Pediatric resident physician University of Colorado DENVER, Colorado, United States
Abstract:
Introduction: Patients with single ventricle heart disease (SVHD) experience significant neurologic morbidity, particularly damage to white matter in the brain, causing lifelong neurodevelopmental (ND) abnormalities. Currently there are no clinically available biomarkers of neurologic injury or ND outcomes. Proteomics represents a novel technique to identify new predictors of neurologic injury to risk stratify patients in a precision medicine approach. Our group has previously demonstrated a unique proteomic phenotype of patients with SVHD using a limited cardiovascular protein panel. Here, we present an expanded analysis of 1500 proteins in infants undergoing Stage 2 palliation.
Methods: Infants with SVHD in the inter-stage period were compared to similar-age healthy controls. Venous serum samples were collected, stored at -80C, and run on a custom panel of 1500 proteins in single batch analysis (Somalogic Inc., CO). Data was log transformed and auto scaled. Partial least squares-discriminant analysis (PLS-DA) was performed comparing the proteomic profile of cases vs. controls and t-tests were performed to identify differences in individual proteins (FDR < 0.05). All CHD patients were followed for a period of 36 months. Those who underwent clinical ND testing using the Motor Composite from the Bayley Scales of Infant and Toddler Development were included in a secondary analysis comparing the proteomic profile of children with normal (within 1 SD of the mean) and abnormal (> 1 SD below the mean) motor development.
Results: We analyzed differences in the circulating proteome between SVHD cases (n=33) and controls (n=24). PLS-DA readily discriminated between cases and controls based on their proteomic pattern (Accuracy=0.98, R2=0.96, Q2=0.78) (Fig 1a). 568 proteins differed between groups (FDR < 0.05). The 15 proteins with greatest effect on the proteomic phenotype between the groups are shown (Fig 1b). Thirteen subjects received ND testing during our follow up period. Of those patients, eight had abnormal BSID Motor Composite scores. PLS-DA discriminated between patients with normal vs. abnormal Motor Composites (Fig 1c) with 105 proteins differing between groups (p < 0.05). The 15 proteins with greatest effect on differentiation between the proteomic phenotypes are shown (Fig 1d). Notable proteins include NEFL (p=0.007) and CNTN2 (p=0.007), involved in neuronal signaling and oligodendrocyte/myelin sheath organization, respectively.
Conclusions: We report a clear differentiation in the circulating proteome of patients with SVHD and healthy controls with >500 circulating proteins distinguishing the groups. Additionally, proteomic profiling was sufficient to distinguish those with normal versus abnormal scores of motor development in a sub-cohort of SVHD patients. This preliminary data further identifies promising proteins with biologic plausibility as markers of neurologic injury or impaired development that warrant further investigation.
