Perioperative procedure

All XEN implantations were performed under local anesthesia by a senior glaucoma specialist (B.W.). For trabeculectomy, the procedure was performed by either a senior instructor (B.W.) or a glaucoma clinical fellow under the supervision of a senior instructor (B.W.).

For stand-alone XEN implantation, the supero-nasal conjunctiva was marked 3 mm from the limbus. Xylocaine 2%, 0.05 mL mixed with mitomycin-C (MMC) 20 microgram 0.05 mL (400 microgram per mL) was injected into the subconjunctiva at the marked area to enlarge the subconjunctival space. Clear corneal main-port and side-port incisions were created temporally, and dispersive viscoelastic was injected to fill the anterior chamber. Preloaded injector was inserted through the main incision aiming toward the superior-nasal quadrant, after which the needle was advanced through the sclera into the subconjunctival space. After the surgeon had confirmed the visualization of the needle tip bevel within the subconjunctival space, a stent was introduced. A gonio lens was used prior to or after the introduction of the XEN at the discretion of the surgeon. The stent was ideally anticipated to be 1 mm in the anterior chamber, 2 mm in the scleral tunnel, and 3 mm in the subconjunctival space, and the injector was then removed. The subconjunctival XEN part was tested to ensure it was freely mobile, and the viscoelastic was removed once the implant was properly positioned; adjustment of the implant was performed if necessary. The anterior chamber was reformed with a balanced salt solution, and the subconjunctival bleb formation was checked. All incisions were sealed by hydration at the conclusion of the surgery.

For combined XEN implantation with phacoemulsification (phaco+XEN), we first performed phacoemulsification with intraocular lens implantation under topical anesthesia. XEN implantation was then performed in the same manner as described above.

For trabeculectomy, fornix-based conjunctival incision was performed, and a triangular scleral flap was created. Subconjunctival MMC 400 microgram per mL was applied for 4 minutes after which BSS rinsing was carried out. Paracentesis, internal sclerectomy with Kelly Descemet’s Membrane Punch and peripheral iridectomy were then performed after which the scleral flap was sutured and fluorescein dye was applied to check for appropriate aqueous outflow. The conjunctiva was finally sutured, and the stitches were buried.

Postoperative treatment regimen included topical steroid (prednisolone acetate 1%) every 2 hours for 1 week, and QID for the next 3 weeks; this treatment tapered off gradually. Steroids were adjusted thereafter according to the postoperative degree of inflammation. Topical antibiotic (fluoroquinolone) was applied QID for 4 weeks.

When adjunctive IOP-lowering treatment was required after the surgery, laser suture lysis, reintroducing glaucoma medications, and/or bleb needling could be performed. Needling of the bleb was documented as a post-operative procedure and was not counted as a glaucoma-related secondary surgical intervention (SSI) or adverse event (AE) [9,10]. The use of additional MMC injection at the time of needling was at the discretion of the surgeon.

Assessments

Demographic data, age, gender, diagnosis (POAG or PACG), monocular best-corrected visual acuity (VA; measured in Snellen and converted to decimals), pachymetry (central corneal thickness), ophthalmoscopy (cup-to-disc ratio), visual field (mean deviation on standard automated perimetry (SAP) based on SITA Standard 24–2 protocol), mean medicated preoperative IOP (measured by Goldmann applanation tonometry) and mean preoperative medication count, were assessed at baseline. Date of surgery, surgery type (stand-alone vs combined), implant location, use of antifibrotic agents (dose and timing) and intraoperative complications were recorded intraoperatively. Postoperative follow-ups were performed at day 1, weeks 1 and 2, and months 1, 3, 6, 9, 12, 18 and 24. Mean IOP and number of anti-hypertensive medications were assessed at each postoperative timepoint. Postoperative safety assessments included slit-lamp biomicroscopy, AEs, and requirement for needling and SSI at each postoperative follow-up. Mean IOP reduction, number of medications, mean deviation, cup-to-disc ratio, success rates and safety profiles were analyzed at 3, 6, 12, 18 and 24-month follow-up timepoints.

