Citations

Importance Allogeneic hematopoietic cell transplant (HCT) is curative for hematologic cancers, yet access remains inequitable for racially and ethnically underrepresented and socioeconomically disadvantaged populations, making the goal of having a suitable donor for every patient who needs a transplant challenging. The ACCESS trial broadened access by enrolling patients without matched donors, who instead received an HCT from a mismatched unrelated donor. Objective To compare baseline characteristics of ACCESS trial participants with participants enrolled in a similar clinical trial and a patient-reported outcome (PRO) protocol cohort. Design, Setting, and Participants This cross-sectional study included adult participants (aged ≥18 years) from 3 cohorts—the ACCESS trial (2021-2024), BMT CTN 1703 trial (2019-2021), and Center for International Blood and Marrow Transplant Research (CIBMTR) PRO Protocol observational study (2020-2025)—who completed a baseline PRO survey. The ACCESS and PRO Protocol cohorts were stratified by conditioning intensity (myeloablative [MAC] vs reduced-intensity and nonmyeloablative [RIC/NMA]); all BMT CTN 1703 participants received RIC/NMA. Exposure Hematopoietic cell transplant. Main Outcomes and Measures Racial and ethnic diversity, insurance type, education, and income were compared among cohorts using counts and percentages, and socioeconomic and structural disadvantage were measured using the Social Vulnerability Index and Comprehensive Score for Financial Toxicity–Functional Assessment of Chronic Illness Therapy. Results Baseline surveys were completed by 208 participants in the ACCESS trial (median [range] age at transplant, 62.3 [20.4-78.9] years; 108 male [51.9%]), 122 participants in the PRO Protocol study (median [range] age at transplant, 63.9 [21.1-78.0] years; 67 male [54.9%]), and 342 participants in the BMT CTN 1703 trial (median [range] age at transplant, 66.9 [20.7-78.6] years; 218 male [63.7%]). Participants in ACCESS were more racially and ethnically diverse, with 15 (7.2%), 25 (12.1%), 46 (22.2%), 110 (53.1%), and 11 (5.3%) of Asian, Black or African American, Hispanic or Latino, White, and other race and ethnicity, respectively, compared with 4 (3.3%), 2 (1.6%), 8 (6.6%) 104 (85.2%), and 4 (3.3%), respectively, in the PRO Protocol and 10 (3.0%), 0, 16 (4.8%), 302 (91.0%), and 4 (1.2%), respectively, in the BMT CTN 1703 trial. Participants in ACCESS were more likely to have Medicaid (36 [18.1%]) vs PRO Protocol (8 [6.7%]) and BMT CTN 1703 (16 [5.1%]) participants and reported lower education (some college or an associate’s degree: 103 [49.5%] vs 73 [59.8%] in the PRO Protocol; postcollege education: 34 [17.3%] vs 35 [29.2%] in the PRO Protocol) and household income (<$40 000 annually: 25 [24.0%] vs 8 [11.6%] in the PRO Protocol and 7 [38.9%] in the BMT CTN 1703 trial). Median Social Vulnerability Index scores were highest among participants in the ACCESS MAC group (median [range], 0.72 [0.01-0.97] vs 0.61 [0.16-0.78] in the PRO Protocol MAC group), and 16 participants [27.6%] in the ACCESS MAC group reported moderate to severe financial toxicity. The ACCESS participants lived closer to transplant centers, especially in the RIC/NMA group (median [IQR], 28 [14-75] miles vs 47 [16-96] miles for BMT CTN 1703 participants and 49 [21-104] miles for PRO Protocol participants). Conclusions and Relevance This cross-sectional study of clinical trial participants and a clinical cohort found that the ACCESS trial enrolled a more racially and ethnically diverse and socioeconomically disadvantaged population. Trial designs that broaden eligibility could expand access to HCT, highlighting the need for systemic interventions to ensure equity.

Compared with two-dimensional MR elastography (MRE), three-dimensional MRE demonstrated better test-retest repeatability, between–field-strength reproducibility, and between-day reproducibility at 3 T for measurement of liver stiffness in adults with severe obesity but failed more frequently.

