Objective. Focal fat infiltration is frequently visible on magnetic resonance imaging (MRI) of the spine in patients with ankylosing spondylitis (AS) and likely reflects postinflammatory tissue metaplasia. To support the concept of coupling between inflammation and new bone formation, we tested the hypothesis that focal fat infiltration at a vertebral corner is more likely to evolve into a de novo syndesmophyte.Methods. MRI scans were obtained at baseline and radiographs were obtained at baseline and 2 years in 100 AS patients from 2 cohorts: a clinical trial cohort (n ؍ 38) and an observational cohort (n ؍ 62). In the clinical trial cohort, patients were randomized to receive anti-tumor necrosis factor (anti-TNF) therapy or placebo for 12-24 weeks and then open-label treatment for 2 years. In the observational cohort, patients received either standard therapy (n ؍ 36) or anti-TNF therapy (n ؍ 26) for 2 years. Vertebral corner inflammation and fat infiltration were assessed independently by pairs of readers who were blinded with regard to the radiographic findings.Results. New syndesmophytes developed significantly more frequently in vertebral corners with fat in both the clinical trial (10.2%) and the observational (6.5%) cohort as compared to those without either Conclusion. Our data lend support to the hypothesis that inflammatory lesions evolve into new bone through a process of tissue metaplasia that includes fat infiltration.A hallmark of ankylosing spondylitis (AS) is the development of new bone in the spine, which typically leads to ankylosis across the disc spaces. Bone proliferation typically starts at the vertebral rim and grows in the direction of the anulus fibrosus, parallel to the vertebral axis. Ossification may extend across the vertical length of the anulus and becomes visible on radiographs as a syndesmophyte.Until recently, it was hypothesized that ankylosis occurred in response to inflammation and following a reparative process that included cartilage metaplasia. However, data from animal models of spondylarthritis (SpA), principally ankylosing enthesitis, directly challenged this hypothesis by demonstrating that anti-tumor necrosis factor ␣ (anti-TNF␣) therapy did not prevent the development of ankylosis (1). An alternative hypothesis was then proposed whereby pathogenetic factors such as altered gene expression, biomechanical factors, and bacteria trigger the concomitant expression of inflammation and new bone formation, which then proceed