Outcomes and analyses

The main outcomes measured were mean IOP reduction and the number of IOP-lowering medications needed at each timepoint after XEN implantation or trabeculectomy.

Other outcomes included proportions of eyes achieving overall success and complete success, proportion of eyes that required needling, the amount of needling required per eye, AEs (by counts and percentages), and subgroup analysis of mean IOP reduction and the number of IOP-lowering medications in POAG and PACG eyes.

Surgical success

The definitions of success and failure were as follows:

Overall success: eyes achieving ≥20% reduction in IOP from the preoperative baseline with or without medication;

Complete success: eyes achieving ≥20% reduction in IOP from the preoperative baseline without medication; and

Failure: eyes not fulfilling overall success criteria, or loss of light perception.

Postoperative IOP measurements and number of IOP-lowering medications were compared with preoperative values and were analyzed with paired t-test. Comparison between the XEN45 and trabeculectomy groups was performed with two-sided t-test, and descriptive statistics were performed for the summarization of all endpoints. Proportions and AEs were analyzed with Chi-square and Fisher’s exact test while subgroup and other outcomes were analyzed with two-sided t-test and Mann-Whitney U test. Failure was examined with Kaplan-Meier survival analysis, and P values <0.05 were considered to be statistically significant. Statistical analysis was performed with SPSS V.25 (SPSS Inc., Chicago, IL, USA).

Intraocular pressure

IOP was significantly lower in the XEN group at day 1 postoperatively (p = 0.001), and in the trabeculectomy group from months 6 to 24 (all p<0.01) except for month 18 (p = 0.094). Final IOP at month 24 was higher in the XEN45 than in the trabeculectomy group at 14.6 and 12.5 mmHg (p = 0.008), respectively. Each postoperative timepoint had a statistically significant reduction in IOP compared with baseline preoperative IOP in both groups (all p<0.001), Fig 1. Percentage of IOP reduction from month 6 onward ranged from 28.7–34.3% in the XEN group and 42.7–46.7% in the trabeculectomy patients. Comparing the percentage IOP reduction between the two groups, the XEN group achieved greater percentage IOP reduction on day 1 (p = 0.001), then, the percentage difference converged from week 1 to month 3 (all p>0.05) and eventually the trabeculectomy group achieved greater percentage IOP reduction on month 6 (p = 0.017), month 9 (p = 0.005) and month 24 (p = 0.045) but the difference was not found on month 12 (p = 0.062) and month 18 (p = 0.205). Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Mean intraocular pressure of the XEN implant and trabeculectomy groups.

Preoperative IOP was comparable in the two groups. Postoperative IOPs were significantly different from month 6 onward except month 18. Final mean IOP was 14.6 mmHg in the XEN and 12.5 mmHg in the trabeculectomy group.

https://doi.org/10.1371/journal.pone.0256362.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mean intraocular pressure percent reduction of the XEN implant and trabeculectomy groups.

Postoperative IOPs were significantly different on day 1, month 6, month 9 and month 24. Final mean IOP reduction was 32.4% in the XEN and 44.4% in the trabeculectomy group.

https://doi.org/10.1371/journal.pone.0256362.g002

Number of medications

The preoperative medication count in XEN patients was comparable to trabeculectomy counterparts, 2.2 vs. 2.4 (p = 0.434), and no difference was detected at any postoperative timepoint (all p>0.05). Each postoperative timepoint had a statistically significant reduction in the number of IOP-lowering medications compared with baseline preoperative values in both groups (all p<0.001). Final number of medications was 0.5 vs. 0.8, respectively (p = 0.225), Fig 3. The number of medications reported for all the visits were representative of the entire cohort.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Mean number of medications between the XEN implant and trabeculectomy groups.

