Polygenic risk scores, commonly abbreviated as PRS, summarize the combined influence of thousands or millions of genetic variants on a person’s likelihood of developing a particular condition. Constructing them is a formidable statistical challenge. Modern biobanks measure hundreds of thousands of individuals at millions of genetic positions, and the number of potential predictors vastly exceeds the number of study participants. In this high-dimensional regime, ordinary regression breaks down, and researchers must rely on penalized regression methods that shrink or eliminate weak signals. The most famous of these, the Lasso introduced by Tibshirani in 1996, adds a penalty proportional to the sum of absolute coefficient values, driving the coefficients of uninformative variants all the way to zero and thereby selecting a sparse subset of predictors.

Despite its elegance, the Lasso faces difficulties when the number of variables dwarfs the number of observations, as it does in genomic data. Feature selection can be unstable: small perturbations in the data may cause the method to pick different variants, and correlated variants can compete with one another in ways that obscure the true signal. UniLasso addresses these problems with a two-stage strategy. In the first stage, the method performs simple univariate regressions, examining each genetic variant one at a time for its individual association with the trait of interest. The signs and magnitudes of these univariate coefficients are then used to guide the second stage, a multivariate penalized regression fitted on the full set of variants simultaneously. Variants with strong univariate evidence receive favorable treatment in the penalty, while those with weak or contradictory evidence are discouraged from entering the model.

The intuition behind this design is that univariate screening provides a stable, data-driven prior about which variants matter. By anchoring the multivariate fit to this prior, uniLasso stabilizes feature selection across datasets and reduces the variance of the resulting coefficient estimates. The authors emphasize that the univariate stage does not replace the multivariate fit; it merely informs it. The final model is still learned from individual-level data in the target population, which means the method can adapt to the specific ancestry, cohort composition, and phenotype definitions of the dataset at hand. This two-stage architecture also confers a computational advantage: the initial univariate screen is embarrassingly parallel and can be distributed across many processors, making the approach feasible for datasets with more than one million genetic variants measured on hundreds of thousands of people.

To test the method at true biobank scale, the team applied uniLasso to the UK Biobank, a population-based repository containing genetic, health, and lifestyle data from roughly half a million participants across the United Kingdom. The researchers adapted the algorithm to handle the sheer dimensionality of the resource, where more than a million variants must be evaluated simultaneously. They also introduced an extension called uniLasso ES, short for external scores, which incorporates summary statistics from pre-existing genome-wide association studies. These external signals, drawn from large meta-analyses that may involve millions of additional participants, guide the regression toward variants with prior evidence of association by informing penalty weights and imposing sign constraints that keep coefficient directions consistent with earlier findings.

The uniLasso ES framework is particularly significant because it bridges two traditionally separate paradigms in polygenic risk prediction. Summary-statistic methods, such as PRS-CS and lassosum2, rely exclusively on aggregated association results and linkage disequilibrium reference panels, avoiding the privacy and logistical hurdles of individual-level data but sacrificing some flexibility. Individual-level methods, by contrast, can model the full joint structure of the data but require access to raw genotypes. UniLasso ES occupies a middle ground: external summary statistics act as a soft guide, shaping the penalty landscape, while the definitive fitting happens on individual-level target data. The external evidence informs the model without dictating it, allowing the method to correct for differences between the discovery population and the target population.

The empirical results reported in the study are striking. Across a range of traits and disease outcomes in the UK Biobank, uniLasso attained predictive performance comparable to the standard Lasso while selecting substantially fewer variants. Sparser models offer tangible benefits beyond aesthetics. A risk score built from a few thousand variants rather than hundreds of thousands costs less to genotype in a clinical setting, reduces the storage and computational burden of scoring large patient cohorts, and, crucially, is easier to interrogate biologically. When a compact set of variants drives a prediction, researchers can more readily trace which genes and pathways contribute to the score, potentially generating new hypotheses about disease mechanisms.

Interpretability has been a persistent criticism of polygenic risk scores. Critics note that scores built from millions of tiny effects are statistical black boxes whose individual components rarely correspond to established biology. By producing models that are sparse by construction and whose selected variants carry coherent univariate evidence, uniLasso nudges the field toward scores that a geneticist can actually read. The method’s reliance on univariate coefficient signs also guards against a known pathology of penalized regression in the presence of correlated predictors, where a variant may enter the model with a counterintuitive sign simply because it is absorbing the effect of a nearby variant. Sign constraints informed by univariate evidence keep the fitted coefficients aligned with the marginal signal.

Benchmarking against competing approaches reinforced the method’s promise. The authors report that both uniLasso and uniLasso ES remained competitive with PRS-CS and lassosum2, two widely used PRS estimation methods that represent the current state of the art. Achieving parity with these established tools while producing markedly sparser models positions uniLasso as a practical alternative for research groups that need both accuracy and parsimony. The scalability of the implementation matters as much as its statistical properties; as biobanks grow toward cohorts of millions and sequencing expands the variant catalogue into the billions, methods that cannot be distributed efficiently will simply fall out of use.

The study arrives at a moment when polygenic risk scores are edging toward clinical application, with health systems beginning to explore their use in screening programs for conditions such as cardiovascular disease, diabetes, and several cancers. The promise of personalized prevention depends on scores that are accurate, portable across ancestries, and affordable to deploy. A method that delivers competitive accuracy with a fraction of the variants addresses all three concerns at once. The work also exemplifies a broader trend in statistics: the creative combination of classical ideas, in this case univariate screening and penalized regression, to meet the demands of modern data scale. As the authors and their collaborators continue to refine the framework, uniLasso may well become a standard tool in the geneticist’s arsenal, helping to convert the torrent of biobank data into models that clinicians and patients can understand and act upon.

Subject of Research: Scalable sparse regression methods for polygenic risk score computation in biobank-scale genomic data

Article Title: Univariate-guided sparse regression for Biobank-scale high-dimensional omics data

Article References: Univariate-guided sparse regression for Biobank-scale high-dimensional omics data. (n.d.). https://doi.org/10.1371/journal.pgen.1012314

Image Credits: AI Generated

DOI: 10.1371/journal.pgen.1012314

Keywords: polygenic risk scores, uniLasso, UK Biobank, sparse regression, Lasso, genome-wide association studies, high-dimensional statistics, summary statistics, predictive modeling, genomics, feature selection, PLOS Genetics

The research team, led by Wenqian Li and colleagues at Jinan University in Guangzhou, China, analyzed data from 21,577 cancer survivors aged 39 to 70 years enrolled in the UK Biobank, one of the world’s largest biomedical databases. Crucially, the investigators excluded anyone who already had the diseases of interest at baseline, ensuring that they were tracking genuinely new-onset conditions rather than pre-existing illness. Incident cases across 39 distinct chronic diseases were identified through linkage to hospital inpatient records and death registries, allowing the researchers to follow participants prospectively over time rather than relying on retrospective self-reports.

What makes the study methodologically notable is its use of two complementary instruments for measuring frailty. The first, the frailty phenotype developed by Linda Fried and colleagues, is a five-point scale based on physical markers: unintentional weight loss, exhaustion, low physical activity, slow walking speed, and weak grip strength. The second, the frailty index, takes a broader approach, aggregating deficits across multiple health domains into a continuous score ranging from 0 to 1. Because these tools capture different dimensions of biological vulnerability—one focused on physical function, the other on accumulated health deficits—the researchers could test whether their conclusions held regardless of how frailty was defined. They did.

Using Cox proportional hazards models, the team estimated hazard ratios linking baseline frailty to the subsequent incidence of each chronic disease. The results were striking in their breadth. Frailty was significantly associated with an increased risk of 28 of the 39 chronic conditions examined. Every one of the eight cardiometabolic diseases studied showed this association, as did five of ten mental and neurological disorders, five of six digestive system diseases, and ten of fifteen other conditions ranging from respiratory to musculoskeletal illness. In other words, frailty did not merely predict one or two unfortunate outcomes; it painted a systemic picture of heightened vulnerability across nearly every organ system the researchers examined.

The magnitude of risk varied considerably by condition, and some of the strongest associations were unexpected. Within the mental and neurological category, depression showed the most pronounced effect: cancer survivors with higher frailty index scores had a 2.39-fold increased risk of developing depression compared with their non-frail counterparts, with a 95 percent confidence interval of 2.19 to 2.62. Bronchiectasis, a chronic condition in which the airways become abnormally widened and prone to infection, carried a hazard ratio of 2.11. Irritable bowel syndrome followed at 1.79, and heart failure—the most strongly affected cardiometabolic outcome—showed a hazard ratio of 1.48. The overall pattern, the authors note, was most pronounced for mental and neurological disorders, a finding that challenges the traditional tendency to view frailty as primarily a physical or geriatric syndrome.

