Netherton syndrome (NS) is a rare congenital barrier disorder caused by pathogenic variation in SPINK5, resulting in severe epidermal dysfunction and immune dysregulation. Although dupilumab is increasingly used in NS, molecular data describing treatment-associated tissue responses remain limited, particularly in genetically complex presentations. We investigated treatment-associated molecular trajectories during IL-4Rα blockade in two siblings with NS carrying heterozygous compound SPINK5 (c.2468dup & c.2243 A > G) and FLG (p.Arg501*) variants. Longitudinal whole-exome sequencing and paired bulk RNA sequencing of peripheral blood and lesional skin were performed before and after six months of dupilumab therapy. Transcriptomic changes were evaluated using effect-size– and rank-based approaches and contextualized through integration with publicly available NS reference datasets. Across both compartments, post-treatment transcriptional profiles exhibit coordinated shifts toward healthy-control–like expression states. Blood transcriptomes show reduced inflammatory, metabolic, and proliferative signatures alongside modulation of FURIN and increase expression of PAPPA, LPAR5, CELSR1, and ARHGEF19. Skin transcriptomes transition from neurosensory and follicular programs toward lipid metabolic organization, keratinocyte differentiation, and cornified-envelope–associated pathways. Cross-compartment integration highlights shared treatment-associated transcripts, including ADAM23, and candidate markers of immune–epithelial recalibration. Together, this exploratory dual-compartment longitudinal analysis provides rare molecular insight into biologic treatment response in a genetically complex NS context and establishes a foundational reference for future genotype-informed and biomarker-oriented studies. Netherton syndrome is a rare inherited skin disorder that causes severe inflammation and a weakened protective skin barrier. Although dupilumab is increasingly being used to treat this condition, its effects on the underlying biological mechanisms are not yet fully understood. To investigate this, we studied two siblings with Netherton syndrome and analyzed blood and skin samples collected before treatment and after six months of therapy. We found that dupilumab was associated with reduced inflammatory activity and gene expression patterns that became more similar to those seen in healthy individuals. These findings suggest that dupilumab may help restore both immune balance and skin barrier function, providing new insight into its mechanism of action and supporting the development of more personalized treatment strategies for people with Netherton syndrome. Bajo-Santos et al. longitudinally profile blood and skin transcriptomes before and after dupilumab treatment in two siblings with genetically complex Netherton syndrome. They observe that IL-4Rα blockade shifts molecular signatures toward healthy-like states, reducing inflammation and promoting epidermal differentiation and barrier-repair pathways.
Traditional dementia research often isolates single aspects, biomarkers or symptoms, overlooking complex interactions shaping disease risk and progression. Lysosomal and mitochondrial dysfunction, reflecting systemic processes, may explain links between Alzheimer’s and metabolic, inflammatory, cardiovascular, ageing, genetic, environmental, and endocrine factors. Barnaghi et al. propose reframing Alzheimer's disease as a systemic disorder rooted in lysosomal and mitochondrial dysfunction, rather than solely a brain-confined condition. This whole-body perspective explains the links to metabolic, cardiovascular, and inflammatory comorbidities and highlights windows for early intervention.
Low von Willebrand factor (VWF; 30–50 IU/dL) is a common clinical phenotype associated with bleeding, but its genetic basis remains poorly defined. We aimed to explore the contribution of rare and common variants to Low VWF and bleeding risk. Whole-exome sequencing was performed in 115 Low VWF cases (33 years, range 18–72) and 139 controls (39 years, range 18–65). Common variant association, gene burden, gene-set burden, and PRS analyses were performed. Only 29 cases (25%) carried known or predicted pathogenic VWF variants. Cases showed a significantly higher burden of rare VWF variants (gnomAD-MAF < 0.01) compared with controls (p = 4.47×10⁻⁶), with no difference in common variant burden. Notably, in cases, the burden of rare VWF variants inversely correlated with VWF:activity (p = 0.007) and VWF:Ag (p = 0.046), indicating that cumulative rare coding variants quantitatively lower VWF plasma levels. Beyond VWF, we investigated loci previously implicated by GWAS and rare/low frequent variants (MAF < 0.05) in ABO and NIPSNAP3B were associated with VWF levels (burden test p = 0.00065 and p = 0.0039), with additional borderline signals in RAB5C, KAT2A, and ACE (p ≤0.087). Among common variants (gnomAD-MAF > 0.05), 27 variants reached nominal statistical significance in 8 genes (ABO, VWF, PXK, OR13C5, STAB2, GIMAP7, TNPO1, SCARA5). Bleeding severity (ISTH-BAT) was significantly associated with the burden of common variants in previously identified GWAS loci (p = 0.0312). This relationship was further supported by a PRS analysis (p = 0.0434). Together, these findings support that Low VWF reflects both the cumulative effect of rare VWF variants and a broader polygenic background involving VWF and non-VWF loci, with common non-VWF variants contributing to the bleeding tendency. Von Willebrand factor (VWF) is a blood protein that helps stop bleeding. Some people have reduced levels of VWF (30–50 IU/dL), known as Low VWF, which can lead to significant bleeding. However, the genetic causes and variability in bleeding symptoms are not fully understood. We analyzed genetic data from 115 individuals with Low VWF and 139 healthy controls using whole-exome sequencing to investigate the role of both rare and common genetic variants. Most patients did not carry a single clearly disease-causing mutation in the VWF gene but instead had a higher burden of rare variants associated with lower VWF levels. We also identified additional genes influencing VWF regulation. Moreover, a higher burden of common risk variants was linked to more severe bleeding. These findings suggest that Low VWF is driven by combined effects of rare and common genetic variation. Seidizadeh et al. performed whole-exome sequencing in 115 Low VWF cases and 139 controls to examine the genetic architecture of low VWF levels and bleeding risk. They show that rare VWF variants, common VWF and non-VWF modifiers, and polygenic factors jointly influence VWF levels and bleeding risk, supporting a continuum model of VWF regulation.
