599 - Disrupted Neural Network Integration Links Early Brain Injury to Later Neurodevelopment in Infants with Congenital Heart Disease

Publication Number: 3580.599

Lauren Harasymiw, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Amy Kuang, University of California, San Francisco, School of Medicine, Daly City, CA, United States; Duan Xu, UCSF Benioff Children's Hospital San Francisco, San Francisco, CA, United States; Aaron Scheffler, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Elizabeth George, UCSF Benioff Children's Hospital San Francisco, San Francisco, CA, United States; Shabnam Peyvandi, UCSF Benioff Children's Hospital San Francisco, San Francisco, CA, United States; Patrick McQuillen, University of California, San Francisco, School of Medicine, San Francisco, CA, United States

Background: Infants with critical congenital heart disease (CCHD) are at high risk for brain dysmaturation and acquired brain injuries, yet the neural mechanisms linking early injury to later outcomes remain incompletely understood. Diffusion MRI (dMRI) and structural connectivity processing can quantify developmental changes in white matter tracts and describe whole-brain neural network organization.

Objective: This study investigated the effects of perioperative brain injury on whole-brain neural network development in infants with CCHD, and whether network development predicts early developmental outcomes.

Design/Methods: 97 term infants with either d-Transposition of the Great Arteries (d-TGA, n = 51) or single ventricle physiology (SVP, n = 46) completed pre- and/or post-operative MRIs, including dMRI, and longitudinal follow-up between 1-4 years using the Bayley Scales of Infant Development (BSID). Brain injury type and severity were scored using volumetric T1- and T2- weighted images. Graph theory metrics-global efficiency (a measure of network integration) and transitivity and modularity (measures of network segregation)-were used to assess network topological changes. Linear mixed-effects models examined the influence of cardiac physiology and brain injury on network development over time, and a joint modeling approach was used to evaluate whether network development predicted BSID scores.

Results: Brain injury was present in 45% of the cohort preoperatively (n = 38) and 35% post-operatively (n = 30, Table 2). While cardiac physiology did not significantly affect perioperative network development, infants with WMI exhibited a slower increase in global efficiency (p = 0.013), with a similar trend observed in those with stroke (Figure 1). Notably, greater perioperative gains in global efficiency were associated with higher cognitive (p < 0.001), language (p < 0.001), and motor (p = 0.007) scores in early childhood across all subjects, independent of brain injury (Table 3). For each 1-standard deviation increase in the predicted trajectory of global efficiency, there was an 8.7-point increase in BSID cognitive score (95% CI: 4.5-13.0).

Conclusion(s): These findings indicate that brain injury, specifically white matter injuries, disrupt normal network maturation in infants with CCHD. Moreover, the trajectory of whole-brain network integration during the early neonatal period may serve as a sensitive marker of future neurodevelopment in this population, particularly for cognitive outcomes. These insights support the use of structural connectome-based metrics as early and influential indicators of brain development.

Table 1. Frequency of brain injury during the perioperative period
1P-values from Chi Square tests. 2Includes white matter injury, intraventricular hemorrhage, stroke, and hypoxic-ischemic injury.

Figure 1. Impact of cardiac lesion and brain injury on the development of network integration during the perioperative period
Linear mixed-effects linear models for global efficiency by cardiac lesion (A), presence of any brain injury (B), presence of white matter injury (C) or presence of stroke (D) after adjustment for gestational age and maternal education.

Table 3. Fixed effects estimates from multivariate linear mixed-effects models predicting Bayley Scales of Infant Development (BSID) composite scores
White matter injury (WMI), intraventricular hemorrhage (IVH). 1Did not reach significance during univariate testing/excluded from multivariate model.