Citations

Objective The aim of the study was to assess the interrater reliability between seven individuals assessing quadriceps femoris cross-sectional area, echo intensity, and corrected echo intensity. Design A retrospective subanalysis of data collected in a randomized controlled clinical trial (IRB 19-008473. NCT ID NCT04302558). Patients aged 13–50 yrs with an anterior cruciate ligament tear requiring surgical reconstruction were assessed for cross-sectional area of the rectus femoris, vastus intermedius, vastus lateralis, vastus medialis, and vastus medialis obliquis, and echo intensity of the rectus femoris. Musculoskeletal ultrasound imaging of the quadriceps was performed by one physical therapist, and quantitative assessments were performed by four physicians and three physical therapists. Results Physicians and physical therapists had good reliability for cross-sectional area of the rectus femoris (median interclass correlation 0.86 doctor of medicine (MD) and 0.84 physical therapist), cross-sectional area of the vastus medialis obliquis (0.82 and 0.82), and echo intensity mean (0.92 and 0.92). Rectus femoris cross-sectional area had the smallest coefficient of variation (26.512.7%), whereas subcutaneous fat had the largest coefficient of variation (93.3% left, 86.4% middle, 86.1% right). Conclusions These findings demonstrate reliability between physicians and physical therapists in interpretation of ultrasound images of the quadriceps. The inconsistency demonstrates the need for a consistent training across disciplines to ensure accurate image interpretation.

Abstract Allogeneic hematopoietic cell transplantation (allo-HCT) is potentially curative for older adults with hematologic malignancies. Concerns on nonrelapse mortality (NRM) in older adults limit allo-HCT utilization. We executed a prospective, observational study BMT-CTN 1704 (Blood and Marrow Transplant Clinical Trials Network) enrolling allo-HCT recipients aged ≥60 years from 49 centers in the United States. We analyzed associations between 13 measurements of older adult health and NRM within 1 year to construct a comprehensive health assessment risk model (primary-CHARM) using multivariate Fine-Gray model and grouped penalized variable selection. Two machine learning (ML) models (Cox and pseudo-value boosting) were also explored. Models’ performances were compared using area under the curve (AUC), with bootstrap and cross-validation sampling to correct for optimism, decision curve analysis (DCA), calibration, and Brier scores. Among 1105 patients with median age of 67 (range, 60-82) years who received allo-HCT, NRM was 14.4% and overall survival (OS) 71.7% at 1 year. Factors statistically selected for inclusion in primary-CHARM were higher comorbidity burden, lower albumin, higher C-reactive protein, older age, higher weight-loss percentage, lower patient-reported performance score, and cognitive impairment. Primary-CHARM scores were independently associated with higher NRM (hazard ratio [HR], 2.72; P < .0001) and worse OS (HR, 2.09; P < .0001). Bootstrap bias–corrected AUC for primary-CHARM was 0.591. Comparing primary-CHARM with HCT-comorbidity index and 2 ML-CHARM models, calibration, Brier score, and DCA analysis favored primary-CHARM. Primary-CHARM, with mostly simple and readily available parameters, risk stratifies older adults for allo-HCT. Adopting primary-CHARM in practice may promote broader use of HCT by quantifying risk and enhance the design of strategies to improve outcomes. This trial was registered at www.ClinicalTrials.gov as #NCT03992352.

Abstract Background Automated data extraction from echocardiography reports could facilitate large-scale registry creation and clinical surveillance of valvular heart diseases (VHD). We evaluated the performance of open-source large language models (LLMs) guided by prompt instructions and chain of thought (CoT) for this task. Methods From consecutive transthoracic echocardiographies performed in our center, we utilized 200 random reports from 2019 for prompt optimization and 1000 from 2023 for evaluation. Five instruction-tuned LLMs (Qwen2.0-72B, Llama3.0-70B, Mixtral8-46.7B, Llama3.0-8B, and Phi3.0-3.8B) were guided by prompt instructions with and without CoT to classify prosthetic valve presence and VHD severity. Performance was evaluated using classification metrics against expert-labeled ground truth. Mean squared error (MSE) was also calculated for predicted severity’s deviation from actual severity. Results With CoT prompting, Llama3.0-70B and Qwen2.0 achieved the highest performance (accuracy: 99.1% and 98.9% for VHD severity; 100% and 99.9% for prosthetic valve; MSE: 0.02 and 0.05, respectively). Smaller models showed lower accuracy for VHD severity (54.1%-85.9%) but maintained high accuracy for prosthetic valve detection (>96%). Chain of thought reasoning yielded higher accuracy for larger models while increasing processing time from 2-25 to 67-154 seconds per report. Based on CoT reasonings, the wrong predictions were mainly due to model outputs being influenced by irrelevant information in the text or failure to follow the prompt instructions. Conclusions Our study demonstrates the near-perfect performance of open-source LLMs for automated echocardiography report interpretation with the purpose of registry formation and disease surveillance. While larger models achieved exceptional accuracy through prompt optimization, practical implementation requires balancing performance with computational efficiency.

The interplay between intracellular and intravascular lipolysis is crucial for maintaining circulating lipid levels and systemic energy homeostasis. Adipose triglyceride lipase (ATGL) and lipoprotein lipase (LPL), the primary triglyceride (TG) lipases responsible for these two spatially separate processes, are highly expressed in adipose tissue. Yet the mechanisms underlying their coordinated regulation remain undetermined. Here, we demonstrate that genetic ablation of G0S2, a specific inhibitory protein of ATGL, completely abolished diet-induced hypertriglyceridemia and significantly attenuated atherogenesis in mice. These effects were attributable to enhanced whole-body TG clearance, not altered hepatic TG secretion. Specifically, G0S2 deletion increased circulating LPL concentration and activity, predominantly through LPL production from white adipose tissue (WAT). Strikingly, transplantation of G0S2-deficient WAT normalized plasma TG levels in mice with hypertriglyceridemia. In conjunction with improved insulin sensitivity and decreased ANGPTL4 expression, the absence of G0S2 enhanced the stability of LPL protein in adipocytes, a phenomenon that could be reversed upon ATGL inhibition. Collectively, these findings highlight the pivotal role of adipocyte G0S2 in regulating both intracellular and intravascular lipolysis, and the possibility of targeting G0S2 as a viable pharmacological approach to reducing levels of circulating TGs.