Samiksha Babbar

Single-nucleotide level changes in our genome are small but mighty predictors for traits and disease, but they will never give us the full story about the plethora of genetic variation. Mapping larger changes to the genome structure and the cross-comparison of genomes can give insight into population-level differences that cause individual differences.

A lot of our understanding of human genetic variation throughout the past two decades can be attributed to the usage of genome-wide association studies (GWAS)^1,2. As a method for the cross-comparison of genomes, we can use GWAS to identify mutational changes of a single base pair, also known as a single nucleotide polymorphism (SNP). GWAS help identify population-level SNPs that may occur more frequently in individuals with diseased traits to predict genetic associations. However, genetic change can form a complex network that encompasses many types of mutations on a larger scale. In fact, over 93% of significant genomic regions from GWAS do not code for proteins and may be part of some other regulatory network³.

Going into the deeper layers of genetic variation, we can come up with some functional explanations of these SNPs. One way is through structural variants (SVs), which are large genetic variations within genomic loci, including deletions, duplications, insertions, and inversions.⁴ They are larger than SNPs and smaller than whole chromosome deletions or duplications⁵. SVs can be difficult to identify, largely due to complexity, size, and the method of sequencing used.⁶ Long-read sequencing, such as whole-genome sequencing (WGS), can capture large structural changes in DNA in a single stretch, rather than in short fragments. This makes it much easier to see the full extent of structural variants.⁷ In a Nature Genetics publication, Chirmade et al.¹ look at how to integrate data from WGS with GWAS to find SVs. This combined mapping improves results and helps identify SNPs that have no initial clarity for causing a specific trait.

The authors developed a web-based visualization tool called GWAS SVatalog¹ to visualize disease-trait associations and linkage disequilibrium (LD), also known as a determination of whether two variants of a gene were inherited together or independently. Using 110 participants of European descent who underwent WGS, they were able to develop an SV reference panel for individuals within a European population. The relationships between each significant SNP and SV were calculated through their corresponding statistic to measure LD, with their merged outcomes visualized on their interface. Over 35,000 SVs were visualized with over 100,000 SNPs significantly associated with the trait/disease from GWAS and 14,000 traits (Figure 1)¹.

Figure 1| Flow chart of the development of GWAS SVatalog: From 101 genomes of individuals who underwent long-read sequencing, they were able to create a reference panel of common structural variants. They integrated this with previously cataloged findings from GWAS and integrated both into a measure of linkage disequilibrium. This led to the development of GWAS SVatalog, providing a view by looking at region, gene, and phenotype. Created on BioRender.

In any population-based study, an important consideration is the generalizability of the variants that are identified. Chirmade and authors state that they used a previous patient dataset from their study of individuals with cystic fibrosis (CF)⁸. They determined that there was no significant difference from a healthy control population apart from the regions in a known genomic region associated with CF. Comparison with public SV datasets that determined the frequency of the less common alleles found that 85% of these matched with previous analyses¹. The SVs that were found were also compared with a well-documented public database for short-read sequencing, gnomAD⁹, and were determined to align with previous long-read versus short-read comparisons. GWAS SVatalog is therefore generalizable and encompasses the cohort it is studying well.

The authors identified candidate structural variants that illustrate how previously reported SNPs within GWAS in LD with SVs and gained insight on their functional impact. SVs that were in weaker LD with GWAS SNPs are likely to have different functions that are more related to the regulation of genes, such as promoting the transcription of RNA. A particular case they used was looking at a depression-related locus near the gene TMEM106B, where GWAS-significant SNPs show complete LD with a nearby SV within regions that influence gene regulation. Previous analyses determine this SNP to have an increased level of dementia¹⁰, and therefore functional follow up could confirm a potential characteristic of patients with this SV and/or SNP.

As we gain complexity in population-level studies, an issue rises where quantity becomes a larger focus than comprehension. Rather than making multiple novel tools, we must focus on creating a balance between data generation and comprehension. GWAS SVatalog can serve as a pilot project for documenting SVs, as it directly focuses on the visualization of documented LD with GWAS SNPs. It is important to build on rather than replicate similar tools, improving the accessibility and interpretability of the relationships that are documented.

The future of GWAS lies in embracing the full spectrum of human variation. Larger, more diverse cohorts and strong whole-genome sequencing resources will be essential for linking structural variants to disease. LD patterns differ substantially from the ancestry or sample that is being analysed, and therefore a tool that emphasizes this should look beyond a group of individuals with the same ancestry, trait, and population. Moreover, finding functional relevance is considered an “afterthought” of population-level studies such as GWAS. Examining LD associations after GWAS is a relevant genomic mapping tool, but a less reliable method to examine SVs. An important direction is to integrate SVs directly within analysis, rather than downstream of GWAS. This will provide unified associations of the reference panel and can potentially include mixed variant types for direct comparison.

Population-level analysis should always be an integrated approach that combines the possibility of all possible structural variants. A study of single-nucleotide changes will never be enough to find the complete picture of human variation. SVs can explain many of the disease associations we aim to find, opening discovery of novel loci and a better understanding of unexplained SNPs.

References

1.   Chirmade, S. et al. GWAS SVatalog: a visualization tool to aid fine-mapping of GWAS loci with structural variations. Heredity. (2025)

2.   Uffelmann, E. et al. Genome-wide association studies. Nat. Rev. Methods Primer 1, 59 (2021).

3.   Maurano, M. T. et al. Systematic Localization of Common Disease-Associated Variation in Regulatory DNA. Science 337, 1190–1195 (2012).

4.   Feuk, L., Carson, A. R. & Scherer, S. W. Structural variation in the human genome. Nat. Rev. Genet. 7, 85–97 (2006).

5.   Freeman, J. L. et al. Copy number variation: New insights in genome diversity. Genome Res. 16, 949–961 (2006).

6.   Mahmoud, M. et al. Structural variant calling: the long and the short of it. Genome Biol. 20, 246 (2019).

7.   Yang, L. A Practical Guide for Structural Variation Detection in the Human Genome. Curr. Protoc. Hum. Genet. 107, e103 (2020).

8.   Eckford, P. D. W. et al. The CF Canada-Sick Kids Program in individual CF therapy: A resource for the advancement of personalized medicine in CF. J. Cyst. Fibros. 18, 35–43 (2019).

9.   Chen, S. et al. A genomic mutational constraint map using variation in 76,156 human genomes. Nature 625, 92–100 (2024).

10. Lee, J. Y., Harney, D., Kwok, J., Larance, M. & Don, A. S. The major TMEM106B dementia risk allele affects TMEM106B protein levels and myelin lipid homeostasis in the ageing human hippocampus. Mol Neurodegeneration, 18, 63 (2023).

Katie Bui

A massively parallel reporter assay identifies genetic variants driven by pleiotropic effects across eight psychiatric disorders.

It is common for individuals to experience symptoms of multiple health conditions, especially when it comes to mental health disorders. While psychiatric disorders have long been suspected to have a genetic basis, genome-wide association studies (GWAS) have since identified numerous genomic loci associated with disease risk.¹ However, the mechanisms by which these genetic variants influence brain development and contribute to multiple disorders have remained unclear.

A GWAS study by the Psychiatric Genomics Consortium investigating eight psychiatric disorders: autism spectrum disorder, attention-deficit/hyperactivity disorder, schizophrenia, bipolar disorder, major depressive disorder, Tourette syndrome, obsessive-compulsive disorder, and anorexia nervosa, revealed widespread pleiotropy, meaning that single genetic variants can influence multiple traits.^1,2 109 of the 136 genomic loci tested were found to be linked to multiple disorders. ^1,2 Building onto this understanding, Lee and colleagues in Cell experimentally investigated how these variants functionally alter brain development to confer risk across multiple psychiatric conditions.³

            As most risk variants lie in non-coding regions, this poses a challenge in identifying which variants truly influence gene expression and which are statistically associated due to proximity on the chromosome (linkage disequilibrium).³ To tackle this, Lee and colleagues used a massively parallel reporter assay (MPRA) to functionally test 17,841 genetic variants in human neural progenitor cells, highlighting those that likely drive regulatory changes in gene expression (Figure 1).³ MPRA works by synthesizing short DNA fragments from GWAS loci containing either the risk or protective allele, attaching them to a promoter, a reporter gene, and a barcode.³ Gene expression levels are then measured to determine how strongly each fragment activates transcription and the effect of the expression of the risk allele relative to the protective. Using this high-throughput strategy enables parallel testing of the regulatory impact of thousands of variants compared to traditional approaches.³

Figure 1. Massively Parallel Reporter Assay (MPRA) enables high-throughput functional testing.³

Only 683 (~4%) variants of the 15,902 that passed quality control were determined as expression-modulating variants (emVars), or those with allelic effects on gene expression, across 103 loci. ³ This finding emphasizes the importance of functional testing as it narrows the hundreds of variants typically in linkage disequilibrium at each locus (many of which are likely not drivers of the diseases). The researchers also discovered 1,478 enhancer elements displaying significant regulatory activity. ³ Remarkably, approximately 9% of these active elements involved Alu repeats, the transposable sequences often overlooked due to their repetitive nature.³ Alu repeats displayed strong enhancer activity driving gene regulation and transcription factor binding, emphasizing their important role in psychiatric risk.