Preoperative medications in XEN were comparable to trabeculectomy, at 2.2 versus 2.4. The postoperative numbers of medications were not different at any timepoint. The final numbers of medications were 0.5 in the XEN group and 0.8 in the trabeculectomy group.

https://doi.org/10.1371/journal.pone.0256362.g003

Surgical success

Overall success in the XEN group was 80.4%, 79.2%, 77.1%, 72.7% and 71.4% at 3, 6, 12, 18 and 24 months, respectively, and 78.9%, 78.9%, 74.5%, 73.6% and 73.3% at 3, 6, 12, 18 and 24 months respectively in the trabeculectomy group (Fig 4). Overall success was similar in the two groups (all p>0.05). Complete success in the XEN group was achieved in 69.6%, 71.7%, 66.7%, 65.9% and 62.9% of cases at 3, 6, 12, 18 and 24 months, respectively, and 70.2%, 70.2%, 65.5%, 64.2% and 62.2% at 3, 6, 12, 18 and 24 months respectively in the trabeculectomy group (Fig 5). Complete success was comparable in the two groups (all p>0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Overall success was defined as postoperative IOP reduced ≥ 20% from the preoperative baseline, with or without medications.

Overall success of XEN implant and trabeculectomy were comparable at all timepoints. At month 24, the overall success was 71.4% and 73.4% respectively.

https://doi.org/10.1371/journal.pone.0256362.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Complete success was defined as postoperative IOP reduced ≥ 20% from the preoperative baseline, without any medications.

Complete success of XEN implant and trabeculectomy were comparable at all timepoints. At month 24, the complete success rates were 62.9% versus 62.2% respectively.

https://doi.org/10.1371/journal.pone.0256362.g005

Survival analysis

Fig 6 is presented as a Kaplan-Meier survival estimate graph with right censoring. In the XEN group, 5% of eyes had failed (SE 1%) at month 3, 10% had failed (SE 2%) at month 6, 18% had failed (SE 3%) at month 12, 27% had failed (SE 4%) at month 18, and 46% had failed (SE 6%) at month 24. In the trabeculectomy group, 5% of eyes had failed (SE 1%) at month 3, 10% had failed (SE 2%) at month 6, 18% had failed (SE 3%) at month 12, 29% had failed (SE 4%) at month 18, and 48% had failed (SE 5%) at month 24. No difference in failure between the two groups was noted (Log-rank p = 0.942). Eyes that failed to meet success criteria at month 3, 6, 12 or 18 were considered as failure, even if later on they met the success criteria.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Kaplan-Meier curve showing the probability of achieving any success criteria in the XEN implant and trabeculectomy groups.

https://doi.org/10.1371/journal.pone.0256362.g006

Visual acuity

The XEN group had Snellen VA in decimals of 0.67±0.23 at baseline, and 0.64±0.19, 0.64±0.20, 0.67±0.19 and 0.66±0.17 at month 6, 12,18 and 24, respectively, and VA did not change significantly compared with preoperative VA in this group during the postoperative follow-up timepoints. Overall, no patient lost >2 Snellen lines equivalent VA in the XEN45 group. In the combined group (Phaco+XEN), VA improved from 0.45±0.16 to 0.53±0.23, 0.62±0.17, 0.62±0.17 and 0.60±0.08 at month 6, 12, 18 and 24, respectively; however, these results were not statistically significant.

In the trabeculectomy group, Snellen VA in decimals was 0.50±0.27 at baseline, and this changed to 0.49±0.26, 0.47±0.27, 0.44±0.28 and 0.43±0.25 at months 6, 12, 18 and 24, respectively. VA suffered a statistically significant deterioration from the baseline in the trabeculectomy group at month 24 (p = 0.037), but this was without clinical significance when translated into Snellen equivalent VA. Two patients lost >2 Snellen lines equivalent VA in the trabeculectomy group: one was related to the development of epiretinal membrane and cataract by month 12 postoperatively, and the other was related to branch retinal vein occlusion by month 12 postoperatively.