Myopia is increasingly prevalent worldwide, and several refractive surgical options are available, including photorefractive keratectomy (PRK), laser-assisted in situ keratomileusis (LASIK), femtosecond laser-assisted LASIK (FS-LASIK), small-incision lenticule extraction (SMILE), laser-assisted subepithelial keratectomy (LASEK), and phakic intraocular lens (PIOL) implantation. This network meta-analysis compared their visual, refractive, and corneal outcomes. PubMed, Embase, Web of Science, Cochrane CENTRAL, and Scopus were searched through May 15, 2025. Randomized and comparative studies of adults undergoing refractive surgery for myopia or myopic astigmatism were included. Bayesian network meta-analysis using Markov chain Monte Carlo (MCMC) simulation estimated mean differences, 95% credible intervals, and surface under the cumulative ranking curve (SUCRA) values at 1, 3, and 6 months. The study was registered in PROSPERO (CRD420250651487). Seventeen studies including more than 1500 eyes were analyzed. At 1 month, uncorrected distance visual acuity (UDVA) did not differ significantly among LASIK, SMILE, PRK, and FS-LASIK, although SMILE ranked highest by SUCRA (0.86). At 3 months, LASIK showed superior UDVA versus PRK (mean difference [MD], −0.01; p = 0.0410), SMILE (MD, −0.02; p = 0.0267), and FS-LASIK (MD, −0.03; p = 0.0001). By 6 months, UDVA and corrected distance visual acuity (CDVA) were comparable across procedures. FS-LASIK best preserved central corneal thickness (CCT) at 1 and 3 months, whereas PRK showed early thinning (MD, −28.00 μm; p = 0.0150). SMILE reduced spherical error at 1 and 3 months, while PRK ranked best at 6 months. No procedure is superior across all outcomes; surgical choice should be individualized. Chen et al. compare major refractive surgical procedures for myopia correction. No single technique is superior across all outcomes, supporting individualized surgical selection. Myopia, or short-sightedness, is increasingly common worldwide, and many adults choose refractive surgery to reduce dependence on glasses or contact lenses. Several procedures are available, including photorefractive keratectomy (PRK), laser-assisted in situ keratomileusis (LASIK), femtosecond laser-assisted LASIK (FS-LASIK), small-incision lenticule extraction (SMILE), laser-assisted subepithelial keratectomy (LASEK), and phakic intraocular lenses (PIOLs), but direct comparisons among all options remain limited. In this systematic review and network meta-analysis, we compared visual, refractive, and corneal outcomes across commonly used refractive procedures. The results showed that no single procedure performed best for all outcomes. Visual outcomes were generally similar by 6 months, while early differences were observed in uncorrected visual acuity, corneal thickness preservation, and refractive error correction. These findings support individualized procedure selection based on corneal thickness, refractive profile, visual recovery expectations, and patient-specific clinical factors.
BACKGROUND:Spontaneous preterm birth, defined as delivery before 37 weeks of gestation, remains a global cause of neonatal illness and death, yet treatments for preterm labor remain limited. We hypothesized that KV7 channels in the uterus (myometrium), could be a therapeutic target for preventing or treating preterm labor. The aim was to characterize KV7 channel subtypes in human myometrium present after labor onset, and to use a preterm birth mouse model to provide proof-of-principal evidence that activating these channels in vivo can delay preterm delivery. METHODS:This experimental study integrated studies of human myometrial tissue (term and preterm pregnancies, n = 159 across experiments), with a RU486-induced preterm birth mouse model (C57BL/6 J, n = 6-9 per experimental group), and ex vivo fetal ductus arteriosus preparations (CD-1 mice, n = 10 per group). Molecular and protein profiling, isometric tension recordings, and functional pharmacology were used to characterize KV7 mediated regulation of human and mouse uterine contractility. In vivo dosing studies in mice assessed the effects of KV7 activators (retigabine and ML213) on delivery timing (n = 6-9). RESULTS:We show that KV7 channels and ancillary units (KCNE1-5), are expressed in pregnant human myometrium, prior to and after the onset of labor. KCNQ4 and KCNE4 transcripts are also present in myometrium taken at preterm gestations ( ± labor). KV7.2-5 activators markedly reduce spontaneous contractions in human and mouse myometrium in vitro and significantly delay preterm birth in vivo with limited impact on fetal ductus arteriosus function. CONCLUSIONS:These findings support KV7 activation as a promising approach to suppress uterine contractility and hence delay preterm birth, highlighting a potential new direction for either drug repurposing or therapeutic development.