Abstract Glioblastoma (GBM), isocitrate dehydrogenase wild-type (IDHwt) and central nervous system diffuse large B-cell lymphoma (CNS-DLBCL) are aggressive brain tumors with overlapping MRI features, yet distinct treatment approaches. Noninvasive tools are needed to aid in differential diagnosis. Deep learning on T1 postcontrast and T2-weighted MRI sequences were used to differentiate GBM and CNS-DLBCL. A three-stage temporal study design was utilized. Model development was performed on 146 patients with CNS-DLBCL and 146 age-matched, sex-matched, and MRI year–matched patients with GBM diagnosed at Mayo Clinic between 1998 and 2019. Models were tested on independent temporal test cohorts. Initial testing included 240 independent GBM diagnosed at Mayo Clinic between 1998 and 2019. The prospective test cohort included 37 patients with CNS-DLBCL and 256 patients with GBM diagnosed at Mayo Clinic after January 1, 2020, and 36 patients with CNS-DLBCL diagnosed at an external institution. Of the patients diagnosed at Mayo Clinic, 47% had MRIs generated from non-Mayo institutions. Two different model approaches were compared: (i) ensemble approach using area under the receiver operating characteristic curve (AUC) and cross-validation for model selection and (ii) loss approach minimizing cross-entropy loss and cross-validation to evaluate prediction performance. The AUCs on the prospective test cohort were 0.84 [95% confidence interval (CI), 0.78–0.90] and 0.83 (95% CI, 0.77–0.88) for the ensemble and loss approaches, respectively. Stability of ensemble prediction improved with the increasing number of models. Stratified AUC analysis demonstrated consistent performance across sex and age. We utilized a robust temporal study design and applied 2 different analytic approaches to develop a classification model. The findings confirm the feasibility of using MRI-based deep learning models to differentiate GBM from CNS-DLBCL. Significance: GBM, IDHwt and CNS-DLBCL are aggressive brain tumors with overlapping MRI features, yet distinct treatment approaches. Noninvasive tools are needed to aid in differential diagnosis. We developed MRI-based deep learning models to differentiate GBM, IDHwt from CNS-DLBCL using a rigorous three-stage temporal design that included prospective validation. The model AUC on a prospective cohort was 0.84.

Abstract Randomized controlled trials (RCTs) provide high internal validity but often rely on restrictive eligibility criteria that limit generalizability and complicate real-world trial emulation. We propose AERO (AI Agent for Adaptive Eligibility Refinement and Optimization), an agentic framework that systematically adapts clinical trial eligibility criteria for application to electronic health record data. AERO integrates external clinical knowledge sources and large language model–based reasoning to classify criteria as strict inclusion, safety exclusion, confounder, or operational artifact. We evaluated AERO by emulating the WARCEF trial using Mayo Clinic Platform data restricted to the pre-trial completion period. Emulation with optimized criteria yielded a hazard ratio of 1.561 (p = 0.0605), consistent with the original neutral trial finding (HR = 1.01, p = 0.91). An ablation analysis demonstrated that eligibility handling decisions materially influence observed treatment effects. These results highlight the importance of systematic, knowledge-informed eligibility refinement in real-world evidence generation.

Introduction: Diabetes, a modifiable risk factor for dementia, is thought to impart increased risk of dementia through cerebral small vessel disease. While the importance of diabetes in midlife is well-documented, the influence of different glycemic levels, duration of diabetes, and Alzheimer’s disease genetic risk (apolipoprotein [ APOE ] ε4 genotype) in late-life on dementia risk remains uncertain. This could have implications for individualized prevention strategies. In this study, we investigated whether a higher glycemic level and longer duration of diabetes in late-life is associated with a higher risk of dementia, and whether associations vary by APOE ε4 genotype. Methods: We performed a prospective cohort analysis with the Atherosclerosis Risk in Communities Study using up to 11 years of follow-up (baseline, 2011-2013). We defined diabetes as glycated hemoglobin A1c (HbA 1c ) ≥6.5%, self-reported physician diagnosis, or use of any diabetes medication. Among those with diabetes, we categorized participants as having a lower glycemic level (HbA 1c <7%) or higher level (HbA 1c ≥7%), and duration as short (<5 years) or long (≥5 years). We estimated hazard ratios (HR) for incident dementia by HbA 1c levels and by diabetes duration in separate multivariable-adjusted Cox regression models. We tested for statistical multiplicative interaction on the relative risk of dementia by fitting interaction terms between diabetes measures and APOE ε4 genotype in separate regression models. Results: In 5030 participants without dementia (mean age, 75.4 years; 58.6% female; 21% Black; 28% with diabetes), there were 1167 (23.2%) incident dementia cases over a median follow-up of 9.0 years. Compared to no diabetes, a lower glycemic level and higher level were both associated with elevated dementia risk ( P -trend <.0001; Table). A higher glycemic level was associated with an elevated risk of dementia relative to a lower level (HR=1.47 [95% CI: 1.22-1.77]; Table). Compared to no diabetes, both shorter duration and longer duration were associated with an increase in dementia risk ( P -trend <.0001; Table). Longer duration was associated with an increased risk of dementia relative to shorter duration (HR=1.57 [95% CI: 1.24-1.98]; Table). Associations for glycemic level and diabetes duration did not significantly differ by APOE ε4 genotype (all P >.05). Conclusions: Diabetes management in older adults that includes lower glycemic goals may help reduce dementia risk.

