Infants exposed to adverse prenatal conditions of intrauterine growth restriction (IUGR), are at high risk for neurological morbidities such as cerebral palsy, mental retardation, a wide spectrum of learning disabilities and developmental behavioural disorders with an association of neuropsychiatric disorders later in life [1]. Experimental studies of IUGR in the rat, induced by protein or caloric restriction, have shown extensive effects on brain development including white and grey matter structural changes [1]. Diffusion tensor imaging (DTI) has been widely used to study brain development. Nevertheless the lack of specificity in DTI derived parameters (i.e. diffusivity values (Mean: MD, axial: D_ax and radial: D_rad) and fractional anisotropy (FA)) makes difficult to properly link changes with physiological process. Recently, new models aimed to be more specific such as the neurite orientation dispersion and density imaging (NODDI [2]) model have been developed giving estimation of the intra-neurite volume fraction (ficvf), the cerebrospinal volume fraction (fiso) and an index of fiber dispersion (orientation dispersion index: ODI). Such a model is applicable to white and grey matter, the aim of this work was to establish offspring cerebral structural changes following low protein diet in rats during gestation by using DTI and NODDI at 9.4T as well as immunohistochemistry.Sprague Dawley rats received low protein diet (LPD, 8% casein) or normal diet (CT, 12% casein) during gestation. At P4, P10 or P21 (n=8 rats per group and per age) rats were sacrificed and brains were formalin-fixed for subsequent genomic analysis of oligodendrocyte differentiation at P4 as well as histology and ex-vivo MRI at P10 and P21. MR experiments were performed on an actively-shielded 9.4T/31cm magnet (Agilent) equipped with 12-cm gradient coils (400mT/m, 120ms) with a 2.5 mm diameter birdcage coil. A multi-b-value shell protocol was acquired using a spin-echo sequence with the following parameters: FOV(mm²)/matrix size/number of slices/thickness (mm) = 20×20/96×64/12/0.6 at P10 and 27×27/128×64/14/0.6 at P21, axial slices, 3 averages with TE/TR = 45/2000 ms. A total of 96 DWI were acquired, 15 of them as b₀ reference images. The remaining 81 were separated in 3 shells with the following distribution (# of directions/b-value in s/mm²): 21/1750, 30/3400 and 30/5100. All 81 directions were non-collinear and were uniformly distributed in each shell. Acquired data were fitted using the NODDI toolbox [2]. The diffusion tensor (DT) was spatially normalized to the study-specific DT template using DTI-TK [3]. The regions of interest (ROI) were drawn on the DT study-specific template and were then transformed back to the subject space in order to compute ROI-averaged estimates of DTI and NODDI maps. Four different brain regions were identified on the DT-template: cortex (Cx), corpus callosum (CC), external capsule (EC) and cingulum (Cg). For statistics (LPD vs. CT), a Mann Whitney test was used (significance: *P<0.05).

At P10, white matter fibers (CC and EC) exhibited significantly increased D_ax and ODI whereas FA was decreased in LPD rats compared to CT. Indeed, fiso was reduced (significantly in CC and tendency in EC) and ficvf increased (significantly in EC and tendency in CC) in LPD. In Cx and Cg no significant change was observed between the groups.

At P21, no significant change was observed between the groups in white matter except increased in LPD D_rad in Cg and decrease of ficvf in LPD EC. However, significant grey matter changes were observed with increased D_ax, FA and fiso as well as decrease of ODI in LPD cortex compared to CT. The density of myelinated fibers and the number of mature oligodendrocytes were significantly reduced in the LPD group in the white matter at P10 and in the cortex at P21.

In this study we characterized LPD microstructural consequences on pup rats at P10 and P21. At P10, structural changes were mainly found in the white matter with axonal degeneration and subsequent myelination/fiber compaction defect as depicted by increased D// and ODI as well as FA decreased. In the other hand at P21, white matter tracts were almost normalized whereas cortical structure showed abnormal diffusion parameters than can potentially result from altered development of dendritic arborization (leading to increased FA and D// as well as decreased ODI). To conclude DTI and NODDI give accurate probing of the cerebral impairment following IUGR with good immunohistological correlation, providing evidence of these new diffusion models power translational animal models. The possible translation of these technics from pre-clinical to clinical scanners makes this study of high interest for the neonatologist community.