Citations

Abstract Aims Lipoprotein(a) [Lp(a)] is a causal risk factor for cardiovascular events. However, the effect of Lp(a) on coronary plaque composition and high-risk plaque features has not been fully characterized. This study aimed to investigate the association between Lp(a) and coronary atherosclerotic plaque phenotype at the plaque level. Methods and results This study included 710 patients who underwent coronary computed tomography angiography (CCTA) and had Lp(a) measured between 2008 and 2024. CCTA scans were analysed with a previously validated artificial intelligence-based algorithm (AI-QCT, Cleerly Inc.). The association of Lp(a) with non-calcified and calcified plaque volumes and HRP features was evaluated at the plaque level using generalized estimating equation models adjusted for traditional cardiovascular risk factors. Among the 710 patients, the mean age was 56 ± 10 years and 379 (53%) were male. In total, 3642 plaques were identified. In the adjusted plaque-level analysis, elevated Lp(a) (≥150 nmol/L) was associated with an increase in non-calcified plaque volume per plaque (P = 0.009), but not with an increase in calcified plaque volume (P = 0.81). Furthermore, elevated Lp(a) (≥150 nmol/L) was strongly associated with the presence of high-risk plaque at the plaque level (adjusted odds ratio: 1.78, 95% CI: 1.25–2.54, P = 0.001), whereas low-density lipoprotein-cholesterol and high-sensitivity C-reactive protein were not (P > 0.05). Conclusion In this plaque-level CCTA study of 3642 individual plaques, elevated Lp(a) levels were independently associated with increased non-calcified plaque volume and with the presence of high-risk plaque, but not with calcified plaque volume. These findings elucidate the impact of Lp(a) on unstable plaque phenotypes, even in patients without prior cardiovascular events.

Background: Agentic artificial intelligence (AI) systems, characterized by autonomous goal-directed behavior, multi-step reasoning, task decomposition, and tool use, are increasingly proposed for healthcare applications. However, their autonomy raises concerns regarding transparency, accountability, and human oversight. While explainable AI (XAI) has been widely studied in traditional predictive models, less is known about how explainability is implemented within agentic architectures. Objective: To map the emerging literature on explainable agentic AI (XAAI) in healthcare and characterize the types, scope, and forms of explainability used in these systems. Methods: A scoping review was conducted following PRISMA-ScR guidelines. PubMed, Embase, IEEE Xplore, and ACM Digital Library were searched through November 2025. Eligible studies described healthcare-related agentic AI systems incorporating explicit explainability mechanisms. Data were extracted on system architecture, explainability type (intrinsic, post hoc, hybrid), explanation scope (local, global), explanation form, and reported clinical outcomes. Results: Nine studies met the inclusion criteria. All systems demonstrated core agentic features, including autonomy, task decomposition, and tool integration, often within multi-agent frameworks. Explainability was predominantly intrinsic and workflow-native, typically delivered through textual reasoning traces and example-based grounding in retrieved clinical evidence. Feature-based and global explanations were comparatively rare and largely confined to hybrid architectures. Across domains including radiology, neurology, psychiatry, and biomedical research, XAAI systems were reported to improve performance and interpretability relative to baseline models in the included studies. However, these findings were derived from heterogeneous, predominantly experimental or retrospective studies, and structured human-in-the-loop oversight was infrequently described. Conclusions: Current XAAI systems appear to emphasize process transparency and evidence grounding rather than mechanistic model-level attribution. The available evidence remains limited and heterogeneous, and findings should be interpreted as early trends rather than established characteristics. Further progress will require standardized evaluation frameworks, clearer reporting of oversight mechanisms, and validation in real-world clinical settings to support safe and trustworthy integration of agentic AI into healthcare practice.

ABSTRACT Pancreatic islet transplantation can restore endogenous insulin production and offers a potential cure for diabetes, but its clinical impact has been limited by the inability to preserve large numbers of functional islets for timely use. Cryopreservation could provide an on-demand supply, yet conventional methods cause ice formation, cell injury, and loss of insulin secretion. We developed CryoMesh, a vitrification platform that combines a low-toxicity cryoprotectant with a thermally conductive, biocompatible mesh to enable ultra-rapid cooling and rewarming for ice-free cryopreservation. This approach supports long-term, clinical-scale preservation of both human pancreatic and stem cell–derived islets while maintaining viability, architecture, mitochondrial integrity, and glucose-responsive insulin secretion. Human islets preserved for up to one year restored normoglycemia in diabetic mice, with complete recovery of function and no increase in immunogenicity or loss of potency. The combination of high viability, recovery, and functional preservation at clinical scale has not been achieved previously. By decoupling islet isolation and manufacture from transplantation, CryoMesh enables extended quality, potency, and safety testing, cost-effective batch production, and global banking and distribution. These capabilities remove a major barrier to curative cell therapy for diabetes and establish a generalizable strategy for preserving complex multicellular therapeutics. One Sentence Summary A clinical scale vitrification platform enables long-term banking of human pancreatic and stem cell–derived islets for transplantation.

