“Major Depressive Disorder (MDD) is associated with trait X” we write when reporting results of observational studies. Replace “trait X” with, for instance, Type 2 Diabetes or Body Mass Index (BMI) and this simply means that people with depression are more likely to have diabetes or higher BMI than those without. But this doesn’t tell us which comes first: is depression a risk factor or a consequence of trait X?
Randomised controlled trials (RCTs) can provide answers about cause and effect. But RCTs are not always feasible or ethical: we cannot increase people’s BMI to see whether they develop depression. Thus, innovative methods using observational data have been developed to infer potential causal relationships between human traits. One such method is Mendelian randomisation (MR; see past blogs by Crick 2023 and Souama 2025; Sanderson et al. 2022). This leverages genetic data, which are randomly assigned at conception, to naturally “randomise” people to higher or lower risk for traits like MDD. By comparing outcomes across these groups, we can get clues about potential cause-and-effect links.
But, MR is not a perfect RCT. Even when using family data, where conditions are more controlled, the key assumption that “everything else is equal” is only partially true. In general populations, other factors can interfere. Still, with careful design and analysis, MR can give useful clues about cause-and-effect relationships. This is what Pasman et al. (2025) did in their recent study that applied MR at an unprecedented scale to study MDD’s potential causal links with more than 100 traits.
This study by Pasman et al. applies Mendelian randomisation at an unprecedented scale to identify possible causes and consequences of depression.
Methods
Pasman et al. applied MR to test the causal connections between MDD and 137 traits selected from over 200 probable risk factors or consequences of depression.
For 89 traits, “two-sample” MR (TSMR) was applied. This uses data from large Genome-Wide Association Studies (GWAS) that estimate the effects of millions of genetic variants on human traits. For instance, data on the genetic base of MDD were retrieved from a GWAS including 166,773 cases and 507,679 controls.
For an additional 48 traits, for which previous GWAS were not available, the relationship with MDD was analysed with the “one-sample” MR approach (OSMR). This integrates genetic data with actual trait and MDD measures in the same study sample. Here, data from 105,567 participants of the UKBiobank cohort were used. Both causal directions were tested: from MDD to other traits and from other traits to MDD.
Results
MR analyses revealed widespread potential causal connections between MDD and multiple traits spanning somatic diseases, inflammation, metabolism, physical activity, risk behaviours, and socioeconomic factors. TSMR analyses identified 57 causal links from MDD to other traits and 27 in the reverse direction, with 26 overlapping. This indicates potential bidirectional links whereby these traits may act both as risk factors and consequences of MDD.
It’s important to note that MR estimates reflect MDD “liability” – the underlying genetic risk of developing depression – rather than its full clinical manifestation. MR does not track whether someone actually develops the disorder, but instead estimates how increases in genetic risk for MDD might influence other traits. Results should therefore be interpreted as the potential causal effects of MDD liability on these traits, whether or not MDD fully develops. This helps avoid the misconception that MR shows a direct trajectory from having MDD to later outcomes.
Potential casual effects of MDD liability included increased risk of:
- Almost all somatic diseases e.g. gastroesophageal reflux disease, irritable bowel disease, Crohn’s disease, peripheral and coronary artery disease.
- Death by suicide.
- Higher inflammation as measured by C-reactive protein.
- Intake of zinc supplements.
Bi-directional links were found for traits that may act both as risk factors or consequences of MDD. These included insomnia, lower cognitive functions and educational attainment, increased BMI and type 2 diabetes, almost all functional measures (e.g. higher loneliness and disability) and risk behaviours (e.g. higher likelihood of smoking initiation).
OSMR analyses further supported the bi-directional causal connection between MDD liability and negative functional outcomes, including increased risk of hospitalisation and higher level of health dissatisfactions and disability in important activities of daily living (e.g. mobility and self-care).
All results were complemented with a wide array of extensive sensitivity analyses aimed at establishing the robustness of the reported findings.
Genetic risk for depression appears to have widespread effects, influencing physical health, behaviour, and daily functioning, with some traits acting bidirectionally
Conclusions
The study by Pasman et al. (2025) applied MR analyses at unprecedented scale to examine the potential causal relationships between MDD and more than 100 traits, spanning both risk factors and consequences of the disorder. Analyses revealed potentially causal (and often bi-directional) links between MDD and traits across multiple domains, including somatic diseases, inflammation, metabolism, physical activity, risk behaviours, and socioeconomic factors.
Taken together, these findings suggest that genetic risk for depression is linked to a broad range of health, functional, and psychosocial outcomes. In particular, the authors highlighted how the findings:
underscore MDD as a crosscutting risk factor across medical, functional and psychosocial domains.
Depression liability spans multiple domains, linking mental health with medical, functional, and social outcomes.
Strengths and limitations
This study has several important strengths. A wide array of relevant traits was carefully selected based on previous literature and the MR analyses were conducted with strong methodological rigor. However, there are two limitations that are not directly addressed in the main article that I would like to highlight here.
The first is shared by all studies that rely on genetic data – the overwhelming majority of data comes from samples of European ancestry. This means that this study’s findings may not generalise to populations with different ancestral and cultural backgrounds. BMI provides a clear example: Consistent with previous MR studies using European samples, Pasman et al. found that higher BMI may increase the risk of developing MDD. However, other research has shown the opposite pattern in East-Asian ancestry samples, whereby a genetically driven increase in BMI is related to a lower risk of MDD (Meng et al. 2024).
