The search strategy for this meta-analysis yielded 725 publications, and 429 studies were excluded because of duplication. After reading the titles and abstracts, 57 studies were excluded. 239 possible full-text studies were carefully reviewed (Animal study [n=4]; Case report ; Letter and Response [n= 8]; Review and meta-Analysis [n=64]). Finally, 78 trials were included for quantitative analysis (14-91) (Fig. 1). The characteristics of included lectures are summarized in Table 1.
Efficacy of XEN standalone surgery for the management of glaucoma
6554 eyes from 65 studies and 4385 eyes from 42 studies were included for IOP and number of antiglaucoma medications (NOAM) analysis before and after XEN standalone procedure. There were no significant differences in IOP or NOAM between prospective and retrospective study (SFig1 and SFig2). Therefore, we combined them in the further analyses.
The total study sample included 3432 eyes before surgery and 3122 eyes after surgery. Overall analysis showed IOP had an obvious improvement after XEN stent implantation (SMD: 1.69, 95% CI 1.52 to 1.86, p value < 0.001) (sFig3). Fewer NOAM was also achieved in glaucoma patients after XEN standalone procedure (SMD: 2.11, 95% CI 1.84 to 2.38, p value < 0.001) (sFig4). According to the follow-up duration, studies were divided into six categories: 6m, 12m, 18m, 24m, 36m and 48m. Considering IOP, no big difference was found at different time point (sFig5). As for NOAM, the difference became less and less with time goes on although significance still was not reached (sFig6). 6 studies addressing the IOP-lowering effect of XEN in Asian populations (168 eyes in before-surgery group and 160 eyes in after-surgery group), 6 studies addressing the treatment effect of XEN in Austrian populations (529 eyes in before-surgery group and 529 eyes in after-surgery group), 8 studies addressing the treatment effect of XEN in North American populations (351 eyes in before surgery group and 284 eyes in after surgery group) and 45 studies addressing this association in European populations (2384 eyes in before surgery group and 2149 eyes in after surgery group), were included in the stratified analyses by ethnicity (SFig7). Subgroup was further done by developed vs. developing country (SFig8). No statistic difference was found in different gene background and medical care the patients could get. NOAM reduction had no difference between developed and developing country (SFig9), while the minimal difference was found in Austrian population when doing ethnicity subgroup analysis (SFig10). Heterogeneity was high in most of the stratified analyses.
Given that differences in the pseudo-exfoliation glaucoma (PEXG) and primary open angle glaucoma (POAG) could potentially bias the current meta-analysis, analyses by different glaucoma were also conducted. Three studies with 237 POAG eyes and 118 PEXG eyes were included. Interestingly, no different treatment effect was found in these analyses on IOP and Medication (SFig11-14). Furthermore, analysis was conducted in patients with or without prior interventional therapies and patients with pseudophakic and phakic eyes. IOP before and after XEN surgery, medication before and after procedure as well as bleb needling rate shown no difference in pseudophakic and phakic eyes (SFig15-19).
Efficacy and safety of XEN combined with cataract surgery for glaucoma patients
In some centers, cataract surgery was done at the same time when XEN stent was being implanted (phaco-XEN). In glaucoma patients IOP dropped significantly after phaco-XEN surgery no matter ethnicity (SFig20) or follow-up duration (SFig21). Medication needed for lowering IOP also had a clear reduction (SFig22). Further comparison was done between XEN standalone surgery and phaco-XEN surgery on IOP and medication. After procedure, there was no significant difference in IOP (SMD: -0.01, 95% CI -0.09 to 0.08, p value 0.894)(sFig23) and NOAM (SMD: 0.09, 95% CI -0.04 to 0.23, p value 0.170)(Fig2) between two group. Stratified analysis was also done by ethnicity and follow-up duration for IOP. In Asian population a clear difference of after-surgery IOP was found between two procedures (SMD: 0.57, 95% CI 0.23 to 0.91), which was absent in European, North American and Australian studies (SFig24). On different follow-up time points, patients in different procedures shared a similar IOP (Fig3). Nevertheless, lower IOP before surgery was found in phaco-XEN group when baseline was analyzed (SMD: 0.31, 95% CI 0.15to 0.47, p value < 0.001), especially in European population (SFig25). Patients whose IOP achieved <18 mmHg, <15 mmHg, <12 mmHg or had a reduction >20% from baseline were counted and RR of success rate was obtained, which showed no difference in efficacy of treatment between XEN alone and phaco-XEN (data not shown). For considering complications, bleb needling rate was compared. Although similar IOP reduction was found in XEN alone and phaco-XEN group, bleb needling rate was significantly high in XEN standalone group (RR: 1.45, 95% CI 1.06to 1.99, p value 0.019)(Fig4).
Efficacy and safety comparisons between XEN standalone and trabeculectomy procedure
Besides comparing to phaco-XEN, XEN standalone procedure was also compared with trabeculectomy surgery. A preference of assigning patients of higher IOP to trabeculectomy group was found when checking baseline (SMD: -0.33, 95% CI -0.49 to -0.16, p value < 0.001) (Fig5). After surgery, IOP showed no difference between two groups (Fig6), while patients underwent trabeculectomy had lower bleb needling rate (RR: 2.42, 95% CI 1.33 to 4.43, p value 0.004)(Fig7).
Sensitivity analyses and publication bias
When leave-one-out sensitivity analyses were conducted, all the results remained statistical robustness (Table 2, SFig26-33). Egger and Begg test was applied to test publication bias. Publication bias was found in IOP and medication comparison before and after phaco-XEN surgery. Publication bias was also found in after-surgery medication comparison of XEN standalone vs. phaco-XEN groups (Table 2). By trim and fill method, both the results of fixed and random effects model are the same with original result (Appendix 3, SFig34-36), except for after-surgery medication comparison of XEN standalone vs. phaco-XEN groups.