As GWAS index single nucleotide polymorphisms (SNP, variants with the strongest statistical association) are often assumed to be the most likely causal variants at a locus, the authors evaluated the number of index SNPs that showed functional activity. ³ They determined 3% of the 100 index SNPs tested showed regulatory function in MPRA, highlighting that index SNPs are not always causally responsible for disease associations and statistical association alone is insufficient to conclude causality. ³ This finding underscores the need for functional validation in future GWAS follow-up studies.

            Lee and colleagues then compared disorder-specific variants (associated with one disorder) to pleiotropic (associated with more than three disorders) (Figure 2).³ Pleiotropic variants tended to be active across various regions of the developing brain impacting multiple neuronal subtypes including excitatory and inhibitory neurons.³ The study revealed that pleiotropic variants often target hub genes, which are genes with high connectivity to other genes in co-expression and protein-protein interaction (PPI) networks.³ This suggests that pleiotropy may arise from disruption of the regulatory nodes in these hub genes, leading to the disruption of numerous brain functions and transcription factors interacting with proteins. The resulting effects therefore caused multiple overlapping psychiatric symptoms across disorders. In contrast, disorder-specific variants tended to exert effects within more restricted cell types.³

Figure 2. Pleiotropic variants target highly connected hub genes, while disorder-specific variants target restricted cell types.³

The researchers’ findings were validated using CRISPR, targeting emVars within the RERE (genetic signaling) and DCC (axon guidance) loci.³ Disruption of variants in RERE and DCC loci selectively reduced expression of predicted target genes without affecting nearby genes, confirming the accuracy of MPRA in determining functional regulatory variants. In vivo CRISPR perturbation revealed that a pleiotropic gene (Anp32e) mainly regulated highly connected neuronal genes, whereas a disorder-specific gene (Kmt5a) mainly affected a smaller set of glial genes.³ This difference reflects what clinicians observe: widely connected genes tend to drive overlapping symptoms, while genes with more restricted expression tend to contribute to disorder-specific risk. Together with MPRA, statistical fine-mapping statistics such as Sum of Single Effects (SuSiE), FINEMAP, and Causal Variants Identification in Associated Regions (CAVIAR) were conducted, to estimate the importance and likelihood of certain variants.^6,7,8 However, these tools could not reliably differentiate emVars from non-functional variants in complex regions, especially in loci with extensive linkage disequilibrium, and various variants prioritized by the algorithms lack detectable regulatory activity in MPRA.⁹ These findings highlight different aspects of causality when comparing statistical and functional approaches, and underscore the need to integrate experimental evidence with statistical models when interpreting variants in psychiatry.

Overall, Lee and colleagues revealed that shared genetic risk across psychiatric disorders stems from the disruption of highly connected regulatory networks across neuronal lineages and developmental stages. Their findings are crucial to the understanding of psychiatric diseases and developing therapeutic treatments. Future research should expand on functional studies across developmental stages and cell types. Advances such as massively parallel CRISPR-based screening at endogenous loci may provide opportunities to directly test regulatory function and overcome the constraints of the current episomal assays.

References

1. Sullivan, P. F. et al. Psychiatric genomics: An update and an agenda. Am J Psychiatry 175, 15–27 (2018).
2. Cross-Disorder Group of the Psychiatric Genomics Consortium. Genomic Relationships, Novel Loci, and Pleiotropic Mechanisms across Eight Psychiatric Disorders. Cell 179, 1469–1482.e11 (2019).
3. Lee, S. et al. Massively parallel reporter assay investigates shared genetic variants of eight psychiatric disorders. Cell 188, 1409-1424 (2025).
4. Sey, N.Y.A., et al. (2020). A computational tool (H-MAGMA) for improved prediction of brain-disorder risk genes by incorporating brain chromatin interaction profiles. Nat. Neurosci 23, 583–593.
5. Ziffra, R.S. et al. Singlecell epigenomics reveals mechanisms of human cortical development. Nature 598, 205–213 (2021).
6. Fulco, C.P. et al. Activity-by-contact model of enhancer-promoter regulation from thousands of CRISPR perturbations. Nat Genet 51, 1664–1669 (2019).
7. Benner, C.,  et al. (2016). FINEMAP: efficient variable selection using summary data from genome-wide association studies. Bioinformatics 32, 1493–1501 (2016).
8. Zou, Y., et al. Fine-mapping from summary data with the ‘‘Sum of Single Effects’’ model. PLoS Genet 18, e1010299 (2022).
9. Hormozdiari, F., et al. Identifying causal variants at loci with multiple signals of association. Genetics 198, 497–508 (2014).

Raina Cui

Gestational diabetes mellitus reprograms the oocyte epigenome, revealing how maternal metabolism can shape disease risk across generations.

Gestational diabetes mellitus (GDM) is associated with long-term metabolic consequences in offspring, including increased risks of obesity, insulin resistance and cardiovascular disease later in life^1-3. In mouse models, GDM impairs glucose and insulin tolerance and increases body weight in female offspring, with these effects transmitted to a second generation through the maternal lineage. Writing in Nature Communications, Guo and colleagues⁴ trace these phenotypic effects back to the germline, showing that GDM reprograms the epigenetic landscape of oocytes. Their findings suggest that maternal metabolism can shape offspring health before fertilization occurs, reframing intergenerational metabolic disease as a problem rooted in the oocyte itself.

Previous research has shown that GDM influences the epigenome of fetal tissues, including blood and placental cells⁵. In contrast, DNA methylation in oocytes is normally established and maintained with tight regulation to support proper development after fertilization⁶. Hence, many studies of GDM have focused on epigenetic changes that occur during pregnancy or early embryonic development, when the embryo is directly exposed to the maternal environment. Whether maternal metabolic disturbances can directly alter the epigenetic state of oocytes themselves has remained unclear. This unresolved question motivates the work by Guo and colleagues, who examine how GDM affects DNA methylation in oocytes prior to fertilization.

Guo et al. addressed this by employing a mouse model mimicking GDM through induced maternal hyperglycemia (Fig. 1a). Oocytes from first-generation (F1) female offspring of GDM-exposed mothers exhibited widespread hypermethylation, reflecting a global increase in gene-silencing marks that may disrupt normal gene regulation and compromise developmental potential. Consistent with this, immunofluorescence revealed elevated levels of 5-methylcytosine (5mC), and whole-genome bisulfite sequencing confirmed a genome-wide rise in DNA methylation, particularly at CG sites. To distinguish prenatal from postnatal effects, the authors performed cross-fostering experiments by raising F1 pups with metabolically normal surrogate mothers. This intervention failed to rescue the hypermethylation phenotype, supporting a direct effect of gestational diabetes on oocyte epigenetic programming.

Furthermore, they observe a subset of these methylation patterns transmitted to second-generation (F2) oocytes, which exhibit elevated methylation levels especially in gene promoters associated with key metabolic pathways, including insulin signaling and type 2 diabetes. Although global methylation patterns were partially reset in the second generation, the selective maintenance of altered regions suggests that GDM can imprint stable epigenetic marks rather than inducing transient noise. These epigenetic changes correlated with physiological consequences: female offspring displayed impaired glucose tolerance, insulin resistance and reduced oocyte maturation rates, alongside decreased litter sizes and lower blastocyst formation rates.

Having established that GDM reshapes the oocyte methylome, Guo et al. sought to identify the underlying regulatory mechanism. They found increased expression of Ezh2, a key component of the Polycomb repressive complex 2 (PRC2), along with higher levels of H3K27me3, a histone mark that silences genes by compacting chromatin. Ezh2 expression rose progressively during oocyte maturation and was particularly enriched in growing follicles, highlighting its role in development. This finding was unexpected since EZH2 primarily modifies histones rather than DNA directly. This suggested that altered chromatin regulation may link maternal metabolic state to widespread DNA methylation alterations.

To test whether EZH2 drives changes in the oocyte methylome, the authors manipulated its activity. Reducing EZH2 lowered DNA methylation, while increasing it had the opposite effect. EZH2 appeared to guide DNA methyltransferases (DNMTs) to chromatin, particularly influencing DNMT1. Although GDM did not substantially alter DNMT1 expression, it promoted DNMT1 accumulation in the nucleus, an effect reversed by Ezh2 knockdown. These results identify EZH2 as a key mediator linking altered chromatin state to aberrant DNA methylation through DNMT1-dependent methylation.

