Citations

A comprehensive, machine-interpretable taxonomy of artificial intelligence (AI) performance metrics provides a shared vocabulary and structured foundation to facilitate rigorous, transparent, and responsible evaluation of medical AI systems.

Background: Transanal total mesorectal excision (taTME) facilitates sphincter preservation in low rectal cancer. Preliminary results and functional outcomes from the North American multicenter phase II taTME trial (NCT03144765) have been published. Three-year outcomes are reported. Methods: Patients with stage I to III rectal adenocarcinoma were enrolled across 11 centers (2017–2022). Endpoints included TME grade, 3-year overall survival, disease-free survival, cancer-specific survival, recurrence, late (beyond 90 d) complications, and stoma-free survival. Surveillance followed National Comprehensive Cancer Network guidelines, with pelvic magnetic resonance imaging (MRI) at 3 years to assess for pelvic sidewall recurrence when possible. Kaplan–Meier analysis assessed oncologic outcomes. Results: One hundred patients underwent taTME for tumors at a median of 5.8 cm from the anal verge, with intersphincteric resection in 36% and handsewn anastomosis in 54%. Median follow-up was 50.9 months. Late complications occurred in 32 patients, including 21 severe (Clavien-Dindo ≥3). Late anastomotic complications occurred in 5 patients; one required surgical revision. Estimated 3-year overall survival, cancer-specific survival, and disease-free survival were 93.7%, 96.6%, and 84.6%, respectively. Distant recurrence occurred in 15 patients (median: 17.6 months), most commonly in the lung and liver. Locoregional recurrence occurred in 2 patients. Preoperative N stage, lymph node yield, tumor deposits, and perineural invasion were associated with recurrence. Among 100 patients diverted at taTME, 97% were stoma-free at 3 years; 4 patients later underwent secondary diversion. Conclusions: Three-year outcomes of this multicenter trial support the safety and efficacy of taTME in resectable mid and low rectal tumors, with outcomes comparable to contemporary studies. Longer-term oncologic and functional outcomes are needed.

Abstract Liver metastases from well-differentiated thyroid carcinomas are rare. We report 3 patients, 1 with follicular and 2 with oncocytic thyroid cancers, treated for liver metastases with yttrium-90 radioembolization. The 2 patients with oligometastatic disease received ablative radiation doses, while 1 patient with diffuse metastatic liver disease was treated palliatively with 2 sessions of reduced nonablative radiation dose due to a large hepatopulmonary shunt fraction diagnosed at work-up. The 2 cases of oligometastatic disease had a complete local response by imaging criteria, while the third case had a pronounced local partial response, without adverse events. Future guidelines should include radioembolization as a viable option for locoregional therapy of liver metastases of differentiated thyroid cancer not responsive to radioiodine in patients that do not qualify for or refuse hepatic surgery.

Abstract Recent therapeutic advances, including bispecific antibodies (bsAbs), have improved outcomes for patients with relapsed/refractory multiple myeloma (RRMM). However, early mortality (EM) remains a significant challenge, even in the era of novel immunotherapies. We conducted a multicenter, retrospective analysis of real-world patients with RRMM treated with commercial bsAbs to evaluate the incidence and causes of EM. Between May 2022 and June 2025, a total of 441 patients were treated across 7 countries. EM, defined as death within 12 months of therapy initiation, occurred in 148 patients (34%), with a median time from bsAb initiation to death of 2.87 months (interquartile range, 1.38-6.85). The leading cause of EM was progressive disease (72%), followed by infection (13%); regardless of the cause of death, 82% had active disease at the time of death. Our findings highlight that despite encouraging efficacy, EM remains high among real-world bsAb recipients, emphasizing the need for improved disease control strategies and supportive care interventions.

