medRxiv : the preprint server for health sciences最新文献

Molecular subtyping of cancer is traditionally defined in transcriptomic space, yet routine clinical deployment is limited by the availability and cost of sequencing. Meanwhile, histopathology captures rich morphological information that is known to correlate with molecular state but lacks a principled, mechanistic bridge to gene-level representations. We propose a graph-constrained learning framework that aligns morphology-derived signals with a fixed, data-driven gene network discovered via hierarchical Monte Carlo screening. We can derive new gene sets for classification using random sampling, and use the coexpression network of that graph to enforce the learning of a pure morphology model without using gene expression. The resulting model performs subtype prediction using morphology alone, while being explicitly forced to operate through a gene-structured latent space. Structural alignment is enforced during training. For Moffitt classification in pancreatic cancer using PANCAN and TCGA datasets, the model has a reported 85% AUC using an alternative gene set's network structure, while the alternate gene set itself has an 84% AUC in all patients that were classified with subtyping with pancreatic cancer in the dataset. This framework demonstrates that virtual transcriptomics can provide biologically grounded molecular insights using only routine histopathology slides, potentially expanding access to precision oncology in resource-limited settings.

Accurate classification of BRCA1 and BRCA2 variants is essential for cancer risk assessment and therapy selection, yet over one-third remain variants of uncertain significance (VUS). Here, using 120,660 real-world cancer genomic profiles with BRCA1 or BRCA2 variants from a >800,000-sample cohort, we develop machine learning models that predict pathogenicity using clinical and tumor-derived features, including a pan-cancer homologous recombination deficiency signature, co-mutated genes, zygosity, and cancer type. Trained on classified variants from ClinVar, our models achieved near-perfect performance, with validation ROC-AUC of 1.000 for BRCA1 and 0.989 for BRCA2 variants with ≥5 observations, translating to strong benign or pathogenic evidence for VCEP classification. Applying these models to 1,073 BRCA1 and 1,639 BRCA2 VUS, we strengthened or enabled classification of 39.48% BRCA1 and 50.52% BRCA2 assessable variants. This approach transforms underutilized tumor profiling data into evidence that can be directly integrated into variant classification, providing a scalable framework for other tumor profiling datasets and cancer genes associated with defined tumor genomic features.

Actigraphy is a popular behavioral sleep assessment tool in research and clinical practice. Hierarchical hand-scoring approaches remain the standard for actigraphy rest interval estimation, but can be impractical for large cohort studies and suffer from reproducibility problems. We developed a semi-automated pipeline (actiSleep) to set rest intervals consistent with best-practice hand-scoring algorithms incorporating event marker, diary, light, and activity data. To evaluate actiSleep performance, we used data from an observational study of 51 adolescents (14-19yr), with and without family history of bipolar disorder. Participants completed 2 weeks of wrist actigraphy and daily sleep diary. We first hand-scored records using a standardized hierarchical algorithm incorporating event marker, diary, light, and activity data. We then compared the hand-scored rest intervals to those from actiSleep and two automated activity-based algorithms ('Activity-Merged', 'Activity-Only'). Activity-Only used activity-based sleep estimation and Activity-Merged joined closely adjacent rest intervals. For rest onset, rest offset, and rest duration, all algorithms had strong mean agreement with hand-scoring: actiSleep estimates were within 1-3 minutes, Activity-Merged within 2-4 minutes, and Activity-Only within 7-14 minutes. However, actiSleep had notably better (narrower) margins of agreement with hand-scoring, as evidenced by Bland-Altman plots, and greater positive predictive value and true positive rates for rest detection, especially in the 60 minutes surrounding the onset and offset of the rest interval. The actiSleep algorithm successfully estimates actigraphy rest intervals comparable to hand-scoring while avoiding pitfalls of activity-only algorithms. actiSleep has potential to replace hand-scoring for research in adolescents but requires further testing and validation in other samples.

Serological surveillance is fundamental to infectious disease research and informed public-health decision making. Immunoassays used in the study of pathogen-specific immunity have historically relied on the collection of venous blood. While critical for many public-health applications, this sample collection method is invasive and resource intensive. The costs and logistical barriers associated with venous blood collection are exacerbated in resource-limited regions, and the shift to less invasive sampling methods would increase sample availability for pathogen surveillance and study of pathogen-specific immunity. To this end, we have developed and optimized a self-collected, saliva-based immunoassay capable of quantifying pathogen-specific antibody binding in saliva samples. Using samples collected from geographically and epidemiologically diverse regions of the world, we compared antigen-specific IgG levels in paired plasma and saliva samples. We observed that levels of IgG against multiple pathogens of public health concern - including SARS-CoV-2 and dengue virus (DENV) - were highly correlated in plasma and swab-collected saliva. In addition, the decay of maternally derived antibodies in saliva samples collected from infants was readily observed using this immunoassay, demonstrating the assay's sensitivity and potential for use in measuring antibody kinetics. We posit that this assay represents a climate stable, non-invasive tool that can aid in the surveillance and study of pathogen-specific immunity across a broad range of public-health indications.