Together, this work supports a model in which maternal GDM reprograms the oocyte epigenome through EZH2-dependent recruitment of DNMT1, establishing aberrant DNA methylation patterns before fertilization that can persist across generations (Fig. 1b). Conceptually, these findings push developmental programming further upstream of previous observations, positioning the oocyte as an early site where metabolic disease risk may be encoded. This aligns with emerging evidence from both mouse and human studies suggesting that maternal metabolic states can leave lasting epigenetic imprints in gametes and early embryos, influencing offspring health. For example, recent work on maternal obesity and high-fat diet models has shown alterations in sperm^7-8 and oocyte⁹ DNA methylation and histone modifications, supporting the idea that germline epigenetic plasticity may be broadly responsive to metabolic stress.

At the same time, important questions remain. The extent to which these mechanisms operate in humans is unclear, as oocyte studies in women are inherently limited. The reversibility of these epigenetic changes is also unknown, as is whether interventions targeting maternal metabolism or EZH2 activity could prevent or reverse germline reprogramming. Moreover, the specific gene networks affected and their contribution to metabolic disease susceptibility in offspring remain to be defined. There may also be additional chromatin or noncoding RNA–mediated pathways acting in parallel, which were not explored in this study.

Guo et al. identify EZH2 as a potential point of intervention for limiting the epigenetic consequences of GDM, raising the possibility that targeting chromatin regulators during sensitive windows of germ cell development could reduce transgenerational transmission of metabolic disease. By uncovering a molecular link between maternal hyperglycemia and germline epigenetic regulation, this work underscores the importance of maternal metabolic health prior to and during pregnancy while also highlighting the challenges in translating these mechanistic insights to clinical practice.

Figure 1| The Epigenetic Effect of GDM in Oocytes. (A) Guo et al.⁴ schematic of mice breeding to obtain and analyze F1 and F2 generations. STZ = Streptozotocin; GDF1 = gestational diabetes F1; NGDF1 = non-gestational diabetes F1; GDF2 = gestational diabetes F2; NGDF2 = non-gestational diabetes F2. Image adapted from ⁴ using BioRender. (B) Proposed mechanism of action involving EZH2 recruiting DNMT, leading to cytosine hypermethylation. The Polycomb Repressive Complex 2 (PRC2) is responsible for adding the repressive H3K27me3 mark and includes: EZH2, SUZ12, and EED. EZH2 recruits DNMTs to chromatin, so that DNA methylation is established in regions marked by H3K27me3. Image adapted from ¹⁰ using BioRender.

References

1. Kong, L., Nilsson, I. A. K., Gissler, M. & Lavebratt, C. Associations of maternal diabetes and body mass index with offspring birth weight and prematurity. JAMA Pediatr. 173, 371–378 (2019).
   
2. Ornoy, A., Ratzon, N., Greenbaum, C., Wolf, A. & Dulitzky, M. School-age children born to diabetic mothers and to mothers with gestational diabetes exhibit a high rate of inattention and fine and gross motor impairment. J. Pediatr. Endocrinol. Metab. 14, 681–689 (2001).
   
3. De Sousa, R. A. L. Animal models of gestational diabetes: characteristics and consequences to the brain and behavior of the offspring. Metab. Brain Dis. 36, 199–204 (2021).
   
4. Guo, X. et al. Gestational diabetes mellitus causes genome hyper-methylation of oocyte via increased EZH2. Nat. Commun. 16, 55499 (2025).
   
5. Dluski, D. F., Wolińska, E. & Skrzypczak, M. Epigenetic changes in gestational diabetes mellitus. Int. J. Mol. Sci. 22, 7644 (2021).
   
6. Stewart, K. R., Veselovska, L. & Kelsey, G. Establishment and functions of DNA methylation in the germline. Epigenomics 8, 1399–1413 (2016).
   
7. Guo, T. et al. Paternally multi-generational high-fat diet causes obesity and metabolic disorder through intergenerational DNA methylation. Front. Nutr. 12, (2025).

8. Keyhan, S. et al. Male obesity impacts DNA methylation reprogramming in sperm. Clin Epigenet 13, (2021).

9. Chao, S. et al. Maternal obesity may disrupt offspring metabolism by inducing oocyte genome hyper-methylation via increased dnmts. eLife 13, (2024).
10. Tan, J. Z., Yan, Y., Wang, X. X., Jiang, Y. & Xu, H.E. EZH2: biology, disease, and structure-based drug discovery. Acta Pharmacol. Sin. 35, 161–174 (2014).

Samy Danial

Recent advancement in sequencing technology provides insight into how the tumor microenvironment and cell type interactions lead to resistance to CTLA-4 and PD-L1 immunotherapy modulators.

Despite advances in cancer management and treatment, one of the key barriers to effective cancer treatment is the development of treatment resistance to a wide range of therapeutic agents⁷. Resistance is broadly classified as intrinsic or extrinsic¹. Intrinsic pathways refer to genetic factors within the tumor that contribute to resistance, while extrinsic pathways refer to the role of non-tumor cells in mediating resistance.

Single cell sequencing technologies have been used to study cancer cell resistance pathways in immunotherapy. These studies provided useful information on the genetic profile of individual tumor cells and the identification of tumor subclones, which are examples of intrinsic factors. In addition, by characterizing non-tumor cell types within the tumor microenvironment, single cell methods can provide information on extrinsic resistance mechanisms. However, because single cell sequencing methods require tissue dissociation, it is not possible to characterize the spatial context of the tumor microenvironment, which is another key extrinsic factor in predicting therapy response². To address this limitation, other modalities, such as High Plex Imaging platforms, study the spatial interactions between tumors and neighboring cells along with pre-selected biomarkers^3-5. However, due to the limited selection of biomarkers, this approach is limited in its ability to provide a comprehensive spatial and molecular integrative approach to study mechanisms of therapeutic resistance.

A groundbreaking study led by Quek et al uses a newly developed Multi-Model Tool Integration Kit (MIT) to study response heterogeneity and immunotherapy resistance pathways in metastatic melanoma patients⁶. MIT works by aligning different single cell sequencing modalities with spatial datasets characterized by Pheno Cycler into an integrated shared dimensional space (Figure 1). The integrated dataset can be used to interpolate molecular subtypes of each cell type characterized by Pheno Cycler in a spatially resolved manner. Due to the larger selection of biomarkers along with integrated spatial information, MIT offers a more comprehensive approach compared to High Plex Imaging Platforms.

Figure 1: Schematic workflow of the MIT workflow. Each patient tumor specimen is divided into two samples: a tumor dissociate that will undergo single cell sequencing studies and a Formalin Fixed Paraffin Embedded (FFPE) tumor block that will be used by PhenoCycler to capture spatial information. The result from the two workflows generates data matrices with information on gene and protein expression patterns and cell location. Adapted from⁶

To uncover mechanisms of immunotherapy resistance, the authors applied this approach to five metastatic melanoma patients treated with immunotherapy, including one case of acquired resistance, two cases of innate resistance, and two responders.

A key finding with this new approach was the difference in distribution of immune cells in responsive and resistant cases. While single cell sequencing showed identical immune composition data between both cases, the new approach demonstrated immune cell distribution differences between responsive and resistant cases. For example, spatial data of responsive patients show immune cells in close proximity to tumor cells, while innately resistant patients have immune cells confined to the stromal regions and not infiltrating the tumor core. In addition, using single cell data, the authors show that the cellular characteristics of immune cells differ between responsive and resistant cases. For example, THY1+ EGFR+ Cancer Associated Fibroblasts (CAFs) and CD14+ macrophages are enriched in resistant tumors compared with responsive tumors, while CD8+ memory cells are enriched in responsive tumors. The enrichment of CAFs suggests a physical mechanism of resistance, where T cells are blocked from entering the tumor site⁶.

The authors also identified key changes in immune and tumor cell molecular composition during tumor progression. For example, CAFs shift towards a pro-tumoral cytokine profile, with reduced lymphocyte recruiting cytokines and increased expression of pro-tumoral cytokines. In patients with acquired resistance, genes involved in T cell memory state were downregulated. Patients with innate resistance had low expression of PD-L1 checkpoint receptors commonly targeted during immunotherapy. These findings suggest new biomarkers that could be screened to predict immunotherapy response.

Finally, the authors demonstrate changes in tumor subclone composition during immunotherapy. For example, the authors revealed that new melanoma subclones dominated following treatment with anti PD-L1 and anti-CTLA-4 immunotherapy agents. These subclones had impairments in antigen processing and presentation pathways, which is crucial for immune cell recognition. This provides key insight into the role of tumor clonal heterogeneity in mediating acquired resistance.

While promising, the superiority of this new approach in uncovering mechanisms of resistance can be validated through larger patient studies. As the authors noted, only five patients were studied with the new method. Larger patient cohorts will also prove useful for machine learning applications, where a model can be trained on large datasets to predict patient response. In addition, studies on other cancer types will provide useful information on overarching and differing mechanisms of resistance to immunotherapy. This will allow researchers to determine whether immunotherapy treatment options can be generalized across different cancer types.

