Prospective, Longitudinal Study of Clinical Outcome and Morphometric Posterior Fossa Changes after Craniocervical Decompression for Symptomatic Chiari I Malformation

Study Population and Ethics

Thirty-eight adult patients (35 women, 3 men) diagnosed with CMI only (n = 15) or CMI and syringomyelia (n = 23) and without previous CMI surgery were enrolled in a clinical protocol and underwent PFDD surgery. The NIH Combined Neuroscience Institutional Review Board approved the research protocol: 10N0143. Consent was obtained from all patients. All patients were evaluated and treated at a single clinical site. The mean age at surgery was 40.6 (SD, 12.6) years (range, 20.3–64.9 years). The protocol specified outpatient study visits and MR imaging 7 times (ie, initial [before the operation], 3 months, 1 year, 2 years, 3 years, 4 years, and 5 years after the operation). MR images and clinical data were collected between 2010 and 2021. MR images were stored in the PACS server. Clinical data were recorded on case report forms, transcribed, and stored in a secure database. All 38 patients attended at least 1 scheduled visit after the operation, and 14 completed the 5-year protocol (Table 1).

Surgical Procedure

All patients underwent the operation in a prone, horizontal position with the head in a neutral or gently flexed posture. The surgical procedure included a suboccipital craniectomy, C1 laminectomy, and sometimes removal of the superior part of the lamina of C2. The bony opening decompressed the inferior cerebellar hemispheres and cerebellar tonsils. A Y-shaped durotomy was performed. The arachnoid was preserved. Intradural structures were not manipulated. An autologous dural graft of the occipital pericranium measuring 4 × 4  cm was sewn to the durotomy edges using a 4–0 braided nylon suture.¹⁸

Image Registration and Midsagittal Plane Selection

We used Matlab (MathWorks), FSL (http://www.fmrib.ox.ac.uk/fsl), and bash scripts in parallel to register 3D T1 MR images to the Montreal Neurological Institute images (MNI152_T1_brain.nii.gz, from FSL) and identify the midsagittal plane. DICOM (.dcm) images were initially converted to NIfTI (.nii.gz) format. The “dcm2nii” Unix command compressed file sizes. FSL commands extracted the brain from MR images. A linear image registration tool (FLIRT command; FMRIB Linear Image Registration Tool; FLIRT; http://www.fmrib.ox.ac.uk/fsl/fslwiki/FLIRT) registered the skull-free image to the atlas image (MNI152_T1_brain.nii.gz). The FSL transformation matrix, including the superior-inferior ROI, was applied to the original images, registering them in the atlas image orientation. The final midsagittal MR image in the midsagittal plane of the atlas was saved as a Matlab file for morphometric measurements.

Morphometric Measurements and Software

We made 9 morphometric measurements using 2 custom software programs developed in Matlab (Table 2, Fig 1, and Supplemental Online Video). In previous studies, these measurements significantly changed after PFDD surgery.^16,22 PFDD removes the opisthion. A described method reconstructed the McRae line on postsurgical images.^16,18

• Download figure
• Open in new tab
• Download powerpoint

FIG 1.

Morphometric measurements on a presurgery MR image (A) and a postsurgery MR image (B) for the same patient 1 year later. 1) CSF area posterior to PCF, 2) anterior CSF area, 3) posterior CSF area, 4) cerebellar height, 5a) posterior dorsal width + (5b) anterior dorsal width = total dorsal CSF width, 6) ventral CSF width, 7) cerebellar tonsillar position, 8) occipital bone length, and 9) cerebellum-clivus bone angle. Note that the syrinx in the presurgical MR image resolved after surgery (Supplemental Online Video).

Clinical Outcomes Measures

The primary clinical outcome measure was the KPS score change between before and after the operation. The KPS was graded on a 0–100 scale at study visits, representing the patient’s ability to perform everyday activities, work, and function without assistance.²³ Exploratory clinical outcomes included the following: 1) the Short-form McGill Pain Questionnaire, a self-reported pain questionnaire used in longitudinal studies;²⁴ 2) the McCormick Clinical/Functional classification, a 4-grade scale ranging from grade I (normal or insignificant functional neurologic deficit) to grade IV (severe deficit, quadriparesis, usually not independent);²⁵ 3) the Functional Ambulatory Score, a 4-grade scale ranging from 1 (ambulant without a walking aid) to 4 (complete motor and sensory paraplegia); 4) the American Spinal Injury Association motor grading scale,²⁶ a 0–100 scale based on a sum of the strengths of 5 upper- and 5 lower-extremity muscles graded from 5 (normal) to 0 (total paralysis); and 5) the Functional Independence Measure, a 13-item questionnaire assessing motor and cognitive functioning on a 1 (total assistance) to 7 (complete independence) ordinal scale.²⁷ The Chicago Chiari Outcome Scale complication component graded complications as 1 (persistent complication, poorly controlled), 2 (persistent complication, well-controlled), 3 (transient complication), and 4 (uncomplicated course).²⁸

Morphologic Outcomes Measures

The primary morphometric outcome measure was the dorsal CSF space, denoted by 5 in Fig 1. The dorsal CSF space (5) summated the anterior-posterior widths of the foramen magnum CSF space (5a) and the retrotonsillar space (5b). Eight other morphometric measures shown in Table 2 were exploratory.

Data Distribution

TPs represented the order of scheduled study visits, including MR imaging scans and clinical data collection. The TP before surgery was –0.5, with the negative sign connoting before surgery. The TP values without negative sign prefixes connoted visits after surgery.

Statistical Analysis

The initial statistical plan of analysis was a repeated measures analysis of variance (RM-ANOVA) or analysis of covariance (RM-ANCOVA), assuming normality. The Box-Cox transformation was applied to the 12 clinical variables with skewed distributions. However, the transformation did not work for 6 of them, and the Friedman test was performed. The Shapiro-Wilk method tested the normality assumption.

For each morphometric and transformed clinical variable, either RM-ANOVA or RM-ANCOVA was used to evaluate the change with time. The TP (years from the operation) variable was treated as categorical because the change in all variables across time did not show linear trends. Age and diagnosis (with and without syringomyelia) were considered covariates, and the significance level of .1 was applied to covariate selection (including 2 interactions: TP × age and TP × diagnosis). For the compound symmetry, the heterogeneous covariance structure was selected based on the Akaike information criterion to account for the correlation among the repeated measures from the same subject and the heterogeneity of time variance. The Dunnett method was applied to multiple comparisons using TP −0.5 or TP 0.5 as the control. The Bonferroni correction was used for multiple comparisons. For the Box-Cox transformed variables, the back-transformed least square (LS) means and 95% CI were reported.

Spearman correlation coefficients were calculated to assess the association between the clinical and morphometric variables based on the following: 1) the relative change: between pre- (TP = −0.5) and postsurgery (TP = 0.5): 100 × (post to pre)/pre, and 2) the measurements at TP = −0.5 and TP = 0.5. The P values were not adjusted for the multiple statistical tests of the exploratory variables. SAS Version 9.4 (SAS Institute) was used for statistical analysis, and P values <.05 were considered statistically significant.