Background/Purpose:  The relapse rate of giant cell arteritis (GCA) in the patients treated by corticosteroids (CS) varied widely in observational series and randomized trials. The purpose of this systematic review was to (i) estimate the prevalence of relapse and (ii) explain the heterogeneity of relapse rate in patients receiving CS alone.

Methods: We searched from PubMed up to April 2017. Two investigators independently extracted data. The prevalence of relapse was pooled by random effect model. Heterogeneity (I² >50%) was explored by meta-regression.

Results: A total of 37 cohorts (6 trials) totalizing 16839 patients were selected.  The overall prevalence of relapse was 43% (95% confidence interval [CI] 0.37-0.50) with a high heterogeneity (I² =90%). The year of publication was positively associated with relapse (37 studies, rate increase of 6.1% for one decade, p =0.02) (Figure 1). The length of CS administration was negatively correlated with the relapse rate (15 studies, rate decrease of 1.1% for one additional month, p =0.04) (Figure 2), but not with the initial CS dose at diagnosis (P =0.46) or time of follow-up (p=0.12). The patients included in the control arms of randomized trials testing immunosuppressive drugs relapsed more frequently (62%, CI 47-75) compared to those in observational studies (40%, CI 33-47) (p =0.006).

Conclusion:  The relapse rate of GCA remains high without improvement across decades. The relapse rate might be related to the duration of CS administration rather than initial dose at induction. Our results challenged the schedules of CS administration used in randomized trials as standard of care and questioned the size of estimated efficacy of methotrexate and tocilizumab.

Figure 1

Figure 2