Background/Objectives: Characterizing spinal cord multiple sclerosis (MS) lesions in MRI is critical for diagnosis, monitoring, and treatment evaluation. However, current automated approaches for lesion detection and segmentation are typically designed for specific MRI contrasts or acquisition sites, limiting their generalizability in real-world clinical settings where imaging protocols vary widely. This work proposes a robust multi-site, multi-contrast segmentation framework for spinal cord lesions. Methods: The segmentation model was trained and evaluated on a large-scale dataset comprising 4428 annotated images from 1849 persons with MS across 23 imaging centers, encompassing six MRI contrasts (T1w, T2w, T2*w, PSIR, STIR, and UNIT1) acquired at 1.5 tesla (T), 3 T, and 7 T. Results: Likert-type assessment performed by neuroradiologist ratings demonstrated superior generalization of the model compared to existing contrast-specific pipelines ( p < 0.01). Additional experiments evaluated robustness across spinal levels, acquisition resolutions, binarization thresholds, and quantitative evaluation on external labeled datasets. Conclusions: The proposed model can achieve accurate and reliable spinal cord MS lesion segmentation across heterogeneous MRI data, addressing a key barrier to clinical translation. The model is available in the Spinal Cord Toolbox v7.2 and higher. Code repository: https://github.com/ivadomed/seg-sc-ms-lesion-multicontrast

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.

PURPOSE GBM AGILE (ClinicalTrials.gov identifier: NCT03970447 ) is a phase II/III Bayesian adaptive platform registration trial testing multiple arms against a common control; the primary end point is overall survival (OS). Regorafenib, a multikinase inhibitor, showed OS benefit in recurrent (RD) glioblastoma in the phase II REGOMA trial and entered GBM AGILE as the first investigational arm. METHODS Patient subtypes included in the regorafenib arm of GBM AGILE were newly diagnosed unmethylated (NDU) and RD glioblastoma. Prospective defined sets of subtypes, or arm signatures, were NDU, RD, and all (NDU + RD). As the first investigational arm in GBM AGILE, regorafenib was equally randomized to the control arm. Treatment in the control arm is temozolomide + radiotherapy (in newly diagnosed) or lomustine (in RD). Efficacy was assessed by OS hazard ratio (HR), arm/control, and demonstrated when the Bayesian probability of benefit (HR <1.00) was ≥98%. Analysis was performed monthly for limited efficacy, which occurs when the Bayesian predictive power is <25% for all signatures, and determines stopping enrollment. Follow-up continued for 12 months after accrual stopped. RESULTS When the predictive power was <25% in all predefined signatures for regorafenib, accrual stopped for limited efficacy. The final analysis did not demonstrate OS improvement in the regorafenib arm in RD nor NDU glioblastoma. Median HRs were 1.05 (NDU), 1.07 (RD), and 1.07 (all) with final probabilities of benefit (HR <1.00) of 0.421 (NDU), 0.312 (RD), and 0.296 (all). Regorafenib was associated with increased toxicity relative to control. CONCLUSION GBM AGILE did not show superiority of regorafenib over control in RD (lomustine) or NDU (temozolomide + radiotherapy) glioblastoma, yet caused increased toxicities. Regorafenib has been removed from National Comprehensive Cancer Network guidelines as a treatment option for RD.

Abstract Objectives The study aimed to compare long-term outcomes of the David and Yacoub valve-sparing aortic root replacement (VSARR) procedures in a nationwide, low-volume setting. Methods All patients undergoing elective VSARR in Denmark between January 2010 and April 2022 were identified using the Danish registries. The primary end-point was all-cause death. Secondary end-points included stroke, reoperation, a composite of death or reoperation, and long-term echocardiographic outcomes. Analyses were performed in both crude and propensity score-matched (PSM) populations and compared with outcomes from a control group comprised of all elective composite root replacement procedures performed at the contributing centres in the study period. Results A total of 160 patients underwent VSARR (median age; David vs Yacoub; 49.7 [SD 14.1] years vs 51.5 [SD 14.3] years; P = .42) with the David (n = 92, 57.5%; median follow-up 8.2 [SD 3.6] years) or Yacoub (n = 68, 42.5%; median follow-up 7.1 [SD 2.4] years) procedure. Ten-year all-cause death was similar (David 9.3% [95% CI: 2.3-15.2] vs Yacoub 8.8% [95% CI: 3.6-20.6]; log-rank P = .54). Reoperation rates (log-rank P = .12), stroke incidence (P = .41), and the composite end-point of death or reoperation (log-rank P = .63) were comparable, with consistent findings in the PSM cohort. Long-term echocardiographic follow-up demonstrated no difference in recurrent regurgitation, stenosis, or valve-related complications. Compared to elective composite root replacement (CRR) (n = 486), VSARR demonstrated lower rate of all-cause death (P = .027) and stroke (P = .026), while reoperation rate was higher (13.8% vs 4.3%; P &lt; .001). However, the risk of reoperation was comparable between procedures when accounting for competing risk of death (log-rank P = .17). Conclusions Long-term outcomes following elective VSARR using the David and Yacoub procedure were comparable. VSARR was associated with lower risk of all-cause death and stroke compared to elective CRR, supporting the two VSARR procedures as durable options when performed in experienced centres with careful patient selection, even in low-volume settings.

