Citations

Abstract Immune checkpoint inhibitors (ICIs) can produce durable tumor control, yet many patients experience primary or acquired resistance, mixed lesion-level responses, or immune-related adverse events. Tissue and blood biomarkers provide biologically informative but spatially incomplete views of a heterogeneous and time-varying tumor–host system. Radiologic imaging complements these assays by repeatedly measuring whole-body tumor burden, organ-specific disease patterns, lesion-level heterogeneity, and treatment-related change. This critical narrative review evaluates imaging-centered artificial intelligence (AI) for precision cancer immunotherapy, including baseline and serial CT, MRI, and PET; radiomics and learned image representations; lesion-level and whole-body heterogeneity; toxicity assessment; and integration with pathology, genomics, circulating tumor DNA, laboratory measurements, and electronic health records. Representative cohorts range from early radiogenomic studies of approximately 100–200 patients to externally validated and longitudinal models involving hundreds or thousands of patients. Rather than treating all reported performance estimates as equally actionable, we appraise the clinical question, cohort design, endpoint, validation strategy, calibration, and intended clinical action. Current evidence supports noninvasive immune phenotyping, risk stratification, longitudinal monitoring, and multimodal biomarker integration. However, most models predict outcomes among patients already receiving ICIs and therefore remain prognostic under treatment rather than predictive of differential treatment benefit. Radiotherapy (RT) combined with ICI serves as a focused imaging use case because lesion identity, dose distribution, fractionation, sequence, and thoracic and normal-tissue exposure must be linked to defined actions. Existing patient-level AI studies estimate efficacy or toxicity after treatment selection; none has prospectively shown that an AI-selected RT strategy improves patient outcomes. Translation will require patient-level leakage control, calibration, geographic and temporal validation, robustness to missing modalities, and prospective evaluation of prespecified model-guided decisions.

Abstract Objective Current clinical risk models were developed for longer-term breast cancer risk stratification and therefore have limited utility for predicting short-term (1-year) breast cancer risk. This study evaluates the performance of a mammogram-based artificial intelligence (MB-AI) risk model for predicting breast cancer diagnosis within 1 year in high-risk women. Methods This retrospective case-control study included 340 women who underwent supplemental MRI screening within 1 year following a negative mammogram, of whom 101 (29.7%) developed breast cancer and 239 (70.3%) remained cancer-free during the same interval. Artificial-intelligence–assigned absolute 1-year risk scores, Tyrer–Cuzick (TC) lifetime and 10-year scores, and Gail lifetime and 5-year scores were analyzed. Model discrimination was evaluated using receiver operating characteristic (ROC) curves. Sensitivity, specificity, and predictive values were estimated at optimal cutoffs. Subgroup analyses examined model performance across genetic, menopausal, reproductive, and tumor-based strata. Results The MB-AI model demonstrated the highest overall discriminative ability (area under the curve [AUC] = 0.812), with strong performance in extremely dense breasts (AUC = 0.976). By comparison, TC lifetime (AUC = 0.625) and Gail lifetime (AUC = 0.668) models showed modest discrimination, while TC 10-year and Gail 5-year models did not discriminate (AUCs ≤0.49). Subgroup analyses showed consistent performance across clinical and pathological categories. The MB-AI model also showed an AUC comparable to that of previously developed AI models. Conclusions A mammogram-based AI model demonstrated improved performance compared with traditional clinical risk models for predicting breast cancer diagnosis within 1 year in women undergoing supplemental screening. These findings support that AI-based approaches may reflect both short-term risk and early imaging features of occult disease and suggest further investigation of AI as a complement to existing risk stratification strategies.

ABSTRACT Background & Aims Metabolic dysfunction‐associated steatotic liver disease (MASLD) is highly prevalent and increasingly recognized as a multisystem cardiometabolic disorder. The global burden, incidence and prognostic implications of coexisting chronic kidney disease (CKD) in MASLD remain uncertain. Methods In this meta‐analysis, we searched MEDLINE and EMBASE from inception through September 25, 2025, for observational studies reporting CKD prevalence and/or incidence among adults (≥ 18 years) with MASLD. The primary outcomes were pooled CKD prevalence, estimated using random‐effects models on the logit scale with the Hartung–Knapp adjustment, and pooled CKD incidence (per 1000 person‐years), estimated using random‐effects models with exact Poisson confidence intervals. Secondary outcomes included all‐cause mortality and cardiovascular, renal and cancer outcomes. Results From 2811 records, 36 observational studies met the inclusion criteria. Twenty‐three studies contributed prevalence estimates (575 615 participants; 56 248 CKD cases) and thirteen studies contributed incidence estimates (27 996 new events). The pooled global CKD prevalence among individuals with MASLD was 15.22% (95% CI 9.69–23.12). The pooled CKD incidence was 22.17 per 1000 person‐years (95% CI 11.44–32.89) in the MASLD population. Hypertension (OR 1.49, 95% CI 1.29–1.72), dyslipidaemia (OR 1.22, 95% CI 1.20–1.24) and diabetes (OR 1.94, 95% CI 1.69–2.22) were associated with higher odds of CKD in the MASLD population. All‐cause mortality rates were 18.28 per 1000 person‐years (95% CI 3.45–33.11) in coexistent MASLD and CKD, more than double those in the MASLD‐only population (7.26 per 1000 person‐years (95% CI 2.97–11.56)). Conclusions Coexistent CKD affects one in seven individuals with MASLD worldwide, conferring a more than two‐fold increased risk of all‐cause mortality compared with individuals with MASLD alone.

