Retrospective evaluation of two-year results with a filtering trabeculotomy in comparison to conventional trabeculectomy by exact matching

Introduction

Since the first description of a guarded filtering procedure in enucleated eyes by Grant in 1958¹, performance of this surgery in patients in 1961 by Sugar², popularization of the term “trabeculectomy” in 1968 by Cairns³, and introduction of mitomycin-C as an antifibrotic to make it more effective⁴, trabeculectomy (TE) has remained a primary surgery in the treatment of glaucoma⁵. Over the years, multiple modifications of this surgery have been explored to improve its effectiveness, to make outcomes more predictable and to reduce postoperative complications and need for interventions. These modifications include, among others, variations in size, localization, and thickness of the scleral-flap, different suture techniques, variable intra- and postoperative treatment with antifibrotics or a combination of these approaches with one another⁶.

In this study, we combined elements of deep sclerectomy^7,8 and trabeculotomy⁹ with TE in an attempt to improve conventional outflow as well as subconjunctival aqueous humor drainage. Encouraged by a pilot study of filtering trabeculotomy (FTO) with a complete success rate of 79%¹⁰, we hypothesized that FTO had a higher success rate and lower complication rate than TE. We applied advanced statistics, exact matching^11,12, to enable a highly balanced comparison of our retrospective data with two-year follow-up.

Methods

Results

A total of 196 patients were included. The unmatched demographic data of FTO and TE had significant differences in preoperative IOP (p=0.017), glaucoma medication score (p=0.001), and pseudophakia (p=0.012). In total 88, eyes (44 in each group) could be matched as exact pairs eliminating key differences in IOP and medications. The IOP at baseline was 22.6±4.7 mmHg in FTO and 22.6±4.7 in TE while on 3.0±0.9 medications in both (Table 1). There were no significant differences between FTO and TE in gender, age, best-corrected visual acuity, type of glaucoma, or surgical side. In FTO, 13 eyes were pseudophakic compared to 3 in TE (p=0.006). Two patients in FTO had a pars plana vitrectomy and one a retinal cryopexy. In TE, there were no prior ocular surgeries other than phacoemulsification.

There were no statistically significant inter-group differences in complete or qualified success at any time (p=0.403 for complete success; p=0.204 for qualified success at 24 months; Figure 1 and Figure 2). The complete success rate in FTO ranged from 79% at 6 to 78% at 12 months, and 59% at 24 months. In TE, it was 81%, 85%, and 66%, respectively. Similarly, IOPs of FTO and TE were not significantly different at any time (12 months: p=0.983, 24 months: p=1.000, Figure 3). At one year, the IOP had declined to 12.4±4.3 mmHg in FTO and 11.3±2.2 mmHg in TE with medications (qualified success). At two years, IOP remained at 13.1±4.1 mmHg in FTO and 12.0±3.5 mmHg in TE (qualified success), respectively. The posthoc analysis showed that we could have detected an IOP difference of 16.1% at one year and an IOP difference of 17.6% at two years with a power above 80%. Cohen’s d was 0.32 at one year and 0.29 at two years, respectively, in both cases classifying as small. The postoperative visual acuity was not significantly different in FTO and TE at any time (p=0.894 after 12 months; p=0.443 after 24 months; Figure 4 and Figure 5).

Figure 1. Kaplan-Meier curve for complete success in filtering trabeculotomy (FTO) and trabeculectomy (TE).

Figure 2. Kaplan-Meier curve for qualified success in FTO and TE.

Figure 3. Mean preoperative vs postoperative intraocular pressure (IOP) of filtering trabeculotomy (FTO) and trabeculectomy (TE) (mean±SD).

Figure 4. Preoperative versus postoperative visual acuity of both groups (mean±SD).

FTO: filtering trabeculotomy; TE: trabeculectomy.

Figure 5. Scatter plots of visual acuity of both groups.

There was a reduction in glaucoma medication from 3±0.9 to 0.6±1.3 in FTO and from 3±0.9 to 0.2±0.5 in TE after 24 months. There was no significant difference between the two groups in glaucoma medications at 24 months with 19% of patients in FTO and 12% of patients in TE using glaucoma drops (p=0.471, Figure 6 and Figure 7).

