Reply to Praprotnik and Tomsic and erratum corrige.

Sirs, We appreciated the interesting comments by Praprotnik and Tomsic (1). The difference from historic cohorts in terms of clinical characteristics evidenced in our study led us to review our original data. In fact there are errata in our manuscript and table as the reported age actually refers to firstline TNF inhibitor (TNFi) initiation (2). For the sake of clarity here we report the age at diagnosis of the non-radiographic (nr-) axial spondyloarthritis (axSpA) group (44.44±14.93 years) and the r-axSpA group (35.37±15.96 years) in our cohort. However, our data remain slightly different from those reported by inception cohorts. The latter are, by definition, a group of individuals identified according to definite criteria at an early and uniform point in the course of disease (3, 4), which is not the case for retrospective studies. Moreover, considering that our cohort was assembled only from tertiary care centres, it is conceivable that a a referral bias may also be responsible for differences from historic cohorts. Nevertheless, concerning age at diagnosis for r-axSpA patients, our data seem to be aligned with a report from a worldwide study (5), whereas the older age in the nr-axSpA group is consistent with a recent report from an Italian tertiary-care cohort (6). A possible further flaw may be the concomitant fibromyalgia (FM) (7). Indeed, it is possible that FM might have been responsible for therapy discontinuation and higher BASDAI at baseline in the nr-axSpA group. Nevertheless, the results form a French study demonstrating a lower retention rate on TNFi in axSpA patients with fibromyalgia seem to support this hypothesis (8). Unfortunately, FM was not assessed in our patients, thus it represents a main limitation of our study. Of note in our multivariate model, the Bath Ankylosing Spondylitis Functional Index (BASFI), which may be influenced by FM (9), confounds the association between nr-axSpA and TNFi discontinuation and is an independent predictor of therapy failure. We would like to take up the suggestion from Praprotnik and Tomsic to underline that functional disability, as assessed by BASFI, may impact the persistence on TNFi of axSpA patients (9).