Background/Purpose: Osteoporosis and vertebral fractures are recognized complications in axial spondyloarthritis (axSpA). Anteroposterior (AP) DEXA is commonly used to assess spinal bone mineral density (BMD) but can become inaccurate in the presence of syndesmophyte formation. The 2015 European League Against Rheumatism (EULAR) imaging guidelines highlight the importance of assessing for osteoporosis in axSpA and suggest use of lateral DEXA to assess spinal BMD [1]. However, it is unclear when lateral DEXA should be performed. The aims of this study were 1) to explore AP-BMD changes with axSpA symptom duration, and 2) to identify when optimum assessment of spinal BMD should include lateral spinal DEXA.

Methods: A cross-sectional study was conducted with axSpA patients fulfilling the ASAS criteria and not using bisphosphonates. Each patient underwent AP-DEXA of the lumbar spine (L1-L4) and total hip. Simultaneous lateral lumbar DEXAs were performed in a random subgroup of patients. AP-BMD was plotted against symptom duration using lowess smoothing. Piecewise linear regression was used to estimate a transition point after which AP-BMD began to increase. The difference between AP and lateral spinal BMDs was compared against symptom duration using scatter plots.

Results: AP-DEXAs were performed in 259 patients, 32 also underwent lateral DEXAs. 75% were male with mean age of 38.8 (SD±12.7) years. Median symptom duration was 16.6 years [interquartile range (IQR) 8, 28.4]. Mean BMI was 28.4 (SD±5.7). TNFi was used by 30% and 7% were taking calcium and vitamin D supplements. The median BASDAI was 6.4 [IQR 4.6, 7.7] and BASFI 6.4 [IQR 3.7, 8.2]. Osteopenia and osteoporosis of the spine were present in 27% and 5% and for the hip 29% and 3%, respectively, on AP-DEXA. In the first decade after symptom onset, patients with longer symptom duration had lower AP-BMD (Fig 1). However after 20 years, AP-BMD was higher with increasing symptom duration. Piecewise regression for spine g/cm² and T-score estimated the transition point to be 13 years (95%CI 2.7, 23.0). The difference between AP and lateral BMD increased with increasing symptom duration (Fig 2).

Conclusion: After 13 years, AP-BMD was higher with increasing symptom duration which likely reflects accrual of pathological new bone in this bisphosphonate-naïve cohort. This was supported by the increasing discrepancy between AP and lateral spinal BMD with increasing symptom duration. AP-DEXA can be used to assess BMD during the first decade of symptom duration, but lateral DEXA should be considered after 13 years, particularly for those with known syndesmophytes.