Daily Cardiology Research Analysis
A large multicenter randomized trial showed that maintaining very-low–tidal volume ventilation with PEEP during cardiopulmonary bypass does not reduce postoperative infections, challenging current practice. An AI system (PanEcho) achieved high accuracy across 39 echocardiography tasks with robust external validation, suggesting practical deployment for automated echo interpretation. A registry analysis of 17,838 aortic stenosis patients found that acute/advanced presentation before valve replace
Summary
A large multicenter randomized trial showed that maintaining very-low–tidal volume ventilation with PEEP during cardiopulmonary bypass does not reduce postoperative infections, challenging current practice. An AI system (PanEcho) achieved high accuracy across 39 echocardiography tasks with robust external validation, suggesting practical deployment for automated echo interpretation. A registry analysis of 17,838 aortic stenosis patients found that acute/advanced presentation before valve replacement doubles post-AVR mortality risk, supporting earlier detection and referral.
Research Themes
- Perioperative ventilation strategies during cardiopulmonary bypass
- AI-enabled echocardiography for comprehensive diagnostics
- Timing and acuity of presentation in aortic stenosis and outcomes after AVR
Selected Articles
1. Maintaining ventilation with very low tidal volume and positive-end expiratory pressure versus no ventilation during cardiopulmonary bypass for cardiac surgery in adults: a randomized clinical trial.
In 1,362 adults undergoing cardiac surgery with CPB, maintaining very-low–tidal volume ventilation with PEEP did not reduce 28-day postoperative infections versus no ventilation (10.0% vs 10.9%; RR 0.92; p=0.58). Antibiotic use was higher in the ventilation-maintained group (IRR 1.08; p=0.02), with no significant differences in other secondary outcomes or adverse events.
Impact: This large, registered, multicenter RCT directly challenges a common intraoperative practice and provides high-level evidence to simplify CPB management without increasing infections.
Clinical Implications: Routine maintenance of very-low–tidal volume ventilation with PEEP during CPB solely to prevent infections is not supported; teams may reasonably discontinue ventilation during CPB while adhering to standard infection prevention bundles.
Key Findings
- Primary outcome: postoperative infection at 28 days was similar (MV− 10.9% vs MV+ 10.0%; RR 0.92; 95% CI 0.67–1.25; p=0.58).
- Antibiotic use was higher in the ventilation-maintained group (incidence risk ratio 1.08; 95% CI 1.02–1.15; p=0.02).
- No significant differences in secondary outcomes or adverse events between groups.
- Ventilation protocol used tidal volume 2.5 mL/kg predicted body weight and PEEP 5–7 cmH2O during CPB.
Methodological Strengths
- Multicenter, single-blind randomized design with large sample size (n=1362).
- Prospective trial registration (NCT03372174) and standardized ventilation protocol.
Limitations
- Single-blind design; potential practice variability across centers.
- Focus on elective surgeries in France may limit generalizability to emergent cases or other settings.
Future Directions: Assess alternative intraoperative lung-protective strategies for endpoints beyond infection (e.g., pulmonary complications, ICU stay), and evaluate subgroups (e.g., severe COPD) where ventilation during CPB could be beneficial.
PURPOSE: Cardiopulmonary bypass (CPB) during cardiac surgery mechanically circulates and oxygenates the blood, bypassing the heart and lungs. Despite limited evidence, maintaining mechanical ventilation (MV) during CPB is recommended, as ventilator strategies during surgery may reduce the occurrence of postoperative infections. We aimed to determine whether maintaining MV for cardiac surgery would decrease postoperative infections compared with stopping MV during CPB. METHODS: We conducted a multicenter, single-blind, randomized trial among adult patients undergoing scheduled cardiac surgery with CPB in six hospitals in France. During CPB, the tracheal tube was disconnected from the ventilator in the control group (MV- group). In the MV + group, ventilation was maintained during CPB with very low tidal volume ventilation, using a tidal volume of 2.5 mL/kg of predicted body weight, with 5-7 cmH
2. PanEcho: Complete AI-enabled echocardiography interpretation with multi-task deep learning.
PanEcho accurately performed 18 diagnostic classifications (median AUC 0.91) and 21 parameter estimations (median normalized MAE 0.13) across internal and external cohorts. It estimated LVEF with MAE 4.2% (internal) and 4.5% (external), and detected severe aortic stenosis with AUC 0.98–1.00. Performance remained strong with abbreviated protocols (median AUC 0.91) and point-of-care ultrasound (median AUC 0.85).
Impact: Represents a comprehensive, multi-task, externally validated AI covering both diagnostic and quantitative echocardiography at scale, with practical performance in limited-protocol and POCUS settings.
Clinical Implications: PanEcho could serve as an adjunct reader to standardize and accelerate echo reporting, aid triage in resource-constrained settings, and enable rapid screening (e.g., LV/RV dysfunction, severe AS) from routine TTE and POCUS.
Key Findings
- Median AUC 0.91 (IQR 0.88–0.93) across 18 diagnostic tasks; median normalized MAE 0.13 (0.10–0.18) across 21 parameters.
