Research Article: Determinants of Indices of Cerebral Volume in Former Very Premature Infants at Term Equivalent Age

Date Published: January 26, 2017

Publisher: Public Library of Science

Author(s): Aurelie Naud, Emmanuelle Schmitt, Maelle Wirth, Jean-Michel Hascoet, Umberto Simeoni.


Conventional magnetic resonance imaging (MRI) at term equivalent age (TEA) is suggested to be a reliable tool to predict the outcome of very premature infants. The objective of this study was to determine simple reproducible MRI indices, in premature infants and to analyze their neonatal determinants at TEA. A cohort of infants born before 32 weeks gestational age (GA) underwent a MRI at TEA in our center. Two axial images (T2 weighted), were chosen to realize nine measures. We defined 4 linear indices (MAfhlv: thickness of lateral ventricle; CSI: cortex-skull index; VCI: ventricular-cortex index; BOI: bi occipital index) and 1 surface index (VS.A: volume slice area). Perinatal data were recorded. Sixty-nine infants had a GA (median (interquartile range)) of 30.0 weeks GA (27.0; 30.0) and a birth weight of 1240 grams (986; 1477). MRI was done at 41.0 (40.0; 42.0) weeks post menstrual age (PMA). The inter-investigator reproducibility was good. Twenty one MRI (30.5%) were quoted abnormal. We observed an association with retinopathy of prematurity (OR [95CI] = 4.205 [1.231–14.368]; p = 0.017), surgery for patent ductus arteriosus (OR = 4.688 [1.01–21.89]; p = 0.036), early onset infection (OR = 4.688 [1.004–21.889]; p = 0.036) and neonatal treatment by cefotaxime (OR = 3.222 [1.093–9.497]; p = 0.03). There was a difference for VCI between normal and abnormal MRI (0.412 (0.388; 0.429) vs. 0.432 (0.418; 0.449); p = 0,019); BOI was higher when fossa posterior lesions were observed; VS.A seems to be the best surrogate for cerebral volume, 80% of VS.As’ variance being explained by a multiple linear regression model including 7 variables (head circumference at birth and at TEA, PMA, dopamine, ibuprofen treatment, blood and platelets transfusions). These indices, easily and rapidly achievable, seem to be useful but need to be validated in a large population to allow generalization for diagnosis and follow-up of former premature infants.

Partial Text

Despite decreasing incidence, cerebral lesions remain an important issue for very premature infants’ neurodevelopment outcome [1, 2]. Early recognition of infants with cerebral lesions, at risk of delayed or abnormal neurodevelopment is important both for accurate counseling of parents and for selection of infants that may benefit from specific follow-up, allowing early behavioral interventions or rehabilitation services [3]. Conventional magnetic resonance imaging (MRI) at term equivalent age (TEA) has been suggested to be able to reliably predict the outcome of very premature infants [4]. An abnormal MRI in and of itself will lead to more vigilance in follow up, but a normal MRI is not always associated with a normal developmental outcome. Therefore, it is important to improve MRI specificity of former preterm infants with apparent “normal” MRI. Peterson et al. found that premature birth was associated with long term specific anatomical reductions in brain volume [5]. Other studies have shown a correlation between cerebral growth and alteration of neurodevelopment outcome in former premature infants [4–7]. Inder et al. demonstrated that premature infants with moderate to severe disability at one year of age have significant reduction in gray matter volume, both cortical and nuclei, and an increase in cerebrospinal fluid volume [6]. This correlation is stronger when white matter injury is associated [7].

In this study, we defined five cerebral indices from simple and reproducible linear measures. The good inter-investigator reproducibility of these indices is consistent with measures easily and rapidly achievable. These indices might be very useful when evaluating specific situations. For instance, Ragan et al. showed that the ventricular volume in infants presenting with hydrocephaly had a strong correlation with a linear index easier to measure than cerebral volume itself: FOHR, which adds the two ventricular diameter of frontal and occipital horns of the lateral ventricles (equivalent to our index, DBVP), divided by twice the transversal diameter of cortex (equivalent to our index, DTc) [19]. Pineda et al studied brain structure differences in neonatal intensive care between open ward and private room in preterm infants with qualitative brain measures, brain metrics (bifrontal, biparietal, transcerebellar diameters, ventricular size and interhemispheric distance), diffusion measures, and volumetry [23]. These methods require trained personnel, specific equipment and cannot be applied in clinical routine. Kidokoro et al. [24] and Inder et al. [25] used a standardized scoring system to evaluate cerebral white matter and cortical gray matter abnormalities. Our study is a quantitative global MRI analysis. We objectively measured parenchyma and skull length and created indices to reflect changes in brain volume independent of changes in brain shape. We studied five indices: four linear indices representing different brain regions and a surface index. All these data are easily obtained when performing standard MRI at TEA.

The screening of children at risk of presenting with neurodevelopmental disorders is important for accurately answering parents’ concern and for selection of infants who may benefit from early intervention. Our study defined indices, easily and rapidly achievable in routine MRI, which may improve MRI specificity. The surface index seems to be a good surrogate for cerebral volume. This study need to be validated in a large population to allow generalization for the follow-up of former premature infants.