ESPE Yearbook of Paediatric Endocrinology

Front Endocrinol (Lausanne). 2025;16:1569355. doi: 10.3389/fendo.2025.1569355

Brief Summary: This prospective observational study on 50 preterm and full-term neonates utilized a steroid metabolomic signature approach, with metabolomic clustering revealing distinct adrenal steroid profiles associated with neonatal health outcomes, highlighting the complex interplay between steroidogenesis and clinical risks. K-means clustering was employed to distinguish three clusters among all neonates in the study, based on steroid profiling using gas chromatography-mass spectrometry of 24-hour urine collections. These three clusters differed in all adrenal steroid synthesis pathways, but not in prematurity, gestational age or birth weight.

The greatest variability among the clusters was related to DHEA and 17-hydroxyprogesterone metabolites. A decreased excretion for both C19 and C21 steroids was observed in Cluster 1, an increased excretion in Cluster 3, and moderately elevated excretion in Cluster 2. With regard to cortisol and cortisone derivatives, Cluster 2 exhibited high levels of excretion, while Cluster 3 demonstrated an intermediate level of excretion; however, Cluster 1 presented a significantly decreased result, with a statistical disparity when compared with Cluster 2. The study observed that neonates in Cluster 1 exhibited a marked reduction in steroids, indicating a profoundly diminished adrenal output. Collectively, these enzymatic patterns suggest a global suppression of neonatal steroidogenesis in Cluster 1, with emergency cesarean sections also significantly more frequent in Cluster 1 compared to Cluster 3. Ultimately, in Cluster 3, where steroidogenesis appears to be preserved, clinical outcomes were favorable.

These findings demonstrate the potential of metabolomic signatures to facilitate the stratification of neonates according to their adrenal steroid profiles, thus offering a promising avenue for the delivery of personalized neonatal care. The authors propose that individualized postnatal steroid replacement regimens could be optimized using each neonate’s steroid metabolomic fingerprint, moving beyond uniform protocols based solely on gestational age. By leveraging AI-driven analytics, clinicians could more accurately identify, classify, and predict high-risk neonatal phenotypes, thus enhancing diagnostic precision and ultimately improving patient outcomes.