Chiari-II malformation: a biometric analysis

Chiari-II malformation is a complex set of anatomic abnormalities of the posterior fossa, brainstem, and cerebellum seen in myelomeningocele. Previous studies have tried to understand this condition and its clinical consequences by studying a few fixed anatomic landmarks as dependent variables and clinical symptoms as outcome variables. These studies have led to conflicting and limited results. As a first step towards a more comprehensive study of the structural-functional relationships in Chiari-II malformation, we undertook a biometric case-series analysis of a group of individuals with MMC, using both traditional methods to study size, and thin-plate spline methodology to study shape.

The traditional analysis was a retrospective, randomly obtained case-series of 25 individuals with myelomeningocele and Chiari-II malformation. MRIs were digitized for biometric analysis of the component structures and compared to 25 controls. Landmarks on the digitized images were measured using Scion Image. Data were statistically analyzed using SPSS. The thin-plate spline analysis used a subset of these individuals (n = 14). Landmarks on digitized MRIs were identified and marked with TPSDig software. Landmark data was then imported into Splus to generate a series of shape variables. Statistical analysis using Splus was then undertaken.

Traditional analysis revealed relationships between vermian herniation and pontomedullary junction position (p < 0.05) as well as between the height (and overall size) of the posterior fossa and degree of vermian herniation (p < 0.05). The degree of brainstem herniation/elongation did not correlate with the degree of vermian herniation. The shape analysis revealed that there is as much shape variability within the group of Chiari-II patients as there is between this group and unaffected controls. Thin-plate spline analysis shows that the position of the caudal-most aspect of the vermis is the point of greatest brain rearrangement; likewise, the opisthion is the point of greatest bony rearrangement. Partial least-squares analysis shows that in Chiari-II patients, a strong correlation (r = 0.821) exists between the extent to which the brain is warped (specifically, in regard to vermian descent) and the extent to which the bone is warped (specifically, in regard to the displacement of the opisthion) (p < 0.05).

Traditional biometric analysis of size confirms that the degree of vermian herniation and cervicomedullary junction herniation are independent variables in Chiari-II malformation. Posterior fossa size is an important factor in explaining the variability of vermian herniation. The relationship between IPN position and vermian herniation suggests the possibility of a common etiology. The phenomenon of cervicomedullary junction descent appears to be independent, suggesting a different etiology. Analysis of shape indicated that shape changes in bone and brain are related, and that the Chiari-II-associated abnormalities vary greatly in their degree of abnormality. The extreme morphological variability argues against the use of mean differences as a statistical technique in analyzing morphological abnormalities in the structures we investigated.