Introduction Large-scale production of mammalian cells is pivotal for applications in biotechnology, regenerative medicine, and therapeutic manufacturing. However, current bioreactor technologies face significant technical and economic challenges, including excessive cell aggregation, shear stress-induced cell death, batch-to-batch inconsistencies, and limited scalability. We propose that engineering a cell-friendly microenvironment can enhance culture efficiency. Previously, we developed alginate hydrogel microtubes (AlgTubes) that significantly improved cell density and growth rates; however, AlgTubes lack adhesion sites essential for anchorage-dependent cells and frequently break, causing cell leakage and production inconsistencies. Methods To address these limitations, we reinforced AlgTubes with collagen nanofibers, creating collagen-alginate hybrid hydrogel microtubes (ColAlgTubes). Collagen was integrated to form a dense nanofiber network interwoven with the alginate mesh, with the dual aim of enhancing mechanical properties and providing cell adhesion sites. ColAlgTubes were fabricated to maintain cell mass within a 400 μm diameter to ensure efficient nutrient exchange and waste removal. Results ColAlgTubes successfully formed a reinforced hybrid hydrogel structure in which the collagen nanofiber network improved the mechanical integrity of the alginate matrix while supplying adhesion sites for anchorage-dependent cells. The optimized 400 μm diameter microenvironment supported high cell viability, rapid proliferation, and exceptional yields of 5 × 10 8 cells/mL. Discussion These results demonstrate that ColAlgTubes overcome the key limitations of AlgTubes by combining structural reinforcement with biological functionality. The improved mechanical properties reduce tube breakage and cell leakage, while collagen-derived adhesion sites broaden compatibility to anchorage-dependent cell types. With their scalability, cost-effectiveness, and high efficiency, ColAlgTubes represent a transformative solution for large-scale mammalian cell production across biotechnology, regenerative medicine, and therapeutic manufacturing.

Abstract Background Accurate and efficient dose calculation is essential for online adaptive planning in proton therapy. Deep learning (DL) has shown promising dose prediction results for pencil beam scanning proton therapy (PBSPT) in recent years, but existing DL‐based dose prediction methods still suffer from limited generalizability and an inability to effectively handle outlier clinical cases. This may lead to inaccurate dose delivery to targets or excessive irradiation to organs at risk (OARs), thereby compromising the safety and efficacy of online adaptive proton therapy. Purpose To design a physics‐aware and generalizable AI‐based PBSPT dose prediction method that incorporates underlying physics to enhance generalizability, particularly in handling outlier clinical cases. Methods This study analyzed PBSPT plans of 103 prostate (93 for training and 10 for testing) and 78 lung cancer patients (68 for training and 10 for testing) from our institution, with each case comprising CT images and structure sets. Using the doses generated by our Monte Carlo‐based dose engine as the reference standard, we compared three methods: the region of interest (ROI)‐based method, the beam mask and sliding window method, and the proposed noisy probing dose method, which rapidly generates a low‐statistics dose via uniformly weighted spots on an expanded spot‐placement target volume without optimization. To evaluate the generalizability of these methods to rare treatment planning scenarios, 12 cases with uncommon beam angles or prescription doses were used to assess their performance, which was evaluated using dose‐volume histogram (DVH) indices, 3D Gamma passing rates (3%/2 mm/10%), and Dice coefficients for dose agreement, while prediction times were measured to gauge model efficiency. Results The proposed noisy probing dose method consistently outperformed the ROI‐based and beam mask baselines across all evaluated metrics, with more accurate dose agreement and superior generalizability. For DVH indices, the noisy probing dose method achieved the smallest deviation in clinical target volume (CTV) dose coverage: in prostate cancer, CTV D98 deviation was reduced by 45% (from 0.53 ± 0.22 Gy [RBE] for ROI‐based) and 29% (from 0.41 ± 0.28 Gy [RBE] for beam mask) to 0.29 ± 0.06 Gy [RBE]; in lung cancer, similar improvements were observed, with CTV D98 deviations reduced to 0.34 ± 0.12 Gy [RBE]. The 3D Gamma passing rates improved to 99.65% ± 1.15% for prostate targets and 97.04% ± 1.17% for lung targets. The dice coefficients of the 90% iso‐dose lines were also the highest with the noisy probing dose method (prostate: 0.983 ± 0.005; lung: 0.967 ± 0.01). For the 12 outlier cases, the noisy probing dose method maintained superior generalizability, yielding higher 3D Gamma passing rates (prostate targets: 96.79% ± 0.83%, OARs: 94.29% ± 1.01%; lung targets: 93.38% ± 1.34%, OARs: 93.95% ± 1.32%), demonstrating robust generalizability to rare clinical scenarios. The dose predictions for all testing cases were completed within 0.3 seconds. Conclusions A novel noisy probing dose method was proposed for PBSPT dose prediction in prostate and lung cancer patients. By embedding more proton‐specific physics, this method demonstrated an improvement in the generalizability of dose prediction.