Schizophrenia (SCZ) and bipolar disorder (BD) are severe psychotic disorders with overlapping clinical manifestations, leading to high rates of misdiagnosis. This study aims to identify disorder-specific neurophysiological biomarkers using electroencephalography and contrastive machine learning to improve differential diagnosis. Resting-state electroencephalography was recorded from 52 patients with BD, 65 with SCZ, and 75 healthy controls. Temporal variability networks were constructed using sample entropy. Contrastive variational autoencoders decomposed these networks into components shared with healthy controls and components specific to each disorder. Based on disorder-specific components, predictive models for clinical symptoms were constructed. Additionally, spatial pattern network filters were implemented to extract discriminative features for the classification of BD and SCZ patients. Here we show pronounced differences in disorder-specific network components between SCZ and BD, especially in frontal-central/parietal connectivity, which were not discernible in the original or shared networks. These disorder-specific components correlate significantly with clinical assessment scores and support predictive modeling of symptom severity. By applying spatial pattern network filters to the disorder-specific components, we achieve 96.154% accuracy in distinguishing SCZ from BD, substantially surpassing conventional approaches. This integrative framework, combining dynamic network analysis with contrastive machine learning, provides a powerful methodology for extracting neurophysiological biomarkers and paves the way for biologically grounded diagnostics in psychotic disorders. Schizophrenia and bipolar disorder are serious mental illnesses that can look very similar, making them hard for doctors to tell apart. This study used an objective brain index (EEG) combined with an artificial intelligence method to find differences between the two conditions. The researchers analyzed brain activity patterns and isolated features unique to each disorder. They discovered that these unique patterns were linked to patients’ symptoms and could predict how severe those symptoms were. Most importantly, the method distinguished between the two disorders with over 96% accuracy. This work could lead to a more reliable, biology-based tool to help doctors diagnose patients correctly, reduce misdiagnosis, and guide more personalized treatment decisions in the future. Jiang, Ye et al. combine resting-state EEG and contrastive variational autoencoders to identify disorder-specific neurophysiological biomarkers in schizophrenia and bipolar disorder. The approach achieves over 96% accuracy in distinguishing the two disorders by revealing distinct frontal-central/parietal connectivity patterns.
Supraventricular tachycardia (SVT) is a common paroxysmal arrhythmia whose intermittent nature limits detection by conventional electrocardiogram (ECG) monitoring. Mobile artificial intelligence (AI)–enabled ECG systems offer a scalable alternative, yet their clinical value and real-world effectiveness for SVT detection remain insufficiently evaluated. In this retrospective real-world study, we analyzed 3,566,626 single-lead ECG recordings from 84,242 users collected in China between August 2019 and May 2025. We assessed the system’s ability to identify SVT events, patterns of patient-initiated specialist review, clinical workflow efficiency, and the potential impact on patient time and costs. Here we show that the mobile system accurately identifies SVT events in real-world use (ROAUC 0.866; 95% CI, 0.863-0.869), with acceptable discrimination and reliable risk estimation (Brier score 0.123). Among recordings identified as high risk and subsequently reviewed by cardiologists, 96.30% are confirmed as SVT. The combination of automated analysis and patient-initiated specialist review reduces the number of recordings requiring cardiologist assessment per confirmed case by 98.8%. Compared with repeated hospital-based evaluations, this approach reduces patient costs by 79.1% and time burden by 95.7%. This study shows that an AI-ECG mobile system combined with patient-initiated specialist review provides an efficient approach for opportunistic detection of SVT in daily life. This strategy may complement clinician-led care by improving the efficiency of cardiac monitoring and triage. Supraventricular tachycardia (SVT) is a type of abnormal heart rhythm that can occur suddenly and may be difficult to detect because episodes are often brief and unpredictable. We investigated whether a portable single-lead electrocardiogram (ECG) device, a small device that records the electrical signals generated by the heart through electrodes placed in contact with the skin, combined with artificial intelligence could help identify SVT in everyday life. We analyzed more than 3.5 million ECG recordings from over 80,000 users in China and evaluated how the system worked together with patients and cardiologists. The system showed good ability to identify possible SVT events and helped cardiologists focus on recordings with a higher likelihood of abnormal rhythms. This approach may provide a more convenient and efficient way for people to monitor heart rhythm outside hospitals and support earlier medical attention when needed. Fan, Zhang et al. evaluate the health-system utility of an AI-enabled ECG mobile system for detection of supraventricular tachycardia using over 3.5 million recordings in China. The system supports real-world cardiac monitoring through AI-assisted triage and patient-initiated review, offering a framework for future digital health tools.
BACKGROUND:A patient undergoes multiple examinations in each hospital stay, where each provides different facets of the health status. These assessments include temporal data with varying sampling rates, discrete single-point measurements, therapeutic interventions such as medication administration, and images. While physicians are able to process and integrate diverse modalities intuitively, neural networks need specific modeling for each modality complicating the training procedure. METHODS:We demonstrate that this complexity can be significantly reduced by visualizing all information as images along with unstructured text and subsequently training a conventional vision-text transformer. Our approach, Vision Transformer for irregular sampled Multi-modal Measurements (ViTiMM), simplifies data preprocessing and modeling by unifying clinical measurements, medications, X-ray images, and electrocardiography scans into a single visual representation. RESULTS:ViTiMM outperforms current state-of-the-art methods in predicting in-hospital mortality, phenotyping, and decompensation on two datasets, the MIMIC-IV and COVID Data for Shared Learning (CDSL) dataset. CONCLUSIONS:We hope our work inspires advancements in multi-modal medical AI by reducing the training complexity to (visual) prompt engineering, thus lowering entry barriers and enabling no-code solutions for training. The source code is publicly available at https://github.com/Cardio-AI/ViTiMM .
