Neonatal hyperinsulinism: a retrospective study of presentation and management in a tertiary neonatal intensive care unit in the UK.

Abstract

Objective: Reports of hyperinsulinism typically focus on infants managed by highly specialised services. However, neonates with hyperinsulinism are initially managed by neonatologists and often not referred to specialists. This study aimed to characterise the diversity in presentation and management of these infants.

Setting: Level 3 neonatal intensive care.

Patients: Neonates with hyperinsulinism, defined as blood glucose <2.8 mmol/mL and insulin level >6 pmol/L.

Design: 7-year retrospective study (January 2015-December 2021).

Results: 99 cases were identified: severe-treated with diazoxide (20%), moderate-clinically concerning hyperinsulinism not treated with diazoxide (30%), mild-biochemical hyperinsulinism (50%). Birth weight z-score was -1.02±2.30 (mean±SD), 42% were preterm, but neither variable correlated with clinical severity. The severe group received a higher concentration of intravenous glucose (27±12%) compared with the moderate (15±7%) and mild (16±10%) groups (p<0.001). At diagnosis, the intravenous glucose intake was similar in the severe (7.43±5.95 mg/kg/min) and moderate (5.09±3.86 mg/kg/min) groups, but higher compared with the mild group (3.05+/2.21 mg/kg/min) (p<0.001). In the severe group, term infants started diazoxide earlier (9.9±4.3 days) compared with preterm (37±26 days) (p=0.002). The national congenital hyperinsulinism service was consulted for 23% of infants, and 3% were transferred.

Conclusions: This study highlights the diversity in clinical presentation, severity and prognosis of neonatal hyperinsulinism, irrespective of birth weight and gestational age. More infants were small rather than large for gestational age, and the majority had transient hyperinsulinism and were not referred to the national centre, or treated with diazoxide. Further research is required to understand the breadth of neonatal hyperinsulinism and optimal management.