Overall, this groundbreaking study provides insight into the future of drug development for combating immunotherapy resistance. For example, the upregulated CAF expression of key pro tumoral cytokine factors can be exploited to develop novel antagonistic agents targeting these cytokines. In addition, by integrating immune cell spatial information, clinicians can more accurately determine whether immunotherapy is a suitable initial line of option for therapy. For example, patients with weak immune cell infiltration and/or increased CAFs are less likely to be suitable candidates for immunotherapy. This is supported by the innate resistant case where the patient studied had low immune cell infiltration at baseline.

References

1.) Sharma, P. et al. Primary, adaptive and acquired resistance to cancer immunotherapy. Cell 168, 707–723 (2017).

2.) Quek, C. et al. High-dimensional single-cell transcriptomics in melanoma and cancer immunotherapy. Genes 12, 1629 (2021).

3.) Phillips, D. et al. Immune cell topography predicts response to PD‑1 blockade in cutaneous T‑cell lymphoma. Nat. Commun. 12, 6726 (2021).

4.) Gouin, K. H. et al. An N‑cadherin‑2–expressing epithelial cell subpopulation predicts response to surgery, chemotherapy and immunotherapy in bladder cancer. Nat. Commun. 12, 4906 (2021).

5.) Schürch, C. M. et al. Coordinated cellular neighborhoods orchestrate antitumoral immunity at the colorectal cancer invasive front. Cell 182, 1341–1359 (2020).

6.) Quek, C. et al. Single-cell spatial multiomics reveals tumor microenvironment vulnerabilities in cancer resistance to immunotherapy. Cell Reports 43, 114392 (2024).

7.) Kaye, S.B. Mechanisms of drug resistance in cancer chemotherapy. Med. Pediatr. Oncol. 14, 35–38 (2008).

Alina Elahie

A population-scale map of structural variants in Qatar reveals medically actionable genetic effects that standard approaches miss, highlighting the growing importance of inclusive genomics for precision medicine.

Precision medicine promises to tailor prevention and treatment strategies to individual genetic profiles, yet much of this vision relies on data from populations of European ancestry and on genetic variants that are relatively easy to detect¹. A study by Aliyev et al. provides a comprehensive analysis of structural variants (SVs) in over 6,000 individuals from the Qatari population, leveraging Qatar’s advanced genomics infrastructure and high consanguinity rates to reveal genetic diversity with direct implications for disease, diagnostics, and therapy¹. The authors link SVs to medically relevant traits such as kidney function, body composition, and extreme obesity, highlighting the clinical importance of investigating underrepresented populations¹. Using whole-genome sequencing (WGS) and detailed phenotyping, they show that SVs contribute substantially to these traits, often independently of single-nucleotide variants (SNVs)¹.

By organizing large DNA rearrangements across thousands of genomes, the study maps a rich landscape of deletions, duplications, and other SVs¹. Many of these variants are underrepresented in global reference datasets yet common in Qatar, frequently affecting genes implicated in Mendelian disease and complex traits¹.This work elevates SVs from an afterthought to a central player in precision medicine for Arab and neighbouring populationsand fills a geographic and demographic gap in human genomics research^1,2.

The Qatari cohort brings this biological potential of SVs into focus. Unlike SNVs, many SVs involve the loss or gain of larger segments of DNA, leading to substantial changes in the number of functional copies of a gene, known as gene dosage (Fig. 1)³. Deletions spanning coding regions can eliminate protein production, duplications can increase gene dosage, and more complex rearrangements can disrupt regulatory elements^3,6,8. Large studies of genetically diverse populations suggest that each person carries a small number of rare SVs that alter the dosage or structure of several genes^3,4.  In populations with high consanguinity, long stretches of DNA inherited from a shared ancestor, known as autozygosity, increase the likelihood that rare variants occur in homozygous form, occasionally causing a gene to become completely inactivated in otherwise healthy individuals^3,4,5,8.

To appreciate the magnitude of this hidden variation, it is useful to consider the broader spectrum of human genetic diversity (Fig. 1)³. While precision medicine has largely focused on SNVs (<50 bp), SVs (≥50 bp) encompass both balanced variants, such as inversions, and unbalanced variants, including deletions and duplications³. Although numerically less frequent than SNVs, the larger size of SVs often results in greater functional impact, from eliminating protein production to altering gene regulation^1,3.

Figure 1: The spectrum of human genetic variation. Genomic diversity spans from single-nucleotide variants (SNVs) to large-scale chromosomal rearrangements³. While precision medicine has historically focused on SNVs (<50 bp), Aliyev et al. highlight the clinical importance of SV’s (≥50 bp)³. These include balanced variants (no net DNA change, such as inversions) and unbalanced variants (dosage-altering deletions and duplications)³. Although SVs are numerically less frequent than SNVs, their size often leads to greater functional impact, including gene knockouts and regulatory disruptions observed in the Qatari population^1,3. Advances in WGS enable detection of complex SV classes that are often missed by microarrays or exome sequencing^1,3,6. Figure adapted from Collins, R. L. & Talkowski, M. E.³.

A key strength of this study is its population context. High rates of consanguinity in Qatar increase the likelihood that rare variants, including SVs, appear in homozygous form^3,8. The authors identify over 180 genes disrupted by homozygous SVs and demonstrate that these disruptions have measurable biological consequences¹. By measuring protein levels, they confirm that gene knockouts reduce or eliminate protein production, linking genotype to molecular phenotype¹. This connection moves beyond cataloguing genetic variation and towards understanding its functional impact.

Importantly, the study also highlights the importance of SVs outside of traditional protein-coding regions, which are often overlooked in clinical genetics¹. By examining individuals with extreme trait values, the authors identified homozygous deletions with large effects on kidney function, body composition, and obesity¹. For example, a deletion in a regulatory region on chromosome 19 is linked to obesity, showing how changes in non-coding DNA can influence key biological processes. Focusing only on protein-coding genes risks missing important disease-causing variants¹.

From a clinical perspective, 3.2% of individuals carry variants in medically actionable genes, and nearly one-third of these would have been missed if only SNVs were considered¹. These findings underscore the need to incorporate SVs into clinical genomics pipelines. For instance, by updating screening arrays or developing sequencing approaches that capture larger DNA rearrangements, enabling more comprehensive detection of disease-relevant variants. Currently, Qatar uses a population-specific screening array (QChip1) that primarily detects small variants and misses many structural alterations⁷. The SV catalogue generated in this study helps identify which missing variants are most relevant locally, informing the design of next-generation screening tools^1,7.

Technology limitations also remain important. Short-read sequencing, even at high coverage, misses many complex SVs and performs poorly in repetitive regions⁹. In contrast, long-read assemblies and regional pangenomes are already revealing tens of megabases of sequence absent from standard references, along with additional SVs of clinical relevance^6,8,9. Looking ahead, long-read sequencing promises improved SV detection, refined breakpoints, and discovery of complex rearrangements⁶.

By examining DNA changes alongside gene activity and cellular behavior, the study provides a clearer picture of how SVs influence human health and biology. This study illuminates a previously hidden layer of human genetic diversity, demonstrating that SVs, particularly in non-coding regions can exert profound effects on health and disease¹. By focusing on a population historically underrepresented in genomics, this work advances precision medicine for the Arabian Peninsula while providing insights relevant to global health².

References

1. Aliyev, E. et al. The biomedical landscape of genomic structural variation in the Qatari population. Nat. Commun. 17, 1–15 (2026).

2. Cole, A. J. et al. The landscape of genomic structural variation in Indigenous Australians. Nature 624, 610–617 (2023).

3. Collins, R. L. & Talkowski, M. E. Diversity and consequences of structural variation in the human genome. Nat. Rev. Genet. 26, 443–462 (2025).

4. Daw Elbait, G. et al. A population-specific major allele reference genome from the United Arab Emirates population. Front. Genet. 12, 660428 (2021)

5. Mezzavilla, M. et al. Ancestry-related distribution of runs of homozygosity and functional variants in Qatari population. BMC Genom. Data 23, 73 (2022).

6. Ramaswamy, S. et al. Middle Eastern genetic variation improves clinical annotation of the human genome. J. Pers. Med. 12, 423 (2022).

7. Rodriguez-Flores, J. L. et al. The QChip1 knowledgebase and microarray for precision medicine in Qatar. npj Genom. Med. 7, 3 (2022).

8. Saleheen, D. et al. Human knockouts and phenotypic analysis in a cohort with a high rate of consanguinity. Nature 544, 331–336 (2017).

9. Scott, A. J., Chiang, C. & Hall, I. M. Structural variants are a major source of gene expression differences in humans and often affect multiple nearby genes. Genome Res. 31, 2249–2257 (2021).

Melissa Elgie

Cutting-edge analysis of high-grade gliomas (HGG) classifies these aggressive brain or central nervous system cancers into three groups based on their molecular and clinical characteristics, providing insight on effective treatment regimens for patients¹.