Flooding is an increasing threat in rapidly growing cities, yet evidence on how heavy rainfall affects access to healthcare within cities and whether impacts are greater for vulnerable populations remains limited. This study quantifies how flooding changes walking access to healthcare in Kampala, Uganda, and how these changes are distributed across areas that differ in their population-level risk of undernutrition and poor maternal and child health outcomes. We conducted a cross-sectional ecological geospatial modelling study across Kampala and its whole population (approximately 1.8 million residents). We estimated walking travel time to the nearest public or private not-for-profit healthcare facility and hospitals under baseline and flood conditions. Flood scenarios were based on a hydrodynamic model simulating 15 rainfall events ranging from 20 to 100 mm over 1, 3, or 6 hours. Travel speeds accounted for caregivers walking with young children. Outcomes were population-weighted changes in travel time, summarised at parish level and compared across vulnerability groups. Here we show that flooding increases travel time across all scenarios, with larger effects for hospital access. Population-weighted mean increases in travel time range from 11.5 to 19.3 minutes for all facilities and from 23.4 to 56.3 minutes for hospitals across scenarios. In high-intensity storms, increases are greater in more vulnerable areas, particularly for hospital access, with largest disruptions in peripheral areas. Flooding reduces access to healthcare in Kampala and disproportionately affects more vulnerable populations. These findings show how extreme rainfall can widen inequalities in access to maternal and child health and nutrition services and support planning. Floods can make it harder for caregivers with young children to reach healthcare facilities. We studied Kampala, Uganda, to estimate how floods change walking time to healthcare. We combined maps of land, roads, water, and elevation with realistic walking speeds on wet surfaces and tested fifteen flood situations, from lighter, longer storms to intense, short ones. Here we show that travel times increased in every case: by about 19 minutes on average to the nearest clinic and 56 minutes to the nearest hospital for the most intense one-hour storms. Delays were greatest in high-vulnerability neighbourhoods and outer areas, while central areas changed less. These results can guide service placement. Lubbers et al. conduct a cross-sectional geospatial modelling study across Kampala to estimate how heavy rainfall affects healthcare access. They show that flooding reduces access to healthcare and disproportionately affects more vulnerable populations, resulting in widened inequalities.
Clinical imaging is routinely used pre-operatively for cochlear implantation, yet lacks the resolution and contrast necessary to visualize the fine intracochlear structures critical for individualized intervention. To address this limitation, an ensemble deep learning model was developed to automatically segment cochlear micro-anatomy from standard clinical scans. The model was trained and validated using an independent internal dataset comprised of paired synchrotron and clinical scans of the same cochleae across various acquisition protocols. Performance was evaluated quantitatively on an unseen internal test dataset and a multi-institutional external test dataset. The deep learning model achieves accurate segmentation of the scala tympani (ST) and scala vestibuli (SV) across all tested modalities, with a mean Dice similarity coefficient of 0.895 ± 0.024 and 0.891 ± 0.029, a max Hausdorff distance of 0.396 ± 0.100 mm and 0.444 ± 0.280 mm, and an average Hausdorff distance of 0.008 ± 0.003 mm and 0.008 ± 0.003 mm, respectively. The model achieves performance metrics superior to previously published scalar models and demonstrates strong viability on the multi-institutional external dataset. Furthermore, anatomical measurements on the automatic segmentations closely match those obtained from high-resolution ground truth segmentations, measured using scalar volume and lateral ST length, confirming reliable estimation of clinically relevant metrics. By bridging the gap between high-resolution imaging and routine clinical imaging, this work could provide a practical solution for patient-specific cochlear implant surgical planning through electrode selection and post-operative assessment through image fusion, advancing the goals of atraumatic insertions and more effective hearing restoration. Micuda et al. develops and validates an ensemble deep learning model that segments cochlear micro-anatomy using paired synchrotron and routine clinical scans. The model accurately delineates the scala tympani and scala vestibuli across imaging configurations, outperforming prior methods and preserving clinically relevant measurements. Imaging used before cochlear implant surgery often does not have enough detail to visualize the small structures inside the cochlea that are important for surgical planning and patient outcomes. To address this limitation, a deep learning model was developed to automatically extract the small anatomical structures from routine clinical scans. The model was developed using paired high-resolution and clinical scans of cadaveric cochleae. It was subsequently evaluated quantitatively on a cadaveric dataset, assessed for clinical plausibility on an international external patient dataset, and compared against previously published models. The model achieved high accuracy on both objective and cochlear-specific metrics across all imaging configurations, closely matched the high-resolution reference images, and outperformed similar models reported in the literature. This deep learning model has the potential to be used in clinical practice to improve surgical planning for cochlear implantation, support patient-specific implant selection, and aid in the assessment of the implant position after surgery.
