Comprehensive analysis of spinal sagittal balance, spinal bone mineral density, and pulmonary function in severe ankylosing spondylitis using quantitative computed tomography

Ankylosing spondylitis (AS) is frequently complicated by osteoporosis and progressive sagittal malalignment and may be accompanied by respiratory limitation. However, in radiographically severe AS, the cross-sectional relationships among sagittal alignment, spinal volumetric bone mineral density (vBMD), and imaging-derived pulmonary metrics remain incompletely characterised. We investigated these associations using quantitative computed tomography (QCT) and CT-based lung volumetry. This retrospective, cross-sectional study included 135 patients with radiographically severe AS (modified New York sacroiliitis grade 3–4) evaluated at a single center between 2021 and 2023. Lumbar trabecular vBMD was measured via QCT, sagittal alignment parameters were assessed on standing radiographs, and pulmonary metrics (TLVin, TLVex, ΔTLV) were derived from paired inspiratory/expiratory chest CT scans. Correlations were adjusted using the Benjamini–Hochberg false discovery rate (FDR). Multivariable linear regression examined associations with sagittal vertical axis (SVA) and pelvic tilt (PT), while multivariable logistic regression was employed to identify independent predictors of osteoporosis (defined as vBMD < 80 mg/cm3). Spinal vBMD and SVA differed significantly between radiographic grades (grade 3 vs 4). In FDR-adjusted correlations, vBMD was inversely associated with age (ρ = −0.58, q < 0.001) and BMI (ρ = −0.28, q = 0.007). Height showed robust associations with alignment (PT: ρ = −0.37, q < 0.001; TK: ρ = −0.39, q < 0.001; SS: ρ = 0.34, q < 0.001) and with TLVin (ρ = 0.33, q < 0.001), while BMI was inversely associated with TLVex (ρ = −0.27, q = 0.011). In prespecified multivariable models, the SVA model explained modest variance (R2 = 0.119) with no independent covariate at p < 0.05; PT was independently associated with height (B = −0.33°/cm, p = 0.001; R2 = 0.248). In the prespecified osteoporosis model, age (OR per 10 years 5.37), grade 4 (OR 4.35), and smoking history (OR 3.23) were independently associated with osteoporosis; discrimination within the derivation cohort was good (AUC 0.874). CT-volumetry showed high reproducibility (intra-/inter-observer ICCs 0.95–0.99). In radiographically severe AS, spinal vBMD is strongly influenced by age and radiographic severity. The identification of a high-risk osteoporosis profile, which is characterized by advanced age, grade 4 sacroiliitis, and smoking history, supports targeted bone health surveillance in these subgroups. While CT-derived lung volumetry is technically reproducible, its cross-sectional associations are predominantly anthropometric. Future prospective studies incorporating longitudinal data and standard spirometry are warranted to clarify the clinical utility of these imaging biomarkers.