Inflammatory Subphenotypes After Cardiopulmonary Bypass in Neonates With Congenital Heart Disease

Tuesday, March 24, 2026

11:15 AM - 12:15 PM Central Time

Location: Connections Central - RST 03

DB

Daniel Balcarcel, MD, MBMI

Children's Hospital of Philadelphia

Disclosure(s): No relevant financial relationship(s) to disclose.
FC

Fang Chen, PhD

Researcher
University of Pennsylvania

Disclosure information not submitted.
VG

Vanessa Gonzalez, MBA

Researcher
University of Pennsylvania

Disclosure information not submitted.
JF

Joseph Fraietta, PhD

Assistant Professor of Microbiology
University of Pennsylvania

Disclosure information not submitted.
AP

Asif Padiyath, MD

Assistant Professor of Anesthesiology and Critical Care
Children's Hospital of Philadelphia

Disclosure information not submitted.
WG

William Gaynor, MD

Professor of Surgery
The Children's Hospital of Philadelphia

Disclosure information not submitted.
NY

Nadir Yehya, MD, MSCE

Children's Hospital of Philadelphia, Pennsylvania

Disclosure information not submitted.
MG

Monique Gardner, MD

Assistant Professor of Anesthesiology and Critical Care
The Children's Hospital of Philadelphia

Disclosure information not submitted.

Introduction: Neonates undergoing cardiopulmonary bypass (CPB) for congenital heart disease (CHD) repair are at risk of death and prolonged ICU stays. Prediction of poor outcomes currently relies on non-modifiable clinical factors, limiting opportunities to individualize care. In ARDS and sepsis, inflammatory cytokines delineate biologically distinct subgroups with prognostic and therapeutic implications. We hypothesized that analogous latent classes would emerge after CPB and that a hyperinflammatory phenotype predicts worse outcomes.

Methods: We analyzed serum inflammatory biomarkers, including cytokines and injury-response proteins, collected 24–48 hours after CPB in 137 neonates with CHD. Thirteen biomarkers were selected based on prior literature. Data were log-transformed, batch-corrected using ComBat with empirical Bayes, and standardized. We excluded biomarkers with high correlation. Latent profile analysis (LPA) was performed using tidyLPA in R, estimating models with 1–4 classes. Model fit was assessed by AIC, BIC, adjusted BIC, and entropy. Clinical characteristics were compared across classes.

Results: The best-fitting model included three classes, representing subgroups with generally low (n=37), intermediate (n=80), and high (n=20) levels of inflammatory biomarkers. The high-inflammation subgroup had higher rates (p < 0.001) of ICU-30 composite events (50%; defined as death, ICU stay ≥30 days, or ICU readmission within 30 days), post-operative ECMO (25%), and seizures (25%). In contrast, neonates in the low-inflammation subgroup experienced fewer ICU-30 composite events (5.4%), with no ECMO or post-operative seizures (both 0%). High-inflammation neonates were older at surgery (median 4.7 vs. 3.9 days; p=0.04), more often had single-ventricle anatomy (45.0% vs. 16.2%; p=0.05), and had longer CPB times (median 156 vs. 114 minutes; p=0.02), but not significantly longer cross clamp or hypothermic circulatory arrest times.

Conclusions: We identified a hyperinflammatory subgroup among neonates after CPB for CHD surgery that was strongly associated with worse clinical outcomes, while a distinct low-inflammation subgroup experienced markedly better outcomes. These findings support inflammation as a potential driver of adverse outcomes and could guide future targeted interventions.