Paraneoplastic cerebellar degeneration associated with anti-Yo antibodies (Yo-PCD) is a rare neurological syndrome affecting patients with breast and gynecological cancers. Although T cell–mediated immune responses against Purkinje cell antigens are thought to drive neuronal injury, the cellular mechanisms underlying disease remain poorly understood. We aimed to define the immune landscape associated with this disorder across the blood and central nervous system. We performed single-cell RNA sequencing with paired T-cell receptor sequencing on immune cells isolated from blood and cerebrospinal fluid from six patients with Yo-PCD. Samples from individuals with multiple sclerosis, idiopathic intracranial hypertension and patients with ovarian cancer without neurological disease served as additional controls. Immune cell states, clonal expansion, and compartment-specific transcriptional programs were analyzed. Patients with Yo-PCD exhibit impaired immune tolerance programs in myeloid cells and regulatory T cells, including reduced expression of the transforming growth factor-β pathway. Myeloid cells display distinct compartment-specific activation states, with major histocompatibility complex class I programs in blood and class II in cerebrospinal fluid. Moreover, cytotoxic CD4-positive T cells, are selectively clonally expanded in cerebrospinal fluid, span progressive differentiation states, and share T-cell receptor clonotypes with CD4 T cells, consistent with local functional reprogramming. These findings identify compartmentalized immune responses as a defining feature of Yo-PCD and establish cytotoxic CD4-positive T cells as candidate mediators of disease. This work provides a baseline for developing biomarkers and immune-targeted therapeutic strategies in this devastating neurological disorder. Paraneoplastic cerebellar degeneration is a rare disease in which the body’s immune system attacks the cerebellum, a part of the brain that controls balance and coordination. It can develop in some people with breast or gynecological cancers, but the reasons are not well understood. In this study, we examined individual immune cells from blood and from the cerebrospinal fluid, the fluid surrounding the brain and spinal cord in patients with this disease. We compared these cells with those from patients with ovarian cancer without neurological disease, as well as from people with multiple sclerosis or intracranial hypertension. We found that cancer-related inflammation was present in patients with both cancers, but only those with paraneoplastic cerebellar degeneration showed signs that the normal mechanisms controlling immune responses had broken down. They also had a type of immune cell called cytotoxic CD4 T cells that accumulated in the cerebrospinal fluid and may contribute to damage of the cerebellum. Our findings provide new insight into why the immune system attacks the brain in this disease and highlight cytotoxic CD4 T cells as potential targets for future treatments. Petitpré, Tran, Wucher et al., use single-cell RNA and T-cell receptor sequencing of blood and cerebrospinal fluid immune cells from patients with anti-Yo paraneoplastic cerebellar degeneration to define the disease-associated immune landscape. They find impaired immune tolerance, distinct immune responses in blood and the nervous system, and expansion of cytotoxic CD4 T cells that may drive disease.
BACKGROUND:Eating jetlag is an emerging concept defined as meal timing regularity between free and work days. We evaluated the association between eating jetlag and the risk of cardiovascular disease in a French cohort. METHODS:Data of 104,806 participants (79% women, mean age=42.7 years (SD 14.6)) were collected from the NutriNet-Santé prospective cohort (2009-2023). Cardiovascular disease was self-reported and validated against medical records. Eating jetlag intensity was calculated using times of first and last caloric intakes in repeated 24-hour dietary records (5.8 (SD 3.2) records on average)), as the absolute difference between weekends and weekdays, and its associations with cardiovascular disease risk were assessed through multi-adjusted Cox models. Eating jetlag direction was assessed using 3 categories: "Advance", "Maintenance", "Delay", in weekends compared to weekdays. RESULTS:During a median follow-up of 8.1 years, eating jetlag was associated, in male participants, with higher cardiovascular disease (HR = 1.14 [1.05-1.23]) and coronary heart disease (HR = 1.21 [1.11-1.33]) risks. In males, participants in the "Advance" group had higher risks of cardiovascular and coronary heart diseases (HR = 1.44 [1.08-1.93], and HR = 1.68 [1.19-2.39], respectively) compared to those in the "Maintenance" group; while those in the "Delay" group had a higher coronary heart disease risk (HR = 1.36 [1.05-1.77]). CONCLUSION:Our results suggest a role for meal timing regularity between weekends and weekdays in CVD etiology, independently of sleep duration and diet quality, specifically in males. If confirmed in other studies, meal timing regularity could be a strategy to mitigate cardiovascular disease risk. TRIAL REGISTRATION:NCT03335644.