Acute kidney injury (AKI) and chronic kidney disease (CKD) are two interconnected clinical conditions, both defined by degree of functional impairment, but with heterogeneous clinical trajectories. Using new transcriptomic technologies, recent studies have described the cellular diversity in the healthy and injured kidney at the single cell level. Here, we used single nucleus transcriptomics to investigate the molecular diversity and commonalities in kidney biopsies from over 150 participants with AKI and CKD enrolled within the Kidney Precision Medicine Project (KPMP) and did so at the patient participant level. Using an unsupervised approach, we identified two multi-cellular programs associated with clinical and histopathological features of acute injury and chronic damage, respectively. We found that these programs are expressed across patients with AKI and CKD, supporting shared, rather than distinct, underlying molecular mechanisms. These programs capture tissue-level compositional changes towards adaptive and failed-repair states in tubular epithelial cells, as well as intra-cellular molecular changes characteristic of stress in all cell types. We identified subunits of the NFkB and AP-1 complexes, as well as members of the STAT family, as putative upstream regulators of the acute and chronic programs. We were able to map these continuous molecular measures of acute injury and chronic damage to urine and plasma protein profiles obtained at time of biopsy. These non-invasive protein signatures were predictive of renal outcomes in an independent cohort of 44 thousand participants from the UK biobank. In summary, unbiased identification of cellular programs in kidney disease biopsies defined molecular programs of injury cutting across conventional disease categorization and established a non-invasive molecular link to long term patient outcomes.

Objective: Posttraumatic stress disorder (PTSD) after a traumatic birth is a serious but overlooked maternal morbidity, affecting ∼20% of women following medically complicated deliveries. PTSD can undermine maternal caregiving. Rapid screening tools suited to busy obstetric settings are lacking. We developed and evaluated a brief screener, derived from the 20-item PTSD Checklist for DSM-5 (PCL-5), to identify PTSD related to childbirth.

Study design: We enrolled 107 women with traumatic childbirth. Participants completed the PCL-5 and the gold-standard clinician diagnostic interview for PTSD (CAPS-5); depression was measured with the Edinburgh Postnatal Depression Scale (EPDS). Bootstrap resampling with LASSO regression identified PCL-5 items most associated with PTSD. Firth logistic regression models estimated diagnostic accuracy. Sensitivity, specificity, area under the ROC curve (AUC), and Youden's J statistic determined performance and optimal cut-off.

Results: A six-item version of the PCL-5 (PCL-5 R6), statistically derived from the full scale, showed excellent discrimination for PTSD compared with clinician evaluation (AUC = 0.95; 95% CI, 0.89-1.00). A cut-off score of 7 yielded high sensitivity (0.96) and good specificity (0.83), with an overall diagnostic efficiency of 0.86, detecting most PTSD cases while minimizing false positives. The PCL-5 R6 correlated moderately with the EPDS (rho = 0.53), showing that a depression screen alone cannot reliably detect PTSD.

Conclusions: A short, 6-item PCL-5 provides a valid, efficient tool for detecting childbirth PTSD. Its brevity and accuracy make it practical for integration into routine postpartum care, enabling timely mental health screening.

Background: Cognitively unimpaired (CU) adults vary substantially in their risk of developing mild cognitive impairment (MCI), yet most subtyping approaches focus on downstream neurobiological or cognitive markers rather than upstream, modifiable risk factors. We aimed to identify clinically meaningful subgroups of CU adults defined by integrated comorbid, behavioral, and social risk profiles, and to evaluate heterogeneity in both incident MCI risk and cardiometabolic treatment effects.

Methods: We conducted a prospective cohort study of 121,322 CU adults aged ≥50 years from the All of Us Research Program. Baseline comorbidities, lifestyle behaviors, and social determinants of health were jointly modeled using the Bayesian Mixed Integrative Data Subtyping framework, which integrates binary and continuous modalities via modality-specific likelihoods and shared latent constructs. Subtype-specific risk of incident MCI was assessed using multivariable Cox proportional hazards models adjusting for demographics and baseline medication use. A double/debiased machine learning interactive regression model with inverse probability of censoring weights to mitigate bias from informative censoring was implemented to estimate the average treatment effects of antihypertensive agents, Glucagon-Like Peptide (GLP) receptor agonists, and non-GLP antidiabetic medications on time to MCI.

