Citations

Background Grade group 5 (GG5) prostate cancer (PCa) carries a less favorable prognosis after standard-of-care (SOC) therapy, necessitating novel therapeutic approaches. High-dose rate brachytherapy (HDRBT) and androgen deprivation therapy (ADT) may modulate immune response in GG5 PCa, particularly in tumors with increased immune content. This study evaluated whether the addition of nivolumab to SOC was associated with improved disease control in patients with high-volume GG5 PCa, including those with oligometastatic disease. Methods In this non-randomized phase II trial, 31 patients with localized or oligometastatic GG5 PCa and >30% positive biopsy cores were evaluated between September 2018 and April 2021. Patients received four doses of nivolumab (240 mg every 2 weeks) beginning 4 weeks prior to HDRBT, alongside ADT, HDRBT, and external beam radiation. The primary endpoint was to evaluate whether the 2-year freedom from biochemical recurrence (FFBR) rate would exceed a prespecified historical control rate of 75%. Results Among the 31 patients, the median follow-up was 38.8 months (IQR 31.0–46.5 months). The addition of nivolumab to SOC RT with ADT was associated with a 2-year FFBR rate of 90.3% (95% CI 74.3% to 98.0%) (median FFBR not reached), exceeding the prespecified historical control rate of 75% (one-sided p value from binomial test=0.024). Definitive and probable nivolumab-related toxicity included 6.3% acute grade 2 and 6.3% acute grade 3 adverse events (AEs), with no grade 4+ AEs observed. A higher Decipher immunosuppression score at diagnosis correlated with early pathologic response (p=0.005) and was independently associated with time to metastatic failure (p=0.044). Conclusions Nivolumab combined with SOC was associated with encouraging FFBR in this high-risk GG5 PCa population and may represent a promising therapeutic intensification strategy. The Decipher immunosuppression score may serve as a predictive biomarker for response. These findings warrant further investigation in randomized trials.

Combined histopathological and metabolomic analysis of the same tissue specimen can reduce biological variability compared to independent analyses of similar samples. However, conventional workflows are incompatible: tissue fixation leads to metabolite loss through dilution and chemical reactions with fixative, while metabolomic extraction typically destroys the specimen. We propose a method that reconciles these conflicting requirements by enabling metabolite recovery from fixative while preserving tissue for histopathology. In this approach, tissue is initially fixed in a small volume of fixative; an aliquot of the fixative is then collected for nuclear magnetic resonance-based metabolomic analysis before additional fixative is added for standard processing. As a proof-of-concept, mouse kidneys from the same subject were analyzed using two protocols: standard perchloric acid extraction (ensuring complete metabolite recovery but destroying tissue) and metabolite extraction from fixative. Several metabolites were fully recoverable from the fixative, others partially, while some were lost due to chemical reactions. Despite some limitations, this strategy may be useful for analyzing precious specimens, such as human biopsies, that must remain intact for further studies. Moreover, using the same specimen for dual analyses reduces the number of animals required and enhances statistical robustness.

Objective To examine the potential errors of a general large language model (LLM) (ie, Claude 3.5 Sonnet) on data extraction from randomised controlled trials (RCTs). Design and setting An empirical study comparing Claude 3.5 Sonnet extractions against a human-performed verification dataset. The extraction tasks for Claude 3.5 Sonnet were based solely on original RCT portable document format (PDF) files. For PDFs that could not be directly extracted by Claude 3.5 Sonnet, optical character recognition was employed to convert them into text format before extraction. Participants A random sample of 664 trials was selected from a well-established trial bank and a final data pool was established based on rigorous manual cross-checking as a reference standard. Data sources PubMed, EMBASE, Scopus, Web of Science (all databases) and the Cochrane Central Register of Controlled Trials (CENTRAL) up to February 2023. Eligibility criteria for selecting studies RCTs on children involving medication and adverse events. Main outcome measures Claude 3.5 Sonnet was applied to extract the basic information (eg, trial design, population information and source of funding) and adverse outcomes (ie, name of adverse events, number of events). Claude 3.5 Sonnet outputs were compared against the final data pool and all errors were recorded. Results are presented as error rates and with 95% CI, estimated using a generalised linear mixed model. Results For the 664 trials, a total of 23 069 data cells were extracted via Claude 3.5 Sonnet, with 10 624 for basic information and 12 445 for adverse outcomes. The overall error rate for data extraction was 6.6% (95% CI 5.4% to 8.2%), with 5.7% (95% CI 5.2% to 6.1%) in basic information and 7.6% (95% CI 4.9% to 11.8%) in adverse outcomes. When stratified the 1542 total errors by error types, misallocation (assigning data to incorrect fields; 57.1%, 881/1542) and missed or omitted data (incomplete extraction of available data; 23.2%, 357/1542) accounted for the two most frequent errors, with misallocation occurring more in basic information (53.3%, 470/881), while missed or omitted data occurred more in adverse outcomes (96.1%, 343/357). Post hoc analysis examining the association between trial reporting quality (assessed using Consolidated Standards of Reporting Trials (CONSORT) 2025 and LLM data extraction error rates indicated that higher CONSORT adherence was associated with lower extraction error rates. Conclusions The data extraction error of Claude was relatively low, but it alerts LLM applications in evidence synthesis. Detailed checking for LLM outputs should be the primary consideration for evidence synthesisers.