BACKGROUND:Body mass index fails to capture variation in fat and muscle distribution that determines metabolic health and disease risk. MRI enables radiation-free quantification of regional body composition, yet scalable open-source tools applied in pooled cohorts with differing acquisition protocols have been lacking. METHODS:MRSegmentator, an open-source nnU-Net-based pipeline, was applied to quantify visceral adipose tissue (VAT), abdominal subcutaneous adipose tissue (ASAT), gluteofemoral adipose tissue (GFAT), trunk musculature, and the liver mask used for liver fat-fraction estimation in 45,851 adults from the German National Cohort (n = 26,877, 3 T multi-centre Siemens) and UK Biobank (n = 18,974, 1.5 T Siemens). Population-scale compartment volumes were segmented from stitched in-phase gradient-echo (GRE) images in both cohorts; liver fat fraction was calculated from fat-only and water-only images. The annotated development data comprised NAKO T2-HASTE and UKB Dixon reconstructions. A single pooled model was applied without site-specific adaptation. A separate two-reader agreement study used 50 scans from these annotated development-sequence domains. Associations between BMI-adjusted body composition and cardiometabolic conditions were estimated using generalized linear mixed-effects models. Incremental discrimination beyond age, BMI, and waist-to-hip ratio was assessed. RESULTS:Five-fold participant-stratified internal cross-validation against curated human-in-the-loop development references comprising UKB Dixon and NAKO T2-HASTE yielded a mean Dice of 0.91. In a separate 50-scan reader study on these annotated development-sequence images, overall reader-reader Dice was 0.937 and overall algorithm-reader Dice was 0.908. The trained pipeline was then used to segment compartment volumes from stitched in-phase GRE inputs in both cohorts, while liver fat fraction was calculated from fat-only and water-only images; direct sequence-matched validation on NAKO GRE was not performed. VAT showed the strongest positive associations with cardiometabolic conditions, while GFAT showed inverse associations, most prominently for type 2 diabetes (OR 0.69, 95% CI 0.66 to 0.72). Disease-specific body-composition phenotypes were identified, with type 2 diabetes characterized by elevated VAT, reduced GFAT, and increased liver fat. MRI-derived compartments modestly improved discrimination for type 2 diabetes and hyperlipidemia beyond anthropometric measures. CONCLUSIONS:A single open-source deep-learning pipeline enabled pooled body-composition phenotyping in two cohorts and captured distributional variation in fat and muscle beyond BMI. High agreement in internal cross-validation (mean Dice 0.91) and the separate two-reader study support the annotated development-sequence analysis, while the population-scale application identified distinct disease-associated phenotypes and modest incremental discrimination beyond conventional anthropometry.
Systemic autoinflammatory diseases (SAIDs) are a diverse group of rare disorders with partially overlapping clinical features. However, their genetic makeup and treatment responses vary widely. Although inflammasome dysregulation is central to many SAIDs, recent studies suggest that additional immune pathways may also contribute to disease activity. We integrated bulk transcriptomic and plasma proteomic data from SAID patients and negative controls, generated using the NovaSeq 6000 and SomaScan platform, respectively. We assessed the expression patterns of four inflammation-related genes (IL1B, IL6, IL18 and BLNK) and evaluated differential expression between patients and controls, and between initial and follow-up samples. We constructed protein-protein interaction (PPI) networks of the transcripts and proteins, then performed a functional enrichment analysis to identify gene sets that were significantly enriched in each PPI network. Transcriptomic profiling identified 1,805 differentially expressed transcripts. BLNK showed significant inverse correlations with IL1B and IL18, whereas BLNK and IL6 were correlated with B-cell proportions. Among the top 50 upregulated transcripts, three enriched groups were identified: associated with secretory granules, haemoglobin complexes and adaptive immune responses. Differential proteomic analysis identified 1217 differentially abundant proteins; the top 50 upregulated proteins were categorised into four groups related to complement activation, the acute-phase response, neutrophil migration and extracellular matrix organisation. Finally, we observed transcriptomic and proteomic expression changes during follow-up. Our analyses reveal the diverse inflammatory signatures observed across patients. These findings underscore the complexity of SAID pathophysiology and could help refine hypotheses regarding disease mechanisms and patient stratification. Systemic autoinflammatory diseases (SAIDs) are rare disorders that occur when the immune system becomes overactive. This results in recurring symptoms such as fever, inflammation and joint pain. These diseases are difficult to diagnose and treat because patients often exhibit overlapping symptoms, and their genetic backgrounds and responses vary widely. Rather than focusing on individual genes or markers, we examined patterns of gene activity and blood proteins in a large cohort of patients. Our research identified patterns of gene activity and protein levels shared across SAIDs and associated with innate and adaptive immune-related genes and proteins, including B-cell markers. These findings demonstrate the complexity of the SAID mechanism and will help to guide future studies of patient heterogeneity and treatment responses. van Wijngaarden et al., integrate bulk transcriptomic and plasma proteomic data from patients with systemic autoinflammatory diseases and controls to characterise inflammatory pathways and monitor molecular changes over time. They identify distinct transcriptomic and proteomic inflammatory signatures, including associations between BLNK, cytokine-related genes and immune cell populations, revealing disease heterogeneity and molecular changes during follow-up.
Clinical trials are essential for developing new cancer treatments and improving patient care. However, participation remains low across oncology and is particularly challenging for people with pancreatic cancer, who often experience rapid disease progression, limited treatment options, and short timeframes for trial enrolment. In this Review, we examine the factors that limit clinical trial participation in pancreatic cancer, including patient identification and referral, eligibility assessment, molecular testing, and trial delivery. We discuss how restrictive eligibility criteria, inconsistent integration of trial processes into routine care, variability in molecular testing, and logistical barriers reduce opportunities for participation. We also consider practical strategies to improve access, including earlier and more systematic trial identification, broader clinical eligibility criteria, timely molecular testing, more flexible trial designs, and decentralised models of care. Improving participation will require coordinated changes across clinical pathways, trial design, and healthcare systems to ensure that trial availability translates into equitable patient access.
