MR imaging of term infants with hypoxic-ischaemic encephalopathy as a predictor of neurodevelopmental outcome and late MRI appearances

Morbidity attributable to hypoxic-ischaemic injury (HIE) in the perinatal period remains problematic, and timely and accurate assessment of the degree of injury is required for clinical management and prognosis. Conventional MR sequences typically appear normal in the first 48 h post HIE. While diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) maps register the injury earlier, perhaps within the first 24 h, it has been suggested that there may be a propensity at that early stage to underestimate the lesion severity or extent.

To assess whether MR imaging that included DWI, measured ADC values and T1- and T2-weighted sequences ultimately correlated with either neurodevelopmental outcome or with late MR imaging at 2 years of age. In addition, we wished to compare the performance of MR imaging with cranial US imaging.

All infants presenting with HIE who had an MRI within 10 days of life were eligible for enrollment and subsequently underwent a full neurodevelopmental assessment at 2 years of age. All children underwent repeat MRI at this time. All neonates had at least one cranial US study. The US findings were categorized as normal, abnormalities confined to the cerebral cortex and subcortical white matter, isolated central grey matter hyperechogenicity, and central hyperechogenicity combined with cerebral cortical/subcortical changes. All MRI studies were retrospectively reviewed by three radiologists. The patterns of injury on the early DWI and ADC maps and early T1- and T2-W studies were recorded as diffuse, central, watershed or atypical. The patterns of signal abnormality were assessed using a scoring system that yielded four separate scores [basal ganglia (BG), watershed (W), BG/W and summation (S)] for the three sets of images, a total of 12 scores in all. The appearance of the posterior limb of the internal capsule (PLIC) on T1-W inversion recovery sequences and of the corpus callosum on all sequences was also documented. After detailed neurodevelopmental assessment at 2 years of age, infants were classified into two groups according to whether they had a favourable or unfavourable outcome.

Of the 26 infants, 6 infants died before formal assessment at the age of 2 years. A further 5 infants had moderate to severe cerebral palsy in addition to severe cognitive impairment. The remaining 15 infants were categorized in the favourable outcome group. The US appearance performed well in terms of predicting final outcome (P = 0.005). The pattern of ischaemia seen on early MRI was a significant predictor of outcome (P < 0.0001). The BG, BG/W and S scores of the diffusion imaging were significantly associated with outcome (P < 0.0001, P < 0.0001 and P = 0.0005 respectively). DWI was predictive of outcome group (P < 0.0001), as were the early T1- and T2-W sequences (P = 0.002) and cranial US (P = 0.005). Assessment of the PLIC in infants with watershed or atypical patterns of ischaemia was found to be less reliable in predicting outcome. The measured ADC value in the PLIC was significantly reduced in those children who had an unfavourable outcome (P = 0.03).

While early MRI performed better than cranial US, the sonography findings were useful. The pattern of ischaemia on early MRI was a good predictor of prognosis. All infants with watershed or atypical patterns had a favourable outcome. The majority of infants with central patterns of ischaemia had an unfavourable outcome and all infants with a diffuse pattern had an unfavourable outcome. DWI was predictive of outcome group, as were early T1- and T2-W sequences and cranial US.