Imagine being tasked with organizing a library of books into groups based on their common genres, target audiences, and number of pages. This task would leave many people wondering where to begin. Such a task is low stakes if it’s just about being orderly, but if it could change the course of someone’s life, things get a lot more serious. Add into the mix that the books being organized are extremely heterogenous, and you get the dilemma facing clinicians in the treatment of high-grade gliomas (HGG), a motley group of cancers that are among the most dangerous to human health². Fernandez et al. tackle this challenge in their recent study in which they categorized HGG into three subgroups¹. Their characterization of this complex type of cancer may have direct implications for how such tumours are treated in the future, streamlining clinical interventions and improving the patient experience¹.

Fernandez et al. sought to determine whether the molecular and clinical differences in HGG could be explained by genetic mutations in the body’s DNA repair and copying systems, which are responsible for cutting out mismatched DNA sequences and replacing them with the correct DNA and replicating DNA, respectively¹. To that end, they studied 162 HGG samples collected from 152 patients from the International Replication Repair Deficiency Consortium database, which they analyzed in terms of their genetic characteristics and clinical presentation¹. The researchers hypothesized that in general, primary mismatch-repair-deficient HGG (priMMRD-HGG, a specific type of glioma characterized by a particular genetic profile) could be stratified into one of three groups based on the mutations they harbour: priMMRD-1 (deficiencies in DNA repair and in the enzyme responsible for copying DNA), priMMRD-2 (just DNA repair deficient), or priMMRD-3 (DNA repair deficiency and mutation in a metabolic enzyme) (Figure 1)^1,3.

Figure 1: Summary of characteristics of glioma subgroups as defined by Fernandez et al.¹. The characteristics of each glioma subgroup (priMMRD-1, priMMRD-2, and priMMRD-3) are presented, including their molecular and genetic profiles, typical age of onset, and response to immunotherapy¹. Figure created in BioRender.com.

Critically, the group found that priMMRD-1 HGG samples were characterized by inactivation of DNA repair genes due to the presence of two mutated copies of DNA repair genes compared to priMMRD-2 and priMMRD-3, which harboured only one mutated copy of DNA repair genes¹. Given that priMMRD-1 tumours were seen to arise at a significantly younger age compared to the other two categories of HGG, the researchers proposed that this could be a determining factor in age of onset¹. A next course of action could be to determine external factors beyond the number functional genes present that could impact age of onset, such as potential immune system deficiencies¹.

The study also elucidates differences in immune system activation across the three HGG groups, with priMMRD-1 and priMMRD-2 exhibiting increased expression of 108 key immunity genes compared to priMMRD-3¹. Given that gliomas are known to suppress the human immune system, the authors take a logical approach in inquiring about the immune profiles of these cancers to possibly determine if immune system-enhancing interventions could combat this effect^1,4,5. To promote the immune system’s anti-tumour activity, researchers could leverage epigenetic changes to DNA, a process that naturally occurs to chemically modify DNA without changing its sequence⁵. Epigenetics-based therapeutics could be combined with immune checkpoint inhibitors (ICI, an immunotherapy) to oppose the immunosuppressive abilities of gliomas^1,5.

In analyzing treatment responses in each HGG group, Fernandez and colleagues investigated samples derived from patients who had been treated with ICI and found that priMMRD-1 patients exhibited better survival compared to priMMRD-2 and priMMRD-3¹. The authors point out in their paper that priMMRD-1 gliomas respond well to immunotherapy, and this presents a more favourable treatment option for children whom clinicians want to spare from harsh radiation therapy (Figure 1)¹. This contrasts with the priMMRD-2 group, which has a variable response to immunotherapy, to which the authors suggest a combination treatment regimen that could include immunotherapies¹. The final group, priMMRD-3 gliomas, does not respond well to immunotherapies, so the group suggests that targeted inhibitors could be used alongside immunotherapy to increase survival for those diagnosed with this type of glioma¹.

While the authors’ work appears to promote research into immunotherapy advances, there are limitations to implementing these treatments for gliomas, which could explain the less favourable response of priMMRD-2 and priMMRD-3 to immunotherapy^1,6. Such obstacles include difficulties with therapies navigating to the brain and the immunosuppressive tumour environment⁶. Despite these challenges, Fernandez et al. do not count immunotherapy out, and this may be for good reason^1,7. An up-and-coming treatment known as oncolytic virus therapy uses viruses to selectively infect cancer cells⁷. This approach to glioma treatment can stimulate the immune system through the release of cellular material from the tumour after the virus kills the cancer cells⁷. These cellular components, such as proteins, can be picked up by antigen-presenting cells, which signal to immune cells to destroy the cancer⁷. Indeed, options such as oncolytic virus therapy clearly illustrate the potential that lies within the immune system to mount lethal, targeted attacks on glioma cells⁷.

This body of work by Fernandez et al. classified high-grade gliomas into three subgroups based on their molecular and clinical characteristics to provide insight on optimal treatment practices for each group¹. Fernandez et al. propose that priMMRD-HGG should be considered for WHO classifications of central nervous system tumors in the future¹. Based on their work, the authors appear to suggest a goal to work toward: investigating how patients with the three different subgroups of HGG respond to different therapy regimens in clinical trials¹. This could take the form of uncovering novel immunotherapeutic avenues^6,7 . The insights from this study equip researchers with new knowledge on how to leverage glioma’s unique molecular intricacies to tailor medicine to specific patients¹.

References

1.           Fernandez, N. R. et al. Patterns of hypermutation shape tumorigenesis and immunotherapy response in mismatch-repair-deficient glioma. Nature Genetics 2025 58:1 58, 132–142 (2025).

2.           Higginbottom, S. L., Tomaskovic-Crook, E. & Crook, J. M. Considerations for modelling diffuse high-grade gliomas and developing clinically relevant therapies. Cancer Metastasis Rev. 42, 507 (2023).

3.           Pirozzi, C. J. & Yan, H. The implications of IDH mutations for cancer development and therapy. Nature Reviews Clinical Oncology 2021 18:10 18, 645–661 (2021).

4.           Xu, S., Tang, L., Li, X., Fan, F. & Liu, Z. Immunotherapy for glioma: Current management and future application. Cancer Lett. 476, 1–12 (2020).

5.           Riyas Mohamed, F. R. & Yaqinuddin, A. Epigenetic reprogramming and antitumor immune responses in gliomas: a systematic review. Medical Oncology 2025 42:6 42, 213- (2025).

6.           Sadowski, K. et al. Revolutionizing Glioblastoma Treatment: A Comprehensive Overview of Modern Therapeutic Approaches. Int. J. Mol. Sci. 25, 5774 (2024).

7.           Zhang, X. et al. Oncolytic virus therapy for glioma: current clinical trials and overcoming key obstacles. Int. Immunopharmacol. 166, 115547 (2025).

Hannah Fraser

A large-scale genome wide association study (GWAS) connects psychiatric disease risk with drug targets, guiding genomics informed drug development.

For many patients with psychiatric disorders, identifying an effective medication often involves prolonged trial-and-error prescribing¹. Could human genetics help break this cycle? A recent study by Hatoum et al.² suggests this may be possible by revealing overlap between genetic risk loci and the molecular targets of existing psychiatric medications². 

Psychiatric disorders place a major burden on healthcare systems worldwide, yet drug development for these conditions lags behind many other therapeutic areas³. Current treatments for disorders such as schizophrenia and depression primarily target dopamine and serotonin pathways that were identified decades ago. Many widely used psychiatric drugs were not originally designed to treat mental illness¹. For example, early antidepressants were first identified during tuberculosis drug trials and were developed without a clear understanding of the underlying disease biology⁴. As a result, psychiatric treatment remains largely nonspecific, forcing patients to cycle through multiple medications before finding one that is effective. Even then, therapies may provide limited benefit or cause significant side effects¹.

These limitations in current treatments have led researchers to look for new, biologically backed methods of drug development. In recent years, genome-wide association studies (GWAS) have identified hundreds of genetic variants linked to psychiatric disease risk². GWAS survey the genomes of large populations to detect variants that occur more frequently in individuals with a disease than in those without it, providing new insight into the biological processes underlying these conditions⁵. However, translating these statistical associations into therapeutic applications remains challenging, as many GWAS signals do not represent true causal variants, and predictive value may vary across populations⁶. To bridge this gap, researchers must determine whether GWAS risk loci map onto molecular pathways targeted by existing therapies.

Hatoum et al. provide a proof-of-principle that GWAS can be used to evaluate psychiatric drug targets². They directly compared genetic risk signals with known drug-target relationships. The study used a drug-set enrichment strategy, which tests whether medications whose targets overlap GWAS risk loci are overrepresented among approved psychiatric drugs. Unlike the traditionally used post GWAS gene-set enrichment analysis, this approach evaluates overlap at the level of drugs rather than individual genes (Figure 1). A key advantage of drug-set enrichment analysis is that it directly evaluates whether genetically implicated pathways correspond to existing medications, making the results more clinically meaningful.

