Background and aims We aimed to investigate the effect of clinical risk factors on white matter structural integrity at term equivalent age (TEA), as measured by fractional anisotropy (FA).

Methods Diffusion tensor imaging of sufficient quality was available for 182 infants (gestational age (GA) <28 weeks), scanned at TEA in two centres. FA values of 14 regions (posterior limb of the internal capsule (PLIC), cerebral peduncles, corpus callosum, sagittal stratum, superior/posterior corona radiata, posterior thalamic radiation, left and right side for all) were automatically calculated (ExploreDTI, Leemans ISMRN 2009) using an atlas-based approach (Oishi, NeuroImage 2011). Subjects with overt parenchymal injury were excluded. Clinical characteristics tested against FA in the multivariable linear regression analysis for each region were GA, gender, intra-uterine growth retardation (IUGR), hypotension, mechanical ventilation >7 days, morphine (yes/no), post-haemorrhagic ventricular dilatation (PHVD), surgery, postmenstrual age at scanning (PMA) and participating centre. A cut-off value of p < 0.004 (0.05/14) was used to correct for multiple comparison.

Results Statistically significant positive associations with FA were found for PMA in PLIC (right), Superior Corona Radiata and Posterior Corona Radiata (left) and GA in Corpus Callosum, Sagittal Stratum (left), Posterior Thalamic Radiation (left). Statistically significant negative associations with FA were found for Surgery in Sagittal Stratum (right) and IUGR in Posterior Thalamic Radiation (left). Centre was significantly associated with FA in 8/14 brain regions.

Conclusion The microstructure of the preterm brain at TEA depends on GA and PMA at scan, IUGR and surgery. A possible interaction between surgery and morphine warrants further investigation.