Optic disc and visual field

Cup-to-disc ratio compared with baseline was stable in both groups throughout the postoperative follow-up timepoints. The XEN group had cup-to-disc ratio of 0.73±0.10 at baseline, and this changed to 0.73±0.09 at month 24, respectively. In the trabeculectomy group, cup-to-disc ratio was 0.76±0.13 at baseline, changing to 0.77±0.10 at month 24, respectively. No rapid progression of cup-to-disc ratio was detected in any patient.

Also, no statistically significant difference in MD was detected compared with baseline in the two groups throughout the postoperative follow-up timepoints. The XEN group had MD of -9.11±6.93 dB at baseline, and this altered to -9.99±6.54 dB, -9.77±6.65 dB, -9.71±6.14 dB and -9.84±6.41 dB at months 6, 12, 18 and 24, respectively. In the trabeculectomy group, MD was -9.67±5.06 dB at baseline, and this shifted to -9.71±7.60 dB, -10.67±8.12 dB, -11.12±6.03 dB and -11.51±8.50 dB at months 6, 12, 18 and 24, respectively. Rapid deterioration of MD, caused by hypotony maculopathy and choroidal detachment, was detected in one trabeculectomy patient at month 6. The MD of this patient was -13.01 dB at baseline, and this declined to -27.60 dB at the month 12 timepoint. All the visual fields data is complete for every patient at every time-point except for 2 participants in the XEN group at month 24, and 2 participants in the trabeculectomy group at month 24.

Needling revision

In the XEN group, needling was required in 10 patients (17.5%) during the follow-up period. The mean number of needling events at 24 months was 0.21±0.49 with a range of 0 to 2, and the mean and median time to the first needling were 132±130 days and 89 days. 4 (40.0%) patients were successful after the needling while 6 (60.0%) patients failed and needed further medications or SSI. In the trabeculectomy group, needling was also required in 11 patients (19.3%) in the follow-up period. The mean number of needling events at 24 months was 0.23±0.50 with a range of 0 to 4. 8 (72.7%) patients were successful after the needling while 3 (27.3%) patients failed. Anti-glaucoma medications were started initially in both groups and if the IOP could not be controlled, needling was indicated. The mean and median time to the first needling were 98±93 days and 68 days. Although the number of patients requiring needling was not different in the two groups, the trabeculectomy patients required earlier needling than their XEN counterparts.

Subgroup analysis between POAG and PACG

Subgroup analysis comparison of POAG (N = 42) and PACG (N = 15) patients in the XEN group was performed and the results are shown in Table 2. This information pertains to all patients in the cohort.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of XEN45 and trabeculectomy in POAG and PACG.

https://doi.org/10.1371/journal.pone.0256362.t002

Baseline characteristics including mean IOP and number of medications were similar in POAG and PACG patients. Mean IOP and number of medications were significantly lower than the baseline in both groups at all timepoints (all p<0.001). Mean IOP and mean number of medications were not statistically different in the POAG and PACG patients at all timepoints (all p>0.05). Overall success and complete success were comparable in the POAG and PACG patients at all timepoints (all p>0.05).

Subgroup analysis comparison of POAG (N = 42) and PACG (N = 15) patients in the trabeculectomy group was performed. Baseline characteristics including mean IOP and number of medications were similar in POAG and PACG patients. Mean IOP and number of medications were significantly lower than the baseline in both groups at all timepoints (all p<0.001). Mean IOP and mean number of medications were not statistically different in the POAG and PACG patients at all timepoints (all p>0.05). Overall success and complete success were comparable in the POAG and PACG patients at all timepoints (all p>0.05).

