Citations

Background: Cardiomyopathy occurring during pregnancy or in the postpartum period is a leading case of maternal death in the US. The most commonly described form of pregnancy related cardiomyopathy (PRCM) associated with left ventricular systolic dysfunction is peripartum cardiomyopathy, defined as left ventricular ejection fraction (LVEF) <45% on echocardiography. PRCM symptoms often overlap with those of normal pregnancy and the lack of validated screening tools limits effective recognition, contributing to diagnostic delays. Recently, several artificial intelligence enabled electrocardiogram (AI-ECG) models have been developed and evaluated for PRCM detection. Aim&Hypothesis: We sought to investigate the pooled diagnostic performance of AI-ECG models for detecting PRCM. We hypothesized that AI-ECG models would be effective for detecting PRCM. Methods: A systematic search of PubMed, Scopus, and Cochrane was performed by combining search terms “artificial intelligence”, “machine learning”, “deep learning”, “algorithm”, “ECG”, “EKG”, “electrocardiogram”, “peripartum cardiomyopathy”, “pregnancy”, “pregnancy related cardiomyopathy” , to identify studies investigating the use of AI-ECG models for PRCM detection. Studies included those utilizing standard 12-lead ECGs as well as single lead ECGs from various devices. For inclusion in the meta-analysis, studies had to provide complete confusion matrix data for the AI-ECG model. Inverse variance random effects model meta-analysis was performed. Pooled performance estimates with corresponding 95% CIs are presented in forest plots and a summary receiver operating characteristics (sROC) curve. Results: Our search identified a total of 201 studies (PubMed: 64, Embase: 117, Cochrane: 20) from which 6 studies involving 11 AI-ECG models and 3,006 patients were included in the meta-analysis (Fig. 1) . Following random effect bivariate meta-analysis, the AI-ECG models yielded a pooled sensitivity of 0.721 (95% CI 0.676–0.762); specificity of 0.931 (95% CI 0.932–0.938); diagnostic odds ratio (DOR) of 44.658 (95% CI 23.843–83.644); and AUC of 0.896 (95% CI 0.87–0.91), for PRCM detection (Fig. 2A-C, 3) . Conclusion: AI-ECG models appear to be effective tools for detecting PRCM and may represent a viable solution to reduce diagnostic delays consequently improving maternal outcomes. Additional studies are needed to evaluate its impact on clinical outcomes in pragmatic settings.

INTRODUCTION: Cardiovascular disease is a leading cause of maternal mortality. Our aim was to evaluate the feasibility and quality of artificial intelligence (AI)-guided focused cardiac ultrasound (FoCUS) by novice scanners to screen for left ventricular (LV) systolic dysfunction in pregnant individuals. METHODS: Patients were prospectively enrolled during routine obstetric appointments. FoCUS images were acquired by individuals without scanning or clinical experience using AI-guided image acquisition software (UltraSight) on a Philips Lumify device. Standard two-dimensional imaging views were attempted: parasternal long axis (PSLA), parasternal short axis mid-level (PSSA Mid), apical four chamber (A4C), three chamber (A3C), and two chamber (A2C). Two experienced cardiologists reviewed all images and graded diagnostic quality: poor, adequate, good, excellent. RESULTS: A total of 101 pregnant participants were enrolled; median age was 32 (interquartile range [IQR], 29, 35) years. The median gestational age was 23 (IQR, 16, 31) weeks; 16% first trimester, 45% second, and 39% third. The most feasible views were the PSLA and PSSA Mid, 98% and 97% with adequate or greater image quality, respectively. A3C and A2C were least feasible, acquired in 34% and 40% with adequate or greater image quality in 28% and 35%, respectively. Average scan time was 6 minutes, 46 seconds ± 0.2. The LV ejection fraction could be determined for the entire cohort and was normal (50–75%) for all. CONCLUSIONS/IMPLICATIONS: Implementation of AI-guided FoCUS in pregnant individuals is feasible, even in the absence of an experienced sonographer. This approach could allow for low-cost screening and early diagnosis of LV systolic dysfunction during pregnancy.

Introduction: Cardiac sarcoidosis (CS) results in HFrEF, but its natural history and response to GDMT remains uncertain. Methods: We queried a registry of 631 CS patients evaluated at our institution to identify those with LVEF < 35% on TTE within 30 d of diagnosis and a follow-up TTE 12-18 months later (external CS diagnoses were excluded). Demographics, LVEF, NT-proBNP, GDMT, immunosuppression, CIED, and outcomes (hyperkalemia, LVAD/transplant, CV hospitalization, death) were recorded. Results: Overall, 25/631 (4%) of CS patients had an LVEF <35% at initial presentation (36% female, 84% white; median age of 58 y; probable/not biopsy-proven CS in 52%). Mean baseline LVEF was 27% (σ=7%), NT-proBNP 1736 pg/mL (n=11), potassium 4.3 mEq/L (n=18), and median NYHA Class II (54%). Patients were on a mean 1.7 GDMT agents at CS diagnosis (Figure). Most were treated with immunosuppression (88%) for mean 0.75 y, usually prednisone (88%, mean 16 mg daily) and fewer with mycophenolate (32%, mean 1875 mg daily), methotrexate (28%, mean 18mg weekly), or TNF-α inhibitor (4%). 64% were on trimethoprim-sulfamethoxazole prophylaxis. At follow-up (median 317 d), 3 patients (12%) required LVAD and 1 (4%) required heart transplant. Mean LVEF improved to 37% (excluding LVAD/transplant, σ=11%, p=0.001). CIED prevalence increased from 48 to 81% (p=0.03). Overall GDMT did not change (mean 2.0 agents, p=0.2) and only beta-blocker dose changed (64 to 138 mg mean daily metoprolol equivalents). Potassium did not change (p=0.4) and hyperkalemic events were infrequent (16%). No patients died, but unplanned hospitalization occurred in 36% of patients. Conclusions: CS patients with HFrEF are usually on partial GDMT at diagnosis, presumably due to earlier recognition of HFrEF, but intensification of GDMT may be limited. Despite improved LVEF which may be due to GDMT, immunosuppression, and/or CIED, over 1/3 patients were hospitalized within 18 mo, highlighting the high-risk profile of HFrEF in CS.

