ObjectivesTo analyse whether time-varying treatment with tumour necrosis factor inhibitors (TNFi) in radiographic axial spondyloarthritis (r-axSpA) has a differential impact on structural damage progression on different spinal segments (cervical versus lumbar spine).MethodsPatients with r-axSpA in the Swiss Clinical Quality Management cohort were included if cervical and lumbar radiographs were available at intervals of 2 years for a maximum of 10 years. Paired radiographs were scored by two calibrated readers according to the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS). The relationship between TNFi use and progression in the cervical and the lumbar spine was analysed using generalised estimating equation models and adjustment for potential confounding. Radiographic progression per spinal segment was defined as an increase of ≥ 1 mSASSS unit or by the formation of ≥ 1 new syndesmophyte over 2 years.ResultsMean ± SD symptom duration was 13.8 ± 9.8 years. Mean ± SD mSASSS progression per radiographic interval was 0.41 ± 1.69 units in the cervical spine and 0.45 ± 1.45 units in the lumbar spine (p = 0.66). Prior use of TNFi significantly reduced the odds of progression in the cervical spine by 68% (OR 0.32, 95% CI 0.14–0.72), but not in the lumbar spine (OR 0.99, 95% CI 0.52–1.88). A more restricted inhibition of progression in the lumbar spine was confirmed after multiple imputation of missing covariate data (OR 0.43, 95% CI 0.24–0.77 and 0.85, 95% CI 0.51–1.41, for the cervical and lumbar spine, respectively). It was also confirmed with progression defined as formation of ≥ 1 syndesmophyte (OR 0.31, 95% CI 0.12–0.80 versus OR 0.56, 95% CI 0.26–1.24 for the cervical and lumbar spine, respectively).ConclusionDisease modification by treatment with TNFi seems to more profoundly affect the cervical spine in this r-axSpA population with longstanding disease. Site-specific analysis of spinal progression might, therefore, improve detection of disease modification in clinical trials in axSpA.