Figure 1. Workflow of the GWAS-based drug enrichment analysis used by Hatoum et al.² Psychiatric GWAS summary statistics were used to identify significant single nucleotide polymorphisms (SNPs). SNPs were then mapped to nearby genes using proximity-based methods. Genes that were identified were linked to drug targets using the DGIdb and Connectivity Map databases. Drug-set enrichment analysis was then performed to test whether medications targeting GWAS associated genes were overrepresented among psychiatric drugs. Results were outputted as odds ratios (OR), representing the strength of enrichment among psychiatric drug classes.

Using this approach, the authors saw significant enrichment for medications used to treat schizophrenia, bipolar disorder, major depressive disorder, and substance use disorders, but not for ADHD, PTSD, generalized anxiety disorder, or insomnia². This suggests that many drugs already used in psychiatry act on biological pathways implicated by genomic studies. Schizophrenia displayed the strongest enrichment, with an odds ratio (OR) exceeding 27, where an OR greater than 1 indicates enrichment. This means that drugs whose targets overlapped with schizophrenia risk loci were far more common among approved treatments than would be expected by chance².

This enrichment extended beyond dopamine-related targets that historically dominated schizophrenia drug development. Although dopamine-associated genes accounted for a substantial portion of the enrichment, removing these genes did not eliminate the overall signal. This suggests that molecular schizophrenia risk is not driven solely by dopamine signalling, but that additional biological pathways are implicated. This may expand the range of potential therapeutic targets and support the development of treatments that move beyond traditional dopamine-based approaches. Similarly, the study found overlap between GWAS-associated risk genes and drug target pathways, including glutamate signaling in schizophrenia, calcium channel genes in bipolar disorder, and opioid and nicotinic receptor pathways in substance use disorders². Together, these findings suggest that GWAS can support current drug targets while also pointing to new ones.

This work has important implications for translating genetic discoveries into clinical practice. One promising application of GWAS is drug repurposing, in which existing psychiatric medications developed for one condition are identified as potential treatments for another⁷. Because repurposed drugs have already undergone clinical testing, they can reach patients faster and at lower cost⁷. GWAS can be used to identify drugs whose molecular targets overlap genetically supported disease pathways, helping prioritize promising candidates for repurposing. Drug development in psychiatry is expensive and often unsuccessful¹. Incorporating genetic data into the early stages of research and development could help narrow down which targets are worth pursuing and lower the financial barrier of developing new treatments.

Although using GWAS to prioritize and evaluate psychiatric drug targets is promising, several limitations remain. Psychiatric diagnoses often include patients with differing underlying biology, making drug target selection more difficult². Most GWAS datasets also rely heavily on individuals of European ancestry, which limits how broadly these findings can be applied. Sex-specific genetic effects are frequently overlooked because GWAS pools male and female samples, even though sex differences are well documented in psychiatric disease risk². GWAS results alone cannot guide treatment decisions, and experimental validation and clinical trials are required before these findings can be translated into the clinic².

Hatoum et al.’s study shows that GWAS results can be used for more than statistical risk prediction. By linking genetic risk signals to existing psychiatric medications, the group shows how human genetics can be used to ask more practical drug development questions². As genetic datasets continue to grow, this type of approach could help guide more targeted research and reduce the reliance on trial-and-error treatment strategies. While further experimental and clinical testing will still be required, this work highlights one way genetics can begin to play a more direct role in shaping future psychiatric therapies.

References

1.         Smoller, J. W. Psychiatric Genetics and the Future of Personalized Treatment. Depress Anxiety 31, 893–898 (2014).

2.         Hatoum, A. S. et al. Psychiatric genome-wide association study enrichment shows promise for future psychopharmaceutical discoveries. Commun Med 5, 176 (2025).

3.         Brewster, P. R., Bari, S. M. I., Walker, G. M. & Werfel, T. A. Current and Future Directions of Drug Delivery for the Treatment of Mental Illnesses. Adv Drug Deliv Rev 197, 114824 (2023).

4.         Hillhouse, T. M. & Porter, J. H. A brief history of the development of antidepressant drugs: From monoamines to glutamate. Exp Clin Psychopharmacol 23, 1–21 (2015).

5.         Uffelmann, E. et al. Genome-wide association studies. Nat Rev Methods Primers 1, 59 (2021).

6.         Scheinfeldt, L. B., Schmidlen, T. J., Gerry, N. P. & Christman, M. F. Challenges in Translating GWAS Results to Clinical Care. Int J Mol Sci 17, 1267 (2016).

7.         Woodward, D. J. et al. Identification of drug repurposing candidates for the treatment of anxiety: a genetic approach. Psychiatry Res 326, 115343 (2023).

Sarah Hammond

A large-scale, multi-ancestry investigation reveals novel loci and biological pathways involved in the pathogenesis of anxiety, highlighting the importance of using diverse genomes in research.

Over 300 million people worldwide suffer from an anxiety disorder, yet the genetic basis of these conditions remains poorly understood¹. Despite the prevalence of anxiety disorders having increased by more than 55% in the last three decades, genomic research in this area remains underrepresented as compared to other psychiatric conditions such as major depressive disorder (MDD) and post-traumatic stress disorder (PTSD)¹. Anxiety disorders affect individuals worldwide, so it is essential that genetic studies capture the genomic variability present across different ancestries. However, most genomic research to date has one striking limitation: studies are largely based on Caucasian European populations. Presently almost 80% of participants in published genome-wide association studies (GWAS) are of European descent². This lack of diversity narrows our view of human genetic variation, limits discovery, and underserves many populations. Beyond ethical and public health considerations, there is strong scientific justification for using multi-ancestry data, including more accurate effect-size estimates and broader generalizability². The study by Friligkou et al. takes a major step toward addressing these gaps and demonstrates the discovery and progress that can be found using multi-ancestry genomic data to study the genetic bases of anxiety disorders³.

The authors analyzed a strikingly large and diverse dataset, drawing on 1,266,780 participants (97,383 anxiety cases) from multiple databases encompassing five ancestries: European, African, admixed American, South Asian, and East Asian. To better understand the pathogenesis of anxiety, they integrated genome-wide, transcriptome-wide, and proteomic-wide association analyses (Figure 1). The goal of these studies was to find an association of either common genetic variants, changes in gene expression, or changes in protein regulation with anxiety.

The authors conducted ancestry-specific and cross-ancestry GWAS, ultimately identifying 41 loci associated with anxiety including 10 novel loci and one specific to individuals of African ancestry (Figure 1C). These discoveries were possible only because the dataset encompassed multiple ancestries. Per the authors, the novel findings of these GWAS quadruple the gene discoveries reported by previous studies, which highlights how the integration of multi-ancestry data from multiple databases can enrich the analysis.

Beyond identifying novel genes, the authors investigated how gene and protein expression levels are altered in anxiety disorders using a transcriptome-wide association study (TWAS) and a proteome-wide association study (PWAS) (Figure 1A-B). This multi-omics approach offers insight into the biological processes underlying anxiety. The tissue-specific and cross-tissue TWAS identified 211 transcriptome-wide associations with the strongest association being to a variation of a DRD2 locus in the cerebellar hemisphere. Several genes – including CTNND1, KHK, and NEK4 – were identified as being associated with anxiety in both the PWAS and TWAS, strengthening the evidence that they play a role in the pathogenesis of anxiety.

Figure 1 | Manhattan plots of PWAS (A), TWAS (B) and GWAS (C) statistics related to anxiety. Friligkou et al show which genes (labeled) have convergent evidence across all analyses. The x axis shows the genomic location of the gene on chromosomes 1-22. The y axis shows −log10(P value) obtained from two-sided statistical tests. A higher point indicates that there is a stronger statistical significance for the association of the gene with anxiety. Dashed lines represent the significance threshold from the Bonferroni multiple testing correction. The TWAS data shown is obtained from the multi-tissue analysis and the GWAS data shown is obtained from the European cohort. Adapted from Figure 3³.

Notably, the DRD2 gene encodes the dopamine D2 receptor, and variation in its expression has been found to be associated with anxiety⁴. DRD2 is already a major therapeutic target for antipsychotics, and it has been proposed as a target for anxiety disorders⁵. From both PWAS and TWAS investigations, they found the strongest evidence for association of the CTNND1 gene with anxiety. This association has been previously described, and this gene has also been associated with depression⁵. CTNND1 encodes a protein which regulates important processes in the central nervous system, and mice without CTNND1 were found to have anxiety-like behaviors⁵. Another gene of interest, KHK, identified across GWAS, TWAS and PWAS, is involved in fructose metabolism, and animal models have previously demonstrated that there may be a role in early-life fructose exposure to depression and anxiety⁶. Lastly, the NEK4 gene which is involved in cell cycle regulation and cell division, has also previously been identified as a drug candidate gene for bipolar disorder and MDD⁷. The involvement of many of these genes in multiple psychiatric conditions highlights the pleiotropic nature of anxiety disorders, prompting the authors to investigate what other conditions may have genetic overlap with anxiety.