Poststenotic dilatation in renal artery stenosis, evaluated at abdominal CT, might be related to stenotic kidney damage and inflammation, particularly to monocyte chemoattractant protein-1 release.

INTRODUCTION: Nosocomial infections (NIs) in cirrhosis are associated with high mortality but could be preventable. Logistic regression (LR) models have failed to identify high-risk patients. We aimed to develop machine learning (ML) models to predict NI. METHODS: The CLEARED consortium consists of prospectively enrolled cirrhosis inpatients from >120 centers. Using day-of-admission clinical data, 3 ML approaches (random forest [RF], extreme gradient boosting, and neural networks [NNs]) were used to predict NI. Data were split 80:20 for training and testing stratified by the outcome. Models were compared using area under the receiver operating characteristic curve (AUC). RESULTS: In total, 8,263 patients (55.90 ± 13.34 years; 64.1% men) from 127 centers in 37 countries were included. NI developed in 869 (10.5%), a median of 6 (4–11) days of postadmission. Major NIs were respiratory (29.6%) and urinary tract infection (15.7%), and spontaneous bacterial peritonitis (13.5%). NIs occurred more frequently in patients from low/low-middle income countries and those with severe liver disease, alcohol etiology, and admission infections. NIs were associated with inpatient mortality (31.9% vs 8.1%, P < 0.001) and liver transplantation (4.8% vs 1.9%, P < 0.001). Although the RF model (AUC 0.69) showed good calibration (Brier score 0.09), outperforming extreme gradient boosting, neural network, and LR models (AUC 0.66 for all; LR comparison P = 0.043), no model achieved AUC ≥0.80 for clinical utility. At 10% predicted probability threshold, the RF model demonstrated only 75.4% sensitivity, 52.9% specificity, and 15.9% positive predictive value (PPV). DISCUSSION: NIs cannot be accurately predicted from day-of-admission data using ML models, even in a large, prospective, global cirrhosis cohort. Every hospitalized patient with cirrhosis should receive protocolized infection control measures.

Chimeric antigen receptor (CAR) T-cell therapy is an effective treatment for hematologic malignancies. However, it is limited by high costs, risk of severe toxicities such as cytokine release syndrome and neurotoxicity, and heterogeneous patient responses. The current therapy monitoring depends largely on subjective symptom assessment, routine laboratory tests, and basic vital signs, without real-time, quantitative evaluation of CAR T-cell expansion or activation in clinical practice. This lack of timely immune monitoring hampers individualized care and contributes to increased treatment costs. To address this need, we present a proof-of-concept, label-free rapid optical imaging (ROI) biosensor with automated machine learning analysis for direct quantification of CAR T-cells from whole blood. This microfluidic platform integrates red blood cell (RBC) removal, CAR T-cell capture, and imaging-based quantification on a single chip, eliminating the need for centrifugation, staining, and operator-dependent interpretation. For validation, 50 μL whole blood samples spiked with Jurkat cells expressing CD19 CARs underwent RBC depletion by agglutination and microfiltration. The remaining blood components were then incubated on a sensor chip functionalized with recombinant CD19 protein. Captured CAR T-cells were imaged by brightfield microscopy and automatically enumerated using a machine learning algorithm trained on fluorescence-validated cells. The CD-19 cells’ capture performance was validated by flow cytometry and fluorescence imaging. The trained machine learning model validated at 88% sensitivity and 96% specificity. Buffer and whole blood calibration curves were established across clinically relevant concentrations (1–1000 cells/µL) with triple replicates. The results showed high correlation (0.975 and 0.990 R2) between the spiked concentration and the detected CAR T-cells, with a 95% certainty limit of detection (LOD) and quantification (LOQ) of 0.6 and 1.1 cells/µL for spiked buffer, and 14 and 67 cells/µL for spiked whole-blood, respectively.