Stratified analyses added further nuance to the picture. The association between frailty and the risk of anxiety and depression was stronger among cancer survivors younger than 60 years, suggesting that frailty may be particularly consequential for mental health in midlife survivors rather than only in the elderly. Meanwhile, prostate cancer survivors exhibited a significantly higher risk of bronchiectasis and diverticular disease in connection with frailty, hinting that cancer type may modify how frailty translates into specific comorbidities. These subgroup findings matter because they imply that a one-size-fits-all surveillance strategy may miss important risk patterns; a younger breast cancer survivor and an older prostate cancer survivor may face very different frailty-related trajectories.

The biological plausibility of these associations is supported by a growing literature on premature aging in cancer survivors. Previous research has suggested that cancer and its treatments can accelerate biological aging through mechanisms including chronic inflammation, telomere shortening, mitochondrial dysfunction, and sarcopenia—the loss of skeletal muscle mass and function. Frailty, in this framework, is not simply a label for weakness but a measurable manifestation of cumulative physiological decline. Studies cited by the authors connect frailty to cardiovascular outcomes, dementia risk, venous thromboembolism, metabolic liver disease, chronic obstructive pulmonary disease, osteoporosis, and even psoriasis and glaucoma in general populations. The new study extends this evidence specifically to cancer survivors, a group already known to bear a disproportionate comorbidity burden as they age.

Sensitivity analyses confirmed the stability of the results, lending weight to the central conclusion that frailty is prospectively associated with an increased risk of diverse chronic diseases in this population. The study was conducted under the principles of the Declaration of Helsinki using UK Biobank data under Application Number 300908, with ethical approval from the relevant UK research ethics committees and written informed consent from all participants. The authors declared no conflicts of interest, and the work was supported by the National Natural Science Foundation of China and several Guangzhou municipal research programs.

Still, some caveats are worth keeping in mind. As with any observational cohort, the study demonstrates association rather than proven causation; it remains possible that subclinical disease contributes both to frailty measurements and to later diagnoses, or that shared underlying mechanisms—such as systemic inflammation—drive both. The UK Biobank population, which tends to be healthier than the general population, may also limit generalizability, and the frailty phenotype and index were measured only at baseline, leaving open questions about how changes in frailty over time might alter risk. Nonetheless, the prospective design, the large sample, the exclusion of baseline disease, and the consistency across two frailty measures collectively strengthen the case that the observed links are meaningful.

The clinical implications could be significant. With cancer survival rates improving steadily across decades, the number of people living long after a cancer diagnosis continues to climb, and their long-term health needs are increasingly recognized as a distinct field of care. The authors argue that routine assessment of frailty may help identify survivors who need targeted interventions and proactive clinical monitoring to prevent long-term comorbidities and enhance quality of life. In practical terms, that could mean incorporating grip strength tests, gait speed measurements, and deficit-based frailty indices into survivorship follow-up visits, then directing frail patients toward exercise programs, nutritional support, mental health screening, and earlier cardiometabolic surveillance. If frailty can be treated as a modifiable risk factor rather than a fixed fate, the study suggests, some of the chronic disease burden that shadows cancer survival might be prevented before it begins.

Subject of Research: The association between baseline frailty and the incidence of chronic diseases among cancer survivors

Article Title: Frailty and risk of common chronic diseases among cancer survivors: a prospective cohort study

Article References: Li, W., Zhang, Z., Lu, C., Xu, Y., Han, S., Zhan, Y., Ge, G., Wang, Z., Lan, X., Zhang, X., Lu, H., & Guo, J. (2026). Frailty and risk of common chronic diseases among cancer survivors: a prospective cohort study. Journal of Cancer Survivorship. https://doi.org/10.1007/s11764-026-02118-x

Image Credits: AI Generated

DOI: 10.1007/s11764-026-02118-x

Keywords: cancer survivors, frailty, chronic diseases, UK Biobank, cardiometabolic disease, depression, heart failure, prospective cohort study, comorbidity, premature aging, survivorship care, frailty index

The brain care score, developed at the McCance Center for Brain Health, condenses twelve modifiable factors into a single number: four physical measures (blood pressure, hemoglobin A1c, cholesterol, and body mass index), five lifestyle behaviors (nutrition, alcohol intake, smoking, physical activity, and sleep), and three social-emotional dimensions (stress, relationships, and meaning in life). In the UK Biobank adaptation used here, the score ranges up to 19 points, with higher values indicating better brain care. Previous studies have linked higher scores to favorable neuroimaging markers and reduced risks of dementia, stroke, and late-life depression. What remained unknown was whether the score also tracks with the fine-grained architecture of the retina, an outgrowth of the central nervous system that shares vascular, metabolic, and inflammatory biology with the brain.

To find out, the team led by investigators at The Chinese University of Hong Kong analyzed optical coherence tomography (OCT) scans from 28,657 UK Biobank participants aged 40 to 69. OCT is a non-invasive imaging technique that uses light interference to map retinal layers at micrometer resolution, and the researchers examined eight distinct layers plus overall macular thickness. After rigorous quality control and statistical adjustment for age, sex, ethnicity, education, deprivation, cardiovascular disease history, refractive error, and intraocular pressure, a clear pattern emerged: each 5-point increase in the brain care score was associated with thicker retinal nerve fiber layer, thicker ganglion cell–inner plexiform layer, thicker photoreceptor segments, and greater average macular thickness.

The effect sizes are small in absolute terms—a few tenths of a micrometer per 5-point score increase, translating to roughly 1.2 percent difference in nerve fiber layer thickness and about half a percent for the ganglion cell layer relative to cohort means. The researchers are careful to stress that these are population-level structural differences, not differences an ophthalmologist could detect in a single patient or act on clinically. Yet the pattern was remarkably consistent. Participants in the lowest score quartile had significantly thinner layers across the board compared with those in the highest quartile, all trend tests were highly significant, and the findings held up in sensitivity analyses that accounted for nonlinear aging effects, restricted the sample to European participants, and replicated the associations using repeat OCT visits years later.

Perhaps the most provocative result concerns retinal biological age. The team trained a support vector machine on 60 OCT-derived structural metrics from thousands of healthy participants to predict chronological age from retinal structure alone, then defined an OCT age gap as the bias-corrected predicted age minus actual age. Positive values indicate a retina that looks older than it should. After full adjustment, each 5-point increase in brain care score was associated with a 0.326-year reduction in this retinal age gap, with the mean gap falling progressively from 0.39 years in the lowest score quartile to −0.08 years in the highest. In other words, people who take better care of their brains carry retinas that read as younger on a machine-learning clock.

Dissecting the score into its components revealed where the signal lives. The physical and lifestyle domains drove the associations, while the social-emotional domain showed no independent link to retinal structure. Favorable blood pressure, hemoglobin A1c, and body mass index were each tied to thicker specific layers and a smaller retinal age gap, with effects on the age gap ranging from about a quarter to nearly half a year. Non-smoking, moderate alcohol intake, regular aerobic exercise, and at least seven hours of nightly sleep each contributed modest reductions in retinal biological age. Interestingly, total cholesterol below 190 milligrams per deciliter was actually associated with a slightly larger retinal age gap, echoing a growing literature on the complex, sometimes U-shaped relationships between lipids and neurological health.

To probe the biology behind these correlations, the researchers turned to metabolomics. For a subset of 14,656 participants with nuclear magnetic resonance profiling of 249 plasma biomarkers, they ran exploratory mediation analyses asking whether circulating metabolites statistically carry part of the association between brain care score and retinal thickness. The candidate pathways that emerged centered on lipid metabolism, unsaturated fatty acids, and branched-chain amino acids. One principal component reflecting HDL-enriched lipid profiles mediated a small fraction of the association with nerve fiber and ganglion cell layer thickness, while another dominated by omega-3 polyunsaturated fatty acids showed positive indirect effects on photoreceptor layer thickness—a plausible finding given that docosahexaenoic acid, the dominant long-chain fatty acid in photoreceptor outer-segment membranes, is essential for photoreceptor function.