Figure 6. Preoperative versus postoperative glaucoma medications of both groups (mean±SD).

FTO: filtering trabeculotomy; TE: trabeculectomy.

Figure 7. Scatter plots of IOP of both groups.

Few patients required medications postoperatively.

Postoperative complications in FTO included 5 eyes (12%) with a high IOP, hyphema in 5 cases (11%) and hypotony (IOP ≤ 5 mmHg¹⁶) with choroidals in 4 eyes (9%). The most common postoperative challenge in TE was a flat bleb in 6 eyes (14%), hyphema in 3 eyes (7%), high IOP in 2 eyes (5%), bleb leakage in 2 eyes (5%), and hypotony in 1 eye (2%). A total of 16, mostly reversible, complications occurred in each group (37%; Table 2).

The number of postoperative interventions was the same in the two groups (p=0.087). Bleb needling, conjunctival suture, and scleral flap suture were the most common interventions. There was no statistically significant difference between both groups in early or late interventions (Table 3), except 5-FU and laser suture lysis, which was performed more often in FTO.

Discussion

TE with MMC remains a primary surgery in the management of advanced glaucoma⁵ despite its potential for serious complications that include choroidal effusions, maculopathy, blebitis, endophthalmitis, and suprachoroidal hemorrhage¹⁷. Numerous modifications have been explored over the years to reduce the rate of these. They include a smaller scleral flap¹⁸, limbus-versus fornix-based conjunctival closure¹⁹, releasable flap sutures^20,21, a combination of trabeculectomy with deep sclerectomy²², different concentrations and exposure times of MMC²³ or sutureless tunnel trabeculectomy without iridectomy²⁴. The iridectomy, which is part of traditional trabeculectomy with a trabecular block excision, can cause hyphema, inflammation, posterior synechiae, iridodialysis, and cataracts^25,26. FTO addresses some of these issues by creating a more spread-out intake of aqueous humor, thereby reducing iris aspiration and avoiding the need for an iridectomy. The trabeculotomy and sclerectomy^8,27,28, that are part of the FTO, were meant to remove some of the post-trabecular outflow resistance^29–31.

We used matching, a nonparametric method of controlling the confounding influence of pretreatment variables in observational data³². Before matching, FTO and TE had significant differences. We have previously used coarsened exact matching, propensity score matching, and, more recently, exact matching. Coarsened exact matching applies multiple imputations to fill in missing data to not distort any relationships contained in the data while enabling the inclusion of all observed data from moderately uneven groups. Such was the case when we compared patients with a primary IOP indication to patients with a mixed indication for both cataract removal and IOP reduction^33–35. Propensity score matching is helpful to compare even more divergent groups, for instance, patients undergoing tube shunt surgery with patients undergoing trabectome surgery^36,37. By contrast, exact matching is well suited to compare similar pathological conditions and similar treatments, for instance, phaco-iStent or phaco-trabectome^38,39. A downside of exact matching is that a certain number of datasets must be excluded from the analysis because the algorithm accepts only identical primary criteria matches. Overall, our data loss was acceptable because of the high similarity that already existed at baseline. We were able to retain a large number of eyes (about 45%) as identical pairs of preoperative IOP and medication to focus on the preeminent questions of success in IOP and medication reduction. We found that FTO was as successful as TE with a similar reduction of IOP and medications. Both had a similar intervention and complication rate, notwithstanding numerical hypotony within the first six weeks after surgery. We observed a remarkably low rate of hyphema compared to ab interno trabeculectomy⁴⁰ that occurs when the IOP is at or below episcleral venous pressure allowing blood to reflux into the anterior chamber. This could indicate reduced patency of collector channels in advanced glaucoma that qualifies for filtering surgery, which has been observed ex vivo⁴¹.