Campylobacter jejuni is a leading cause of bacterial gastroenteritis, and fluoroquinolone-resistant strains represent a major public health concern. Still, C. jejuni can also cause invasive disease, particularly in immunocompromised individuals, highlighting the need to elucidate the adaptive mechanisms behind bloodstream invasion and persistence. This study aimed to characterize the within-patient genomic evolution of an invasive C. jejuni strain. Whole-genome sequencing was performed on same-patient isolates from stool (n = 1) and blood (n = 2), followed by in-depth genomic comparisons and antimicrobial susceptibility testing. Here we show that rapid within-patient microevolution is driven by amino acid substitutions and small inactivating indels. Most mutated loci have predicted or reported functions related to motility (including flagella- and energy taxis/chemotaxis-associated proteins), adherence, cell shape/envelope organization and host interactions (e.g., immune evasion and resistance to blood environment). All isolates are resistant to ciprofloxacin (MIC = 16 mg/L) due to the canonical GyrA Thr86Ile substitution. After about one month of infection (including a 21-day ciprofloxacin treatment), ciprofloxacin MIC increased to 128 mg/L and a newly acquired resistance to moxifloxacin (MIC > 32 mg/L) was observed. This expanded resistance profile correlates with the emergence of the GyrA Asp90Gly substitution (previously unreported in C. jejuni) and a 1-bp deletion in the cmeABC promoter region (previously demonstrated to increase efflux pump expression). Dynamic phase variation of several loci was also observed during bloodstream persistence, including ON-phase switching of the cell invasion protein A (CipA). These genomic findings provide insight into within-host population shifts and the emergence of antibiotic-resistant clones, contributing to better understanding the dynamics associated with C. jejuni bloodstream invasion and persistence. Campylobacter jejuni (C. jejuni) is a species of bacteria that commonly causes foodborne illness. In rare cases, especially in people with weakened immune systems, it can spread from the gut into the bloodstream causing severe infection. We wanted to understand how this bacterium changes inside the body during this type of infection and how it becomes more resistant to antibiotics. We compared the complete DNA of bacteria collected from the stool and blood of the same patient over about one month, together with laboratory tests of antibiotic susceptibility. We found that the bacterium rapidly accumulated genetic changes potentially linked to survival in the bloodstream and developed increased resistance to important antibiotics during treatment. These findings improve our understanding of how C. jejuni adapts during invasive infection and may help improve diagnosis, treatment, and future surveillance of antibiotic-resistant strains. Borges et al. investigate the within-host evolution of an invasive Campylobacter jejuni strain using stool and blood isolates collected from a patient over one month. Observed genetic changes may be associated with bloodstream persistence and the emergence of increased antimicrobial resistance, suggesting their possible role in adaptation and invasive disease.
Larger body size is strongly associated with type 2 diabetes, but the mechanisms linking the two remain uncertain. Proposed explanations include shared genetic susceptibility, effects of fat storage capacity and fat distribution, early-life influences, social and environmental factors, and reverse causation through insulin resistance affecting weight. We aim to clarify the timing and interplay of overweight, obesity, and type 2 diabetes at the population level. We analyze longitudinal data from the Health and Retirement Study, a nationally representative survey of older adults in the United States followed over time. The analytic sample consists of 7663 European-origin individuals with mean age at baseline as 57.24. We model transitions between weight status and diabetes status using Continuous Time Markov and Hidden Markov Chain models to evaluate how much alternative mechanisms beyond excess body weight account for observed associations. We show in this cohort, excess body weight is the factor most strongly associated with transitions into type 2 diabetes. Early-life conditions, genetic propensity to type 2 diabetes, and education play a secondary role by moderating the effects of excess body weight on type 2 diabetes risks. These findings indicate that, among older U.S. adults, excess body weight is the dominant correlate of T2D incidence in population transition patterns, while other proposed mechanisms contribute more modestly. This supports prevention strategies that prioritize maintaining healthy body weight to reduce type 2 diabetes risk. Overweight individuals are more likely to develop type 2 diabetes (T2D), but the mechanisms behind this association are not fully understood. This study examines whether patterns of T2D onset are explained primarily by excess body weight, or whether early-life conditions, genetic susceptibility, and education also account for the relationship. We use the Health and Retirement Study, a longitudinal survey following more than 20,000 U.S. older adults since 1992. Using repeated interviews, we track changes in weight status and diabetes status over time and compare how well different explanations align with observed transitions. Excess body weight shows the strongest association with subsequent T2D, while early-life factors, genetic propensity, and education contribute more modestly, mainly by shaping the strength of the weight–T2D link. Huangfu et al. use longitudinal Health and Retirement Study data and Markov and hidden Markov models to examine the association between obesity and T2D after age 45. They find excess weight strongly predicts diabetes onset, while genetic risk, early-life factors, and education play smaller roles in the association between obesity and T2D.
