Patterns of Brain Injury in Neonates Exposed to Perinatal Sentinel Events

OBJECTIVES: We studied (1) the pattern of brain injury in term neonates with encephalopathy with evidence of a preceding hypoxic sentinel event, (2) prenatal and perinatal risk factors, and (3) the correlation between neuroimaging findings and developmental outcomes. METHODS: We identified, among 500 term neonates with encephalopathy who were studied with MRI between 1992 and 2005, 48 infants with evidence of a preceding acute hypoxic event, and we reviewed their MRI scans retrospectively. Prenatal and perinatal data were compared with those for term normal low-risk infants. Neurodevelopmental outcomes were assessed at a minimum of 12 months. RESULTS: Five patterns of brain injury were identified, as follows: pattern I, basal ganglia and thalami lesions associated with severe white matter damage (n = 6; 14%); pattern II, basal ganglia and thalami lesions with mild or moderate white matter changes (n = 24; 56%); pattern III, isolated thalamic injury (n = 2; 5%); pattern IV, moderate white matter damage only (n = 1; 2%); pattern V, mild white matter changes or normal findings (n = 10; 23%). No scan showed evidence of long-standing injury. The internal capsule was abnormal in 93% of infants with patterns I and II, and 86% of those infants died or developed cerebral palsy. Infants with patterns III and IV had developmental delay and diplegic cerebral palsy, respectively. Pattern V was associated with normal outcomes. Case infants were significantly more often of African descent, born to pluriparous or hypertensive mothers. Uterine rupture followed previous cesarean section in 8 of 11 cases. Cord prolapse accompanied undiagnosed breech presentation in 4 of 9 cases. CONCLUSIONS: Basal ganglia and thalami lesions are the imaging signature in term neonates exposed to hypoxic-ischemic sentinel events. Patterns of central gray matter and secondary white matter injury were associated with higher risks of severe morbidity and death. Affected infants did not seem intrinsically different from our low-risk population. These data support the need for anticipating sentinel events and expediting delivery.