Midazolam dose correlates with abnormal hippocampal growth and neurodevelopmental outcome in preterm infants

Objective Very preterm-born neonates (24–32 weeks of gestation) are exposed to stressful and painful procedures during neonatal intensive care. Analgesic and sedation therapies are essential, and opiates and benzodiazepines are commonly used. These medications may negatively impact brain development. The hippocampus may be especially vulnerable to the effects of pain and analgesic and/or sedative therapies and contribute to adverse outcomes. The effect of invasive procedures and analgesic–sedative exposure on hippocampal growth was assessed, as was that of hippocampal growth on neurodevelopmental outcome. Methods A total of 138 neonates (51% male, median gestational age = 27.7 weeks) underwent magnetic resonance imaging and diffusion tensor imaging (DTI) scans, early in life (postmenstrual age [PMA] = 32.3 weeks) and at term-equivalent age (PMA = 40.2 weeks). Volumes and DTI measures of axial diffusivity, radial diffusivity, and mean diffusivity (MD) were obtained from the hippocampus. Cognitive, language, and motor abilities were assessed using the Bayley Scales of Infant Development–III at 18.7 months median corrected age. Models testing the association of invasive procedures with hippocampal volumes and DTI measures accounted for birth gestational age, sex, PMA, dose of analgesics/sedatives (fentanyl, morphine, midazolam), mechanical ventilation, hypotension, and surgeries. Results Total midazolam dose predicted decreased hippocampal volumes (β = −1.8, p   0.5 each). Lower cognitive scores were associated with hippocampal growth (β = −0.31, p = 0.003), midazolam dose (β = −0.27, p = 0.03), and surgery (β = −8.32, p = 0.04). Interpretation Midazolam exposure was associated with macro- and microstructural alterations in hippocampal development and poorer outcomes consistent with hippocampal dysmaturation. Use of midazolam in preterm neonates, particularly those not undergoing surgery, is cautioned. Ann Neurol 2016;79:548–559