The authors are appropriately cautious about these mediation results. Because the brain care score, the metabolites, and the retinal measurements were all captured at the same point in time, no temporal ordering can be established, and the indirect-effect estimates cannot confirm genuine biological causation. The principal components themselves may be sample-dependent, and the proportion of the total effect explained by any single metabolic pattern was modest—ranging from under 3 percent to about 11 percent. Still, the convergence on lipid and amino acid pathways is biologically coherent: excess branched-chain amino acids have been implicated in oxidative stress and inflammation in animal models of diabetic retinopathy, and lipoprotein subclass composition has repeatedly surfaced in studies of age-related macular degeneration.

The broader significance of the study lies in what it says about the retina as a sentinel organ. Thinner retinal nerve fiber and ganglion cell layers have previously been linked to cognitive decline, reduced brain volumes, Alzheimer’s disease, and incident dementia, and photoreceptor thinning has been associated with morbidity and mortality. By showing that a practical, modifiable brain health index tracks with both retinal structure and a machine-learned retinal age, the findings reinforce the idea that the same vascular and metabolic forces that shape brain aging also leave fingerprints in the eye—fingerprints that a routine OCT scan, already common in eye clinics, can capture.

Limitations remain, and the researchers enumerate them candidly. The observational design leaves room for residual confounding, the cohort’s healthier-than-average volunteers and predominantly European ancestry may limit generalizability, and the OCT-based age clock correlates only moderately with chronological age, meaning it captures just one facet of retinal aging. The brain care score itself is still a prototype awaiting systematic refinement. What the study delivers is a hypothesis-generating map: a demonstration that brain care and retinal health travel together, that specific metabolic pathways plausibly connect them, and that longitudinal studies with repeated measurements should now test whether improving one’s brain care score actually slows the thinning of the retina—and, by extension, perhaps the aging of the brain behind it.

Subject of Research: Associations between the brain care score and retinal layer thickness and retinal biological age in the UK Biobank

Article Title: Brain care score and retinal health: structural and metabolic insights from the UK Biobank

Article References: Yu, J., Zhang, Y., Gao, Y. L., Ho, M., Kam, K. W., Gong, B., Young, A. L., Pang, C. P., Tham, C. C., Yam, J. C., & Chen, L. J. (2026). Brain care score and retinal health: structural and metabolic insights from the UK Biobank. GeroScience. https://doi.org/10.1007/s11357-026-02579-z

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02579-z

Keywords: brain care score, retina, optical coherence tomography, UK Biobank, retinal biological age, metabolomics, macular thickness, neurodegeneration, lipid metabolism, omega-3 fatty acids, GeroScience, brain health

A team led by Adam Socrates of King’s College London and the Icahn School of Medicine at Mount Sinai, together with Jessie Baldwin of University College London, veteran psychosis researcher Robin Murray, Paul O’Reilly of Mount Sinai, and Jean-Baptiste Pingault of King’s College London, set out to test a deceptively simple question: if you measure someone’s inherited liability to psychopathology, does it predict their exposure to environmental risk? The answer, drawn from one of the largest and most systematic screenings ever attempted in this field, is a qualified but striking yes.

The researchers harnessed data from the UK Biobank, a vast biomedical database containing genetic and health information from roughly half a million British adults. Crucially, they restricted their sample to 307,493 participants who had no recorded diagnosis of schizophrenia, bipolar disorder, or major depressive disorder, and who were not taking antipsychotic medication. This design choice matters: by excluding people already diagnosed with serious mental illness, the team could ask whether genetic risk predicts environmental exposure before any disorder emerges, rather than simply detecting the downstream consequences of being ill.

On the genetic side, the investigators computed nine polygenic scores — numerical summaries of the small genetic variants scattered across a person’s genome that collectively contribute to a trait or condition. These scores covered attention deficit hyperactivity disorder, schizophrenia, bipolar disorder, major depression, neuroticism, educational attainment, and other genetically influenced characteristics. The scores were generated using PRS-CS, a sophisticated statistical method that leverages large-scale genome-wide association data to sharpen the predictive power of polygenic scores, improving on older approaches that count risk variants more crudely.

On the environmental side, the team assembled an extraordinary catalogue of 49 distinct risk factors, spanning five broad categories: victimisation and adverse life events, socioeconomic circumstances, behavioural and lifestyle factors, cognitive and educational measures, and perception-based or subjective factors such as how people appraise their own lives. Each polygenic score was then tested against each environmental factor in a grid of 441 standardised linear regression models, using HC1 robust standard errors — a statistical safeguard that protects against distortions caused by uneven variability in the data, a common problem in large biobank samples.

Because so many tests were run, the risk of false positives was substantial. To guard against this, the researchers applied false discovery rate correction within each polygenic score, a procedure that adjusts the statistical threshold so that the expected proportion of spurious findings stays controlled. Even under this conservative standard, 212 associations survived. That is a remarkable yield, and it suggests that the overlap between genetic liability and environmental exposure is not a statistical artefact but a pervasive feature of the data.

The breadth of the associations varied by genetic score. Polygenic scores for ADHD, educational attainment, major depression, schizophrenia, and bipolar disorder showed the widest-reaching profiles, correlating with environmental factors across multiple domains. Yet the pattern was not uniform: each score carried its own signature of environmental associations, hinting that different forms of inherited liability travel along different social and behavioural routes. A genetic predisposition toward ADHD, for example, may nudge individuals toward different life circumstances than a predisposition toward depression, even when both ultimately relate to elevated psychiatric risk.

One of the study’s most provocative findings emerged from a secondary analysis. When the team separated environmental factors into those involving perception or subjectivity — how people interpret and report their experiences — and those that are more objective, they found that seven perception-related factors showed stronger pooled associations with polygenic scores for schizophrenia, major depression, ADHD, bipolar disorder, and neuroticism than the 25 objective factors did. In other words, the genetic signal seemed to flow more strongly through the lens of subjective experience than through externally verifiable circumstances. Four conceptually tighter matched comparisons — pairing subjective and objective measures of the same underlying construct — produced a similar pattern, although it was not entirely consistent across all comparisons.

The authors are careful about interpretation, and the caveats deserve emphasis. The effects observed were small, and the study is observational: it establishes correlation, not causation. A polygenic score predicting environmental exposure does not mean genes directly cause adversity. Instead, the findings point to what behavioural geneticists call gene–environment correlation, the process by which inherited tendencies influence the situations people encounter. A person genetically inclined toward impulsivity may, for instance, drift into riskier social settings; a person with inherited cognitive tendencies may attain different levels of education and income; and inherited differences in temperament may colour how people perceive and report the events of their lives.

This mechanism has profound implications for psychiatric research. Many celebrated studies of environmental risk — childhood adversity, urban upbringing, socioeconomic deprivation — implicitly assume that these exposures are independent of genetic liability. If they are not, some portion of the apparent environmental effect may actually reflect inherited confounding, meaning that genetic risk inflates both the exposure and the outcome. The new findings suggest that measured environmental risk is partly correlated with inherited liability through behavioural, social, socioeconomic, cognitive, and perception-related pathways. Future studies of environmental effects on mental health, the work implies, should routinely adjust for or otherwise account for polygenic liability, or risk overstating purely environmental causes.

There is also a subtler lesson about measurement. The stronger links between genetic scores and subjective, perception-based factors raise the possibility that some of what researchers record as environmental exposure is filtered through the same psychological tendencies that genetics influence. Two people may live through objectively similar events yet encode them very differently, and those differences in appraisal are themselves partly heritable. That does not make subjective reports unreliable — how people perceive their lives is genuinely consequential for mental health — but it complicates the tidy division between what happens to us and what we are.

The study’s scale and rigour lend it unusual weight. Running 441 pre-specified models with robust standard errors, correcting for multiple testing, and probing results with matched comparisons reflects a level of methodological discipline that the field has often lacked. The use of a sample free of major psychiatric diagnoses strengthens the argument that genetic risk shapes environmental exposure in the general population, not merely among the ill. And the open-access publication means the full analysis is available for scrutiny and reuse by other researchers.

Still, the work is a beginning rather than an endpoint. The UK Biobank is a predominantly British, largely European-ancestry sample, and polygenic scores derived from European genome-wide studies lose accuracy in other populations, so the findings will need replication in more diverse cohorts. The environmental measures, though numerous, are self-reported and cross-sectional, limiting what can be said about the direction of effects over time. Longitudinal designs — following genetically characterised individuals from childhood — will be essential to disentangle whether genetic liability truly precedes environmental exposure, and through which specific pathways.

What the study delivers now is a conceptual correction with viral potential: the tidy story in which genes and environment are separate contributors to mental illness is wrong, or at least incomplete. Our inherited makeup quietly helps write the circumstances of our lives — the neighbourhoods, the stresses, the perceptions — that in turn feed back into mental health. Understanding that loop, rather than pretending it does not exist, may be the key to designing interventions that genuinely break the cycle of psychiatric risk.

