BACKGROUND: Diagnosing cardiac amyloidosis (CA) on echocardiography can be challenging due to the imaging overlap between CA and more prevalent causes of a hypertrophic phenotype. This study sought to (1) evaluate the performance of artificial-intelligence (AI) derived measurements incorporated into the established multiparametric echocardiographic scoring system to detect CA; (2) develop and validate an AI-based deep-learning model for video-based detection of CA on echocardiography. METHODS: The study population comprised 5776 patients (CA, 2756; controls, 3020). The training data set included patients from the UK National Amyloidosis Center and Taiwan MacKay Memorial Hospital (CA, 2241; controls, 2130). External test data sets were obtained from the US Duke University Health System (CA, 334; left ventricular hypertrophy controls, 668) and Japan National Cerebral and Cardiovascular Center (CA, 181; left ventricular hypertrophy controls, 222). RESULTS: The multiparametric echocardiographic score computed using AI-derived measurements achieved an accuracy of 79.5% (sensitivity, 75.4%; specificity, 81.5%) in the United States cohort and 79.7% (sensitivity, 81.6%; specificity, 78.1%) in the Japan cohort. The deep-learning model demonstrated accuracies of 96.2% (sensitivity, 96.8%; specificity, 95.7%) and 95.8% (sensitivity, 97.3%; specificity, 94.3%) in the internal validation and internal test sets, respectively. External validation of the deep-learning model showed accuracies of 87.5% (sensitivity, 86.6%; specificity, 87.9%) in the United States and 88.4% (sensitivity, 92.3%; specificity, 85.3%) in the Japanese cohort. Subgroup analysis demonstrated that the deep-learning model showed robust discrimination of CA from other hypertrophic phenocopies: CA versus hypertension (area under the curve [AUC], 0.92 [95% CI, 0.91-0.94]), CA versus hypertrophic cardiomyopathy (AUC, 0.91 [95% CI, 0.87-0.94]), CA versus aortic stenosis (AUC, 0.93 [95% CI, 0.90-0.95]), CA versus chronic kidney disease (AUC, 0.93 [95% CI, 0.91-0.95]). The deep-learning model was able to classify a greater proportion of patients compared with the AI-derived multiparametric echocardiographic score and achieved superior diagnostic accuracy (AUC, 0.93 [95% CI, 0.91-0.95] versus AUC, 0.88 [95% CI, 0.85-0.90]; P<0.001). CONCLUSIONS: Both the multiparametric echocardiographic score computed from AI-derived measurements and the fully automated deep-learning model can accurately identify patients with CA in globally diverse cohorts, with the deep-learning model providing superior performance.
Introduction: Hepatocellular carcinoma (HCC) can develop in individuals with metabolic dysfunction-associated steatotic liver disease (MASLD) without cirrhosis, especially in those with elevated liver enzymes. However, there is currently no low-cost, scalable screening tool for this population. To address this need, we investigated the association between serial changes in FIB-4 (fibrosis-4) scores over a 3-year period and the risk of developing HCC in a large, diverse cohort of patients with MASLD without cirrhosis. Methods: This study utilized a nationwide cohort from the National Health Insurance Research Database (NHIRD), including 810,698 patients with MASLD who had demographic and laboratory data at baseline and at least 3 years of follow-up. FIB-4 scores were analyzed to assess transitions between low-risk (<1.45), indeterminate-risk (1.45-2.67), and high-risk (>2.67) categories over time. Competing risks for HCC and mortality were modeled to estimate the sub-distribution hazard ratios (SHRs). Results: Changes in FIB-4 scores over time provided superior predictive accuracy compared to one-time measurement. Individuals with persistent high FIB-4 (high-to-high group) had a 14-fold higher risk of developing HCC (adjusted SHR [aSHR] 13.91, 95% CI: 11.94-16.20) compared to those with stably low FIB-4 (low-to-low group), while those with improved scores (high-to-low and high-to-indeterminate groups) had a lower risk compared to the high-to-high group. Worsening FIB-4 scores, as shown in the low-to-indeterminate group (aSHR 2.22, 95% CI: 1.93-2.57 or the low-to-high group (aSHR 4.75, 95% CI: 3.38-6.68), were associated with progressively increased risks of HCC compared to the low-to-low group. In contrast, when compared to the high-to-high group, those with improved FIB-4 scores, including the high-to-indeterminate group (aSHR 0.41, 95% CI: 0.35-0.48) and the high-to-low group (aSHR 0.25, 95% CI: 0.15-0.42), exhibited reduced HCC risks. Conclusion: Serial FIB-4 measurements offer greater accuracy in identifying individuals at high risk for HCC in non-cirrhotic MASLD Asians with elevated liver enzymes, with HCC surveillance warranted in high-risk groups.
