Browse papers

176 papers

A perioperative multi-modal fusion and deep learning-based prognostic system for upper tract urothelial carcinoma: a multi-institutional study.

Peng X et al. · Jun 25, 2026

Objectives Precise perioperative risk stratification for upper tract urothelial carcinoma (UTUC) is essential. We developed a multimodal prognostic model integrating perioperative clinical data, radiomics, and deep learning (DL) features from baseline CT urography to improve survival prediction and guide adjuvant management. Materials and methods We retrospectively enrolled 623 patients from six institutions, divided into training, internal validation, and independent external validation sets. Four single-modal models (clinical, radiomics, 2D DL, and 2.5D DL) were developed, and an integrated combined model was constructed by fusing their prognostic scores. Performance was evaluated using the C-index, area under the curve (AUC), calibration curves, and decision curve analysis (DCA). Results The combined model consistently outperformed all single-modal models across all cohorts. C-indices reached 0.758 (95% CI: 0.712-0.804), 0.725 (95% CI: 0.651-0.798), and 0.704 (95% CI: 0.631-0.777) in the training, internal validation, and external validation sets, respectively, numerically surpassing the best single-modal models. Notably, our 2.5D DL model (C-index: 0.705) demonstrated a consistent incremental improvement over the 2D DL model (C-index: 0.681) in capturing prognostic information. In external validation, the combined model achieved a 3-year AUC of 0.766. DCA indicated the comprehensive model exhibited excellent calibration and provided the highest net benefits. Conclusion This multimodal system, featuring a robust 2.5D DL strategy, improves overall survival prediction in UTUC. It offers a valuable tool for accurate perioperative risk stratification immediately after radical nephroureterectomy, demonstrating particularly reliable value for 3-year intermediate-term clinical decision-making. Critical relevance statement This multimodal system advances clinical radiology by fusing perioperative clinical data, radiomics, and DL features from CTU images, enhancing risk stratification accuracy to guide postoperative adjuvant management for UTUC. Key points Current prognostic models for UTUC lack accuracy, creating an unmet clinical need for precise perioperative risk stratification to guide adjuvant management. A multimodal prognostic model fusing clinical, radiomic, and DL features from baseline CT urography consistently outperformed single-modality models in predicting overall survival.

Medicine

Pharmacovigilance Signal Detection and Mutually Exclusive Driver Mutations of the PI3K/AKT Pathway in Breast Cancer Treated With Capivasertib.

Luo Z et al. · Jun 25, 2026

Background and objective This study is aimed at comprehensively evaluating the real-world safety profile and underlying molecular mechanisms of the AKT inhibitor capivasertib by integrating pharmacovigilance data, computational biology, and multiomics analyses. Methods Adverse event (AE) reports from the FAERS database were analyzed using disproportionality algorithms (ROR, PRR, BCPNN, EBGM) to detect significant safety signals. To elucidate potential toxicological mechanisms, we employed network toxicology and molecular docking, further validated by 100 ns molecular dynamics (MD) simulations. Additionally, bulk genomic cohorts (e.g., TCGA, METABRIC) and single-cell RNA sequencing (scRNA-seq) datasets were utilized to assess mutation patterns and delineate key targets within the tumor microenvironment (TME). Results Analysis of 22,143 AE reports yielded 133 significant safety signals, predominantly involving metabolism (e.g., hyperglycemia), the gastrointestinal system (e.g., nausea, stomatitis), and dermatological conditions (e.g., rash). Pathway enrichment highlighted the PI3K-AKT, HIF-1, and EGFR signaling networks. Integrative analyses identified critical toxicity-related modulators, notably AKT1, IGF1, PTEN, TP53, and GSK3B. Crucially, MD simulations robustly confirmed the thermodynamic stability of the capivasertib-GSK3B complex. Genomic profiling revealed pronounced mutual exclusivity among PIK3CA, AKT1, and PTEN alterations. Furthermore, scRNA-seq analysis demonstrated that GSK3B overexpression defines a highly aggressive, proliferative malignant subpopulation that profoundly reshapes intercellular communication with stromal fibroblasts. Conclusion By seamlessly bridging real-world pharmacovigilance with advanced structural biology and single-cell transcriptomics, this study delineates the comprehensive safety landscape of capivasertib. Our findings provide crucial clinical alerts for AE monitoring and offer deep mechanistic insights to optimize personalized therapeutic management in breast cancer.