The second limitation concerns key assumptions required for valid interpretation of MR results. These assumptions become particularly difficult to evaluate when complex traits like MDD are used as risk factors. One core assumption, called “exclusion restriction”, requires genetic variants used in MR to affect the outcome only through the risk factor of interest. This assumption is relatively plausible for simple traits, such as blood levels of a specific protein, where relatively few genetic variants are involved and their biological pathways are well defined. In contrast, MDD is influenced by hundreds of associated genetic variants linked to multiple genes, functions, pathways and traits. However, some of the genetic variants most strongly associated with MDD are also associated with BMI (Adams et al 2025). In this scenario, it is extremely difficult to evaluate the exclusion restriction assumption; even if sensitivity analyses such as those performed by Pasman et al. could not find evidence of its violation. Therefore, the results from MR analyses using complex polygenic traits as risk factors require extra caution in their interpretation.
Overall, the findings of this study, especially those testing MDD as a risk factor, require further validation from independent studies addressing the same research questions with different methodologies, an important scientific process known as “triangulation” (Treur et al. 2024).
Genetic studies of depression need careful interpretation, especially for complex traits and diverse populations.
Implications for practice
The findings from this study have no immediate implications for clinical practice, but nonetheless they provide important and highly relevant indications that could help shape the future of depression care. The study supports what epidemiological research has long suggested: depression affects not only mental health but also physical, functional and social wellbeing. This helps explain why depression is projected to be one of the leading causes of disability by 2030, with a major impact on society and public health systems.
Beyond mental health, MDD increases the risk of developing physical illnesses like coronary artery disease and type 2 diabetes, which contributes to the excess mortality among individuals with depression and other psychiatric disorders. Findings of the present study indicated that this connection is likely causal, with genetic vulnerability for MDD also affecting multiple somatic diseases. These considerations should encourage us to move towards a full re‑conceptualisation of depression as a whole‑body, rather than brain‑only, systemic disorder, whose evaluation and treatment cannot be separated from the intertwined management of physical health.
The extensive causal links between multiple traits acting simultaneously as risk factors and consequences of MDD highlight once more the complex, multifactorial nature of this disorder. Any attempt at its prevention or reduction of its impact at the population level will need to address multiple dimensions – from biology (e.g., metabolic health), to lifestyle (e.g., physical activity, smoking reduction) and social factors (e.g., loneliness).
Finally, Pasman et al.’s study provides a useful roadmap for future research. MR results are not definitive proof of causality, but they can help prioritise which traits are most likely to influence depression. We are overwhelmed with results from observational research stating that “depression is associated with trait X”, but we have limited resources to test this robustly. Nevertheless, Pasman et al.’s findings represent a great resource to prioritise risk factors with stronger evidence of potential causal effects on MDD to be carried forward in intervention studies, and possibly to discard those with lower potential.
Depression is a whole-body disorder, with complex causes and consequences spanning health and social systems.
Statement of interests
Yuri Milaneschi – After the merging of two departments of psychiatry under the same unique institution (Amsterdam UMC) Yuri Milaneschi became a colleague of the first author of the study. Nevertheless, Yuri was previously not involved with the study presented here or its peer-review evaluation.
Editor
Edited by Éimear Foley. AI tools assisted with language refinement and formatting during the editorial phase.
Links
Primary paper
Pasman, Joëlle A., Bergstedt, Jacob, Harder, Arvid, Gong, Tong, Xiong, Ying, Hägg, Sara, Fang, Fang, Treur, Jorien L., Choi, Karmel W., Sullivan, Patrick F., & Lu, Yi. (2025). An encompassing Mendelian randomization study of the causes and consequences of major depressive disorder. Nature. Mental health, 3(9), 1002–1011. https://doi.org/10.1038/s44220-025-00471-x
Other references
Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium. Electronic address: [email protected]; Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium. Trans-ancestry genome-wide study of depression identifies 697 associations implicating cell types and pharmacotherapies. Cell. 2025 Feb 6;188(3):640-652.e9. doi: 10.1016/j.cell.2024.12.002. Epub 2025 Jan 14. PMID: 39814019; PMCID: PMC11829167.
Meng, X., Navoly, G., Giannakopoulou, O., Levey, D. F., Koller, D., Pathak, G. A., Koen, N., Lin, K., Adams, M. J., Rentería, M. E., Feng, Y., Gaziano, J. M., Stein, D. J., Zar, H. J., Campbell, M. L., van Heel, D. A., Trivedi, B., Finer, S., McQuillin, A., Bass, N., … Kuchenbaecker, K. (2024). Multi-ancestry genome-wide association study of major depression aids locus discovery, fine mapping, gene prioritization and causal inference. Nature genetics, 56(2), 222–233. https://doi.org/10.1038/s41588-023-01596-4
Sanderson, E., Glymour, M. M., Holmes, M. V., Kang, H., Morrison, J., Munafò, M. R., Palmer, T., Schooling, C. M., Wallace, C., Zhao, Q., & Smith, G. D. (2022). Mendelian randomization. Nature reviews. Methods primers, 2, 6. https://doi.org/10.1038/s43586-021-00092-5
Treur JL, Lukas E, Sallis HM, Wootton RE. A guide for planning triangulation studies to investigate complex causal questions in behavioural and psychiatric research. Epidemiol Psychiatr Sci. 2024 Nov 7;33:e61. doi: 10.1017/S2045796024000623. PMID: 39506622; PMCID: PMC7616800.