Subject of Research: Gene–environment correlation between polygenic scores for psychopathology and environmental risk factors

Article Title: Genetic risk of psychopathology predicts environmental risk

Article References: Socrates, A., Baldwin, J. R., Murray, R. M., O’Reilly, P. F., & Pingault, J.-B. (2026). Genetic risk of psychopathology predicts environmental risk. Schizophrenia. https://doi.org/10.1038/s41537-026-00805-3

Image Credits: AI Generated

DOI: 10.1038/s41537-026-00805-3

Keywords: polygenic scores, gene-environment correlation, psychiatry, UK Biobank, schizophrenia, major depression, ADHD, bipolar disorder, neuroticism, environmental risk factors, behavioural genetics, psychosis

The study’s scale is unprecedented for this age group. The team, led by researchers at the University of Cambridge and the University of Bergen, analysed age-standardised and sex-standardised BMI at eleven timepoints, from six weeks to eight years of age, in up to 62,276 children from the Norwegian Mother, Father and Child Cohort Study (MoBa). Across those timepoints they identified 369 genome-wide significant associations, which collapsed into 152 independent signals — a dramatic increase over the 25 loci found in the largest previous study of measured childhood BMI. Ninety-four percent of the 116 loci newly identified for BMI before age five had never been reported before, and none of the signals showed evidence of sex-specific effects.

A striking feature of the data is how much the genetic architecture of BMI changes with age. SNP-based heritability estimates rose after birth to a peak around the ‘adiposity peak’ at age one, then declined toward the ‘adiposity rebound’ around age five to six, mirroring the natural trajectory of body fat in early life. Genetic correlations told a similar story: the correlation between BMI in infancy and BMI in later childhood was only modest, at 0.39, and the correlation between childhood and adult BMI was lower still, at 0.19. In other words, the genetic variants that make a baby chubby are largely not the same ones that determine adult body size.

To push discovery further, the researchers used a statistical technique called genomic structural equation modelling to fuse three childhood-related traits into a single ‘childhood adiposity’ factor: BMI at age eight in MoBa, recalled comparative body size at age ten from 444,345 UK Biobank participants, and a genome-wide study of age at menarche, a trait tightly linked to early-life adiposity. This yielded an effective sample size of 599,924 individuals and 526 additional independent signals. A polygenic score built from the combined 624 signals explained up to 9.2 percent of variation in childhood BMI at age nine in the independent Avon Longitudinal Study of Parents and Children — outperforming scores based on adult BMI for every timepoint below age five.

When the team clustered the 624 signals by how their effects unfolded across childhood, two clear trajectories emerged. A ‘Transient’ group of 115 variants influenced BMI only during infancy, peaking between six weeks and about twelve months, while a ‘Persistent’ group of 507 variants continued to act into later childhood. The distinction matters clinically: in a disease-wide analysis of UK Biobank data, transient signals did not raise the risk of any adult disease and appeared mildly protective against type 2 diabetes and hypertension, whereas persistent signals raised cardiometabolic risk in ways that vanished once adult BMI was accounted for. The message is that infant fat that later resolves carries no long-term penalty — and may even confer benefit.

Gene-mapping analyses linked 434 high-confidence genes to the signals, and among them were eleven components of the leptin–melanocortin pathway, the master neuroendocrine circuit of appetite regulation whose disruption causes severe early-onset obesity. Signals tied to BSX, GNAS, LEPR and PCSK1 showed childhood-specific effects, while incretin pathway genes — GLP1R and GIPR, the very receptors targeted by today’s blockbuster weight-loss drugs — also emerged with effects confined largely to childhood. Notably, none of the GLP1R signals was associated with BMI beyond early childhood, and one PCSK1 variant even showed opposite effects on BMI in infancy and adulthood. Because the authors found that rare variants in these same pathways act more strongly in children than in adults, they suggest that drugs targeting the leptin–melanocortin pathway, such as leptin and setmelanotide, may be more effective in children than in adults.

Single-nucleus RNA sequencing data from the human hypothalamus added a neurobiological dimension. Childhood adiposity signals were enriched in 260 of 452 mapped hypothalamic cell populations, all of them neuronal — and nineteen of these were enriched only for childhood, not adult, signals. Five of the child-specific populations sit in the arcuate nucleus, the hypothalamic hub of energy homeostasis, and another five in the mammillary bodies, a region better known for memory. These cell populations were marked by expression of leptin–melanocortin and incretin components such as LEPR, POMC, MC4R, GLP1R and GIPR, hinting at brain circuits that regulate body fat specifically during development.

Whole-genome sequencing of 479,615 UK Biobank participants then extended the search to rare variants. Burden tests implicated six genes — ADCY3, CALCR, MC4R, MRAP2, POMC and MYH13 — with rare protein-coding variants that showed stronger associations with recalled childhood adiposity than with adult BMI. Four of the six encode central components of the leptin–melanocortin pathway, and the associations for ADCY3 and MRAP2 provided the first population-scale evidence that single copies of damaging variants in these severe-obesity genes measurably shift childhood body size. A novel association at MYH13, a muscle gene, affected childhood adiposity but showed no link to adult BMI at all.

The Norwegian cohort’s parent–child trios also allowed the team to probe genomic imprinting — the parent-of-origin silencing of certain genes. Evidence of imprinting emerged at signals near known imprinted regions, including a maternal effect at KLF14 on infancy BMI that reverses direction in adulthood, a childhood-specific paternal effect at GNAS, and a novel paternal-only association near ZDBF2, a gene that in mice regulates neonatal feeding and growth. The authors argue that these findings extend the classic parental-conflict hypothesis of imprinting, which was developed for fetal growth and puberty timing, to the regulation of childhood adiposity itself.

The study has limitations the authors acknowledge: the cohorts were of northern European ancestry, BMI in infancy is an imperfect proxy for fat mass, and the hypothalamic cell atlas was built from adult autopsy tissue. Larger, more diverse studies with directly measured infant body composition will be needed. Even so, the conclusion is hard to escape. With 624 signals — comparable to the 941 found for adult BMI in more than 700,000 people — childhood adiposity genetics proved at least as informative as its adult counterpart, and far more revealing about core energy-balance biology. As the authors put it, childhood appears to be a more sensitive window for detecting variation in the key endocrine and neuropeptide pathways that regulate how our bodies store fat, a finding that could reshape both obesity drug development and early-life prevention strategies.

Subject of Research: Genome-wide association and sequencing analysis of common and rare genetic variants influencing adiposity across childhood

Article Title: Genome-wide mapping of common and rare variant effects on adiposity across childhood

Article References: Kentistou, K. A., Sundfjord, J., Karimi, R., Kaisinger, L. R., Hofmeister, R. J., Fragoso-Bargas, N., Lupu, A. E., Zhao, Y., Tadross, J. A., Steuernagel, L., Dowsett, G. K. C., Lockhart, S., Brüning, J. C., Liu, J., Cortes, A., Lo, Y., Davitte, J., Clement, L., Havdahl, A., … Johansson, S. (2026). Genome-wide mapping of common and rare variant effects on adiposity across childhood. Nature Genetics. https://doi.org/10.1038/s41588-026-02772-y

Image Credits: AI Generated

DOI: 10.1038/s41588-026-02772-y

Keywords: childhood obesity, genome-wide association study, adiposity, leptin-melanocortin pathway, incretin signaling, hypothalamus, rare variants, genomic imprinting, polygenic scores, MoBa cohort, UK Biobank, BMI trajectories

The core challenge any model of genetic variation must confront is linkage disequilibrium, the nonrandom association of variants across the genome. Neighboring SNPs, or single nucleotide polymorphisms, tend to be inherited together, but so do many distant ones, and these long-range correlations carry much of the structure that makes a population recognizable in the data. Classical simulators built on the coalescent model capture these patterns by tracing observed variation back to latent genealogies shaped by demographic history, mutation, and recombination. That approach is expressive but computationally punishing, which is why practical tools such as msprime rely on Markovian approximations. A second tradition, descending from the product-of-approximate-conditionals framework of Li and Stephens, dispenses with explicit genealogies and fits the data distribution directly, naturally yielding hidden Markov models. HMMs have been enormously successful in haplotype phasing, genotype imputation, and ancestry inference, but their chain structure forces information between distant SNPs to pass through every intermediate hidden variable, weakening long-range dependencies at each step.