Importance Sex and racial or ancestral disparities in Alzheimer disease remain incompletely understood; autopsy studies that examine amyloid, tau, and genetic factors are scarce. Objective To test whether neuritic plaque burden and cognitive outcomes differ by sex and whether sex modifies the effects of apolipoprotein E ε4 ( APOEε4 ), informant-reported race, and African ancestry. Design, Setting, and Participants This was a cross-sectional study using postmortem neuropathological data from the Biobank for Aging Studies, University of São Paulo, São Paulo, Brazil. A total of 2268 autopsies from a population-based, diverse clinicopathological sample were collected between April 2004 and March 2025. Exposures Sex, informant-reported race (Black, White), African ancestry proportion, and APOEε4 carrier status. Main Outcomes and Measures Neuritic plaque burden (Consortium to Establish a Registry for Alzheimer’s Disease [CERAD] score), and cognitive function (Clinical Dementia Rating–Sum of Boxes [CDR-SB]). Ordinal logistic regression examined association of sex with CERAD scores and 2- and 3-way interactions among sex, race, ancestry, and APOEε4 ; adjusting for age, education, vascular factors, and Braak stages. Linear models related pathology to CDR-SB, adding copathologies. Results The analysis included 2268 autopsies (median [IQR] age, 74.8 [63.8-83.3] years; 1152 [51% male] and 1116 [49%] female; 802 [35%] Black and 1466 [65%] White; other race groups not included owing to small numbers); female individuals were older than male individuals and more likely to exhibit cognitive impairment (CDR global score ≥0.5). Female individuals had higher plaque burden than male individuals (unadjusted odds ratio [OR], 1.97; 95% CI, 1.67-2.29; P < .001), and this association remained significant in adjusted models for sociodemographic and vascular factors and APOEε4 status (adjusted OR, 1.65; 95% CI, 1.33-2.20; P < .001). APOEε4 carriers of both sexes had an approximately 4-fold greater odds of plaques. Significant 2-way interactions were found between sex, APOEε4 status, race, and ancestry on CERAD scores. Black noncarriers (OR, 0.47; 95% CI, 0.34-0.67) and noncarriers of African ancestry (OR, 0.57; 95% CI, 0.43-0.76) were least likely to have high plaque burden, whereas this protection was weakened in ε4 carriers. No significant 3-way interaction was detected. Among individuals with a CERAD score of 2 or higher, female individuals were more likely than male individuals to reach Braak stage V-VI than male individuals (probability ratio, 1.25; 95% CI, 1.13-1.38; P = .002). Adding Braak stage to multivariable models attenuated the female-male difference in plaques and interaction of sex and plaque on CDR-SB was no longer significant. Conclusions and Relevance The findings indicate that female sex, APOEε4 , and both race and African ancestry were jointly associated with amyloid in this study population. Excess amyloid among women may partly explain their greater tau burden and steeper cognitive decline. These findings highlight the importance of incorporating sex, race, and ancestry into biomarker thresholds, risk stratification, and the design of preventive or disease-modifying trials for Alzheimer disease.

Abstract Aims Left ventricular hypertrophy (LVH) is a strong predictor of cardiovascular disease. We previously compared supervised machine learning techniques to classify cardiac magnetic resonance (CMR)-derived LVH using electrocardiogram (ECG) and clinical variables in 37 534 UK Biobank participants, obtaining an area under the receiving operating curve (AUROC) of 0.85, but with limited specificity and requiring external validation. In this study, we develop a deep learning (DL) model to improve classification with external evaluation in the Study of Health in Pomerania (SHIP). Methods and results We analysed 12-lead ECGs of 48 835 participants from the UK Biobank imaging study. The dataset was split into a training set (70%), validation set (15%), and test set (15%) for performance evaluation. The model architecture was a fully convolutional network, for which the input was the participants’ median ECG and clinical variables and the predicted indexed left ventricular mass (iLVM) as the output. A subsequent logistic regression model was used to recalibrate iLVM predictions. In UK Biobank, 717 (1.5%) participants had CMR-derived LVH and the AUROC for the DL model was 0.97. The ECG components most predictive of LVH were the QRS complex and ventricular rate. The DL model outperformed our supervised algorithms, previous DL modelling efforts and clinical ECG benchmarks. There was modest generalizability of the DL model to 1423 participants in SHIP (AUROC 0.78), with differences in clinical profile, ECG acquisition, and CMR labelling as important factors. Conclusion Our findings support the feasibility of scalable DL-based screening tools for the prediction of LVH from the ECG, whilst highlighting the need for model development using larger datasets with greater diversity to ensure generalizability.