Introduction: Severe tricuspid regurgitation (TR) is associated with considerable morbidity/mortality in an elderly population with multiple comorbidities. There is interest in transcatheter interventions to manage severe TR in this population. Understanding right heart (RH) anatomy has implications for patient/device selection for transcatheter intervention. Hypothesis: We analyzed RH anatomy by CT of patients referred for TR therapy to determine transcatheter tricuspid intervention device sizing needs. Methods: Retrospective Mayo Clinic study of 29 patients considered for intervention of severe TR from 3/1/2016-12/15/2020; echocardiogram and CT angiogram obtained in all patients. Results: Mean age was 83±8 (83% female), 100% had atrial fibrillation, and 62% had chronic kidney disease ≥3a. 97% were symptomatic, 93% were prescribed loop diuretics, and 24% had device leads. Mean tricuspid annular plane systolic excursion 16.8±4.5 mm, effective regurgitant orifice area (EROA) 81±33 mm 2 , and cardiac index 2.6±0.6 L/min/m 2 . 41% had at least moderate right ventricular (RV) dysfunction with a mean RV systolic pressure of 46±16 mmHg. Patients receiving intervention had significantly larger EROA (101±33 v. 63±22 mm 2 , P=0.033), shorter tricuspid leaflet tenting length (6.5±3.0 v. 8.9±2.7 mm, P=0.042), and smaller annuloplasty diagnostic perimeter during diastole (120.1±16.6 v. 131.1±7.4 mm, P=0.041). Intervention patients tended to have better right ventricular function, smaller RV and IVC size, and more severe symptoms. Max tricuspid annulus diameter in systole and diastole was 51±5 and 53±7 mm, respectively. Conclusions: Severe TR patients referred for transcatheter intervention present with severe RH enlargement with a large proportion having tricuspid annulus dimensions outside the range for current devices available in clinical trials. These data have implications for device development/selection and procedural feasibility.

Introduction: Subclinical cardiovascular involvement in COVID-19 patients has not been well described. 2D-Speckle Tracking Echocardiography derived global longitudinal strain (GLS) and systolic and early diastolic strain rate (SRs, SRe) measurements are more sensitive than standard echocardiographic parameters to diagnose subclinical mechanical dysfunction in patients with normal Left Ventricle Ejection Fraction (LVEF). Hypothesis: Evaluate subclinical myocardial mechanical function and reserve in active COVID-19 systemic disease patients with normal LVEF vs controls and between survivors and non survivors of COVID-19 in the longitudinal domain of contractility. Methods & Results: 166 adult patients with active COVID 19 having normal LVEF were included and compared to 89 healthy volunteers. Baseline parameters were recorded. Peak GLS, SRs and SRe, were measured offline. Mean age was 62.53 ± 18.96 years and 87 (52%) were males and Mean LVEF was 62±5%. COVID-19 patients had higher GLS compared to controls (-20.93%±0.30 vs. -18.48%±0.40; p value <0.0001), SRs was similar (-0.97 s -1 ±0.10 vs. -0.98 s -1 ±0.02, p value 0.2528) and SRe was lower (0.85 s -1 ±0.0 vs. 1.05 s -1 ±0.02; p value <0.0001) respectively. After adjusting for age and sex, GLS, SRs and SRe was significant (p<0.001). GLS, SRs and SRe were similar in survivors vs. non survivors (p NS), both groups had elevated biomarkers (cardiac troponin, NT-pro BNP, CRP), but non survivors had higher levels (3.22±0.18 vs 3.62±0.28, p = 0.2371, 5.86±0.19 vs 7.15±0.27, p = 0.0004, 3.75±0.14 vs 4.31±0.21, p = 0.0422, 7.64±0.17 vs 8.26±0.25, p=0.0487 respectively) (Table 1). Conclusions: In our cohort, COVID-19 patients had better systolic reserve and abnormal relaxation compared to controls. There was no significant difference in strain (GLS) and strain rate (SRs & SRe) amongst survivors and non survivors in the longitudinal domain of contractility despite of higher inflammatory biomarkers in non-survivors.