Deep learning promised a way out, and a wave of generative adversarial networks, variational autoencoders, restricted Boltzmann machines, and most recently diffusion models has been applied to genetic data. These models can produce artificial genomes that look plausible in principal component analyses and allele frequency summaries. Yet the UCLA-led team argues they carry structural liabilities for genomics. Generative adversarial networks do not define a probability distribution over the data at all, making likelihood-based inference impossible. Restricted Boltzmann machines define one, but evaluating it requires computing a partition function over an exponentially large space of configurations. Variational autoencoders expose only a lower bound on the marginal likelihood. Diffusion models can in principle compute exact likelihoods but scale poorly to dense SNP data and have not been evaluated on imputation. Crucially, none of these approaches supports efficient estimation of conditional probabilities, the mathematical operation at the heart of genotype imputation, and none offers an objective way to judge whether training has converged.

GPC’s answer is architectural. The model builds on hidden Chow-Liu trees, a class of latent variable models in which every observed SNP is paired with a hidden variable, and the hidden variables are wired together not in a chain but in a tree learned from the data itself. The Chow-Liu algorithm, dating to 1968, computes pairwise mutual information between all variables and connects them in the maximum-weight spanning tree, placing strongly correlated SNPs adjacent to one another regardless of their positions along the chromosome. Where a hidden Markov model would route the dependence between two distant but correlated SNPs through every latent variable in between, losing correlation at each hop, the learned tree carries it over a single edge. In the tree learned from a 10,000-SNP region of chromosome 15 in the 1000 Genomes dataset, the researchers found 4,599 leaf nodes, 397 nodes with degree greater than two, one hub connected to 161 other SNPs, and 18.3 percent of edges spanning more than 1,000 positions apart, a topology fundamentally unlike the unbranched chain of an HMM.

Expressiveness alone would be useless if the model were computationally intractable, and this is where probabilistic circuits enter. A probabilistic circuit represents a joint distribution as a directed acyclic graph of input, sum, and product nodes, with product nodes encoding factorizations and sum nodes encoding weighted mixtures. When the graph satisfies two structural properties, smoothness and decomposability, any marginal probability can be computed exactly in time linear in the circuit’s size. Conditional probabilities, the quantities needed for imputation, follow as simple ratios of two marginal queries, each evaluated in a single feedforward pass. Sampling artificial genomes proceeds by ancestral traversal of the circuit, also linear in size. By compiling hidden Chow-Liu trees into this circuit representation and training them with expectation-maximization on GPUs using the PyJuice package, the team trained models with more than 88 million parameters, with each EM epoch taking under two seconds on the 1000 Genomes data and full training completing in roughly two to six hours on a single NVIDIA RTX A5000 graphics card.

The practical payoff shows up most clearly in genotype imputation, the task of inferring variants a person was not directly genotyped, which underlies much of genome-wide association science. The researchers evaluated GPC across three settings using the 1000 Genomes Project, the UK Biobank, and a high-coverage 1000 Genomes release. In the general setting, where models trained on diverse ancestries impute into similarly diverse test genomes, GPC used for direct conditional imputation achieved a 27.5 percent average improvement in the squared correlation between imputed and true genotypes over the next best method, a restricted Boltzmann machine, and a striking 174 percent improvement for low-frequency variants with minor allele frequency below one percent. Direct imputation through the model itself also beat GPC’s own artificial genomes used as reference panels for the standard tool Impute5, by about 15.5 percent overall, apparently because it targets the imputation objective directly rather than injecting the noise of an intermediate simulation step.

The gains were largest precisely where existing infrastructure is weakest: populations underrepresented in public reference panels. Because large public datasets are overwhelmingly of European ancestry, imputation into African, admixed, or other non-European populations suffers from distributional mismatch. In population-specific experiments, GPC’s direct imputation improved on the next best deep generative method by 33 percent on average, and by 279 percent for low-frequency variants. Remarkably, in the 1000 Genomes experiments GPC trained on ancestry-matched private data outperformed Impute5 running on real European reference genomes, achieving a 12.3 percent average improvement in squared correlation and a 42.1 percent improvement for rare variants. In a realistic array-based scenario, imputing 12,551 SNPs missing from the HumanOmni5Exome genotyping array, GPC outperformed the European-panel Impute5 benchmark by 96.5 percent on average and by more than twelvefold for low-frequency variants. The entire held-out array imputation took about one second in a single conditional query.

Privacy, the original motivation for artificial genomes, also fared better under GPC. Using the nearest-neighbor adversarial accuracy metric, which scores values near 0.5 as the ideal balance between utility and privacy, GPC came closest to the ideal across nearly all splits of both datasets. The failure modes of the competitors were instructive. Restricted Boltzmann machines produced synthetic haplotypes that each sat closer to some individual real genome than to any other synthetic one, meaning samples clustered around training individuals, a memorization pattern that could allow synthetic genomes to be traced back to specific people. The Wasserstein generative adversarial network showed the opposite pathology, with artificial genomes occupying a region largely disjoint from the real data, sacrificing utility. The hidden Markov model failed in the most extreme form, producing genomes perfectly separable from real ones in both directions, achieving maximal privacy by failing to model the data at all. GPC’s artificial genomes were not perfectly indistinguishable, and the authors caution that improved performance on this metric does not guarantee protection against membership inference or attribute inference attacks.

Limitations remain, and the authors are candid about them. GPC currently operates on single genomic regions of roughly 10,000 to 15,000 SNPs, bounded by the memory cost of constructing the Chow-Liu tree, which grows with the square of the SNP count; reaching genome-wide scale will require hierarchical or distributed approaches. The present implementation handles haploid rather than diploid genomes, and like all generative models GPC inherits biases from its training data, leaving fairness across diverse cohorts an open problem. Still, the study marks a notable convergence of two research traditions that have mostly run in parallel: the tractable probabilistic models beloved of population geneticists and the expressive deep architectures driving modern machine learning. By learning the dependence structure of the genome directly from mutual information and then compiling it into a form where exact inference is cheap, GPC suggests that the trade-off between realism, computational tractability, and privacy in synthetic genomic data may be less inevitable than it once appeared. Code and experiments are publicly available as the field moves toward equitable genomic tools for populations long left out of reference panels.

Subject of Research: A tractable deep generative model, Genetic Probabilistic Circuits, for modeling human genetic variation data

Article Title: GPC: An expressive and tractable deep generative model for genetic variation data

Article References: Anand, P., Liu, A., Dang, M., Fu, B., Wei, X., Van den Broeck, G., & Sankararaman, S. (2026). GPC: An expressive and tractable deep generative model for genetic variation data. PLOS Genetics, 22(10), e1012321. https://doi.org/10.1371/journal.pgen.1012321

Image Credits: AI Generated

DOI: 10.1371/journal.pgen.1012321

Keywords: genetic probabilistic circuits, artificial genomes, genotype imputation, linkage disequilibrium, probabilistic circuits, hidden Chow-Liu trees, population genetics, deep generative models, genomic privacy, 1000 Genomes Project, UK Biobank, underrepresented populations

The team set out to quantify, systematically and at genome-wide scale, how well the two dominant study designs converge. Cross-sectional designs compare genetic associations between individuals of different ages at a single time point, asking whether a variant’s effect differs between a 45-year-old and a 65-year-old sampled simultaneously. Longitudinal designs instead track the same individuals over time, modeling within-person change directly. In principle, if aging is the only force at work, both approaches should estimate the same quantity. In practice, they rest on strong assumptions: that genetic effects do not vary by birth cohort, that sampling is comparable, and that participation in the study does not itself depend on age and phenotype.

Using the UK Biobank, the researchers analyzed 31 non-binary health-related phenotypes spanning cognition, physiology, clinical indicators, anthropometrics and lifestyle behaviors. The cross-sectional arm drew on up to 498,845 participants aged 40 to 69 at baseline; the longitudinal arm followed up to 99,459 of them across repeat assessments conducted between 2012 and 2024, with a mean follow-up of roughly 11 years. At the phenotypic level, cross-sectional age effects explained about 70.6 percent of the variance in longitudinal estimates—substantial, but far from the near-perfect agreement that equivalence would demand. The largest discrepancies appeared in behavioral traits. For the number of medications taken, the cross-sectional model estimated an age effect three times larger than the longitudinal one. For smoking and alcohol use, the two designs even disagreed about the sign: traits that decline with age within individuals falsely appeared to increase with age when compared across individuals.

