Types of high-riding vertebral artery: a classification system for preoperative planning of C2 instrumentation based on 908 potential screw insertion sites.

BACKGROUND CONTEXT Our recent studies indicated that a high-riding vertebral artery (HRVA) is a common variant posing a risk of injuring the vessel during C2 instrumentation. However, several different types fit in the current definition of HRVA, which may require a different strategy for C2 screw placement. PURPOSE To classify HRVA and provide a clinical aid for preoperative planning of C2 instrumentation. A secondary goal was to estimate coexistence of each HRVA type with the ipsilateral narrow C2 pedicle (NP). STUDY DESIGN A retrospective observational study involving radiologic measurements of the estimated number of anonymized cervical computed tomography (CT) scans. STROBE checklist was applied. PATIENT SAMPLE 908 potential screw insertion sites (PSIS) of 454 consecutive cervical CT scans were analyzed. The sample size was estimated using ScalaR SP function in RStudio. OUTCOME MEASURES Three types of HRVA based on a series of C2 vertebral morphological parameters including the C2 isthmus height (C2IsH) and C2 internal height (C2InH). Also, the prevalences of each HRVA type and coexisting NP based on the C2 pedicle width (C2PW). METHODS HRVA was defined as C2IsH of ≤ 5 mm and/or C2InH of ≤ 2 mm measured 3 mm lateral to the lateral border of the spinal canal. A narrow pedicle was defined as C2PW of ≤ 4 mm. Measurements were done using Syn.govia software. Interobserver, intraobserver, and inter-software agreement coefficients for C2IsH, C2InH, and C2PW parameters were adopted from our previous study. K-means cluster analysis was applied. RESULTS Prevalence of at least one HRVA was 24.9% (n = 113 subjects) and 16.2% of PSIS (n = 147 sites). Based on the measurements and K-means clustering, the following three types of HRVA have been distinguished: type 1 - isthmic with only C2IsH being reduced and normal C2InH; type 2 - internal with only C2InH being reduced and C2IsH within normal limits; type 3 - isthmo-internal with both C2IsH and C2InH being reduced. Kruskal-Wallis test followed by unadjusted and Bonferroni-adjusted post-hoc multiple comparison analysis detected significant differences across the types. The prevalences of the newly identified types were as follows: 78.2%, 8.8%, and 12.9% for type 1, type 2, and type 3, respectively. 73.9% of type 1 HRVA, 53.8% of type 2 HRVA, and 100% of type 3 HRVA had a concomitant ipsilateral NP. Prediction of the HRVA types by the K-means clustering has been evaluated. Screw placement techniques for each type are proposed and discussed. CONCLUSION We present the first classification system for the high-riding vertebral artery distinguishing three types based on the large homogenous cohort, which may serve as an adjunct to preoperative planning of C2 instrumentation. External validation of this classification scheme shall determine its further clinical utility.