Research Briefing

We investigated the presence of genetic variants for 105 genes previously associated with premature ovarian insufficiency (POI) in over 100,000 women. We found that predicted damaging variants in these genes were commonly found in the heterozygous state in women that had menopause within the normal age range. This suggests that monogenic causes of POI are rare, and that POI is more likely to be a polygenic disorder.

Primary sclerosing cholangitis dramatically increases the risk of colon cancer in patients with inflammatory bowel disease. Our study associates a newly discovered antigen-driven adaptive immune signature with the development of colorectal cancer in patients with primary sclerosing cholangitis and might help explain the high incidence of colon cancer in those patients.

In a person living with with human immunodeficiency virus (HIV), the recruitment of CD4⁺ T cells that recognize a co-pathogen in the central nervous system (CNS) resulted in localized inflammation and sparked HIV replication in the presence of suboptimal CNS anti-retroviral levels, leading to systemic virological failure. This case highlights the importance of considering inflammation and co-infection in HIV remission strategies.

The Human Lung Cell Atlas (HLCA), which includes data from 36 studies, is the most comprehensive representation of cellular gene expression in the human respiratory system to date. The HLCA serves as a reference for future cellular studies of the lung, enabling a better understanding of lung biology in health and disease.

With data from more than 500,000 adults in China, we assessed the association of alcohol use with a range of diseases reported in hospital records, using standard epidemiological and genetic approaches. We found that alcohol consumption was associated with higher risks of over 60 diseases in Chinese men, expanding the scope of known alcohol-related harms.

α-Synuclein seeds have been identified in various tissues in patients with synucleinopathies. We have developed real-time quaking-induced conversion combined with immunoprecipitation, a method that enables the detection of α-synuclein seeds from the serum of patients with synucleinopathies. This method can lay the foundation for the biological diagnosis of synucleinopathies.

We performed multi-ancestry genome-wide meta-analyses of preeclampsia/eclampsia and gestational hypertension, identifying 12 susceptibility loci associated with one or both conditions. These analyses enabled the derivation of genome-wide polygenic risk scores, which predicted the development of these hypertensive pregnancy complications in independent cohorts, independent of clinical risk factors.

A newly composed single-cell transcriptomic atlas of tumor-infiltrating T cells across 16 cancer types revealed previously undescribed T cell states and heterogeneity. A unique T cell stress response state, T_STR, was linked to immunotherapy resistance. Our high-resolution T cell reference maps, web portal, and annotation tool can assist efforts to develop T cell therapies.

We characterized cancer cells, immune responses and the microbiota composition in a cohort of primary colon cancers. Multi-omic analyses defined parameters associated with favorable prognosis, including a score that captured the intratumoral immune response and a specific microbiome signature. This data repository (atlas and compass of immune–cancer–microbiome interactions (AC-ICAM)) is publicly available.

Machine learning models that integrate cardiac troponin concentrations and clinical features to compute the probability of myocardial infarction outperform current care pathways that use fixed troponin thresholds or risk scores. Adoption of these models could reduce inequalities, prevent unnecessary admissions, and accelerate the diagnosis and treatment of patients with myocardial infarction.

An individual-level simulation model has been developed to evaluate the comparative effectiveness of maternal health policy interventions in 200 countries and territories. In addition to assessing global progress, this approach can guide priority setting and provide context-specific guidance for local policymakers.

By using several large population-based and clinical neuroimaging cohorts, we identified a shared neural basis for several mental disorders, namely the neuropsychopathological factor. This genetically determined factor might represent a delayed development of prefrontal brain neural circuits, leading to poor executive function.

Data from circulating tumor DNA (ctDNA) testing were generated for over 1,900 samples across at least 3 time points in a phase 3 clinical trial and used to build a machine learning model to predict patient survival. The model accurately identified patients with a high risk of disease recurrence and could provide a basis for assigning therapies in phase 1/2 clinical trials.

A large meta-analysis of genome-wide association studies (N >40,000) on perivascular space (PVS) burden, an emerging brain imaging marker of cerebral small vessel disease, has revealed 24 genetic risk loci for extensive PVS burden. These findings provide novel insights into the biology and clinical significance of this trait.

Lung adenocarcinomas (LUADs) encompass a broad spectrum of histological appearances. We use multi-region, prospective and longitudinal sampling from the TRACERx dataset to show the relationship between LUAD morphologies and their underlying evolutionary genomic landscape, as well as clinical risk and the nature of metastatic dissemination.

Body mass index (BMI) is a clinically useful obesity measure that shows an association with healthspan and lifespan. We developed ‘biological BMI’ scores, calculated from multi-omics data (proteins and metabolites) and clinical laboratory tests from blood. We show that a biological BMI more accurately reflects metabolic health and is more responsive to lifestyle changes than a classically measured BMI.

We used deep neural networks trained on optical histology and open-source genomic data to predict the molecular genetics of brain tumors during surgery. Our results represent how AI-based diagnostics can provide a valuable adjunct to wet laboratory methods for molecular testing in patients with cancer.

Using a systems-level, multi-omics approach, we reveal several genes associated with arachnoid cysts and identify four phenotypic subtypes of arachnoid cysts, the severity of which correlates with the presence of protein-damaging de novo variants. All candidate genes are expressed in the developing brain and encode molecules implicated in chromatin modification or transcriptional regulation.

Using observational data from over 200,000 participants with up to 32 years of follow-up, we compared the strengths of eight healthy dietary patterns for general health. We found that diets that lowered hyperinsulinemia, chronic inflammation and diabetes risk may offer the greatest protection against chronic diseases.

We developed a compact database, called a ‘Rareservoir’, that contains the rare variant genotypes and phenotypes of 34,523 patients with a rare disease and 43,016 unaffected relatives. We inferred 260 genetic associations with rare disease classes, of which 19 were previously unidentified, and validated etiological roles for ERG, PMEPA1 and GPR156.