The culprit, the authors show, is confounding by birth cohort. In cross-sectional data, age and year of birth are almost perfectly inversely correlated—age equals period minus cohort—so differences between generations can masquerade as aging. Because date of birth and cross-sectional age correlated at roughly −0.99 in the UK Biobank, negative cohort effects systematically inflate cross-sectional age estimates and can even reverse their direction. Significant cohort effects emerged for 28 of the 31 traits, and cohort confounding accounted for a striking 90.8 percent of the variance in the discrepancy between longitudinal and cross-sectional age estimates. The historical and social contexts in which people were born—educational reforms, economic upheavals, public health policies such as the 1965 UK ban on cigarette advertising—leave lasting marks on trait levels that a cross-sectional comparison misreads as aging.

The researchers then turned to the genetic level, implementing a two-step genome-wide strategy that combined scans for marginal genetic effects, cross-sectional gene-by-age interactions and genetic effects on longitudinal change. They identified 57 linkage-disequilibrium-independent variants with significant age-varying effects, most of them detected in cross-sectional analyses (44 variants, in up to 406,226 individuals) and fewer in longitudinal analyses (14 variants, in up to 83,579 individuals). The variants clustered around anthropometric traits such as weight, body fat-free mass and body mass index, metabolic indicators such as basal metabolic rate, and clinical outcomes including medication count and number of cancers. Self-reported health behaviors—physical activity, smoking, sleep—showed few or no significant age-varying genetic effects.

Encouragingly, direction was largely preserved across designs: 84.21 percent of the 57 variants showed consistent interaction directions whether tested cross-sectionally or longitudinally. Among the concordant variants, both attenuation and intensification of genetic effects with age were common, accounting for 50 percent and 35.42 percent of cases respectively. The pattern was trait-specific. Genetic effects on obesogenic traits tended to weaken with age, consistent with earlier reports of declining heritability for body mass index and depression in adulthood—plausibly because accumulated environmental exposures, medication, dietary change and shifting occupational circumstances gradually swamp genetic predisposition. Conversely, traits adversely affected by aging itself, such as cancer count, medication burden and reaction time, more often showed intensifying genetic influence over time. Crossover effects, in which the direction of a genetic association reverses across the lifespan, were rare, appearing in just seven variants with no detectable marginal effect.

Magnitude was another matter. Although cross-sectional and longitudinal estimates of age-varying genetic effects were linearly related, the regression slope was significantly below one, indicating that cross-sectional estimates were systematically larger. Decomposing the discrepancies revealed a familiar hierarchy: gene-by-birth-year interactions explained 70.8 percent of the variance in effect-size differences across variants, selective participation accounted for an additional 11.6 percent, and unmodeled nonlinear age trajectories contributed only 4.2 percent. For several variants—including rs2597355 on depression, rs56299829 on fruit intake and rs17362578 on walking pace—cohort confounding was strong enough to reverse the estimated direction of the age-varying genetic effect under a cross-sectional model. Suggestive gene-by-cohort effects were observed for 10 of the 57 variants, and the authors note that such effects are not mere noise: they capture how societal change reshapes genetic associations across generations, as documented in studies of genetic influences on social outcomes before and after the collapse of the Soviet Union and in response to educational policy reforms.

Participation bias, the second-largest contributor, operates differently in each design. Cross-sectional estimates are vulnerable to selective volunteering at baseline—the UK Biobank’s well-documented healthy volunteer effect—whereas longitudinal models adjust for time-invariant selection but assume that continued participation is independent of changes in phenotype. Because the pressures shaping initial recruitment differ from those governing retention, the two designs absorb different slices of selection bias. The scale of loss to follow-up is considerable: of the 399,661 baseline participants who did not take part in follow-up research, 42,001—about 11 percent—died before the first longitudinal assessment, though most attrition stemmed from other causes. Reweighting the samples to improve representativeness shifted cross-sectional estimates more than longitudinal ones, with sign reversals observed for four traits, and sample representativeness explained 2.8 percent of the discrepancy between designs—small next to the 90.8 percent attributable to cohort effects at the phenotypic level.

Nonlinear aging, by contrast, proved a minor player. Significant quadratic age effects were detected for 17 traits, meaning some aging trajectories genuinely bend rather than run straight, and such curvature can make estimated slopes depend on the age range sampled—the cross-sectional sample tops out at 69 years of baseline age while follow-up reaches 86. Yet deviations from linearity explained under 5 percent of the variance in design discrepancies, and the authors caution that their small-subset estimates are diluted by measurement error, with a dilution ratio of −0.43 for the nonlinear component. The findings replicated at the polygenic level: polygenic score effects on anthropometric and metabolic traits showed consistent directions but divergent magnitudes across designs, with cohort confounding again dominant at 85.2 percent of explained variance—a result with practical implications for anyone deploying polygenic risk scores across age groups or generations.

Where does this leave the field? The authors argue that neither design is immune to its own pathologies, and that robust inference demands integrating both. Cross-sectional analyses offer unmatched statistical power and age range, maximizing discovery of candidate age-varying variants; longitudinal analyses provide direct modeling of within-person change, control against time-invariant confounding and the ability to detect gene-by-cohort effects and nonlinear trajectories that cross-sectional data cannot in principle deliver. The team also illustrates how longitudinally derived genetic instruments could feed into Mendelian randomization, recovering the well-established adverse effect of body mass index on systolic blood pressure, albeit with wide uncertainty given the scarcity of strong instruments for change. As prospective biobanks, electronic health records and whole-genome sequencing expand, the framework will extend to rare variants and richer repeated measures. For now, the message is clear: before declaring that a gene’s influence grows or fades with age, researchers must ask whether they are measuring aging—or merely the echo of the century in which their participants were born.

Subject of Research: Age-varying genetic effects on complex traits and how study design, cohort confounding and participation bias shape their inference

Article Title: Design and model choices shape inference of age-varying genetic effects on complex traits

Article References: Schoeler, T., Wiegrebe, S., Winkler, T. W., & Kutalik, Z. (2026). Design and model choices shape inference of age-varying genetic effects on complex traits. Nature Aging. https://doi.org/10.1038/s43587-026-01232-w

Image Credits: AI Generated

DOI: 10.1038/s43587-026-01232-w

Keywords: genetic epidemiology, age-varying genetic effects, UK Biobank, birth cohort effects, participation bias, genome-wide association study, gene-by-age interaction, longitudinal analysis, cross-sectional design, polygenic scores, Mendelian randomization, complex traits

The cardio-renal-metabolic connection has become one of the defining frameworks of modern chronic disease medicine. Type 2 diabetes, cardiovascular disease and chronic kidney disease are not isolated conditions; they share overlapping risk factors, amplify one another pathophysiologically, and frequently cluster within the same patient. Epidemiological evidence has long shown that people who accumulate multiple cardiometabolic conditions face mortality risks that far exceed the sum of their parts. What has been harder to establish is which measurable traits, captured early and cheaply in a clinical setting, can flag individuals at risk of entering and then progressing through this disease cascade before irreversible damage is done.

Enter the body roundness index. Proposed in 2013 by Diana Thomas and colleagues from a geometric model of the human body, BRI estimates how closely a person’s body shape approximates a circle or an ellipse, using waist circumference and height as inputs. Unlike body mass index, which simply divides weight by height squared and cannot distinguish muscle from fat, BRI is designed to capture central adiposity, the visceral and ectopic fat that wraps around internal organs and drives metabolic dysfunction. A growing body of literature has linked elevated BRI to metabolic syndrome, incident type 2 diabetes, chronic kidney disease, cardiovascular events and all-cause mortality in populations ranging from the United States to China and Japan. What remained unclear was whether BRI could predict not just the onset of a single disease but the full longitudinal progression through the cardio-renal-metabolic continuum.

To answer that question, a team led by researchers at Fudan University in Shanghai analyzed 442,489 UK Biobank participants aged 40 and older who were free of cardio-renal-metabolic diseases at baseline. The investigators examined BRI both as a continuous variable and in tertiles, allowing them to compare people in the highest third of body roundness against those in the lowest. The outcomes of interest formed a natural disease pathway: the first cardio-renal-metabolic disease, abbreviated FCRMD; progression to cardio-renal-metabolic multimorbidity, or CRMM, meaning the accumulation of two or more such conditions; and ultimately all-cause mortality. Hazard ratios were estimated with Cox proportional hazards models, while population attributable fractions quantified how much disease burden could theoretically be traced back to elevated BRI.

