Accuracy of in-utero MRI to detect fetal brain abnormalities and prognosticate developmental outcome: postnatal follow-up of the MERIDIAN cohort.

Abstract

In utero MRI (iuMRI) detects fetal brain abnormalities more accurately than ultrasonography and provides additional clinical information in around half of pregnancies. We aimed to study whether postnatal neuroimaging after age 6 months changes the diagnostic accuracy of iuMRI and its ability to predict developmental outcome. Families enrolled in the MERIDIAN study whose child survived to age 3 years were invited to have a case note review and assessment of developmental outcome with the Bayley Scales of Infant and Toddler Development, the Ages and Stages Questionnaire, or both. A paediatric neuroradiologist, masked to the iuMRI results, reviewed the postnatal neuroimaging if the clinical report differed from iuMRI findings. Diagnostic accuracy was recalculated. A paediatric neurologist and neonatologist categorised participants' development as normal, at risk, or abnormal, and the ability of iuMRI and ultrasonography to predict developmental outcome were assessed. 210 participants had case note review, of whom 81 (39%) had additional investigations after age 6 months. The diagnostic accuracy of iuMRI remained higher than ultrasonography (proportion of correct cases was 529 [92%] of 574 vs 387 [67%] of 574; absolute difference 25%, 95% CI 21 to 29; p<0·0001). Developmental outcome data were analysed in 156 participants, and 111 (71%) were categorised as normal or at risk. Of these 111 participants, prognosis was normal or favourable for 56 (51%) using ultrasonography and for 76 (69%) using iuMRI (difference in specificity 18%, 95% CI 7 to 29; p=0·0008). No statistically significant difference was seen in infants with abnormal outcome (difference in sensitivity 4%, 95% CI -10 to 19; p=0·73). iuMRI remains the optimal tool to identify fetal brain abnormalities. It is less accurate when used to predict developmental outcome, although better than ultrasonography for identifying children with normal outcome. Further work is needed to determine how the prognostic abilities of iuMRI can be improved. National Institute for Health Research Health Technology Assessment programme.