Abstract Background and Aims Stroke and systemic embolism are recognized but poorly characterized complications of atrial fibrillation (AF) and left atrial flutter (AFL) catheter ablation. Their incidence, outcomes, and procedural associations remain insufficiently defined. Methods Retrospective global registry was initiated by the European Heart Rhythm Association Scientific Committee. A secure database captured data from ablation centres worldwide on general AF and left AFL ablation activity, as well as detailed information on patients with symptomatic embolic events between 2017 and 2024. Results Overall, 204 centres reported 335 743 ablation procedures and 550 embolic events (incidence 0.16%). Centre-level rate ranged from 0% to 3.1%. Most events were cerebral (94%), followed by peripheral (5%), and combined (1%). Coronary air embolism accounted for 62% of all peripheral events. Intraprocedural events occurred in 93 patients (17%), within 72 h in 357 (78%) and beyond 72 h in 100 (22%); 128 cases (23%) were diagnosed after hospital discharge. Nonparoxysmal AF (0.23% vs 0.15%; P < .001), centre experience, and ablation energy source (radiofrequency, 0.16%; cryoballoon, 0.16%; pulsed field ablation, 0.25%; laserballoon, 0.88%; P < .001) were associated with embolic risk. Median NIHSS score after stroke was 4 (interquartile range, 2–9). Three-month follow-up was available for 518 patients with embolism (95%): 181 (35%) had sequelae and 16 (3%) died. Conclusions Symptomatic embolic events are infrequent but associated with significant morbidity and mortality. Most events occur early, though a substantial proportion arise later. Higher event rates were observed with laser and possibly with pulsed field ablation. A greater centre experience was associated with a lower embolism incidence.

Abstract Background Current risk assessment tools for guiding direct oral anticoagulant (DOAC) therapy for patients with atrial fibrillation (AF) based on clinical risk factors demonstrate modest predictive performance limiting clinical impact. Additionally, while guidelines recommend periodic reassessment of risk over time, there remains an absence of modelling solutions for capturing evolving risk in AF patients. Methods Using UK electronic health records, we developed and validated the Transformer-based Risk assessment survival model (TRisk), an artificial intelligence model that predicts 12-month thromboembolic and bleeding events in AF patients by leveraging temporal patient journeys up to baseline. A cohort of 411,850 prevalent non-valvular AF patients aged ≥18 years between 2010 and 2020 was identified from 1,442 English general practices. Practices were randomly allocated to derivation (n=1,079) and external validation (n=363) cohorts. TRisk was compared with CHA 2 DS 2 -VASc and CHA 2 DS 2 -VA for thromboembolic event prediction, and HAS-BLED and ORBIT for bleeding prediction, with subgroup analyses by sex, age, and baseline characteristics. A second validation of TRisk was also performed on 16,218 US AF patients between 2010 and 2023. A decision model compared outcomes and healthcare costs for TRisk versus standard care. Findings TRisk achieved higher discrimination for thromboembolic event prediction (C-index: 0.82; 95% confidence interval [CI]: [0.81, 0.83]) as compared to CHA 2 DS 2 -VASc (0.71 [0.70, 0.73]) in UK validation. Application of TRisk to US data yielded similar C-index: 0.82 (0.80, 0.84). For bleeding prediction, TRisk (C-index: 0.70 [0.69–0.71]) outperformed both HAS-BLED (0.63; [0.61, 0.64]) and ORBIT (0.64; [0.63, 0.65]), with comparable US results (0.71; [0.69, 0.74]). The model remained well-calibrated across both populations and performed equitably across subgroups, including by race and during the COVID-19 pandemic. Impact analyses showed TRisk could reduce DOAC prescriptions by 8% in the UK and 7% in the US relative to guideline-recommended approaches, while preventing at least as many thromboembolic events. This refined approach would generate annual healthcare savings of £5.5 million and $456.2 million in the UK and US respectively among patients initiating DOACs, rising to £48.6 million and $1.8 billion when extended to all AF patients on DOACs. Interpretation TRisk enabled more precise prediction for both thromboembolic and bleeding events across AF populations in UK and US compared to established clinical scoring systems. Incorporating TRisk into routine AF care would result in substantial cost savings without compromising the identification of true high-risk patients. Funding None

