. The creation of more representative variant databases and ethical, globally inclusive data-sharing frameworks will be crucial to ensure that advances in genomic medicine benefit all populations.

Differences in LD structure and haplotype blocks across ancestries provide both scientific challenges and opportunities. While new methods must be developed to account for these variations, the potential for new discovery is significant. Given that many common variants predate the migration out of Africa, the patterns of selection, population bottlenecks, and environmental influences across different groups can impact LD structures, complicating risk detection. Additionally, understanding genetic and environmental risks requires large, transcontinental cohorts, as admixture has been a constant feature of human evolution. Validating risk factors across populations will provide a clearer understanding of these dynamics.

Early efforts to diversify GWAS have shown that training PGS models using multi-ancestry summary statistics can improve PGS performance by capturing a broader range of genetic variation. For instance, PGS derived from African ancestry populations perform better in sub-Saharan African populations than those derived from European populations. ^, The availability of large-scale datasets from multiple ancestries, such as the All of Us Research Program, the PGC, and the Million Veterans Program, has enabled these advancements. Global initiatives like the PRIMED Consortium (Polygenic Risk Score Task Force of the International Common Disease Alliance) are also focused on improving the applicability of PGS in diverse populations. Recent advancements in PGS methods aim to enhance prediction accuracy in diverse populations. One example, PGS-CSx, integrates GWAS summary statistics from multiple populations, improving cross-population prediction by using more accurate effect size estimates and leveraging LD diversity. This Bayesian-based method has demonstrated improved schizophrenia risk prediction in non-European populations. Newer methods such as PROSPER and GAUDI, which utilize fused lasso penalty functions, are showing even greater prediction accuracy in African ancestry populations, outperforming PGS-CSx. To further improve PGS performance, better imputation panels, multi-ancestral genotyping arrays, whole-genome sequencing approaches, and integrated models that combine PGS with clinical, lifestyle, and environmental data have shown promise. This is also true for GWAS, which can benefit from more complete etiological models as shown in previous work using stratified GWAS based on stressful life events uncovering new associations otherwise masked in standard GWAS models.

One central question that has always engaged cultural psychiatry is the tension between a view of psychopathology as common across cultures versus a view that highlights cultural differences. This is important not only for nosological reasons and diagnostic validity but also because methods such as GWAS assume a common biology. A widely used framework, borrowed from sociologists and anthropologists, with emic approaches identifying entities and concepts unique to individuals and groups, while etic approaches focus on what is common to all cultures. Teasing these out on a granular scale in the dance of genetic and environmental factors is a major challenge. Allied to this is the ongoing need for transcultural phenotyping, which has been the purview of the producers of subsequent editions of the International Classification of Diseases and the Diagnostic and Statistical Manual. There is a clear need, as championed by bodies such as the Wellcome Trust, for common measures of psychopathology, which will in turn raise questions of measurement of invariance across different parts of the world. There is an ongoing realization that research must go beyond the current diagnostic categories to accommodate transdiagnostic and dimensional and developmental phenotypes.

Specific genetics-related challenges include data sharing, in view of a growing bioresource, country-specific data privacy policies, mistrust by less-resourced investigators given past scientific injustices and biocolonialism, and data nationalism. The perennial challenges of infrastructure and cluster access, funding, training and collaboration opportunities will need to be solved in tractable ways to ensure a true decolonisation of the field.

Accountability and Call to Action

Efforts to promote equitable collaboration in psychiatric genetics, particularly across the global south, are crucial for updating our understanding of the relationship between genetics and mental health across diverse populations. In regions such as Africa, Asia, and Latin America, a young, vibrant, and ambitious demographic often faces significant barriers due to historical and systemic constraints, including inadequate government policies, career instability, and infrastructure deficits. Aspiring researchers in these regions contend with a lack of local funding, insufficient training opportunities, and dependence on foreign funding, as well as inadequate infrastructure ranging from power and broadband connectivity to well-equipped laboratories and biobank services. Specific challenges in genetics research, such as limited access to bioinformatics software, genotyping chips, and imputation panels, further exacerbate the problem. While collaborations with the global north have alleviated some of these issues, there is still much work to be done to achieve equity.

As Ruzycki and Ahmed emphasize, the failure to incorporate equity, diversity, and inclusion in biomedical research is harmful to patients, trainees, and researchers alike. They call for integrating diversity and inclusion throughout all stages of research, from study conceptualization and team composition to data generation and knowledge translation. Similarly, Palk and colleagues highlight ethical challenges in neuroimaging genetics, warning against pitfalls such as genetic essentialism, stigma, and ethics dumping , where ethical considerations are neglected in cross-regional collaborations. The TRUST Code further reinforces that collaborations must be based on the values of fairness, care, respect, and honesty to ensure balanced and ethical international partnerships.

One key challenge is the underrepresentation of non-European individuals in large-scale genomic studies, limiting the development of PGS models that account for the genetic diversity of different ancestral groups. Including these populations in research will not only improve our understanding of disease etiology by capturing different LD patterns and allele frequencies but also address the largest health disparities faced by these groups, which could be exacerbated by their continued exclusion.

To overcome these barriers, funding agencies, journals, and professional groups must take an active role in implementing structural changes. Harmonizing guidelines for sharing genetic material and data, while respecting local laws, would facilitate distributed collaboration across regions. Initiatives must focus on equitable funding distribution, providing infrastructure support, and promoting capacity-building efforts, especially in underrepresented regions. International partnerships should prioritize mutual benefits and create frameworks that allow for the equitable exchange of resources, data, and knowledge.

Increasing global participation in psychiatric genetics, both among investigators and study participants, is essential for attaining truly representative ancestral diversity. Continuous support for existing and novel initiatives—particularly those that facilitate capacity building, global data collection, and equitable collaboration—will help to expand the reach of psychiatric genetics research. By doing so, we can ensure that genetic discoveries are broadly applicable and generalizable, ultimately improving health outcomes for all populations. Institutions must prioritize clinical impact and hold themselves accountable through multi-agency commitments to ensure that mental health research reduces disparities and has both global and local benefits.