ABSTRACT Objective This study compared the effects of phentermine‐topiramate‐ER (mid‐dose 7.5/46 mg) versus placebo on weight loss and cardiovascular disease (CVD) risk outcomes when used as an adjunct to a digitally enhanced lifestyle intervention (DELI). Methods We conducted a 12‐month, randomized, double‐blind, placebo‐controlled trial at a single tertiary academic center in the United States (June 2020–June 2022). Eighty participants with obesity (BMI ≥ 30 kg/m 2 ) were enrolled in the DELI program, consisting of in‐person and telehealth modalities, dietary and physical activity goals, and use of a smartphone application integrated with digital devices (Apple Watch and Bluetooth‐enabled weight scale and blood pressure monitor). Participants were randomized 1:1 to receive either phentermine‐topiramate‐ER ( n = 42) or placebo ( n = 38) in addition to the DELI. Results At 3 months, the phentermine‐topiramate group lost a mean of 10.82 kg versus 4.04 kg in the placebo group (mean difference −6.78 kg; p = 0.002). At 12 months, weight loss was 15.32 kg versus 5.85 kg, respectively (mean difference −9.48 kg; p < 0.001). Participants receiving phentermine‐topiramate‐ER experienced a 3.35% reduction in the estimated atherosclerotic CVD risk compared to baseline ( p = 0.004). Conclusions Phentermine‐topiramate‐ER, when combined with a DELI, produced significant and sustained weight loss and reduced CVD risk in adults with obesity. Trial Registration ClinicalTrials.gov : NCT04408586

Objective: To establish international benchmark values for relevant outcome parameters in robotic Whipple. Summary Background Data: For safe adoption of surgical innovation, robust quality control is essential. Benchmarking is a validated tool for assessing surgical performance. Recent international consensus identified establishing benchmark values for robotic Whipple as top priority. Methods: We analyzed consecutive patients undergoing robotic Whipple between 2020-2023 with a minimum one-year follow-up. Reference centers were required to perform ≥15 cases/year, be scientifically active in the field, and maintain a prospective database. Benchmark criteria included benign or resectable malignant disease without neoadjuvant therapy, arterial resection, major co-morbidities, or significant previous abdominal surgery. Benchmarks were established for 13 outcome parameters. Result: The benchmark cohort comprised 418 patients from 12 centers across four continents. Benchmark values were: conversion rate ≤4.3%, transfusion rate ≤2.1%, 6-month mortality ≤2.2%, major complications ≤23.2%, and CCI® ≤20.9. Clinically relevant pancreatic fistula (grade B/C) and hemorrhage (grade B/C) rates were ≤23.6% and ≤12.7%, respectively. For pancreatic ductal adenocarcinoma (n=123), the benchmark for lymph node yield was ≥20. Higher surgical difficulty was associated with increased overall postoperative morbidity (R 2 =0.38, P =0.019), higher center caseload with reduced pancreas-specific complications (R 2 =0.28, P =0.044). Independent POPF predictors included duct diameter ≤4 mm (OR 1.37, 95% CI: 1.03, 1.82), anticoagulation (OR 2.45, 95% CI: 1.47, 3.99), and indication other than PDAC (OR 2.33, 95% CI: 1.68, 3.27). Conclusions: This study establishes the first international benchmarks for robotic Whipple, demonstrating oncologic outcomes and morbidity comparable to open surgery with the benefits of minimally invasive surgery.