To propose a MRI cerebellar algorithm that may be applied to guide genetic/malformative or biochemical investigations for patients with cerebellar ataxia.

Cerebral MRI of 158 patients with cerebellar ataxia and no supratentorial abnormality were examined according to a new categorization system based on posterior fossa imaging. The clinical and radiological findings were confronted to biochemical and/or genetic results using the MR cerebellar algorithm. Seven groups of cerebellar MRI pattern were described: vermian dysgenesis (n=27), cerebellar hypoplasia (n=15), hemispheric cerebellar dysgenesis (n=6), unilateral hemispheric atrophy (n=5), global cerebellar atrophy (n=84), signal abnormalities (n=11) and normal MRI (n=10). Cerebellar hypoplasia, vermian dysgenesis and hemispheric cerebellar dysgenesis groups were classified as malformative disorders. Global atrophy and signal abnormality groups were classified as metabolic disorders.

In the vermian dysgenesis group, a specific genetic diagnosis was obtained in eight children (8/27) and all of the mutated genes (AHI1 (JBS3), CEP290 (JBS5), TMEM67 (JBS6), and RPGRIP1L (JBS7)) are involved in primary cilia function. In the group of pontocerebellar hypoplasia specific genetic diagnosis was obtained in one patient (PCH2) (1/15). Thus, nine of 42 children classified as malformative disorder had a molecular diagnosis. Global atrophy and signal abnormality groups were classified as metabolic disorders, specific biochemical was obtained in 46/95 children. In global atrophy group, respiratory chain deficiency was diagnosed in 18 children (18/84). In 21 children a congenital disorders of glycosylation type 1a (CDG Ia) was diagnosed (21/84) and infantile neuroaxonale dystrophy (INAD) was diagnosed in one child. In signal abnormalities group, specific biochemical diagnosis was obtained in six out of 11 children, five children with respiratory chain deficiency and one child with sulphite oxidase deficiency. In hemispheric cerebellar dysgenesis and normal MRI groups, no biological diagnosis was found for any of the patients. In the group of unilateral hemispheric atrophy, we hypothesized a clastic prenatal injury.

The proposed MR cerebellar algorithm was useful to guide genetic/malformative or biochemical investigations, allowing an etiological diagnosis in 55 children.