- LVEF estimation MAE: 4.2% (internal) and 4.5% (external); RV systolic dysfunction AUC 0.93–0.94; severe AS AUC 0.98–1.00.
- Abbreviated TTE protocols retained strong performance (median AUC 0.91), and real-world POCUS achieved median AUC 0.85.
- Model trained on 1.2 million videos from 32,265 TTE studies; code/model availability stated.
Methodological Strengths
- Large-scale development with temporal internal validation and multi-cohort external validation.
- Multi-task design integrating classification and regression across 39 echo targets.
Limitations
- Retrospective design with reliance on clinical labels; preprint (not peer reviewed) at time of report.
- Generalizability to non-participating geographies and devices may need further validation.
Future Directions: Prospective clinical utility studies (workflow impact, time-to-report, diagnostic yield), regulatory-grade validation across vendors, and integration with decision-support for guideline-concordant care.
IMPORTANCE: Echocardiography is a cornerstone of cardiovascular care but relies on expert interpretation and manual reporting from a series of videos. We propose an artificial intelligence (AI) system, PanEcho, to automate echocardiogram interpretation with multi-task deep learning. OBJECTIVE: To develop and evaluate the accuracy of PanEcho on a comprehensive set of 39 echocardiographic labels and measurements on transthoracic echocardiography (TTE). DESIGN SETTING AND PARTICIPANTS: This study represents the development and retrospective, multi-site validation of an AI system. PanEcho was developed using a sample of TTE studies conducted at Yale-New Haven Health System (YNHHS) hospitals and clinics from January 2016-June 2022 during routine care. The trained model was internally validated in a temporally distinct YNHHS cohort from July-December 2022, externally validated across four diverse external cohorts, and made publicly available.
3. Acute Valve Syndrome in Aortic Stenosis.
In 17,838 AS patients undergoing AVR, over half presented with acute/advanced symptoms (AVS). Two-year mortality was 17.5% in AVS vs 7.6% (PVS) and 5.8% (asymptomatic), and HF hospitalization was 41.5% in AVS. After adjustment, AVS presentation was associated with a two-fold higher mortality risk (HR 2.2).
Impact: Quantifies the prognostic penalty of delayed or acute presentation in AS at national scale, reinforcing the need for earlier detection, referral, and timely AVR.
Clinical Implications: Screening and timely evaluation of symptomatic AS are critical; healthcare systems should prioritize pathways that reduce time-to-AVR to avoid progression to AVS and its doubled mortality risk.
Key Findings
- Among AVR patients, 51.7% presented with acute/advanced symptoms (AVS).
- Two-year mortality: asymptomatic 5.8%, PVS 7.6%, AVS 17.5%; two-year HF hospitalization: 11.1%, 19.0%, and 41.5%, respectively.
- AVS presentation independently associated with higher mortality after AVR (adjusted HR 2.2; 95% CI 1.8–2.6).
Methodological Strengths
- Very large real-world multicenter cohort (n=17,838) with standardized classification of presentation acuity.
- Time-to-event analysis with adjustment for confounders.
Limitations
- Observational design with potential residual confounding and selection biases.
- Classification of acuity relies on clinical documentation; unmeasured factors may influence outcomes.
Future Directions: Develop and test care pathways that expedite evaluation/AVR in high-risk symptomatic AS, and validate risk models integrating presentation acuity with imaging and biomarkers.
BACKGROUND: To describe the impact of clinical presentation among patients with aortic stenosis (AS) undergoing aortic valve replacement (AVR). METHODS: We analyzed a real-world dataset including patients from 29 US hospitals (egnite Database, egnite). Patients over 18 years old with moderate or greater AS undergoing AVR were included. Patients were classified into 3 groups according to the acuity and severity of clinical presentation prior to AVR: (i) asymptomatic, (ii) progressive signs and symptoms (progressive valve syndrome [PVS]), and (iii) acute or advanced signs and symptoms (acute valve syndrome [AVS]). Mortality and heart failure hospitalization after AVR were examined with Kaplan-Meier estimates, with results compared using the log-rank test. RESULTS: Among 2,009,607 patients in our database, 17,838 underwent AVR (78.6% transcatheter AVR, 21.4% surgical AVR). Age was 76.5 ± 9.7 years, and 40.2% were female. Prior to AVR, 2504 (14.0%) were asymptomatic, 6116 (34.3%) presented with PVS, and 9218 (51.7%) presented with AVS. At 2 years, the estimated rate of mortality for asymptomatic, PVS, and AVS were 5.8% (4.6%-7.0%), 7.6% (6.7%-8.4%), and 17.5% (16.5%-18.5%), respectively, and the estimated rate of hospitalization with heart failure for asymptomatic, PVS, and AVS were 11.1% (9.5%-12.6%), 19.0% (17.8%-20.2%), and 41.5% (40.2%-42.8%), respectively. After adjustment, patients presenting with AVS had increased risk of mortality after AVR (hazard ratio, 2.2; 95% CI, 1.8-2.6). CONCLUSIONS: From a large, real-world database of patients undergoing AVR for AS, most patients presented with AVS, which was associated with an increased risk of mortality and heart failure hospitalization.