Trabeculotomy opening size and IOP reduction after Trabectome® surgery

Single-center retrospective observational case series. All experiments were performed in accordance with the guidelines of GCP and the Declaration of Helsinki. The study was approved by the University Hospital Freiburg Ethics Committee (permit number 235/10_160678). All patients had given informed consent before inclusion into the study.

Ninety-three eyes of 93 patients with chronic open-angle glaucoma suitable for minimally invasive glaucoma surgery were included. Exclusion criteria were a narrow chamber angle (below Shaffer III) in the nasal quadrant and a history of previous glaucoma surgery including laser trabeculoplasty.

All eyes underwent Trabectome surgery (Trabectome; Neomedix, Tustin, CA, USA) either alone or combined with phacoemulsification and posterior camber lens implantation. All surgery was performed between 2009 and 2015 by three surgeons (AA, MN, JFJ). Trabectome surgery was performed in accordance with the manufacturer’s recommendations. A 1.8-mm clear cornea incision in the temporal quadrant was used to approach the nasal TM. The TM was opened over 4 to 5 clock hours starting with an energy of 0.8 W, increasing in 0.1-W steps until the electrosurgical effect allowed smooth removal of the TM without tearing.

Post-operatively, any topical antiglaucomatous therapy was changed to pilocarpine 2%/timolol 0.5% fixed combination preservative-free eye drops (Fotil sine; Santen GmbH, Munich, Germany) twice daily and topical steroids (dexamethasone 0.1% preservative-free eye drops). In case of timolol intolerance, only pilocarpine and steroids were given. The steroids were tapered within 6 weeks. Topical antiglaucomatous therapy was adjusted during follow-up depending on the post-operative IOP and possible side effects. If IOP readings were higher than 18 mmHg, glaucoma medication was extended.

All follow-up data was obtained at a single visit at approximately 24 weeks (range 6 to 48 weeks) post-operatively (Table 1 and Fig. 2a). If no OCT scan was available at 24 weeks, the follow-up visit with OCT scan closest to the 24-week timepoint was chosen. All OCT scans were taken by a single technician. All measurements of the trabeculotomy opening were performed by a single investigator (CvO) who, at the time of data acquisition and analysis, was blinded for the IOP and other medical details of the patients.

A swept source anterior segment OCT device (Casia SS-1000, Tomey, Erlangen, Germany) was used to obtain 256 radial scans of 16-mm length, each consisting of 512 A-scans. The given number of scans resulted in a scan density of 0.7° per scan.

The trabeculotomy (TO) opening was defined as the segment of the chamber angle circumference where the TM appeared clearly open (> ¼ of the height of Schlemm’s canal) in the OCT image (illustrated in Fig. 1a, Supplemental video 1). If anterior synechiae, membranes or re-closure of the TM interrupted the segment, only the open sub-segments were counted.

The intra-observer variability of the TO opening measurement was determined by measuring 15 scans twice in random order and in a blinded fashion. Subsequently, the intra-observer intraclass correlation coefficient (ICC) was calculated and the data was plotted in a Bland–Altman chart.

The anterior chamber depth (ACD) was measured in a modified way with the built-in algorithm of the OCT device on a single horizontal OCT section. First, a line between the nasal and the temporal angle recess was drawn (Fig. 1b). Perpendicular to this line the maximum distance to the posterior surface of the cornea was defined as the ACD.

The IOP was measured with Goldmann applanation tonometry (GAT). The baseline IOP was calculated from the mean of three IOP measurements taken the day before surgery. The follow-up IOP was obtained from a single measurement.

