Association between flexibility assessment based on prone full-spine CT scout view and surgical strategy for old thoracolumbar fracture with kyphotic deformity

Abstract

This study aimed to investigate whether spinal flexibility, assessed by the alignment difference between Full-length Spine CT scout view (FLS-CT) and standing radiographs, was associated with the eventual use of three-column osteotomy (3CO) in patients with old thoracolumbar fracture with kyphotic deformity (OTLFK). This single-center, dual-surgeon retrospective study included patients with OTLFK who underwent corrective surgery between August 2021 and August 2024. Patients were categorized into the 3CO group and the non-three-column osteotomy (N3CO) group according to whether 3CO was ultimately performed. Demographic characteristics, surgical variables, radiographic parameters, and clinical outcomes were compared between groups. Clinical outcomes included Visual Analog Scale (VAS) and Oswestry Disability Index (ODI) assessed at 1 month, 6 months, and final follow-up. To quantify positional flexibility relative to the correction ultimately achieved, the kyphosis flexibility percentage (Kflx%) was retrospectively calculated using local kyphosis Cobb angle (LKCA) measurements from preoperative standing radiographs, preoperative prone FLS-CT, and postoperative standing radiographs. Because its denominator included postoperative standing LKCA, Kflx% was treated as a postoperative target-referenced metric rather than a purely preoperative predictor. Its apparent discrimination for classifying the observed use of 3CO was explored using a receiver operating characteristic (ROC) curve. A total of 109 patients were included, comprising 75 in the 3CO group and 34 in the N3CO group. Compared with the N3CO group, the 3CO group had significantly lower preoperative standing lumbar lordosis and significantly higher preoperative prone LKCA and T1 pelvic angle. Kflx% was significantly lower in the 3CO group than in the N3CO group (34.2% vs. 60.6%, P  < 0.001). Among the variables entered into the multivariate logistic regression model, Kflx% was independently associated with the observed use of 3CO (OR = 0.940, 95% CI 0.910–0.970, P  < 0.001). ROC analysis identified an exploratory, cohort-derived Kflx% threshold of 49.7%, with a sensitivity of 82.7%, specificity of 61.8%, and positive predictive value of 82.9% for classifying patients who ultimately underwent 3CO. VAS and ODI improved significantly after surgery in both groups, without significant between-group differences at the same follow-up time points. In this institutional cohort, lower Kflx% was associated with 3CO use. Kflx% was calculated retrospectively and normalized to the correction achieved after surgery. The 49.7% cutoff is exploratory and should not be regarded as a validated preoperative decision threshold. Prospective evaluation using alignment targets defined before surgery is required.