Early Blood Metabolome Remodeling Reveals Metabolic Signatures of Hypoxic-Ischemic Encephalopathy

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Hypoxic-ischemic encephalopathy (HIE) is a neonatal brain injury in which a definitive diagnosis within the first 6 hours of life is essential for initiating therapeutic hypothermia and improving neurological outcomes. However, early clinical evaluation and currently available biomarkers lack quantitative specificity during this narrow therapeutic window. We are providing insights into how rapid disturbances in systemic metabolites that regulate cerebral energy metabolism, neurotransmitter cycling, redox balance, and membrane integrity generate a definitive biochemical signature of HIE immediately after birth.

We performed quantitative blood-based 1H NMR metabolomics on collected blood samples within ∼1 hour of birth from 81 neonates (HIE, n = 42; non-HIE, n = 39), under optimized handling conditions to preserve metabolic integrity. Metabolite analysis revealed a distinct HIE-associated profile characterized by elevated lactate, alanine, succinate, glutamate, taurine, glycine, choline, and pyroglutamate, alongside significant depletion of glucose and glutamine compared with non-HIE controls (p < 0.05). These coordinated metabolic shifts reflect impaired mitochondrial respiration, enhanced anaerobic glycolysis, excitotoxic amino acid accumulation, altered membrane phospholipid turnover, and oxidative stress.

Multivariate analysis demonstrated clear separation between groups (PLS-DA accuracy = 0.83, R²Y = 0.46, Q² = 0.82), with glutamine, lactate, glutamate, and pyroglutamate as key discriminators. Pathway enrichment highlighted perturbations in glycolysis, the glucose-alanine cycle, glutamate-glutamine metabolism, Warburg effect like metabolic reprogramming, and redox homeostasis. Integration into supervised machine-learning models (Random Forest, XGBoost, SVM, KNN) achieved strong diagnostic performance (AUC = 0.97 ± 0.03; sensitivity ≈ 87%). Collectively, this minimally invasive NMR-to-machine-learning framework enables early, mechanistically grounded risk stratification of neonatal HIE within the therapeutic window.

Article activity feed