The magnitude of the associations was striking. In the highest BRI tertile, the hazard ratio for developing type 2 diabetes reached 6.94, with a 95 percent confidence interval of 6.57 to 7.33, meaning that people with the roundest body shapes were nearly seven times more likely to develop diabetes than those in the lowest tertile. The hazard ratio for progressing to cardio-renal-metabolic multimorbidity was 3.25. These are effect sizes rarely seen for a single anthropometric measurement, and they place BRI in the same conversation as far more elaborate biomarkers that require blood draws, imaging or laboratory infrastructure.

The multistate analyses, however, are what set this study apart from earlier work. Rather than treating disease onset and death as isolated endpoints, multistate models treat the disease course as a network of transitions: from healthy to first disease, from first disease to multimorbidity, and from any state to death. This approach matters because it avoids a well-known statistical pitfall in obesity research. Conditioning on the presence of disease can introduce bias, sometimes producing the misleading impression that obesity protects against mortality in sick patients, the so-called obesity paradox. By modeling transitions explicitly, the researchers could estimate the risk that elevated BRI confers at each step of the disease journey independently.

Those transition-specific results were consistent and sobering. People in the highest BRI tertile had a 2.24-fold higher hazard of transitioning from a healthy baseline to a first cardio-renal-metabolic disease, with a population attributable fraction of 26.22 percent, suggesting that roughly a quarter of first disease events in the population could be attributed to high body roundness. Among those who had already developed one condition, high BRI conferred a 1.58-fold higher hazard of progressing to multimorbidity, with an attributable fraction of 21.21 percent. Even the transition from baseline directly to death showed a modest but statistically significant elevation, with a hazard ratio of 1.05 and an attributable fraction of 2.06 percent. The pattern indicates that BRI exerts its influence primarily by launching and accelerating the disease process rather than by killing directly, a distinction with important implications for where intervention can help most.

The study also examined how body roundness interacts with genetic predisposition and lifestyle, two forces that shape cardiometabolic risk in different ways. Participants with elevated BRI who also carried high genetic risk for cardiometabolic disease, or who adhered to unhealthy lifestyles, faced a greater risk of both disease onset and progression along the continuum than could be explained by either factor alone. This finding underscores a theme that has emerged repeatedly in the polygenic risk score literature: genetic risk scores carry substantial uncertainty at the individual level, and their predictive power improves when combined with measurable physical traits and behavioral data. A tape measure and a questionnaire, in other words, can sharpen what a genome cannot resolve on its own.

Biologically, the link between central adiposity and the cardio-renal-metabolic cascade is well supported. Visceral fat is not a passive storage depot but a metabolically active endocrine organ that secretes inflammatory cytokines, promotes insulin resistance, and generates oxidative stress. Adipocyte dysfunction drives chronic low-grade inflammation, a process that intersects with the aging biology studied under the banner of geroscience, including inflammaging and cellular senescence. These mechanisms provide a plausible pathway by which a rounder body shape translates into damaged blood vessels, failing kidneys and deranged glucose metabolism, and they explain why the associations in this study were strongest for type 2 diabetes, the condition most directly governed by adipose tissue physiology.

The practical implications are considerable. BRI requires only a measuring tape and a formula, making it deployable in primary care settings, community screening programs and low-resource environments where dual-energy X-ray absorptiometry or magnetic resonance imaging of visceral fat are unavailable. The population attributable fractions reported in the study suggest that reducing elevated body roundness could prevent a meaningful share of first cardio-renal-metabolic events and of progression to multimorbidity, a burden that health systems worldwide are increasingly struggling to carry as populations age. Because the strongest effects appeared at the earliest transition, from health to first disease, the findings argue for using BRI as an early-warning indicator rather than waiting for laboratory abnormalities to appear.

Caveats remain, as they do in any observational study. The UK Biobank cohort, while enormous, is not fully representative of the broader population, and waist circumference measurements capture body shape imperfectly. Residual confounding by factors not measured at baseline cannot be excluded, and BRI is a static snapshot in an analysis that did not model changes in body shape over time. Yet the consistency of the results across continuous and categorical analyses, the use of transition-specific modeling, and the sheer scale of the cohort give the findings considerable weight. For clinicians and public health planners, the message is straightforward: how round a body is, measured cheaply and quickly, tells us a great deal about how the story of chronic disease is likely to unfold, and it tells us early enough to change the ending.

Subject of Research: Association between body roundness index and the progression of cardio-renal-metabolic diseases

Article Title: Association between body roundness index and progression of cardio-renal-metabolic diseases using multistate models

Article References: Association between body roundness index and progression of cardio-renal-metabolic diseases using multistate models. (n.d.). https://doi.org/10.1007/s11357-026-02512-4

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02512-4

Keywords: body roundness index, cardio-renal-metabolic diseases, multimorbidity, UK Biobank, multistate models, obesity, type 2 diabetes, visceral fat, genetic risk, lifestyle, mortality, GeroScience

The research team, led by Kevin Y. H. Liang, Thomas M. Zheng, Dandan Tan and colleagues, including senior author J. Brent Richards, employed a technique known as two-sample mendelian randomization. This method exploits the random allocation of genetic variants at conception, which act much like the assignment groups in a randomized clinical trial. Because genetic variants inherited from parents are fixed at birth and are not confounded by later lifestyle factors or reverse causation, they can be used to test whether an exposure, such as a genetic predisposition to Type 2 diabetes, genuinely causes a change in an outcome, such as kidney function. The approach is particularly valuable in fields like diabetes research, where observational associations are notoriously vulnerable to confounding.

In the first stage of the study, the investigators asked a deceptively simple question: which of the thousands of proteins circulating in human blood are changed by a genetic predisposition to Type 2 diabetes? To answer it, they drew on a proteomic genome-wide association study from the Icelandic biobank deCODE, which measured blood protein levels in 35,559 individuals, and combined it with a diabetes genome-wide association study encompassing 80,154 cases of Type 2 diabetes. Running mendelian randomization across the proteome, they identified 71 circulating proteins whose blood levels appear to be shifted by genetic liability to the disease. These proteins, in other words, form a candidate pool of molecular intermediaries that could plausibly carry the damaging signal of diabetes onward to the kidneys.

Identifying proteins that diabetes influences was only half the task. The team next needed to determine which of these diabetes-responsive proteins actually affect kidney health. For this, they used a second application of mendelian randomization, this time relying on cis-acting genetic variants, variants located near the gene encoding each protein that specifically regulate that protein’s circulating level. Such variants provide a cleaner proxy for the protein’s causal effect because they are less likely to influence the kidneys through unrelated pathways. Testing the candidate proteins against three kidney traits, blood urea nitrogen, estimated glomerular filtration rate, and diagnosed chronic kidney disease, using genetic data from up to 1,004,040 participants, the researchers converged on five proteins with credible causal roles: INHBC, GNPTG, LPO, AGRN and CTSD.

One protein stood out from the pack. INHBC, a circulating member of the inhibin-beta family with known ties to inflammatory and fibrotic signaling, showed a striking pattern: higher genetically influenced levels of the protein were estimated to lead to a lower estimated glomerular filtration rate, the standard measure of how well the kidneys filter blood, and to higher blood urea nitrogen, a marker of accumulating waste products that failing kidneys cannot clear. In practical terms, the analysis suggested that elevated INHBC pushes kidney function in the wrong direction on both fronts, making it the most compelling single mediator to emerge from the screen.

A crucial strength of the study lies in its replication strategy. Proteomic measurements are typically made on antibody-based platforms that can differ substantially between laboratories, and any single cohort may harbor idiosyncrasies that distort results. To guard against these artifacts, the team re-ran their mendelian randomization analysis for INHBC using proteomic genome-wide association data from four additional cohorts: the UK Biobank Pharma Proteomics Project, the Fenland study, the Atherosclerosis Risk in Communities study, and EPIC-Norfolk. Across all four independent datasets, the direction of effect was consistent, a result the authors interpret as evidence that their findings are robust to both platform differences and cohort variation. Such cross-cohort concordance is far from guaranteed in proteomic genetics and lends considerable weight to the central claim.

The genetic evidence was then complemented by a more traditional observational analysis. In 37,854 participants from the UK Biobank, individuals with higher measured circulating levels of INHBC faced an increased hazard of receiving a kidney disease diagnosis over follow-up. While observational associations of this kind cannot by themselves establish causation, their alignment with the genetic results creates a coherent picture: the same protein that genetic instruments implicate as a causal driver of reduced kidney function is also elevated in people who go on to develop kidney disease in the real world.