The IOP reduction was expressed as either the difference between the pre-operative and the follow-up IOP, named ΔIOP, or as the percentage IOP reduction, calculated as:

GNU R [11] (version 3.2.2) with additional packages [12, 13] was used for the multiple linear regression model. For additional data analysis of subgroups, either a Student’s t test for normally distributed data (tested with D’Agostino–Pearson omnibus K2 test) or a Wilcoxon rank sum test for non-normally distributed data (e.g. percent IOP reduction) was applied, using Prism 6.0 software (GraphPad, La Jolla, CA, USA). Results were either presented as scatter plots with mean and standard error of the mean (SEM) or box plots with whiskers as proposed by Tukey (1.5× interquartile range) and outliers for minimum/maximum values represented by dots [14].

The relevant data of the study population is summarised in Table 1. An illustration of the parameter TO opening is shown in Fig. 1a. Representative OCT images as used to measure TO opening and ACD are shown in Fig. 1b–d and in supplemental video 1. The mean ± SD TO opening at follow-up was 53 ± 35°, and the mean ACD was 3.24 ± 0.24.

After a follow-up time of 125 ± 66 days, Trabectome surgery lead to a significant reduction in IOP (Fig. 2a) with a baseline IOP of 25.2 ± 4.6 mmHg and a mean follow-up IOP of 16.3 ± 4.1 mmHg (p < 0.0001; paired t test). The number of topical antiglaucomatous medication was significantly reduced from 2.14 ± 1.1 to 1.5 ± 1.2 (Fig. 2b; p < 0.0001; Wilcoxon matched-pairs signed rank test).

Surgical success defined as ≥20% IOP reduction and follow-up IOP ≤21 or ≤18 mmHg was reached by 67 eyes (72%) and 60 eyes (65%), respectively.

To rule out the possibility that an exceptionally high variability of the TO opening measurement blurred the hypothesized correlation with IOP reduction, we determined the intra-observer variability of the TO opening measurement by calculation of the ICC. The intra-observer ICC in a sample of 15 cases measured twice in random order and in a blinded fashion was 0.993, which is considered sufficiently low for not significantly interfering with a potential correlation between TO opening and other parameters. The intra-observer variability data was also plotted in a Bland–Altman chart (supplemental Fig. 1A). The bias and the 95% limits of agreement, calculated from the Bland–Altman chart were 1.7 ± 10.4°, respectively. The slope and 95% confidence interval of the slope of the linear regression analysis were −0.004; −0.05 to 0.04; R² = 0.003. As only one person performed the OCT measurements, no inter-observer variability was calculated.

Thirty-eight eyes (41%) in our study underwent combined Trabectome and cataract surgery. Thus, the lens removal itself might have introduced a systematic error to the ACD measurement. We tested for a systematic change in ACD through cataract surgery with a prospective co-study, using a separate group of 10 eyes of 10 patients, which had all received cataract surgery without any additional procedure. The mean difference (bias) between pre- and post-operative ACD measurements and the 95% limit of agreement was 0.013 ± 0.177 mm (suppl. Fig. 1B). The slope of a linear regression line was −0.057 (95% confidence interval of slope: −0.50 to 0.39; R² = 0.01). Given this data, we assume no significant influence of cataract surgery on our ACD measurement. To further rule out an influence of cataract surgery on our results, a subgroup consisting of all trabectome-only cases (no combined surgery) was separately analyzed (see below).

To test whether a larger TO opening correlated with a higher percent IOP reduction, both parameters were plotted in a scatterplot (Fig. 3a) and linear regression analysis was performed. The slope of the regression line was close to zero (slope = 0.03; 95% confidence interval: −0.08 to 0,14; R² = 0.0025), indicating that no relationship between TO opening and IOP reduction existed. The analysis was repeated for ΔIOP (Fig. 3b), producing a similar result (slope = 0.01; 95%CI: -0.02 to 0.04; R² = 0.0024). The same applied for the IOP at follow-up vs. TO opening (data not shown).

A deep AC might have been connected with a greater chance for the trabeculotomy to remain open post-operatively and, thus, might have led to better IOP reduction. We, therefore, plotted the ACD against both parameters (Fig. 3c and d). The linear regression lines showed a tendency towards larger TO opening and better IOP reduction for deeper ACs. However, for both cases, the 95% confidence interval of the slope included the value zero, suggesting that no significant correlation existed.

