Materials and methods

Patient cohort

For this retrospective analysis, the database of a tertiary neurosurgical center was queried for patients with the ICD10 diagnosis G91.2 (idiopathic normal pressure hydrocephalus), planned for shunt surgery between 2013 and 2023. Patient data were found eligible for inclusion in the analysis if MRI of the cervical and/or lumbar spine prior to shunt surgery was available. Patients were clinically assessed using the NPH-grading scale score [Kubo et al., 2008], which scores cognitive, urinary and gait impairment, with higher scores indicating more severe clinical impairment and a maximum score of 12 points. The NPH-grading scale score was determined preoperatively and at follow-up (on average 976 days postoperatively).

Evaluation of spinal imaging data

Cervical MRI was evaluated by applying the Muhle Grading Scale (Muhle et al., 1998), which comprises grades 0 to 3. Grade 0 corresponds to a normal width of the spinal canal, grade 1 to partial obliteration of the subarachnoid space, grade 2 to complete obliteration of the subarachnoid space, and grade 3 to spinal cord impingement (Fig. 1).

Fig. 1: Muhle grading scale for cervical stenosis: Grade 0: no stenosis, 1: partial obstruction of the subarachnoid space, 2: complete obstruction of the subarachnoid space, 3: myelon impingement. (Modified) Schizas grading Scale for lumbar stenosis: Grade A: minor stenosis with CSF being clearly visible (AP diameter <13 mm), grade B: rootlets still discernible but CSF hardly visible, grade C: compressed rootlets without visible CSF, but epidural fat still visible, grade D: absolute stenosis with epidural fat being not/hardly visible.

Lumbar MRI was evaluated by using a modified version of the Schizas grading scale (Schizas et al., 2010), differentiating grades A to D, with grade A corresponding to minor stenosis (with AP diameter <13 mm according to Papanagiotou and Boutchakova, 2014]), but CSF being still visible within the dural sac. Grade B corresponds to moderate stenosis (with rootlets occupying the whole of the dural sac, but with single rootlets still being discernible), grade C to severe stenosis (with no singular rootlets or CSF within the dural sac being discernible, but with visible epidural fat posteriorly), and grade D corresponding to maximal stenosis (with no rootlets being discernible, as well as neither CSF nor epidural fat being visible) (Fig. 1).

The maximally stenotic segment in each patient was then quantitatively evaluated by determining the dural sac cross sectional area (DSCSA) in the axial plane (in mm2), and the anterior-posterior (AP) diameter in the midsagittal plane (in mm). In addition, the number of stenotic segments per patient, as well as frequencies per segment levels affected were analyzed.

Evaluation of cranial imaging data

Cranial computed tomography (CT) acquired prior to shunt-surgery was evaluated using the Radscale Score (Kockum et al., 2018), which is a composite score of 7 NPH-typical cranial imaging features comprising the Evans-Index, callosal angle, size of temporal horns, narrowness of high-convexity sulci, dilated Sylvian fissures, focally dilated sulci, and periventricular hypodensities. The Radscale-Score comprises maximally 12 points, with higher scores corresponding to more NPH-typical cranial imaging features and was previously found to correlate with NPH related symptom-burden (Kockum et al., 2018).

In addition, absolute values of both, the callosal angle (measured in the coronal plane crossing the posterior commissure orthogonal to the axial plane connecting the anterior and posterior commissure), as well as of the Evans-Index (defined as the maximum width of the frontal horns divided by the maximum intracranial biparietal diameter, with values > 0.30 indicating pathological enlargement of the ventricles) were analyzed. CT data was only included, if preoperative imaging was digitally available and if multiplanar reformations were sufficient in image resolution (slice thickness ≤ 2 mm), see Fig. 2).

Fig. 2: a) Evans-Index, defined as the maximal width of the anterior horns divided by the maximal intracranial biparietal width. Values > 0.30 indicate pathological enlargement of the ventricles. b) Callosal angle (CA) measured between the lateral ventricles in the coronal plane through the posterior commissure and orthogonal to the axial plane connecting the anterior to the posterior commissure (ACPC plane). CA is typically <90° in NPH patients.

Statistical analysis

Section

Descriptive statistics of patient demographics, number of stenoses, frequency of segments affected, and mean values of AP and DSCSA were determined.

All statistical analyses were performed with R [R Core Team (2022). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.], RStudio [Posit RStudio. Posit team (2024). RStudio: Integrated Development Environment for R. Posit Software, PBC, Boston, MA. URL http://www.posit.co/.], and Jamovi [The jamovi project (2022). jamovi. (Version 2.3) [Computer Software]. Retrieved from https://www.jamovi.org].

Section

Hypothesizing, that patients would differ in hydrocephalus related cranial imaging parameters depending on the degree of spinal stenosis, patients with cervical stenosis (CS) and lumbar stenosis (LS) were each subgrouped based on their DSCSA by median split, resulting in subgroups with small/large DSCSA (CS: NsmallDSCSA = 45, NlargeDSCSA = 44, DSCSA cut-off = 85.7 mm2; LS: NsmallDSCSA = 30, NlargeDSCSA = 31, DSCSA cut-off = 75.0 mm2). For each subgroup analysis (small and large DSCSA subgroups, separately for LS and CS), the Evans Index, Radscale-Score and callosal angle were compared by applying Mann-Whitney U Tests, tested one-sided with a significance level of p = .05. Results were adjusted using the Bonferroni-Holm correction for multiple comparisons (k = 3 per family, _p_Holm).

Section

In addition, group differences were analyzed accordingly for cervical (CS) and lumbar stenosis (LS) subgroups based on small/large AP diameters (CS: NsmallAP = 45, NlargeAP = 45, AP cut-off = 6.5 mm; LS: NsmallAP = 30, NlargeAP = 31, AP cut-off = 6.7 mm), comparing Evans Index, Radscale Score and callosal angle applying Mann-Whitney U Tests, tested one-sided with a significance level of p = .05. Results were adjusted using the Bonferroni-Holm correction for multiple comparisons (k = 3 per family, _p_Holm).

Section

Partial correlations were computed in those patients in which both, cervical and lumbar MRI were available to assess associations between Evans Index, Radscale-Score, and Callosal Angle on the one hand, and cervical and lumbar AP-diameter as well as DSCSA on the other hand, while controlling for age. All resulting p-values were Bonferroni-Holm-corrected for multiple comparisons (_p_Holm).