Malignant glioma is considered invariably recalcitrant, thus necessitating development of new therapeutic approaches. Spontaneously arising canine glioma exhibits important molecular and pathological similarities to malignant pediatric glioma, making it an invaluable model for testing novel therapies. We evaluated safety and activity of RNA lipid particle aggregates (LPAs) designed to provoke cancer immunogenicity in 10 dogs with histologically confirmed glioma. Vaccine manufacturing was feasible in all subjects and generally well tolerated. Median survival time was 123 and 155 days in dogs receiving three (Group A) or four (Group B) vaccines, respectively; this was increased from historical reporting of subjects receiving palliative care. Responding animals had a robust cytokine response following infusions. In animals receiving non-specific RNA-LPAs, there was localization of messenger RNA (mRNA) payloads to the brain tumor microenvironment and a rapid shift in immunologic gene signatures consistent with a viral response. These data demonstrate the ability to induce rapid viremic response in the tumor microenvironment following a single peripherally administered mRNA vaccine. Since RNA-LPAs are amenable to repeat boosts, these formulations may overcome challenges of tumor immunosuppression and therapeutic access and are currently under evaluation in first-in-human trials ( NCT04573140 , NCT06389591 ).

ABSTRACT Transformed follicular lymphoma (tFL) is typically associated with chemotherapy resistance and a poor prognosis. There are limited data regarding outcomes after CD19‐directed chimeric antigen receptor T‐cell (CAR T) therapy in relapsed/refractory (R/R) tFL. A total of 923 adult patients with R/R tFL who received commercial CD19 CAR T therapy between 2017 and 2023 were identified in the Center for International Blood and Marrow Transplant Research registry. Median age was 64 years (range: 30–86) and median prior lines of therapy was 4 (range: 1–18). Most patients (78%) received axicabtagene ciloleucel, with 67% of patients having resistant disease at the time of CAR T infusion. At a median follow‐up of 25 months (range: 1–72) from CAR T infusion, the 2‐year overall survival (OS) was 57% (95% CI: 53–60) and progression‐free survival (PFS) was 43% (95% CI: 40–47). The 2‐year cumulative incidences of relapse or progression (rel/prog) and non‐relapse mortality (NRM) were 47% (95% CI: 44–51) and 9% (95% CI: 7–11), respectively. The overall response rate to CAR T was 76%, with a complete response rate of 63%. Grade ≥ 3 cytokine release syndrome (CRS) was observed in 7.1% and grade ≥ 3 immune effector cell–associated neurologic syndrome (ICANS) in 21.6% of patients. Multivariable analysis suggested that resistant disease status at the time of CAR T, use of bridging therapy, and high comorbidity index ≥ 3 were associated with inferior PFS and OS. Older age ≥ 60 significantly increased the risk of NRM. Our study suggests that CD19 CAR T is effective and safe for tFL.

Summary CD19 chimeric antigen receptor T‐cell (CAR‐T) therapy is a pivotal part of the treatment algorithm for large B‐cell lymphoma (LBCL) in the second‐ and third‐line settings based on numerous clinical trials. However, these studies included very few patients with a diagnosis of high‐grade B‐cell lymphoma, not otherwise specified (HGBCL‐NOS) and it is unclear if outcomes are similar to those observed with more common LBCL histologies. Using the Center for International Blood and Marrow Transplant Research (CIBMTR) registry, we identified 111 HGBCL‐NOS patients who received CAR‐T therapy. The cytokine release syndrome (CRS) rate was 74% (7.7% grade 3+), median onset was 4 days (1–17) and immune effector cell‐associated neurotoxicity syndrome rate was 45.2% (22.6% grade 3+), median onset was 7 days (1–17). At 2 years post‐CAR‐T infusion, the probability of overall survival and progression‐free survival were 41.6% and 28.7% respectively. The 2‐year non‐relapse mortality and relapse/progression were 3.1% (95% confidence interval [CI]: 0.6–7.6) and 68.1% (95% CI: 57.9–77.6) respectively. Our analysis illustrates that patients with HGBCL‐NOS represent a particularly difficult group to treat, even with CAR‐T therapy. We observed that one third of patients achieved a durable response; however, the overall results were less favourable, with lower overall response rate, overall survival and progression‐free survival as compared to published reports on LBCL.