Results: Four distinct subtypes were identified: I low-risk healthy aging, II behavioral/social vulnerability, III cardiometabolic-depressive multimorbidity, and IV mixed social-medical vulnerability profiles. Compared with Subtype I, Subtype III demonstrated the highest risk of incident MCI (HR: 3.69, 95% CI: 3.14-4.33), followed by Subtype IV and Subtype II. In treatment effect analyses, antihypertensive use was associated with a modest prolongation of MCI-free survival overall (time ratio:1.04, 95% CI: 1.03-1.06), with the largest benefit observed in Subtype III (time ratio: 1.14, 95% CI: 1.09-1.19). Non-GLP antidiabetic therapies were similarly associated with modest overall delay, with significant benefits in Subtypes I and III. GLP-class therapies were not associated with overall delay but showed a significant association in Subtype III.

Conclusions: Integrative subtyping based on comorbid, behavioral, and social risk factors reveals clinically meaningful heterogeneity in both cognitive risk and treatment response. Aligning dementia prevention strategies with dominant vulnerability pathways may enhance the effectiveness and equity of population-level precision prevention.

Individuals with chronic kidney disease (CKD) have higher rates of hip fracture and post-fracture mortality. Although they may develop age-related osteoporosis similar to those without CKD, they may also exhibit CKD-related metabolic bone disease (MBD), characterized by low, high, or mixed turnover at similar levels of bone mineral density (BMD). Because BMD does not provide information about turnover status, clinical decision-making is challenging. This study evaluated the associations between circulating bone-turnover biomarkers and static histomorphometry in patients undergoing hip-fracture surgery. In this cross-sectional study, we enrolled adults with and without CKD, defined as estimated glomerular filtration rate (eGFR) ≤60 ml/min/1.73m² (CKD-EPI 2021), undergoing hip-fracture surgery. Blood samples, bone specimens from the femoral head or greater trochanter, and demographic and clinical data were collected at the time of surgery. Plasma biomarkers included α-Klotho, bone alkaline phosphatase (BAP), dickkopf-related protein 1 (DKK-1), fibroblast growth factor 23 (FGF23), tartrate-resistant acid phosphatase 5b (TRAP5b), parathyroid hormone (PTH), and sclerostin. Logistic regression models, adjusted for age, gender, eGFR, and osteoporosis, assessed associations with CKD status. Tertiles of osteoblast surface (Ob.S/BS) and eroded surface (ES/BS) were defined in participants without CKD and applied to the full cohort. Multinomial and multivariable linear regression evaluated associations of biomarkers with these histomorphometry parameters. Among 97 enrolled participants (mean age 80 ± 11 years; 67% female), 68% had CKD. Of 75 with complete biomarker and histomorphometry data, 96% demonstrated low bone turnover. CKD was associated with lower trabecular thickness (Tb.Th) and higher osteoid thickness (O.Th), osteoid volume (OV/BV), and osteoid surface (OS/BS), suggesting thinner, largely unmineralized trabeculae. Higher BAP (222.2% difference per doubling; 95% CI 77.2-485.8) and TRAP5b (319.3%; 95% CI 128.3-669.5) were directly associated with Ob.S/BS and ES/BS, whereas sclerostin was inversely associated with ES/BS (-28.9%; 95% CI -44.8 to -7.1). PTH was not associated with bone-turnover measures. These findings suggest that BAP, TRAP5b, and sclerostin may provide useful adjunct information alongside PTH for assessing bone turnover and guiding therapy in patients with and without CKD.

Background: While growing evidence implicates sleep-wake and circadian rhythm disturbances (SCRDs) in the onset and course of mood and psychotic disorders, longitudinal studies using objective measures are limited. This clinical cohort study examined whether actigraphy-derived SCRDs (sleep duration, timing, and efficiency) predicted transition to (i) any full-threshold mental disorders; and then specifically: (ii) full-threshold bipolar or psychotic disorders or (iii) other full-threshold (i.e. depressive or anxiety) disorders, in youth accessing mental health care.

Methods: Actigraphy monitoring was completed for 5-23 days in 250 participants (aged 12-30) presenting to youth-focused early intervention services in Sydney, Australia. Participants were followed longitudinally as part of the Optymise cohort for 6+ months (up to 8 years; median 2.5 years). Logistic regression and Cox proportional hazard models estimated associations between SCRDs and illness progression, after controlling for relevant baseline clinical and demographic covariates (e.g., age, sex, social and occupational functioning, mania-like and psychotic-like experiences, medication use).

Results: Longer sleep duration at baseline predicted higher odds of transition (OR = 2.23 [95%CI = 1.38-3.74]), and shorter time-to-transition (HR = 2.05 [95%CI = 1.23-3.40]) to full-threshold bipolar or psychotic disorders. This effect remained significant after controlling for clinical covariates. Later sleep midpoint predicted transition to any full-threshold mental disorder (OR = 1.46 [95%CI = 1.02-2.17]) at the uncorrected significance level.

Conclusions: Excessive sleep duration may represent an early marker of vulnerability for progression to severe mental illness. Findings support the prognostic utility of objective measures of SCRDs to guide indicated prevention and early intervention.