Introduction: Abnormal sleep has been linked to increased dementia risk, but the pathophysiological processes driving this connection are unclear. To provide insight into the biological pathways underlying the relationship between sleep and brain health, we tested the hypothesis that increased obstructive sleep apnea (OSA) severity and shorter habitual sleep duration are associated with a higher burden of Alzheimer’s disease (AD) pathology, neurodegeneration, and neuroinflammation measured through plasma biomarkers. Methods: ARIC participants who took part in the Sleep Heart Health Study at Visit 4 (1996-98) with stored plasma samples at Visit 3 (1993-95) and Visit 5 (2011-13) were examined (n=261; mean age 61.3 years; 43.3% male). OSA severity, categorized based on the number of apneas and hypopneas per hour of sleep (<5 events/hour [Normal], 5 to <15 events/hour [Mild OSA], or ≥15 events/hour [Moderate/Severe OSA]) was assessed via in-home polysomnography, and average habitual sleep duration (<7 hours, 7 to <8 hours, or ≥8 hours) was self-reported. Plasma biomarkers of amyloid beta (Aβ) 42 and 40, phosphorylated Tau 181 (p-Tau181), neurofilament light chain (NfL), and glial fibrillary acidic protein (GFAP) were measured using Quanterix Simoa immunoassays. Covariate-adjusted linear regression models estimated associations of sleep indices in 1996-98 with standardized plasma biomarker measurements in 1993-95 (3 years earlier) and in 2011-13 (15 years later). Covariate-adjusted linear mixed effects models quantified associations of sleep measures with the 5-year rates of change in standardized biomarker levels (1993-2013). Results: Compared to a habitual sleep duration of 8 hours or more, sleeping less than 7 hours per night was associated with higher NfL, a biomarker of neurodegeneration, 15 years later (β = 0.252; 95% confidence interval [CI]: 0.003, 0.501). Sleep indices were not associated with biomarker levels 3 years earlier. Counter to our hypothesis, Moderate/Severe OSA was associated with slower increases in p-Tau181 (β = -0.102; 95% CI: -0.198, -0.006), a biomarker of AD pathology, over a median of 17.9 years. Conclusions: Short habitual sleep duration may impact dementia outcomes through neurodegenerative and other processes rather than through pathways specific to AD pathology or neuroinflammation. Longitudinal studies with sleep measurements in larger, more diverse samples are needed to further distinguish the pathways linking sleep to dementia.

ABSTRACT Bispecific T cell engagers (BTEs) induce MHC‐independent cytotoxicity by bridging T cells to tumor cells via binding a T cell–activating receptor and a tumor‐associated antigen. BTEs have proven effective in hematologic malignancies and some solid tumors, yet their potential in glioblastoma (GBM) is largely unexplored. We developed a fully humanized BTE (hBTE) targeting interleukin‐13 receptor alpha 2 (IL13RA2), a tumor‐associated antigen widely expressed in GBM and associated with poor prognosis. In vitro, hBTE activated T cells and induced antigen‐dependent cytokine release and cytotoxicity against IL13RA2‐positive GBM cells. In vivo, hBTE showed robust target‐specific activity and markedly prolonged survival in primary and recurrent GBM xenograft models, without detectable off‐target local or systemic toxicity. Beyond GBM, hBTE also exhibited antitumor activity in IL13RA2‐expressing solid tumors, demonstrating selective tumor accumulation and therapeutic efficacy in models of breast cancer brain metastases and extracranial lung cancer. This work highlights the therapeutic potential of BTEs in IL13RA2‐expressing tumors and establishes a strong preclinical rationale for advancing hBTE therapy toward clinical translation in GBM and other tumors.

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.