Abstract Background Accurate synthesis of computed tomography (CT) images from magnetic resonance imaging (MRI) is clinically valuable for cranial applications such as attenuation correction, radiotherapy planning, and image‐guided interventions. However, the heterogeneity of MRI data across magnetic field strengths and pulse sequences limits the generalizability of existing methods, posing a barrier to clinical translation. Purpose This study aims to reformulate cranial CT synthesis as a modular, structurally coupled problem using a deep learning approach, enhancing adaptability across heterogeneous MRI conditions, including different field strengths and sequence protocols. Methods We implemented a cascaded multitask pipeline that jointly models skull segmentation and Hounsfield Unit (HU) regression in anatomically targeted regions. A 3D patch‐based training paradigm shifts the modeling focus from global image translation to localized bone feature extraction. The backbone leverages a residual Mamba‐based state space model within a 3D U‐Net structure to improve spatial representation, while a Transformer U‐Net serves as a widely recognized reference baseline comparison. The model was trained and evaluated in terms of multi‐modal (T1‐weighted and T2‐FLAIR) and cross‐domain transferability from a public 1.5 Tesla brain dataset ( n = 37) to an independent 7 Tesla clinical brain dataset ( n = 44). Performance was assessed using Dice and Jaccard indices for segmentation accuracy and mean absolute error (MAE) for HU regression. Results Quantitative analysis showed that our multitask pipeline, incorporating both morphological and HU map prediction stages, significantly outperformed conventional direct MRI‐to‐CT mapping across all metrics on the public 1.5T dataset (p < 0.05). Consistent performance on an external 7T clinical dataset (p < 0.001) further demonstrated the method's robustness and adaptability across field strengths. Conclusion These findings support that task‐structured training for modality transformation markedly improves both accuracy and generalizability of cranial CT synthesis across heterogeneous MRI conditions. The observed consistency across field strengths validates the robustness of the proposed methodology.

Abstract Cerebral blood flow is stabilized through dynamic adjustments across both macro‐ and microvascular compartments. While dynamic cerebral autoregulation (dCA) quantifies upstream pressure–flow coupling, downstream microvascular responses are less well characterized and may represent a distinct but functionally linked process. This study tested whether a two‐component cascade model, treating dCA and an empirically derived index of microvascular function (MF) as sequential stages can represent the integrated regulation of cortical oxygenation. Data from 41 healthy adults (20–45 years) were analysed. Beat‐to‐beat mean arterial pressure (MAP), middle cerebral artery flow velocity (CBFV) and cortical oxyhaemoglobin (O 2 Hb) were recorded during supine spontaneous oscillations and forced oscillations at 0.05 Hz using repeated sit–stand manoeuvres. Transfer function analysis quantified frequency‐domain gain, phase and coherence for MAP→CBFV (dCA), CBFV→O 2 Hb (MF) and MAP→O 2 Hb (total pathway). The cascade model was computed as the product of dCA and MF transfer functions. The cascade model derived indices showed strong correlations with total pathway gain, phase and coherence measures during both spontaneous and forced oscillations, with improved linear coupling under forced oscillations. These results support the applicability of a two‐component cascade model for integrated cerebrovascular regulation and suggest that serial interactions between macrovascular and microvascular regulatory mechanisms jointly shape the frequency‐dependent propagation of arterial pressure to brain‐tissue oxygenation dynamics. image Key points Cerebral blood flow is stabilized by coordinated regulation across large arteries and microvessels, but their dynamic interaction has not been experimentally modelled. We examined whether a two‐component cascade model linking upstream dynamic cerebral autoregulation and downstream microvascular function can explain how blood pressure oscillations influence cortical oxygenation. In 41 healthy adults, we recorded beat‐to‐beat blood pressure, middle cerebral artery blood flow velocity and near‐infrared spectroscopy‐derived oxygenation during rest and sit–stand manoeuvres, and analysed them using transfer function analysis. Gains, phases and coherences derived from the cascade model closely matched those from the direct blood pressure–O 2 Hb relationship, particularly under forced oscillations during sit–stand, demonstrating the model's physiological relevance. The cascade model provides a mechanistic framework to separate and quantify large‐ and small‐vessel contributions to cerebral blood flow regulation, with potential application in future studies of ageing and cerebrovascular disease.

Background: The Spinal Cord Injury Model Systems (SCIMS) began studying traumatic SCI (TSCI) in 1970. The National Institute of Disability, Independent Living and Rehabilitation Research (NIDILRR) provided pilot funding during the 2021-2026 SCIMS funding cycle to perform an initial study of individuals with nontraumatic spinal cord injury (NTSCI). Objectives: To describe the demographic and etiological profiles and inpatient rehabilitation outcomes of the SCIMS NTSCI pilot cohort. Methods: This was a descriptive study of adults with NTSCI admitted to SCIMS centers who consented to medical records abstraction and completed survey questionnaires near the completion of inpatient rehabilitation. Results: One hundred eighty-seven people with NTSCI consented to participate and provided data. The etiology of patients’ NTSCI was degenerative vertebral causes ( n = 67), tumor-malignant ( n = 29), tumor-benign/unknown ( n = 16), vascular ( n = 28), infection ( n = 26), and other NTSCI causes ( n = 21). The average age was 59.2 years. Over 40% of the population was age 60-74 years. The population was 35% female, 59.5% non-Hispanic White, 21.6% non-Hispanic Black, and 11.4% Hispanic. Age, sex, race, and ethnicity distributions did not differ significantly across etiology groups. Only symptom onset time ( P = .002) and marital status ( P = .005) were significantly different between groups. Vascular disorders onset was less than a day in 60% of cases. Greater than 60% of tumor-malignant, vascular, and other NTSCI individuals were married or living with a partner. Conclusion: The US SCIMS program can recruit a population of NTSCI research participants that is consistent with previously published results. Ongoing study of NTSCI through the US SCIMS could provide important data akin to that available regarding TSCI.