Diversity in a dish: Leveraging organoids to reflect genetic ancestry and sex differences in health and disease

This review addresses how genetic ancestry and biological sex shape variability in disease susceptibility, therapeutic efficacy, and adverse drug reactions. It highlights the persistent lack of diversity in genomic datasets (e.g., GWAS) that hinders fine-mapping of loci and development of accurate PRS, and it underscores documented ancestry- and sex-stratified differences in disease prevalence and pharmacology. The purpose is to evaluate advanced in vitro human models—especially hPSC- and adult stem cell–derived organoids—as scalable, physiologically relevant platforms to model human diversity, uncover genotype–phenotype relationships, improve pharmacogenomic prediction, and inform equitable precision medicine.

The review synthesizes evidence that: (1) GWAS and PRS remain biased, with about 79% of samples of European ancestry, limiting transferability to African, Hispanic, Middle Eastern, and Asian populations. (2) Disease prevalence varies by ancestry and sex across conditions including AD, asthma, breast and prostate cancer, ASCVD, cystic fibrosis, SLE, MS, Parkinson’s disease, osteoporosis, IRDs, and type 2 diabetes. (3) Pharmacogenomic variation across HLA and DMET genes shows strong population structuring; examples include HLA-B*57:01 with abacavir hypersensitivity (rarer in African ancestry), HLA-B*15:02 with carbamazepine toxicity in Southeast Asians, and CYP2C19 loss-of-function alleles affecting clopidogrel response. (4) PRS portability issues can substantially inflate risk estimates, e.g., schizophrenia PRS overestimation by ~10-fold in African populations. (5) Sex influences immune responses and drug metabolism; women experience higher rates of autoimmune diseases, differing cardiovascular risk trajectories, and more frequent ADR reports, exacerbated by historical underrepresentation in trials. (6) Racially biased dosing algorithms (e.g., warfarin) can increase bleeding risk in African American patients due to ancestry-specific variants. The review compiles drug-specific ADR susceptibility across populations and sexes (e.g., docetaxel neutropenia in East Asians; efavirenz neurotoxicity linked to CYP2B6*6; isoniazid hepatotoxicity with NAT2 slow acetylators; cisplatin AKI risk in African Americans; SRI-related ADRs more frequent in women). It also surveys organoid studies demonstrating ancestry- and sex-relevant biology, including African ancestry breast cancer organoids (CRISPR kinome screens), African American prostate cancer organoids, sex-dimorphic neurodevelopment in brain organoids, and endometrial organoids modeling hormone responses.

As a narrative review, the article does not detail primary experimental procedures. Instead, it delineates methodological frameworks and best practices for modeling human diversity using stem cell–derived organoids: (1) Model selection: use both hPSC-derived and adult stem cell–derived organoids to capture long-term culture, genetic manipulation, and tissue-specific fidelity. For sex biology, mitigate hPSC X-chromosome inactivation (XCI) instability by selecting lines with stable XCI and validated XIST expression; consider adult stem cell organoids which retain stable XCI. (2) Strategies to represent diversity: (a) Parallel single-donor organoid lines, supported by micro-engineered hydrogel substrates and organoid arrays for improved standardization and reproducibility; (b) Pooled systems (“village-in-a-dish”), including chimeric organoids (chimeroids) generated by disaggregation and re-aggregation of early organoids from multiple donors, enabling single-cell barcoding to track donor contributions; and (c) Population Organoid Panels (PoP), created by mixing clonal progenitors from multiple donors into embedded cultures to form mosaic organoids with donor-specific genomic validation. (3) Throughput and standardization: employ automation, standardized assays and readouts, and gene editing to introduce or correct variants across ancestries. (4) Sampling considerations: leverage insights from population genetics (1000 Genomes, gnomAD) to determine panel sizes—approximately 20–30 lines for relatively homogeneous groups and 50–100 lines for highly diverse populations (e.g., African or Latin American ancestry) to capture rare and common variants. (5) Biobanking and sourcing: expand and geographically diversify organoid/iPSC biobanks; consider HLA-homozygous iPSC haplobanks (e.g., Japan’s CiRA) to facilitate immune-relevant applications and broader applicability; integrate ethical, culturally sensitive consent and data governance. (6) Translational alignment: prioritize use cases in pharmacology and toxicology (e.g., GI toxicity, DILI) aligned with regulatory qualification pathways (e.g., FDA ISTAND).