The authors demonstrate that anxiety disorders do share much of their genetic basis with other psychiatric and physical conditions. After performing statistical analyses to investigate causation, the authors found that most genetic associations showed shared pleiotropy rather than direct causal influence – only a small subset of traits show some evidence favoring causation. There is a substantial genetic overlap between anxiety and both psychiatric and physical health conditions. They observed a substantial overlap with gastrointestinal and pain-related phenotypes. Interestingly, only about 10% of the influential variants for anxiety pathogenesis did not overlap with those for MDD. This suggests that the genetic basis of anxiety disorders is likely due to disruptions in shared biological pathways rather than a disorder-specific mechanism.

Friligkou et al. provide a clear example of how diverse population datasets can expand the discovery of new genetic associations³. They discovered novel anxiety risk loci and gained insight into biological pathways involved in the pathogenesis of anxiety, surpassing the insights from previous research studies which used only European data. The authors also clearly show how intertwined anxiety disorders are with other psychiatric disorders and even how they may overlap with physical health. Unfortunately, the limited data available for non-European populations reduced the statistical power in these populations and prohibited further ancestry-specific gene discovery. However, with worldwide efforts to expand genomic databases, these large genomic studies will become more robust and able to provide more detailed information on worldwide genomic variation. When genomic research reflects global diversity, we will gain a deeper, more accurate understanding of how biological mechanisms shape anxiety across populations.

References

1.        Javaid, S. F. et al. Epidemiology of anxiety disorders: global burden and sociodemographic associations. Middle East Current Psychiatry 2023 30:1 30, 44- (2023).

2.        Peterson, R. E. et al. Genome-wide Association Studies in Ancestrally Diverse Populations: Opportunities, Methods, Pitfalls, and Recommendations. Cell 179, 589–603 (2019).

3.        Friligkou, E. et al. Gene discovery and biological insights into anxiety disorders from a large-scale multi-ancestry genome-wide association study. Nat. Genet. 56, 2036–2045 (2024).

4.        Ike, K. G. O. et al. The human neuropsychiatric risk gene Drd2 is necessary for social functioning across evolutionary distant species. Mol. Psychiatry 29, 518–528 (2024).

5.        Li, W. et al. Genome-wide meta-analysis, functional genomics and integrative analyses implicate new risk genes and therapeutic targets for anxiety disorders. Nature Human Behaviour 2023 8:2 8, 361–379 (2023).

6.        Hyldgaard Andersen, S., Black, T., Grassi-Oliveira, R. & Wegener, G. Can early-life high fructose exposure induce long-term depression and anxiety-like behaviours? – A preclinical systematic review. Brain Res. 1814, (2023).

7.        Gong, B., Xiao, C., Feng, Y. & Shen, J. NEK4: prediction of available drug targets and common genetic linkages in bipolar disorder and major depressive disorder. Front. Psychiatry 16, 1414015 (2025).

Isabella Harmic

A new triple-marker diagnostic test offers a non-invasive method to screen for gastric cancer and the potential to address the difficulty physicians face to diagnose this disease.

As of 2022, Gastric cancer (GC) is the fifth most diagnosed cancer type worldwide, and responsible for 6.8% of cancer-related deaths¹. It remains relatively common even in first world countries where the eradication of Helicobacter pylori, a known risk factor, has decreased its incidence^1,2. To combat the outstanding burden of GC, Lee et al. have developed a new diagnostic test using circulating tumor DNA (ctDNA) that has the capability to accurately and non-invasively detect early stages of the cancer².

Using ctDNA to find genetic markers of disease is a rising trend in the field of cancer diagnostics. ctDNA is classified as a form of cell-free DNA (cfDNA), or DNA circulating in the bloodstream because of natural cellular processes, but originating from cancer cells (Figure 1)³. As such, this DNA can be easily accessed by a non-invasive blood sample for a series of analysis, one component of which focusing on methylation patterns. Methylation is a covalent marker that is not incorporated into the DNA sequence, rather it functions like an on/off switch, indicating which genes are actively expressed. One of the main advantages of ctDNA analysis is that it can detect cancers without any physical manifestation of symptoms – which is often the case with GC.

Figure 1. Analytic tests performed on cfDNA and its clinical applications. Tumors, by their nature, increase vascularization (blood flow) in the surrounding tissue to allow for greater access to essentials like oxygen and nutrients⁴. As such, it is common that DNA released by cancer cells from processes such as cell death, secretion, etc. makes its way into the bloodstream (left side of the figure; Liquid Biopsy); as diagramed by Dao et al. This circulating tumor DNA (ctDNA) can be accessed from plasma, which is extracted from a blood sample. Several properties of ctDNA can be of interest (middle of the figure; Sample Analysis), including the amount present in the blood which can be related to the size/burden of the tumor, the mutations in the DNA which can provide information on the tumor’s biology, and the methylation patterns present in the ctDNA. All of this analysis provides relevant information that can advise clinical progression, however, this paper focuses on the impacts of methylation, which informs physicians about gene expression levels (right side of the figure; Clinical Application)³. Figure taken from ³.

Often, early stages of GC present asymptomatically until it becomes extremely advanced, whereby the five-year survival rate drops to approximately 30%^2,5,6. Although an early diagnosis increases survival, many physicians hesitate to recommend screening for or even diagnose symptomatic GC because its symptoms replicate many other diseases⁶. Additionally, screening is typically only offered to those of high-risk status, or individuals who were exposed to multiple risk factors for the disease (e.g. H. pylori infection, increased age, etc.)⁵. The current gold standard to screen and diagnose GC is endoscopy, however, this technique possesses its own limitations. Specifically, endoscopy is highly invasive, reducing patient participation, and exhibits reduced accuracy for detecting early GC. In fact, an analysis of endoscopy effectiveness by Pimenta-Melo et al. found that the technique misses 1 in 10 GC cases, the majority of which were in the early stages of development⁷.

Unfortunately, a frequently encountered issue in the use of ctDNA-based diagnostic approaches is decreased sensitivity in early stage cancers with lower tumor cell counts, which produce less ctDNA in the bloodstream³. Despite this difficulty, Lee et al. rose to the challenge of designing an effective GC ctDNA test. They began by selecting two candidate genes from a database of over 1000 genome-wide methylation samples assembled by The Cancer Genome Atlas that exhibited elevated methylation profiles unique to GC samples, but not the other 32 cancer types documented. Their first candidate, GHR, produced the GC-specific methylation pattern they expected. As did their second candidate GLRB, although it also exhibited elevated methylation in colorectal cancer (CRC) samples. The authors validated their findings by measuring the methylation of these same genes in cancer and non-cancerous cell lines. Finally, they tested their genetic markers on a retrospective case-control cohort of 60 GC patient cases and 40 non-affected controls, of which 73% of the cases had stage 1 cancer. They found the methylation status of their chosen genes remained highly predictive, although as a final controlling measure they implemented a third gene with CRC-specific methylation, GATM, as a reference to prevent mistakes in conflating GC with CRC².

Lee et al.’s final triple-marker test, measuring the methylation of GHR, GLRB, and GATM, had a sensitivity of 82% and specificity of 90% in stage 1 GC patients. Moreover, their trend in sensitivity suggested increased accuracy in tandem with later stages of the cancer. Although Lee et al. were not the group to develop the most accurate, non-invasive test for GC, their test does have the highest sensitivity at detecting stage 1 GC, indicating its promise as a tool to screen for early disease. Three other notable high accuracy diagnostic tests have been developed using ctDNA methylation profiles, however, they all suffered decreases in sensitivity with stage 1 GC². The most accurate of the three, a triple-marker test developed by Anderson et al., had an overall sensitivity of 85% and specificity of 86%, but the sensitivity dropped to 50% for stage 1 GC⁸.

However, it would also be untrue to describe Lee et al.’s work as perfect. There are some limitations to the distributions of participants in the case-control study, such as a lack of controlling for H. pylori infections amongst cases, to see if this characteristic methylation profile is present in everyone exposed to the pathogen or only those who develop cancer. Additionally, they recruited a low number of participants from stages 2-4 of GC. In the future, both of these limitations could be addressed by applying their triple-marker diagnostic test to a larger, more generalized cohort of patients to ensure it remains accurate in a wider implementation².

A new technique is needed to bridge the gap between diagnosis and early stage GC.  Particularly one that avoids patient discomfort to allow for generalized screening of the population, not just those of high-risk status. Lee et al.’s triple marker test provides a promising solution. Their test is straightforward, analysing the methylation profiles of only three genes, and non-invasive, requiring just a regular blood sample to conduct. As such, it would be simpler to implement clinically and reach a much greater number of patients. With accessible screening, this test has great potential to increase early disease detection and save countless lives.

References

1.         Bray, F. et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA. Cancer J. Clin. 74, 229–263 (2024).

2.         Lee, Y. Y., An, J., Han, J., Moon, Y. & Lee, S.-I. Plasma-based digital PCR assay for early detection of gastric cancer using multiple methylation biomarkers. Sci. Rep. 16, 1727 (2025).