Background:The Taiwan Registry of Hypertrophic Cardiomyopathy (THIC) is a multicenter national registry containing the clinical and imaging data of patients with hypertrophic cardiomyopathy (HCM) in Taiwan. The aim of the registry is to systematically evaluate the clinical, genetic and biochemical features, possible natural course, and outcomes of HCM and relevant rare diseases that mimic HCM such as Fabry disease (FD) and transthyretin amyloid cardiomyopathy in Taiwan, and to identify their specific "red-flag" signs, which are especially valuable from the perspective of unique genetic mutations or clinical manifestations in Taiwanese patients. Herein, we present the design and initial baseline data from the registry. Methods:The THIC is an observational program that aims to collect prospective and/or retrospective data of patients with HCM in Taiwan. The registry plans to recruit 800 individuals with unexplained left ventricular hypertrophy, including 200 with FD, with a follow-up period of at least 12 months, and the project is expected to run for 5 years. Data on baseline characteristics, laboratory and imaging results, deaths, major adverse cardiovascular, cerebrovascular and renal events are collected. Results:The THIC has been in the enrollment phase since December 2022, and has enrolled 534 patients (age 62.37 ± 13.41 years, male 65.6%) as of March 15, 2025 from 13 centers. At enrollment, 284 of these patients had HCM, 227 had FD, and 23 had ATTR-CM. Family history was found to be an important diagnostic clue; however, common echocardiographic and laboratory data including N-terminal pro-brain natriuretic peptide were not significantly different between the three groups. Conclusions:The THIC will contain comprehensive clinical and imaging data of patients with HCM, FD and ATTR-CM in Taiwan, and provide an opportunity to extend our knowledge on the clinical presentations and long-term consequences of these disease entities. It will aid in understanding patients with unexplained LVH in the context of the genetic background of Taiwanese patients, and in identifying predictors of LVH and important clinical events.
Pleural effusion is common in critically ill adults, and bedside pigtail catheter drainage is often required; insertion without continuous imaging may lead to pneumothorax or catheter malposition. We conducted a single-center retrospective cohort study of adult intensive care unit pigtail pleural drainage procedures (28 December 2020-2018 March 2025) comparing real-time ultrasound-guided insertion with ultrasound-assisted insertion using preprocedure localization/skin marking only (no real-time needle tracking). The primary outcome was procedure-related mechanical complications (pneumothorax or catheter malposition); between-group comparisons used exact tests, and relative risks (RRs) with 95% confidence intervals (CIs) were reported. Among 175 procedures in 145 patients, 89 were real-time ultrasound-guided and 86 were ultrasound-assisted without real-time guidance. Mechanical complications occurred in 0/89 (0.0%) versus 10/86 (11.6%) procedures, respectively (RR 0.05, 95% CI 0.00-0.77; p < 0.001), driven by catheter malposition (0/89 vs 7/86 (8.1%); RR 0.06, 95% CI 0.00-1.11; p = 0.006) and pneumothorax (0/89 vs 3/86 (3.5%); RR 0.14, 95% CI 0.01-2.63; p = 0.117). Median catheter dwell time was 11.0 (8.0-20.0) versus 9.0 (6.0-15.0) days (p = 0.031). Real-time ultrasound guidance was associated with fewer observed mechanical complications in this cohort; given the retrospective design and baseline differences between groups, prospective risk-adjusted validation is warranted.
We present the case of a 12-year-old post-menarchal female (bone age: 13 years; 30th percentile for height) who received an initial gonadotropin-releasing hormone (GnRH) analogue injection for height augmentation to maximize adult height potential prior to imminent epiphyseal fusion. Immediately following the injection, she developed acute-onset abdominal pain, emesis, and diarrhea. Refractory to conservative management, her clinical status prompted pelvic ultrasonography and subsequent contrast-enhanced CT imaging, which revealed significant bilateral ovarian enlargement, multiple cystic masses, and congestion of the left vascular pedicle highly indicative of adnexal torsion. Urgent surgical exploration confirmed bilateral ovarian cyst formation and left-sided ovarian torsion. Following an unremarkable post-operative recovery, GnRH analogue therapy was permanently discontinued. This case underscores the critical need to recognize the heightened risk of rapid ovarian hyperstimulation and subsequent torsion when initiating these medications in post-menarchal or late pubertal females, in whom mature follicles are already present. While GnRH analogues generally maintain a favorable safety profile, clinician vigilance regarding this severe ovarian complication in older pediatric cohorts is vital to prevent diagnostic delays and ensure prompt surgical intervention.
