The relationship between brain structure and function of very preterm infants, and the ability to predict neurodevelopmental outcomes

Abstract

Background: Prediction of motor outcomes and cerebral palsy (CP) in infants born very preterm is needed to identify infants that may benefit from targeted intervention. Brain MRI at term equivalent age in very preterm infants has demonstrated predictive value for CP and adverse motor outcomes. Accuracy is further enhanced when MRI is combined with clinical measures of motor or neurological function. There is a need to determine if MRI and clinical biomarkers earlier than term equivalent age can determine motor outcomes. This would create a new window for possible intervention at a time of greater neuroplasticity and brain development.Aims: To: i) systematically review the evidence for early MRI to identify infants with adverse motor outcomes or CP; ii) validate a structural MRI scoring system of brain injury and growth impairment (Kidokoro et al, 2013) at 29-35 weeks postmenstrual age (PMA, ‘Early MRI’); iii) elucidate motor, neurological and neurobehavioural associations with structural MRI scores at Early and Term (40-42 weeks PMA) MRI; and iv) examine relationships between Early and Term MRI diffusion measures in the corpus callosum, posterior limb of the internal capsule (PLIC) and cerebral peduncle, and 12 month motor outcomes.Research design and methodology: A prospective cohort study was conducted of infants born <31 weeks gestational age. Infants underwent Early and Term 3T MRI without sedation utilising an MRI compatible incubator. Concurrent clinical assessment performed within a week of each MRI consisted of the General Movements assessment (GMs), Hammersmith Neonatal Neurological Examination and the NICU Neonatal Neurobehavioural Scale. The Premie-Neuro was performed following Early MRI and the Test of Infant Motor Performance (TIMP) and a visual assessment at Term. Follow up at 3 months corrected age consisted of the GMs, TIMP and a visual assessment. At 12 months corrected age, infants were evaluated by a paediatrician for evidence of CP using a structured neurological examination. The Bayley Scales of Infant and Toddler Development 3rd edition (Bayley III), Alberta Infant Motor Scale (AIMS) and Neurosensory Motor Developmental Assessment (NSMDA) were conducted.The structural MRI scoring system generated white matter (WM), cortical gray matter (GM), deep GM, cerebellar and global scores. Inter- and intrarater reliability and agreement of each of the MRI subscale scores and the overall global score were evaluated. The relationship between MRI scores and 12 month motor and cognitive outcomes were examined. Associations with concurrent clinical measures were assessed. Fractional anisotropy (FA) and mean diffusivity (MD) were calculated for the corpus callosum, PLIC and cerebral peduncle, where regions were extracted using registration  to the Johns Hopkins University neonatal Atlas. Relationships with 12 month motor and neurological outcomes were examined.Results: The systematic review and meta-analyses revealed that Early structural MRI had satisfactory sensitivity and specificity to determine adverse motor outcomes and CP but that evidence for diffusion MRI was still emerging. Reproducibility was demonstrated for the structural MRI scoring system with good reliability and agreement for the overall score and all subscales except for cortical GM. Early MRI global, WM, and deep GM scores were negatively associated with Bayley III motor (regression coefficient for global score s=-1.31; 95%CI=-2.39, -0.23; p=0.02; r2=0.07), cognitive (s=-1.52; 95%CI=-2.39, -0.65; p<0.01; r2=0.16) and NSMDA outcomes (s=-1.73; 95%CI=-3.19, -0.28; p=0.02; r2=0.09). Early MRI cerebellar scores were negatively associated with the NSMDA (s = -5.99; 95% CI, -11.82, -0.16; p = 0.04; r2=0.08). Associations were reconfirmed at Term MRI and cerebellar scores were also associated with Bayley III and NSMDA outcomes.Structural MRI scores were associated with concurrent motor, neurological and neurobehavioral function at Early and Term MRI. At Early MRI, cerebellar scores demonstrated the strongest associations with clinical measures, displaying associations with neurological and motor items but not neurobehavioural items. At Term MRI, the strongest associations were with motor performance measured with the TIMP. White matter abnormality scores were related to motor and neurological performance at Term but not at Early MRI.Early MRI FA and MD in the defined regions were not associated with motor or neurological outcomes. Term MRI FA was not associated with outcomes. Term MRI MD in the left corpus callosum was associated with neurological outcome. Term MRI MD in the right cerebral peduncle was associated with motor outcome on the AIMS and NSMDA.Conclusions: Early structural MRI is clinically accessible; however, limited reporting of diagnostic accuracy in Early MRI studies currently restricts clinical utility and translation to clinical practice. The Early structural MRI scoring system is valid for use between 29 and 35 weeks PMA, and is the first to incorporate qualitative evaluation of brain injury and evidence of growth impairment; as well as assessment of deep GM and the cerebellum. Early MRI diffusion measures of FA and MD in the corpus callosum, PLIC and cerebral peduncle were not associated with motor or neurological outcomes. The use of automatic segmentation methods to derive brain regions of interest resulted in exclusion of infants with significant structural brain lesions, possibly limiting the ability to find associations between the diffusion measures and 12 month motor outcomes.