- Underrepresentation persists: Although nearly half of GWAS now include non-European participants, ~79% of samples are still European, limiting fine-mapping and PRS accuracy across populations. - Population pharmacogenomics: Strong ancestry-related differences in HLA and DMET gene frequencies influence ADR risk and efficacy (e.g., HLA-B*57:01–abacavir hypersensitivity rarer in African ancestry; HLA-B*15:02–carbamazepine toxicity in Southeast Asians; CYP2C19 variants affecting clopidogrel activation). - PRS portability issues: Schizophrenia PRS can overestimate risk by ~10× in African populations; trans-ancestry meta-analyses improve locus discovery and PRS performance (e.g., 43 novel loci in Japanese studies for CAD). - Sex differences: Women show stronger immune responses, higher autoimmune prevalence, distinct cardiovascular risk profiles (risk underestimation by traditional scores), and report ~2× ADRs versus men. Sex hormones modulate pharmacokinetics across the lifespan. - Drug-specific disparities: Examples include faster clozapine metabolism in sub-Saharan Africans versus Europeans; higher docetaxel-induced neutropenia in East Asians; efavirenz CNS toxicity linked to CYP2B6*6 (more common in African/Asian populations); NAT2 slow acetylator–associated isoniazid hepatotoxicity (notably in West Asians); cisplatin AKI risk elevated in African Americans; SRI-related ADRs reported more often in women. - Organoids as diversity platforms: Organoids recapitulate tissue architecture, enable long-term culture and gene editing, and can model ancestry- and sex-specific biology (e.g., African ancestry breast cancer organoids revealing kinase vulnerabilities; African American prostate cancer organoids reflecting aggressive pathway alterations; brain organoids showing androgen-enhanced excitatory neurogenesis; endometrial organoids modeling hormone responsiveness). - Pooled approaches: Chimeroids and PoP enable scalable, reproducible, multi-donor assays with single-cell readouts, capturing interindividual variability in responses (e.g., differing susceptibility to ethanol and valproic acid in brain chimeroids). - Sample size guidance: To capture genetic diversity, panels may require ~20–30 organoid lines for genetically homogeneous groups and ~50–100 for more diverse populations. - Policy and regulatory context: DEPICT and FDORA mandate diversity action plans in trials; FDA Modernization Act 2.0 encourages human-based models; FDA’s ISTAND can qualify organoid assays for regulatory use. - Potential impact: Regionally adapted pharmacogenetics (e.g., in South Africa) could reduce ADRs by up to 30%; diverse, standardized organoid models can improve translational relevance, safety, and efficacy across populations.

The review argues that health disparities rooted in ancestry and sex cannot be resolved without inclusive preclinical models. By aligning organoid model design with population genetics, pharmacogenomics, and sex biology, researchers can better map genotype–phenotype links and predict therapeutic responses. Organoid arrays, pooled multi-donor systems, and standardized/automated pipelines address scalability and variability barriers, enabling high-throughput, reproducible phenotyping across diverse genetic backgrounds. Integrating diverse donors and sex-balanced cohorts into organoid biobanks improves external validity and supports development of population-appropriate dosing and risk stratification (e.g., avoiding biased algorithms such as European-trained warfarin dosing). The policy environment (DEPICT, FDORA, FDA Modernization Act 2.0) further incentivizes adoption of human-based models, while regulatory pathways (ISTAND) encourage qualification. Together, these elements position organoids as a bridge between genomic diversity insights and precision medicine, with implications for reducing ADRs, improving efficacy, and enhancing clinical trial success among underrepresented populations.

Organoids derived from diverse human donors provide a scalable, physiologically relevant framework to capture ancestry- and sex-linked biological differences in health and disease. By standardizing assays, adopting automation, leveraging organoid arrays, and deploying pooled multi-donor systems (chimeroids, PoP), researchers can interrogate interindividual variability at throughput. Coupling these models with inclusive biobanking, ethical governance, and integration of clinical-genomic data will enhance pharmacogenomic prediction and therapeutic equity. Future work should: (1) expand globally representative organoid/iPSC biobanks; (2) establish consensus protocols and model-omics resources; (3) systematically include sex- and ancestry-stratified analyses; (4) address technical challenges such as XCI stability and scalability; (5) align with regulatory qualification to facilitate translation; and (6) foster north–south collaborations and funding mechanisms to build regional capacity and ensure equitable access to diverse models.

- Narrative review without primary experimental data; conclusions rely on cited literature and illustrative case studies. - Scarcity of diverse donor lines in existing biobanks and over-representation of Northern European ancestry limits current generalizability. - Technical constraints include XCI instability in female hPSC lines, batch effects, and variability across differentiation protocols; scaling single-donor lines is labor-intensive. - Ethical and logistical challenges in sourcing and sharing diverse samples, including culturally sensitive consent and data governance. - Many pharmacogenomic associations and PRS remain less validated in underrepresented populations; broad continental labels (e.g., “African,” “Asian”) can obscure intra-population diversity. - Large sample sizes are needed to capture rare variants and detect population-level effects; pooled models require rigorous donor tracking and validation to avoid bias. - Translational gap persists between in vitro findings and clinical implementation; standardized qualification and regulatory acceptance are ongoing efforts.