Background. Donor-derived cell-free DNA (dd-cfDNA) has emerged as a promising noninvasive biomarker for detecting allograft rejection in solid organ transplantation. However, its diagnostic utility in pancreas transplantation, including in multiorgan recipients, remains underexplored. Methods. We conducted a retrospective analysis of 42 pancreas transplant recipients (38 simultaneous pancreas-kidney, 4 pancreas transplant alone) who underwent for-cause organ biopsy and dd-cfDNA testing between April 2020 and December 2024. dd-cfDNA was measured using the Prospera assay, with a threshold of ≥1% considered high risk and <1% low-risk. Biopsy results were correlated with dd-cfDNA to assess test performance. Results. A total of 50 dd-cfDNA tests and 45 organ biopsies were analyzed. Among simultaneous pancreas-kidney recipients, dd-cfDNA demonstrated a sensitivity of 86.7%, specificity of 52.2%, positive predictive value (PPV) of 54.2%, and negative predictive value (NPV) of 85.7% for rejection. In pancreas transplant alone recipients, all high-risk dd-cfDNA results corresponded to biopsy-proven rejection, yielding 100% sensitivity, specificity, PPV, and NPV. When combining both cohorts, dd-cfDNA testing achieved an overall sensitivity of 88.2%, specificity of 57.7%, PPV of 57.7%, and NPV of 88.2%. Additionally, dd-cfDNA kinetics suggested its potential utility as a surveillance tool posttreatment, potentially reducing the need for closure biopsies. Conclusions. dd-cfDNA is a valuable noninvasive adjunct for monitoring of rejection in pancreas transplant recipients. Further prospective studies are needed to validate these findings and define the optimal role of dd-cfDNA in posttransplant care.

Abstract Background: Cholangiocarcinoma (CCA) is an aggressive hepatobiliary malignancy with marked molecular heterogeneity, limited systemic treatment options, and poor prognosis. Chemoresistance and KRAS mutations remain major therapeutic challenges. Dysregulation of the Hippo pathway, particularly YAP/TAZ hyperactivation, drives CCA progression and resistance. TEAD, the essential transcriptional partner of YAP/TAZ, represents a promising target. IAG933 is a novel oral pan-TEAD inhibitor designed to disrupt YAP/TAZ-TEAD interactions. RMC-6236, a pan-KRAS inhibitor, has also gained attention. Combined TEAD and KRAS inhibition may overcome resistance and improve efficacy in CCA. Methods: We evaluated IAG933 in human (HuCCT1, RBE) and murine (SB1, FAC) CCA cell lines. Cells were treated with IAG933 alone or in combination with GemCis or RMC-6236. Cell viability assays were used to determine IC50 values and calculate combination indexes (CI). TEAD activity was assessed by luciferase reporter and qRT-PCR of YAP/TAZ target genes (CTGF, CYR61, NUAK2). Transcriptomic alterations were analyzed by RNA sequencing. In vivo efficacy was evaluated in syngeneic orthotopic and patient-derived xenograft (PDX) models. Results: IAG933 suppressed TEAD-dependent transcription (IC50 ≈ 500 nM) and downregulated YAP/TAZ target genes. Combination with GemCis or RMC-6236 produced synergistic effects (CI &lt; 1) in all CCA lines. Transcriptomic analysis revealed activation of oxidative stress response pathways, particularly SOD2 upregulation. MitoSOX staining confirmed increased mitochondrial ROS after treatment. In vivo, IAG933 monotherapy suppressed tumor growth similarly to GemCis. Dual TEAD and KRAS inhibition achieved marked tumor regression and significantly delayed tumor regrowth after treatment cessation in KRAS-mutant PDX models. Conclusion: IAG933 effectively suppresses YAP/TAZ-TEAD signaling and inhibits CCA growth. While the combination with GemCis showed limited additional benefit due to strong single-agent efficacy, synergy with RMC-6236 produced durable antitumor effects, particularly in run-out studies, delaying regrowth in KRAS-mutant models. These findings highlight IAG933 as a promising therapeutic candidate targeting the Hippo-TEAD pathway, with potential benefit in KRAS-mutant CCA. Citation Format: Hidemi Nishi,Erik Jessen,Danielle M. Carlson,Amro M. Abdelrahman,Enis H. Ozmert,J. Pedro Fincatto Safi,Brandon A. Wilbanks,Shelby K. Yee,Jack W. Sample,Nathan W. Werneburg,Rory L. Smoot. TEAD inhibition by IAG933 as a novel therapeutic strategy for cholangiocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 3081.