Trabeculotomy ab externo has been applied to adult POAG before but, compared to TE, was noted to have a lower success rate of 70% at one year, presumably due to a reapproximation or regeneration of the disrupted trabecular meshwork⁴². A study by Chihara et al. was in agreement with this, finding that a modified trabeculotomy ab externo lowers IOP to an average near 16 mmHg in a safe fashion⁴³, by not as much as TEs. Ogawa et al. compared a nonpenetrating trabeculectomy with or without trabeculotomy⁴⁴ using a technique that was very similar to the one applied in our study, except without MMC. Despite the absence of this antifibrotic, the authors achieved a two-year IOP of 13 mmHg, not unlike our patients.

Our two-year TE results match Kirwan et al.’s multicenter study of TE with an iridectomy in 428 eyes well⁴⁵. These eyes achieved an IOP of 12.4±4 mmHg and 80% did not have to use glaucoma medications anymore, slightly more than in our study population. The authors performed needling in 17% of patients, not unlike our rates, and numerical hypotony during the first six months occurred in 7.2% of patients, which is in a similar range as our 9% in FTO and 2% in TE. Leakage was observed in 14% compared to 9% in our FTO and 2% in our TE. However, Kirwan et al.’s cataract surgery rate was 31%, much higher than our 1% in both FTO and TE, which might represent a difference in practice pattern or simply easier access to this elective procedure, which is done as an outpatient surgery in the UK.

It is interesting to note that TEs in the Tube Versus Trabeculectomy (TVT) study were performed without an iridectomy⁴⁶. In that study, IOP reduced from 25.6±5.3 mm Hg to 12.7±5.8 mm Hg at one year while the number of glaucoma medications declined from 3.0±1.2 to 0.5±0.9⁴⁷, which is also relatively similar to our results although these investigators included eyes with prior intraocular surgery, including glaucoma procedures. An early complication rate of 37% was observed by these authors, primarily consisting of a shallow or flat anterior chamber in 20% and choroidal effusions in 10%⁶. In another study of TE without an iridectomy by Jea et al., IOP reduced from 26.3±10.9 mmHg to 10.2±4.1 mmHg at two years⁴⁸. The number of glaucoma medications decreased from 2.2±1.6 to 0.5±1.0. A complete success occurred in 76.6% at one year and 66.2% at two years, matching ours.

In all of these studies, TE was performed as an outpatient patient procedure, a practice that emerged in the late 1980s^49,50 and became a standard for most types of eye surgeries and countries^51–54. Even before the now-common use of antifibrotics, there was no significant difference in success or complication rates between inpatients and outpatients⁵⁰, an observation that has been confirmed with antifibrotics as well⁵³. In the country of our study, TE and FTO are only reimbursable as inpatient procedures. It has been argued that meticulous micromanagement after TE for several days may be associated with better long-term outcomes of TE. However, this hypothesis is challenged by the present study and by the findings of others^49,50,53,55. One could argue that the considerably lower cataract surgery rate in our data compared to Kirwan et al.⁴⁵ might indicate a higher threshold for cataract surgery. These would have typically also been done as inpatient procedures to better handle post-cataract surgery bleb care.

Our study was limited by its retrospective nature and nonrandomized design. In patients with surgery in both eyes, we had selected the first eye undergoing glaucoma surgery for our analysis to take advantage of a longer history with more data. This selection might have favored the eye with more severe glaucoma. However, as this occurred in both groups, a similar bias will have occurred. Certainly, randomized controlled trials are a more sophisticated tool to reduce bias when trying to detect differences. However, given that exact matching already allows for a highly balanced comparison of retrospective data, such an effort might be difficult to justify. The indication for postoperative interventions and length of hospitalization was at the discretion of the treating physicians and was not standardized for both groups. Despite reducing confounding through the exact matching of IOP and medications, other confounding factors might have contributed to small differences in early postoperative patient management, as reflected by the fact that FTO patients received more 5-FU injections. Although these patients had a higher rate of numerical hypotony they were hospitalized for a slightly shorter time and experienced results that were not significantly different.

In conclusion, our results are largely in line with other FTO and TE outpatient studies. Combining elements from both yields reasonable two-year rates of surgical success, postoperative complications, and interventions while avoiding an iridectomy.

Data availability