In terms of comparing XEN and trabeculectomy in the POAG group, baseline characteristics including mean IOP and number of medications were similar in both groups. Mean IOP and number of medications were significantly lower than the baseline in both groups at all timepoints (all p<0.001). Mean IOP was lower in the XEN group on postoperative day 1 (p = 0.002) and was lower in the trabeculectomy group from month 6 to 24 (all p<0.05) except month 18 (p = 0.139) and mean number of medications were not statistically different in both groups at all timepoints (all p>0.05). Overall success and complete success were comparable in both groups at all timepoints (all p>0.05).

Comparing XEN and trabeculectomy in the PACG group, baseline characteristics including mean IOP and number of medications were similar in both groups. Mean IOP and number of medications were significantly lower than the baseline in both groups at all timepoints (all p<0.001). Mean IOP was lower in the XEN group on postoperative day 1 (p = 0.041) and was lower in the trabeculectomy group on month 9 (p = 0.047) while other timepoints failed to show the IOP difference between the two groups (p>0.05). The mean number of medications were not statistically different in both groups at all timepoints (all p>0.05). Overall success and complete success were comparable in both groups at all timepoints (all p>0.05).

Of note, the surgery on five eyes in the trabeculectomy group were performed by a glaucoma fellow under the supervision of a senior instructor (B.W.). The final mean IOP (11.6±3.4 vs. 12.5±4.0 mmHg, p = 0.623), mean number of medications (0.8±1.1 vs 0.9±1.4, p = 0.895), overall success (80.0% vs 72.5%, p = 0.708) and complete success (60.0% vs 62.5%, p = 0.848) were similar when performed by a glaucoma fellow or by a senior instructor, respectively. However, the lack of difference may be due to the small number of patients operated by a glaucoma fellow.

Subgroup analysis between phakic and pseudophakic patients

Subgroup analysis comparison of phakic (N = 20) and pseudophakic (N = 37) patients in the XEN group was performed and the results are shown in S1 Table. Four patients in the phakic group who underwent combined Phaco+XEN were excluded from this analysis.

Baseline characteristics including mean IOP and number of medications were similar in the two groups except pseudophakic patients are older than phakic patients (73.9 ± 7.0 vs. 64.1 ± 7.2, p<0.001). Mean IOP and number of medications were significantly lower than the baseline in both groups at all timepoints (all p<0.001). Mean IOP and mean number of medications were not statistically different in the phakic and pseudophakic patients at all timepoints (all p>0.05). Overall success and complete success were comparable in the phakic and pseudophakic patients at all timepoints (all p>0.05).

Subgroup analysis comparison of phakic (N = 30) and pseudophakic (N = 27) patients in the trabeculectomy group showed baseline characteristics including mean IOP and number of medications were similar in the two groups except pseudophakic patients are older than phakic patients (73.5 ± 7.8 vs. 64.8 ± 7.0, p<0.001). Mean IOP and number of medications were significantly lower than the baseline in both groups at all timepoints (all p<0.001). Mean IOP and mean number of medications were not statistically different in the phakic and pseudophakic patients at all timepoints (all p>0.05). Overall success and complete success were comparable in the phakic and pseudophakic patients at all timepoints (all p>0.05).

Adverse events

With regard to safety, complications are shown in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Selected safety profiles of the XEN45 and trabeculectomy groups.

https://doi.org/10.1371/journal.pone.0256362.t003

Intraoperatively, subconjunctival hemorrhage occurred in 6 cases (10.5%) and device fracture in 2 cases (3.5%) during XEN implantation. In the trabeculectomy group, 2 cases (3.5%) of flap lacerations, 2 (3.5%) of conjunctival tear, 1 (1.8%) of bleb hemorrhage and 1 case (1.8%) of hyphema were noted.

In the early postoperative phase (within 1 month) [11], wound leak occurred more in the trabeculectomy group, 0 versus 9 eyes (15.8%); p = 0.002. Subconjunctival hemorrhage occurred more in the XEN group (5 eyes (8.8%) versus 0; p = 0.022).

