A Multi-parametric MRI and Machine Learning Study of Cerebellar Structure in Youth with Neurofibromatosis Type 1.

backgroundNeurofibromatosis type 1 (NF1) is a genetic condition caused by pathogenic variants of the NF1 gene. While alterations in cerebral structural and microstructural differences have been reported in NF1, the cerebellum remains largely unexplored. Since individuals with NF1 are at risk for cognitive difficulties, which are in turn associated with cerebellar processes, understanding the underlying neural mechanisms is critical for intervention development.

methodsYouth (5-16 years) with NF1 (n = 30) and unaffected youth (n = 40) participated in neuropsychological (i.e., neurocognitive, parent report of motor abilities) testing and MRI to ascertain structural cerebellar metrics, including volume and white matter mean diffusivity (MD), fractional anisotropy (FA), neurite density (NDI), and orientation dispersion (ODI). We used ANCOVAs to compare between-groups and support vector modeling to investigate which variables (imaging, neurocognitive, motor) contribute the most to NF1 and unaffected participants distinction.

resultsAfter controlling for total brain volume, white matter volume was larger in the NF1 versus unaffected group with a large effect (partial η²=0.57). Cerebellar MD was higher in the NF1 group, while FA, NDI and ODI were lower in the NF1 group (p's < 0.05). Support vector modeling correctly classified 90.20% of participants as being in the NF1 or unaffected group. Top three weights were white matter volume, mobility ratings, and NDI.

conclusionsDifferences in cerebellar white matter microstructure (compared to unaffected youth) were identified in NF1. Cerebellar white matter volume, NDI, and MD were particularly useful differentiators between NF1 and unaffected youth and may be underlying mechanisms of cerebellum-mediated neurocognitive deficits in NF1.