MRI Based Preterm White Matter Injury Classification: The Importance of Sequential Imaging in Determining Severity of Injury

BACKGROUND The evolution of non-hemorrhagic white matter injury (WMI) based on sequential magnetic resonance imaging (MRI) has not been well studied. Our aim was to describe sequential MRI findings in preterm infants with non-hemorrhagic WMI and to develop an MRI classification system for preterm WMI based on these findings. METHODS Eighty-two preterm infants (gestation ≤35 weeks) were retrospectively included. WMI was diagnosed and classified based on sequential cranial ultrasound (cUS) and confirmed on MRI. RESULTS 138 MRIs were obtained at three time-points: early (<2 weeks; n = 32), mid (2-6 weeks; n = 30) and term equivalent age (TEA; n = 76). 63 infants (77%) had 2 MRIs during the neonatal period. WMI was non-cystic in 35 and cystic in 47 infants. In infants with cystic-WMI early MRI showed extensive restricted diffusion abnormalities, cysts were already present in 3 infants; mid MRI showed focal or extensive cysts, without acute diffusion changes. A significant reduction in the size and/or extent of the cysts was observed in 32% of the infants between early/mid and TEA MRI. In 4/9 infants previously seen focal cysts were no longer identified at TEA. All infants with cystic WMI showed ≥2 additional findings at TEA: significant reduction in WM volume, mild-moderate irregular ventriculomegaly, several areas of increased signal intensity on T1-weighted-images, abnormal myelination of the PLIC, small thalami. CONCLUSION In infants with extensive WM cysts at 2-6 weeks, cysts may be reduced in number or may even no longer be seen at TEA. A single MRI at TEA, without taking sequential cUS data and pre-TEA MRI findings into account, may underestimate the extent of WMI; based on these results we propose a new MRI classification for preterm non-hemorrhagic WMI.