For a more comprehensive analysis of the influence of TO opening and ACD on the IOP reduction, a multiple linear regression model was set up, which included, besides TO opening and ACD, the following covariates: the change in number of eye drops, lens status (phakic, pseudophakic, combined surgery), glaucoma subtype, follow-up time and the baseline IOP. As the IOP reduction expressed as percent reduction from baseline IOP is not normally distributed, ΔIOP was used as the dependent variable instead. The results are summarised in Table 2. There was no correlation of either TO opening or ACD with ΔIOP. Besides baseline IOP, all added covariates, including follow-up time and the change in number of eye drops had no influence on ΔIOP.

As testing for a linear relationship between TO opening and IOP reduction and ACD did not exhibit a significant correlation, we sub-analyzed in a more simple approach whether at least the extremes of the TO opening differed in their corresponding percent IOP reduction and ACD. Thus, the smallest and the highest TO opening quartile (n = 23 each group) were compared (Fig. 4). While TO opening was highly significantly different between the two groups [smallest: 13.9 ± 1.4; largest: 99.4 ± 5.2° (mean ± SEM); p < 0.0001, unpaired t test; Fig. 4a], no significant difference was detected with regard to the IOP reduction [smallest: −31.5 ± 4.0%, largest: −34.3 ± 4.1% (mean ± SEM); p = 0.64; Wilcoxon rank sum test; Fig. 4b]. However, the largest TO opening quartile exhibited significantly higher ACD values than the smallest quartile (smallest: 3.16 ± 0.04 mm; largest: 3.32 ± 0.05 mm; p = 0.031; Wilcoxon rank sum test with Bonferroni–Holm correction for multiple comparison; Fig. 4c).

As cataract surgery itself might have had an influence on IOP reduction as well as on ACD, we separately analyzed the subgroup of trabeculotomy cases without additional cataract surgery (n = 55). Compared to the whole study population, nothing changed regarding the relationship between IOP reduction and TO opening: No significant correlation was found neither in the scatter plot with linear regression analysis[(Fig. 5a (percent IOP reduction) and B (ΔIOP); slope for 5A = 0.11; 95% CI: -0.03 to 0.25; R² = 0.04; slope for 5B = 0.02; 95% CI: -0.02 to 0.07; R² = 0.02] nor in the comparison between the 1st and 4th quartile of TO opening (Fig. 6a and b; 1st quart. IOP reduction: −37 ± 4.7%, 4th quart. IOP reduction: −28 ± 5.1%; p = 0.28). Regarding ACD, no correlation was found with the amount of IOP reduction (Fig. 5c). However, in this subgroup, a deeper AC was correlated with a larger post-operative TO opening in both the linear regression analysis (Fig. 5d; slope = 38.6; 95% CI: 3.7 to 73.6; R² = 0.08; p value for the slope being non-zero: 0.03) and in the comparison between the 1st and 4th TO opening quartile (Fig. 6c; mean ACD 3.07 ± 0.05 vs. 3.3 ± 0.08 mm, respectively; p = 0.02).

The reflux of blood through collector channels into the AC may indicate the patency of the aqueous outflow system distal to Schlemm’s canal. In our study, the surgeon rated the amount of intraoperative blood reflux to no, mild, moderate or strong. When the resulting IOP reduction of these blood reflux subgroups was compared, no significant difference of their mean IOP reduction was seen (p = 0.93, ANOVA; data plot in supplemental Fig. 3).

All eyes in the strong reflux blood group (n = 8) had an IOP reduction of more than 20%, while for the mild reflux group, it was only 27 out of 40 (67,5%) and for the moderate reflux group, 32 out of 43 (74,4%) eyes. These differences were not statistically significant (p = 0.16, Kruskal–Wallis test). The group size of ‘no blood reflux’ was too small (n = 2) to be included in the analysis.