Hormone modeling in preterm neonates: Establishment of pituitary and steroid hormone reference intervals

Context: Reports suggest significant differences in serum levels of hormones in extremely preterm compared with late preterm and full-term infants. Objectives: The purpose of this study was to develop reference intervals (RIs) for 3 pituitary hormones and 5 steroid hormones in serum of preterm infants. Design: Blood samples were collected from 248 (128 male and 120 female) preterm neonates born between 24 and 32 weeks' gestation. Setting: Participants were recruited from 3 neonatal intensive care wards in Melbourne, Australia. Participants: No infant in this cohort had ambiguous genitalia or other endocrine abnormalities. All infants included in the RI determination survived beyond the equivalent of term. Interventions: Serum was analyzed for prolactin, FSH, and LH by automated electrochemiluminescence immunoassay (Roche Cobas 8000-e601). Liquid chromatography coupled with tandem mass spectrometry was used for analysis of 17-hydroxyprogesterone, androstenedione, cortisol, cortisone, and testosterone. Main Outcome Measures: The robust method was applied to define the central 95% RI, after each hormone measure was transformed using a Box-Cox transformation to correct for asymmetry. Results: RIs were established for 8 hormones. Gender-specific intervals were developed for FSH, LH, and testosterone. Cortisone and 17- hydroxyprogesterone required division based on gestational age, with neonates born at <30 weeks' gestation demonstrating higher levels than their older counterparts. Androstenedione, cortisol, and prolactin did not require any division within this cohort for RI assignment. Conclusions: This report provides the first characterization of serum steroids measured by mass spectrometry in preterm neonates, with the additional characterization of 3 pituitary hormones in infants born at ≤32 weeks' gestation. Use of these data allows for correct interpretation of results for very preterm neonates and reduces the risk of incorrect diagnosis due to misinterpretation of data.