In the later-than-one-month post-operative phase, XEN had a lower rate of numerical hypotony (IOP ≤5 mmHg at ≥1 month apart) than trabeculectomy, at 1 eye (1.8%) versus 6 (10.5%); p = 0.041. Ptosis was noted in 6 eyes (10.5%) in the XEN group and in 2 eyes (3.5%) in the trabeculectomy group; p = 0.136. Conversely, bleb leak was detected in 3 eyes (5.3%) of trabeculectomy patients but none in their XEN counterparts. XEN implant extrusion occurred in 3 eyes (5.3%) and XEN device fracture was detected in 1 eye (1.8%). In the XEN group, 7 patients (12.3%) required SSI throughout the follow-up period, and 4 of these required trabeculectomy with MMC, while 3 had the XEN removed due to implant extrusion. In the trabeculectomy group, 2 patients (3.5%) required SSI which was marginally fewer than the XEN group (p = 0.082), and 1 of these had undergone second trabeculectomy with MMC, while 1 required a tube shunt surgery. During the follow-up period, there was no bleb-related infection, endophthalmitis, suprachoroidal hemorrhage, or serious complication e.g. loss of light perception, after each procedure.

Complications and management

In terms of safety, XEN showed a favorable safety profile compared with trabeculectomy with no serious complication occurring up to the 24-month follow-up. XEN also had a lower rate of numerical hypotony (IOP ≤5 mmHg at ≥1 month apart) and bleb leak than trabeculectomy. Most complications were transient and self-limited in both groups.

There were 2 remarkable complications in the XEN implant group which have rarely been mentioned in previous studies: ptosis and nasal bleb. Six eyes (10.5%) had marked ptosis and did not improve over time (S2 Fig). Ptosis might be related to the following causes. First, the Asian interpalpebral fissure is relatively small, and trying to open the lid widely for surgical field with lid speculum could injure the levator aponeurosis. Second, MMC and xylocaine injection at the superior nasal conjunctiva might spread to the superior rectus muscle and be toxic to it. Even though low doses of MMC were used, it could not be washed out as we could in trabeculectomy. The remaining MMC and xylocaine would last longer inside the injected area and might have toxic effects on the superior rectus and levator muscles. As yet, there is no standard dose of MMC for XEN implantation. In management, we observed those cases for 6 months. Two cases underwent subsequent ptosis correction by an oculoplastic surgeon.

Nasal bleb was another complication and occurred in 5 eyes (8.8%) of the XEN group. Patients presented with large cystic bleb extended from the superior to the nasal conjunctival area. Epiphora and Dellen cornea were observed. Nasal bleb could have occurred along with the spread of MMC subconjunctival injection. This symptomatic bleb was tapped with a 30-G needle, but it recurred later. Repeated tapping was performed upon the bleb size and symptoms. Yavuzer demonstrated a technique using a drainage channel with sutures to manage this complication [33]; however, surgical techniques for avoiding this problem are not yet available.

Study limitations

The non-randomized, retrospective nature of this study may have inherently created information and selection bias. Although the clinical data from the chart represent real-world results, uncontrollable factors, such as patients being lost to follow-up and missing information, may have created statistical errors. This study also included both phakic, pseudophakic and combined Phaco+XEN patients in which IOP reduction might be related to cataract extraction. The subgroup of PACG was of relatively small size, so that some differences may not show statistically significant results. The ethnicity in this study was also limited to a Southeast Asian population. Surgeon factor might be another limitation since both procedures were not performed by the same surgeon. However, we believe that this manner is a standard practice in many tertiary eye care training centers. Also, since both eyes were enrolled in 5 patients, possible crossover effect from the use of beta-blocker eye drops might also affect the IOP lowering results. Issues of cornea endothelial cell counts, bleb morphology, quality of life and cost effectiveness were not determined within the study, and assessment of a long-term follow up period of 3 to 5 years is essential for this new implant. Further prospective, multicenter, multi-ethnicity, randomized, controlled trials are required to verify the efficacy and safety of the procedure.