Variants in the SORL1 gene have been identified and associated with increased risk of developing Alzheimer’s disease (AD). However, there is an unmet need for an effective approach to verify pathogenicity and treat individuals carrying SORL1 variants causal of AD before the development of symptoms. We conducted a panel of biochemical and biological assays in cell lines and iPSC-derived neurons to evaluate the pathogenicity of the SORL1 p.D1108N variant and tested for a possible dominant-negative effect by analysis of its interaction and impact on intracellular trafficking of wild-type (WT)-SORL1. We investigated the effect of deletion of exon 23 using cell biological assays and screened a panel of splice-switching antisense oligonucleotides (SSO) for their ability to exclude exon 23. Here, we demonstrate how the SORL1 p.D1108N variant, which maps to a conserved sequence position within a complement-type repeat (CR)-domain and is involved in Calcium ion coordination, is pathogenic and acts through a dominant-negative mechanism by retaining WT-SORL1 in the endoplasmic reticulum ER. We next show how SORL1 protein that lacks the 38 amino acids that correspond to the CR1 domain, encoded by the 114-base-pair exon 23, has similar activity as the full-length receptor. Finally, we identify an SSO that can be used to efficiently exclude exon 23 from SORL1 transcripts. We provide evidence that the SORL1 p.D1108N variant is pathogenic and acts through a dominant-negative mechanism, while demonstrating that therapeutic exon 23 skipping preserves receptor activity. Accordingly, carriers of the p.D1108N variant are likely suitable for an early intervention. Our study thus represents the first step towards a personalized medicine therapy for a subpopulation of patients with SORL1-associated AD. People who have particular variants of the SORL1 gene are more likely to develop Alzheimer’s disease (AD) and dementia. We undertook a study to investigate why this is the case and developed a molecule that could be used to reduce the impact of these variants. This approach could potentially be used as a treatment for people with these variants to reduce the chances of their developing Alzheimer’s Disease. Rosenberg et al. compare the activity of the endosome sorting receptor SORL1 carrying a disease-associated variant with an isoform lacking the disease exon. p.D1108N is shown to be a pathogenic and dominant-negative variant, and ASO-induced exon-skipping represents a therapeutic strategy to avoid Alzheimer’s disease for carriers.
Abstract Modern medicine increasingly extends beyond treating disease to supporting individuals through recovery, adaptation, and long-term functioning, all central to the concept of resilience. Despite growing interest, resilience remains conceptually difficult to structure and compare across disciplines, limiting its value for guiding research and clinical care. In this Perspective, we identify recurring features across many resilience traditions and propose a practical organizing scaffold for studying resilience in medicine. The scaffold specifies four elements that any resilience study should make explicit: the stressor, the system, resilience-relevant resources and processes, and outcomes. Using worked examples, we show how the scaffold clarifies resilience questions and improves comparability across studies. We also provide a step-by-step roadmap to help researchers new to the field design resilience studies, from framing a question through to interpreting results.
Collagen, a major component of the extracellular matrix, undergoes reorganization during tumorigenesis. However, visualizing collagen in tissue currently requires special stains (complicated by variability) or imaging (hindered by reliability, cost, and throughput). Here, we describe an approach to exhaustively detect and quantify collagen in hematoxylin and eosin (H&E)-stained images. Polarization images of picrosirius red-stained slides served as ground-truth training images for development of inferred quantitative multimodal anisotropy imaging (iQMAI), a deep learning model that infers collagen directly from H&E-stained whole-slide images. Individual collagen fibers were extracted after iQMAI inference, and features describing collagen intensity and fiber morphology were computed. iQMAI outputs were measured against polarization imaging and were compared to tissue composition, gene expression, and survival in lung adenocarcinoma, lung squamous cell carcinoma, hepatocellular carcinoma, and pancreatic adenocarcinoma datasets from the cancer genome atlas. iQMAI-derived collagen predictions and fiber features (fiber tortuosity, length, width, and relative angle) are correlated with polarization-based measurements. In pancreatic adenocarcinoma, fiber density and width are negatively associated with the LRRC-15 gene expression signature, and increased fiber width is associated with longer overall survival. iQMAI, a deep learning model identifying collagen in H&E-stained images, allows exhaustive, spatially-resolved quantification of collagen morphology, enabling investigation of the interplay between collagen and the tumor microenvironment. The association of collagen features with immunosuppressive fibroblasts and outcome in pancreatic adenocarcinoma demonstrates the potential of iQMAI. Understanding the relationship between collagen, tumor composition, and disease progression may aid the development of effective oncology therapies. Collagen is a structural protein found throughout the body. In diseases, including cancer, the amount and organization of collagen is altered. However, evaluating collagen in tumor pathology samples has required the use of special stains or imaging approaches, which can be variable and expensive, limiting widespread adoption. We developed a computational model (termed iQMAI) to predict and quantify collagen in images from routinely-stained tissue samples. This tool yields information about the amount of collagen present and the organization of collagen fibers. The collagen distribution measured by iQMAI in pancreatic cancer samples was found to be associated with poor outcomes. The use of this model has the potential to identify collagen patterns associated with outcomes for patients with cancer. Nguyen, Zhang et al. detail a deep learning approach to infer collagen from digitized H&E-stained whole slide images. In pancreatic adenocarcinoma, outputs of this model are associated with an immunosuppressive gene expression signature and overall survival.