Perhaps the most consequential number in the study is a modest one. Using mediation analysis that combines the genetic estimates, the researchers calculated that circulating INHBC levels mediate approximately 1.3 percent of the association between Type 2 diabetes and kidney disease diagnosis, with a 95 percent confidence interval spanning 0.85 to 1.9 percent. On its face, 1.3 percent may sound underwhelming, and the authors are careful not to oversell it. Yet in a disease as widespread as diabetes, which affects hundreds of millions of people globally, even a small fractional contribution can translate into a substantial absolute burden of kidney disease. More importantly, the result establishes a proof of principle that specific, druggable-class molecules can be pinpointed as mediators of diabetic organ damage using human genetics alone, opening a template for discovering the remaining pathways that account for the other 98.7 percent.

The authors are equally candid about the limitations of their work. Although they observed limited evidence for violations of the mendelian randomization assumptions, some of those assumptions, such as the absence of pleiotropic pathways connecting the genetic instruments to kidney outcomes, are fundamentally untestable and can never be fully excluded. Furthermore, the study was not conducted in individuals with confirmed diabetic kidney disease; instead, it relied on independent population-based studies that assessed diabetes and kidney function separately. This distinction matters, because the biology of diabetic kidney disease may involve tissue-specific processes within the kidney itself that circulating protein levels only partially reflect. The authors emphasize that additional functional analyses in disease-specific cohorts will be needed before the findings can be translated into clinical practice.

Even with those caveats, the study arrives at a moment of genuine unmet need. Few interventions currently prevent chronic kidney disease in people living with diabetes beyond tight glycemic control, blood pressure management and newer agents such as SGLT2 inhibitors, and many patients progress to kidney failure despite the best available care. By nominating INHBC and four other proteins as causal candidates, the research provides a shortlist for laboratory follow-up, drug development and biomarker work. If future studies confirm that INHBC drives fibrotic or inflammatory injury in diabetic kidneys, therapies aimed at lowering its circulating levels could one day join the arsenal against diabetic kidney disease. For now, the study stands as a vivid demonstration of how human genetics, massive biobanks and proteomic technology can be woven together to illuminate the hidden molecular bridges between two of the world’s most common diseases.

Subject of Research: Protein mediators of chronic kidney disease in Type 2 diabetes identified through mendelian randomization

Article Title: Protein mediators of chronic kidney disease in Type 2 diabetes: A mendelian randomization study

Article References: Liang, K. Y. H., Zheng, T. M., Tan, D., Sasako, T., Ilboudo, Y., Chen, Y., Butler-Laporte, G., Yoshiji, S., & Richards, J. B. (2026). Protein mediators of chronic kidney disease in Type 2 diabetes: A mendelian randomization study. PLOS Medicine, 23(9), e1004802. https://doi.org/10.1371/journal.pmed.1004802

Image Credits: AI Generated

DOI: 10.1371/journal.pmed.1004802

Keywords: Type 2 diabetes, chronic kidney disease, mendelian randomization, proteomics, INHBC, glomerular filtration rate, UK Biobank, genetic epidemiology, kidney failure, biomarkers, PLOS Medicine, causal inference

The research, published in BMC Medicine by a team led by Sixian Liu and Jieyun Yin of Soochow University together with colleagues at Sichuan University, mined the UK Biobank, one of the world’s richest biomedical databases. The team defined the starting point of their disease model as either a first live birth or the diagnosis of an adverse pregnancy outcome, whichever came first. From there, they tracked women through a sequence of health states: the appearance of a first cardio-renal-metabolic disease, the accumulation of a second condition to reach multimorbidity, and ultimately death. This staged framing matters because it treats chronic disease not as a single event but as a process with distinct, measurable transitions.

Statistically, the investigators relied on multi-state models, a class of survival-analysis techniques that estimate the hazard — the instantaneous risk — of moving from one health state to the next, while adjusting for confounders such as age, socioeconomic status and other baseline characteristics. The approach allowed them to quantify not only whether adverse pregnancy outcomes raise risk, but at which point in the disease cascade that risk is concentrated. It also enabled them to test for interaction, asking whether the effect of a complicated pregnancy depends on what a woman does afterwards in terms of diet, smoking, alcohol, physical activity and body shape.

The headline numbers are sobering. Women with a history of adverse pregnancy outcomes progressed from a healthy initial state to a first cardio-renal-metabolic disease at a rate of 19.84 percent, compared with 15.56 percent among women without such a history — and the adjusted hazards ratio of 3.32 means their risk of making that transition was more than three times higher. That first transition proved to be the steepest part of the curve. Once a first disease had appeared, the excess risk of progressing to full multimorbidity persisted but moderated, with a hazards ratio of 1.41, and the risk of dying after multimorbidity was elevated at 1.65.

The mortality findings add a darker dimension. Women with adverse pregnancy outcomes faced a higher hazard of dying even before any cardio-renal-metabolic diagnosis, with an adjusted hazards ratio of 1.46 for the transition from the initial state directly to death, and a ratio of 1.28 for death after a first disease, though the latter estimate carried wider statistical uncertainty. Across nearly every stage of the model, the progression percentages were higher in the adverse-outcomes group, painting a consistent picture of a population whose pregnancies left a durable physiological signature.

What elevates the study beyond a simple risk tally is its treatment of lifestyle as a modifying force rather than a background variable. The researchers found a significant interaction between adverse pregnancy outcomes and high-risk lifestyle factors, with alcohol consumption, insufficient physical activity and unhealthy body shape emerging as the pivotal players during the transition from the initial state to a first cardio-renal-metabolic disease. In other words, the damage associated with a complicated pregnancy was not a fixed sentence; it interacted with behaviors that women and their clinicians can actually change.

The biological plausibility of such a long shadow is grounded in vascular and metabolic physiology. Hypertensive disorders of pregnancy are thought to reveal, and possibly worsen, underlying vulnerabilities in endothelial function — the health of the thin cell layer lining blood vessels — while gestational diabetes signals impaired glucose regulation that may predate the pregnancy itself. Pregnancy thus functions as a natural stress test for the cardiovascular, renal and metabolic systems. A woman who fails that test may carry forward subtle deficits in insulin sensitivity, blood pressure regulation and kidney filtration that silently accumulate until they cross diagnostic thresholds years or decades later.

The multi-state prediction component of the study carries practical weight. By combining adverse pregnancy outcome status with the number of lifestyle risk factors, the model showed that the probability of disease progression increased as risks stacked up. This has direct implications for how medicine organizes care. Obstetric histories are frequently recorded and then forgotten once a patient transitions to primary or specialty adult care, yet the findings argue that a history of gestational diabetes or preeclampsia should function as a lifelong flag — a reason for earlier screening, tighter monitoring of blood pressure and glucose, and more aggressive counseling on alcohol, activity and weight.

There are, of course, limits to what an observational cohort can prove. The UK Biobank population is not fully representative, and lifestyle factors were measured at baseline rather than tracked continuously across four decades, so changes in behavior over time are imperfectly captured. Residual confounding can never be excluded, and the interaction analyses, while statistically significant, describe associations rather than guaranteed causal pathways. Still, the sheer size of the cohort, the length of follow-up and the consistency of the hazard estimates across transitions give the findings considerable weight.

The broader message is one of opportunity. Cardio-renal-metabolic multimorbidity is among the most burdensome and expensive patterns of chronic disease in aging populations, and it rarely arrives unannounced. This study suggests that for millions of women, the first warning sign appears not in middle age but in a delivery room — in a blood pressure reading or a glucose tolerance test. Treating adverse pregnancy outcomes as the beginning of a long-term prevention program, rather than the end of a pregnancy complication, could shift the entire trajectory of women’s cardiovascular, kidney and metabolic health, and the lifestyle levers identified here offer a concrete place to start.

Subject of Research: The association between adverse pregnancy outcomes, lifestyle factors and long-term cardio-renal-metabolic multimorbidity risk in women

Article Title: Association of adverse pregnancy outcomes history and lifestyle factors with cardio-renal-metabolic multimorbidity: a prospective cohort study

Article References: Liu, S., Long, L., Peng, Y., Guo, N., Gu, Y., Yang, M., & Yin, J. (2026). Association of adverse pregnancy outcomes history and lifestyle factors with cardio-renal-metabolic multimorbidity: a prospective cohort study. BMC Medicine. https://doi.org/10.1186/s12916-026-05224-w

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05224-w

Keywords: adverse pregnancy outcomes, cardio-renal-metabolic multimorbidity, UK Biobank, gestational diabetes, hypertensive disorders of pregnancy, lifestyle factors, chronic kidney disease, type 2 diabetes, cardiovascular disease, multi-state models, epidemiology, women's health