Literature search and study characteristics
Initially, 3051 potentially relevant citations were screened, and 2648 remained after duplicates were removed. The flowchart of the literature search is shown in figure 1. After manually searching the reference lists, our literature search finally identified five published articles including six clinical trials [25-29] with an overall 1733 patients (777 patients received secukinumab vs. 389 patients received a placebo; and 376 patients received ixekizumab vs. 191 patients received a placebo) that could be used in this meta-analysis. All studies were phase III randomized, double-blind, placebo-controlled trials. Secukinumab was evaluated in 4 trials of 3 published articles [25-27], and ixekizumab was used in two articles in the treatment of ankylosing spondylitis [28, 29]. No data concerning brodalumab therapy in ankylosing spondylitis were published through the date of literature retrieval. The ASAS20/40 response rate of treatment for ankylosing spondylitis at week 16 was reported in all six trials, while the ASAS partial remission rate was described in three trials [25, 26]. Similar large variations were observed for the proportion of male sex, ranging from 52% (MEASURE-3) to 83.7% (COAST-W), and the mean ± SD of age, ranging from 40.1±11.6 years (MEASURE-1) to 47.4±13.4 years (COAST-W). Patient characteristics are detailed in Table 1. The methodological qualities of all trials are high in light of the clear declaration of the randomization in patient selection, blinding, and outcomes of all patients in their trials.
Overall treatment effect of IL-17 inhibitors
Amongst the six trials (four trials of secukinumab and two of ixekizumab) focusing on the efficacy of IL-17 inhibitors in ankylosing spondylitis, 1153 patients received IL-17 inhibitor therapy (777 of secukinumab and 376 of ixekizumab), and 580 patients received a placebo (389 patients were used as comparators for secukinumab and 191 for ixekizumab). Pooled analysis demonstrated that at week 16, the primary endpoint of the ASAS20 response rate was significantly increased in patients treated with any dosage and type of IL-17 inhibitor (57.6%, 664/1153) compared to placebo (35.3%, 205/580) (RR=1.63, 95% CI 1.45 to 1.84, p<0.001). Subgroup analysis suggested similar results for the comparison of both secukinumab (58.4%, 454/777) vs. placebo (35.7%, 139/389) (RR=1.64, 95% CI 1.41 to 1.89, p<0.001) and ixekizumab (55.9%, 210/376) vs. placebo (34.6%, 66/191) (RR=1.63, 95% CI 1.31 to 2.01, p<0.001). A χ² test for heterogeneity did not indicate heterogeneity amongst the included trials in synthetic analysis (I²=34%, p=0.18) and subgroup analysis for secukinumab (I²=60%, p=0.06) and ixekizumab (I²=0%, p=0.88) (Fig. 2A). The secondary endpoint of the ASAS40 response rate also had a significant increase in the IL-17 inhibitor regimen (37.1%, 428/1153) compared with that in the placebo treatment (17.6%, 102/580) (RR=2.12, 95% CI 1.75 to 2.56, p<0.001), and the subgroup analysis revealed an increased ASAS40 response rate with secukinumab (36.9%, 287/777) vs. placebo (18.8%, 73/389) (RR= 1.97, 95% CI 1.57 to 2.47, p<0.001) and ixekizumab (37.5%, 141/376) vs. placebo (15.2%, 29/191) (RR=2.49, 95% CI 1.75 to 3.57, p<0.001). There was no heterogeneity across the included trials in pooled analysis (I²=55%, p=0.05) and subgroup analysis of ixekizumab (I²=0%, p=0.59), but an exception was found for secukinumab (I²=66%, p=0.03) (Fig. 2B). Subgroup analysis based on the exposure to TNFis indicated a trend towards a higher ASAS20 response rate in TNFi-naïve patients (61.7%, 230/373) than in TNFi-IR patients (47.7%, 74/155) after IL-17 inhibitor therapy (RR=1.27, 95% CI 1.06 to 1.52, p=0.01), but no trend was observed in the ASAS40 response rate (40.2%,150/373 vs. 29.7%, 46/155) (RR=1.31, 95% CI 1.00 to 1.72, p=0.05) at week 16 (online supplementary figure S1).
Heterogeneity analysis and publication bias
In the present study, since the I² values of the efficacy analysis were more than 50%, as demonstrated above, a Galbraith radial plot was produced to explore the potential sources of heterogeneity. Our results suggested that no study was the major source of heterogeneity (online supplementary figure S2). A sensitivity analysis, by iteratively removing individual studies, was conducted to explore the impact of individual studies on the heterogeneity in this meta-analysis, and our findings suggested that there was no influence on the efficacy analysis after omitting any single-study estimates (online supplementary figure S3).
Safety profile of IL-17 inhibitors
The incidence of adverse events during the placebo-controlled periods was reported in all the included trials. Adverse events at week 16 were evaluated in this meta-analysis. After IL-17 inhibitors treatment, the most frequent adverse events reported were treatment-emergent adverse events (57.2%, 660/1153 vs. placebo 51.4%, 297/578) (RR=1.11, 95% CI 1.01 to 1.22, p=0.03) (Fig. 3A) and non-severe infections (27.4%, 211/770 vs. placebo 15.0%, 58/384) (RR=1.82, 95% CI 1.40 to 2.37, p<0.001) (Fig. 3D). The majority of infections were mild or moderate, with the most frequently reported being upper respiratory tract infections and nasopharyngitis. No discontinuation due to non-severe infections was reported in the included studies. There were no significant differences between IL-17 inhibitors and the placebo with regard to death (0.17%, 2/1153 vs. 0.17%, 1/578) (RR=0.70, 95% CI 0.14 to 3.52, p=0.86) (Fig. 3B), discontinuation due to adverse event (2.5%, 29/1153 vs. 2.1%, 12/578) (RR=1.18, 95% CI 0.62 to 2.26, p=0.62) (Fig. 3C), or serious adverse events (including reactivation tuberculosis, bacterial sepsis, or invasive fungal infections) (2.3%, 27/1153 vs. 3.1%, 18/578) (RR=0.74, 95% CI 0.42 to 1.33, p=0.32) (Fig. 3E). The safety profile of each trial was demonstrated in Table 2. A subsequent Q-test test showed that none of the single studies were homogeneous in this meta-analysis.