Medicine

Adenosquamous carcinoma of the lung: Comparative CT and pathological features versus adenocarcinoma and squamous cell carcinoma.

Yuan Q et al. · Jun 24, 2026

Objective Adenosquamous carcinoma (ASC) of the lung, containing both adenocarcinoma (AC) and squamous cell carcinoma (SCC) components, is associated with aggressive behavior and poor prognosis. Due to overlapping imaging features with AC and SCC, its preoperative diagnosis remains challenging. This study aimed to compare CT and pathological characteristics among ASC, AC, and SCC; identify key differentiators. Methods This retrospective study included 27 patients with pathologically confirmed ASC who underwent surgical resection from November 2018 to January 2025. Forty cases each of AC and SCC, matched for age, sex, and smoking history, were selected for comparison. Clinical, radiological, and pathological features were analyzed. Variables with statistical significance (P Results Compared with SCC, ASC lesions were more peripheral and associated with higher distance ratios (DR), pleural retraction, spiculation, and lymph node metastasis (LNM). Compared with AC, ASC tumors were larger, more often solid, and exhibited lobulation, bronchial cutoff sign, and advanced staging. Among the evaluated analyses, the combined pathology-imaging approach showed the highest discriminatory performance, with AUCs of 0.909 (ASC vs SCC) and 0.900 (ASC vs AC). LNM and DR were more strongly associated with ASC relative to SCC, whereas larger tumor size, ill-defined margins, and advanced stage were more strongly associated with ASC relative to AC. Conclusion Certain CT imaging features, especially DR and LNM, were associated with ASC and may provide useful information for preoperative radiological assessment and subtype stratification.

Medicine

Association between deployment and Gulf War Illness and adverse COVID outcomes in a nationwide cohort of 1990-1991 Gulf Era war veterans in the VA's Million Veteran Program.

White DL et al. · Jun 15, 2026

Background U.S. military veterans deployed to the Persian Gulf to serve in the 1990-1991 Gulf War (GWVs) have expressed concerns about potential increased risk of experiencing adverse COVID-related health outcomes given broad near-unique exposure to multiple diverse toxic agents during deployment and ongoing ~30% prevalence of Gulf War Illness (GWI+), a chronic, medically unexplained multi-symptom disorder associated with inflammation and immune dysregulation. Methods We conducted a retrospective study in the largest nationwide research cohort of 1990-1991Gulf War era veterans (GWEVs) enrolled in the VA's nationwide Million Veteran Program (MVP) to interrogate: 1) if deployed GWVs had increased risk for testing positive for novel SARS-CoV-2 infection (COVID+), COVID-related hospitalization, ICU admission, or death in 2020 (pre-vaccine period); and 2) if deployed GWVs with GWI+ had particularly increased susceptibility. COVID outcomes ascertained using VA's COVID Data Resource; GWI+ determined using CDC severe criteria in surveyed GWV subcohort. Findings Our overall GWEV cohort (N = 136,868) had mean age 60.7 years, with 84% male, 27% African-American, and 22% deployed; n = 26,141 COVID-tested in 2020. COVID test-positivity (COVID+) was slightly higher in deployed GWVs (18.8% vs. 17.2% in non-deployed GWEVs, p = 0.005). In multivariable logistic models, there were neither strong nor significant associations between deployment and testing COVID+, COVID-related hospitalization or ICU admission (adjusted [adj] ORs ranged from 0.90-1.06, all p-values >0.05), nor significant differences in COVID-related mortality. Preliminary analysis in our pre-pandemic surveyed deployed GWV subcohort (N = 1,643 COVID-tested, 39% GWI+, 18.4% testing-COVID+, 10.6% COVID+ hospitalized) suggested GWI+ potentially associated with significantly increased COVID-hospitalization risk (adjOR=2.21, 95%CI: 1.02-4.81, p = 0.04); though no significant excess was observed for COVID-related ICU admission or mortality. Conclusions Our findings demonstrated that overall deployment to serve in the 1990-1991 Gulf War was not associated with increased COVID-related health risks among Gulf War era veterans using VA healthcare system in early pre-vaccine pandemic era. Preliminary findings suggesting association between Gulf War Illness and potential increased COVID-related hospitalization risk among deployed Gulf War veterans in the early pandemic era warrants replication and ongoing evaluation to assess if it persists in the era of COVID vaccination, oral anti-COVID medications and new viral variants.