3.         Dao, J. et al. Using cfDNA and ctDNA as Oncologic Markers: A Path to Clinical Validation. Int. J. Mol. Sci. 24, 13219 (2023).

4.         De Palma, M. & Hanahan, D. Milestones in tumor vascularization and its therapeutic targeting. Nat. Cancer 5, 827–843 (2024).

5.         Karimi, P., Islami, F., Anandasabapathy, S., Freedman, N. D. & Kamangar, F. Gastric Cancer: Descriptive Epidemiology, Risk Factors, Screening, and Prevention. Cancer Epidemiol. Biomarkers Prev. 23, 700–713 (2014).

6.         Xia, J. Y. & Aadam, A. A. Advances in screening and detection of gastric cancer. J. Surg. Oncol. 125, 1104–1109 (2022).

7.         Pimenta-Melo, A. R., Monteiro-Soares, M., Libânio, D. & Dinis-Ribeiro, M. Missing rate for gastric cancer during upper gastrointestinal endoscopy: a systematic review and meta-analysis. Eur. J. Gastroenterol. Hepatol. 28, 1041 (2016).

8.         Anderson, B. W. et al. Detection of Gastric Cancer with Novel Methylated DNA Markers: Discovery, Tissue Validation, and Pilot Testing in Plasma. Clin. Cancer Res. Off. J. Am. Assoc. Cancer Res. 24, 5724–5734 (2018).

Yixin He

The study establishes mosaic loss of the Y chromosome (mLOY) as an important, independent risk factor of cardiovascular death in men, suggesting a new perspective of precision medicine.

Cardiovascular disease (CVD) is a leading threat to male health¹, yet a fundamental scientific question remains unsolved: approximately half of all CVD cases in men cannot be explained by traditional risk factors like blood pressure and cholesterol². A recent study published in the European Heart Journal provides an understanding of why these sex-related disparities in prevalence and outcome may occur³. The authors identify mosaic loss of the Y chromosome (mLOY), a common age-related mutation in men’s blood cells, as a strong and independent risk factor of mortality in patients with CVD. More importantly, the study reveals that this risk is highly dependent on genetic predisposition to interstitial myocardial fibrosis, which is a profibrotic protein signature (figure 1). This work provides a new foundation for identifying high-risk men and developing sex-specific intervention strategies.

Figure 1 | The authors propose that LOY in leukocytes is associated with myocardial fibrosis and contributes to increased cardiovascular disease mortality. The figure is taken from³.

The researchers began by analyzing data from the well-established Ludwigshafen Risk and Cardiovascular Health (LURIC) study⁴. They selected 1,698 male participants who had undergone coronary angiography and stratified them into two groups based on the proportion of Y chromosome loss in their white blood cells, using a clinically defined threshold of >17%. Initial analysis revealed that individuals with low LOY had significantly lower mortality during a follow-up of approximate 10 years compared to those with high LOY. Their mortality rate was comparable to that of women in the same cohort, who typically have lower mortality⁵, suggesting a link between LOY and poor prognosis.

However, this association could be confounded by other factors. For instance, LOY is itself a common marker of aging, and age is the strongest risk factor of CVD. So, is the effect of LOY independent, or does it merely reflect the influence of age and other variables?

To answer this, the team employed a Cox proportional hazards model, a statistical tool that quantifies the independent effect of a specific factor on event risk after controlling for others⁶. They included LOY status in the model while adjusting for over ten traditional risk factors, including age, smoking status, body mass index, and etc. The results demonstrated that high LOY is an independent risk marker, separate from all these conventional factors. It was associated with a 41% higher risk of all-cause mortality (hazard ratio 1.41). Using a more refined model that accounts for competing risks, high LOY was linked to a 49% increased risk of cardiovascular death (hazard ratio1.49) and a striking 165% increased risk of fatal myocardial infarction (hazard ratio 2.65).

After establishing LOY as an independent predictor of CVD, the researchers sought to investigate its underlying mechanisms. Previous research has shown that LOY in hematopoietic cells promotes cardiac fibrosis in mouse models⁷, a condition that increases myocardial stiffness and can lead to arrhythmias⁸. To investigate the interaction between LOY and cardiac fibrosis, the researchers first applied a quantitative metric, the weighted genetic risk score(wGRS) for myocardial fibrosis, measuring each patient’s genetic susceptibility for cardiac fibrosis. They found that the harm from LOY exhibits synergy with this genetic background. In men predisposed to fibrosis (wGRS > 0), LOY increased the risk of cardiovascular death by 114% (hazard ratio 2.14). Conversely, in men not predisposed to fibrosis (wGRS ≤ 0), the excess risk from LOY was completely absent (hazard ratio 1.02). This strongly suggests a functional biological interaction between LOY and fibrotic pathways, indicating that LOY likely exerts its deleterious effect by promoting fibrosis.

Building on these findings, the researchers next investigated the relationship between LOY and myocardial fibrosis at the cellular level. They performed genome-wide methylation analysis within the LURIC cohort, comparing blood samples from men with high and low LOY. This revealed 298 differentially methylated genes, suggesting that LOY may reprogram cell function epigenetically. To verify whether these epigenetic changes alter actual gene expression, the team turned to a previous independent single-cell RNA sequencing dataset from seven male patients with severe degenerative aortic stenosis³. Comparing the transcriptomes of LOY cells and normal cells within this dataset, they found that 37 of the 298 genes showed significant expression differences, including several, like ARAP1, BST1, and RPS5, known to be involved in fibrosis regulation^9-11.

Current technology does not allow for the knockout of an entire Y chromosome in vitro. Therefore, the researchers knocked down three candidate genes in macrophages in vitro to mimic the state of LOY cells. The results showed that knocking down RPS5 was sufficient to drive cardiac fibroblasts to produce excess collagen. The study thus provides a compelling, but indirect, evidence chain linking LOY to the promotion of cardiac fibrosis. However, this experimental design still has certain limitations. Whether simply knocking down these three genes respectively in vitro is sufficient to faithfully mimic the cellular state of LOY remains an open question.

In summary, this research establishes an age-related somatic mutation (LOY) as an independent risk factor of CVD in men and elucidates its likely mechanism of action through interaction with cardiac fibrosis pathways. It paves the way for a deeper understanding of why men are disproportionately affected by CVD and for crafting therapeutic strategies tailored specifically to them. Moving forward, developing blood-based LOY assays for risk stratification and designing drug interventions targeting its downstream pathways (e.g., RPS5) hold promise as novel precision medicine strategies.

References

1          Global burden of 288 causes of death and life expectancy decomposition in 204 countries and territories and 811 subnational locations, 1990-2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet 403, 2100-2132 (2024). https://doi.org:10.1016/s0140-6736(24)00367-2

2          Magnussen, C. et al. Global Effect of Modifiable Risk Factors on Cardiovascular Disease and Mortality. N Engl J Med 389, 1273-1285 (2023). https://doi.org:10.1056/NEJMoa2206916

3          Mas-Peiro, S. et al. Mosaic loss of Y chromosome in monocytes is associated with lower survival after transcatheter aortic valve replacement. Eur Heart J 44, 1943-1952 (2023). https://doi.org:10.1093/eurheartj/ehad093

4          Winkelmann, B. R. et al. Rationale and design of the LURIC study–a resource for functional genomics, pharmacogenomics and long-term prognosis of cardiovascular disease. Pharmacogenomics 2, S1-73 (2001). https://doi.org:10.1517/14622416.2.1.S1

5          Fan, J. et al. The global burden and attributable risk factor analysis of cardiovascular disease between females and males, 1990-2021: findings from the 2021 global burden of disease study. Int J Surg 112, 239-249 (2026). https://doi.org:10.1097/js9.0000000000003397

6          Cox, D. R. Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B (Methodological) 34, 187-220 (1972).

7          Sano, S. et al. Hematopoietic loss of Y chromosome leads to cardiac fibrosis and heart failure mortality. Science 377, 292-297 (2022). https://doi.org:10.1126/science.abn3100

8          de Jong, S., van Veen, T. A., van Rijen, H. V. & de Bakker, J. M. Fibrosis and cardiac arrhythmias. J Cardiovasc Pharmacol 57, 630-638 (2011). https://doi.org:10.1097/FJC.0b013e318207a35f

9          Zhang, X. et al. Matrine attenuates pathological cardiac fibrosis via RPS5/p38 in mice. Acta Pharmacol Sin 42, 573-584 (2021). https://doi.org:10.1038/s41401-020-0473-8

10        Yuan, Y. et al. BST-1 aggravates aldosterone-induced cardiac hypertrophy via the Ca2+ /CaN/NFATc3 pathway. Gen Physiol Biophys 42, 349-360 (2023). https://doi.org:10.4149/gpb_2022063

11        Jiang, L. et al. Downregulation of the CD151 protects the cardiac function by the crosstalk between the endothelial cells and cardiomyocytes via exosomes. PLoS One 19, e0297121 (2024). https://doi.org:10.1371/journal.pone.0297121