Medicine

Utility of sentinel surveillance as an early warning system for emerging pathogens in a low-resource setting: Correlations of dengue, chikungunya, Zika, and COVID-19 trends between sentinel and passive surveillance systems in Puerto Rico, 2012-2023.

Hernandez-Romieu AC et al. · Jun 12, 2026

The representativeness and timeliness of sentinel surveillance for endemic and emerging arboviral and respiratory diseases in low-resource settings are understudied. We compared laboratory-confirmed epidemic dengue, non-epidemic dengue, Zika, chikungunya, and COVID-19 (pre-Omicron and Omicron periods) cases reported in Puerto Rico's Sentinel Enhanced Dengue Surveillance System (SEDSS) with island-wide trends reported by the Department of Health's passive disease surveillance system (PADSS). We plotted trends over time to assess representativeness and used lagged cross-correlations to determine whether SEDSS reporting preceded PADSS. SEDSS trends were representative of island-wide trends for all pathogens. SEDSS preceded reporting in PADSS by up to three, eight, and two weeks for epidemic dengue, Zika, and pre-Omicron COVID-19, respectively. Increasing case trends for chikungunya occurred at broadly similar times in both systems, while temporal concordance was lower for non-epidemic dengue. In Puerto Rico, sentinel surveillance was representative of island-wide trends and could provide early warning for dengue epidemics and emerging diseases, such as Zika and COVID-19.

Medicine

Targeted cleavage site mutations in the Gn precursor enable efficient generation of replication-competent rVSV-based surrogates for emerging nairoviruses.

Jain S et al. · Jun 11, 2026

Orthonairoviruses are rapidly emerging, tick-borne viruses including Crimean Congo hemorrhagic fever virus (CCHFV), a highly pathogenic virus requiring biosafety level 4 (BSL-4) containment. Recently discovered nairoviruses such as Yezo virus (YEZV) cause febrile illness and are spreading across East Asia. No vaccines or therapeutics exist for these emerging nairoviruses. Recombinant vesicular stomatitis virus (rVSV) systems are promising vaccine candidates for CCHFV and enable neutralization studies in lower containment laboratories; however, efficient rescue of rVSV expressing CCHFV glycoproteins has been technically challenging. Nairovirus glycoprotein precursor (GPC) processing requires cleavage by host protease subtilisin kexin isozyme-1 (SKI-1/S1P) to generate mature Gn, with frequent adaptive mutations observed in the RRLL cleavage site during rVSV-CCHFV rescue. Here, we investigated the role of PreGn cleavage in generating replication-competent rVSV-CCHFV. Targeted mutation of the SKI-1 cleavage site, resulting in uncleaved PreGn and immature Gn expression, markedly improved rVSV-CCHFV rescue efficiency when combined with Gc cytoplasmic tail truncation. This approach enabled robust generation of replication-competent rVSV-CCHFV across two genetically distinct strains (IbAr10200 and Turkey). To assess generalizability, we extended analysis to Hazara virus (HAZV) and YEZV. Unexpectedly, we efficiently rescued both rVSV-HAZV and rVSV-YEZV with intact cleavage sites, while cleavage mutations reduced their replication efficiency, indicating virus-specific requirements. Using human convalescent and animal sera, rVSV-CCHFV provided reliable neutralization assays with results comparable to authentic CCHFV under BSL-4 conditions. Animal and human CCHF-positive sera exhibited low, yet occasionally measurable, cross-neutralizing activity against VSV-HAZV. These findings define strategies for generating replication-competent rVSV vectors displaying nairovirus GPC and provide opportunities for studying neutralization in lower containment facilities.

Medicine