Comparison of Two Tube-Modification Techniques in Baerveldt Glaucoma Implantation in Refractory Glaucoma

Introduction

Glaucoma refractory to maximum tolerated medical therapy and laser trabeculoplasty often requires surgical intervention to lower intraocular pressure (IOP) and prevent ultimate vision loss. The most commonly performed procedure is trabeculectomy, which drains the aqueous through an artificial fistulous tract and creates a bleb. However, high risks of bleb-related complications and failure indicate that the use of glaucoma drainage devices (GDDs) may be a preferable alternative procedure.^1,2 The general purpose of these implants is to divert aqueous to the equatorial region through permanent sclerotomy (tube). GDDs are increasingly utilized in the management of refractory glaucoma.^3–6

In the past 2 decades, several types of glaucoma drainage implants have been developed. One of the most commonly used shunts is a non-valved Baerveldt Glaucoma Implant (BGI). Previous studies have concluded that the BGI is more effective than Ahmed glaucoma valves both in terms of surgical success rate and ability to reduce glaucoma medication.^7–9 Because BGI does not have a flow restrictor, hypotony and its associated complications are theoretically more common.^6,10,11 One of the biggest challenges of non-valved drainage device surgery is the prevention of hypotony during the early post-operative period.

Method wherein aqueous flow is blocked to form capsule to prevent further hypotony include ligation (tying a suture around the implant tube), stenting (threading a suture into the tube lumen) and their combination.^6,12-16 Limited data exist between stenting and ligation. Therefore, this study compared treatment outcomes in eyes with refractory glaucoma: stent group, ligature group and combined group.

Patients and Methods

In this study, the medical records of all patients who underwent BGI at Ramathibodi Hospital (Mahidol University, Bangkok, Thailand) between October 2011 and February 2018 were retrospectively reviewed. The study protocol was approved by the institutional review board of the Hospital and adhered to the tenets of the Declaration of Helsinki. The medical records of patients with refractory glaucoma aged over 18 years old who underwent BGI were included. Refractory glaucoma was defined as patient with uncontrolled intraocular pressure more than 21 mmHg despite maximal antiglaucoma medication, previously failed non-seton surgical treatment, or a combination. Data collected included details of the surgical procedure. Data related to best-corrected visual acuity (BCVA) in logMAR unit, IOP and IOP-lowering medications at each follow-up were also recorded.

Surgical Technique

All BGIs were performed by one of the three glaucoma attending physicians or glaucoma fellows under direct supervision. Conjunctival peritomy was performed at an angle of 100–120° with radial incision. Two rectus muscles were identified and secured by silk 2–0. The patency of the Baerveldt tube was confirmed by fluid irrigation before implantation. Then, the device was placed and fixed to the underlying sclera 10 mm away from the limbus. No antimetabolites were used intra-operatively. For aqueous blockage, ligation (7–0 polyglactin [Vicryl^®]), stenting (4–0 chromic catgut) or their combined technique (ligation with 7–0 polyglactin and stenting with 4–0 chromic catgut) was used, depending on the surgeon’s preference. Further, 2–3 venting slits were created proximally to the ligated site.¹⁷ The tube was trimmed to an appropriate length and inserted into the anterior chamber, ciliary sulcus or pars plana under the scleral flap and preserved scleral graft with the bevel facing anteriorly through a 23-gauge needle track. The conjunctiva was closed with a 7–0 polyglactin suture. Patients were administered a combination of a topical antibiotic and steroid four times daily, the dose of which was then tapered over 2 months. Chromic catgut suture was removed at post-operative 4–6 weeks. Once conjunctival wound completely healed and anterior chamber inflammation subsided, chronic catgut suture could be removed. Polyglactin suture was retained to be spontaneously absorbed without adjunctive laser suture lysis.

Outcome Measurements

Treatment failure was defined as an IOP ≥ 21 mmHg on ≥2 consecutive follow-up visits; an IOP lowering of <20% from baseline, despite the use of IOP-lowering medications; need for further surgical intervention; or no light perception of the visual acuity.

Statistical Analysis

Statistical analysis was performed with Stata (ver. 15.1, StataCorp, College Station, Texas, USA). Data are presented as mean ± standard deviation, unless stated otherwise, and compared among three groups using one-way ANOVA or Kruskal–Wallis test depending on data distribution. Categorical data are presented as frequency (percent) and compared among three groups using chi-square test or fisher exact probability test as appropriate. In multivariable analysis to compare the primary and secondary outcome, combined group was excluded for data analysis because of small sample size. The outcomes of treatment were compared between ligature group and stent group. Logistic regression was used to calculate a propensity score (PS), which evaluates confounding by indication. The variables included in the model for PS were age, gender, eye side, diagnosis, previous surgery, lens status, preoperative BCVA (logMAR), preoperative IOP, implant location, tube fixation and Baerveldt model. Repeated measurements with mixed models stratified by PS and eye side was used for comparing the primary outcomes demonstrated with a mean difference with 95% confidence interval (CI). A multilevel parametric survival analysis model stratified by PS and eye side was used for comparing the secondary outcome (treatment failure) between groups and demonstrated with a hazard ratio (HR) with 95% CI. The cumulative hazard function of treatment failure analyzed by mixed-effects exponential regression was demonstrated. Cut off for statistical significance was set at P < 0.05.

Results

In total, 174 eyes (165 patients) implanted with BGI were eligible. Among these, 163 were primary implants and 11 were sequential implants. Nine patients with both eyes were eligible. This study included only primary GDDs (ie primary BGI); sequential GDDs were excluded. Demographic characteristics of patients are shown in Table 1.

Table 1 Patient Characteristics (N = 154, Eye Side = 163)

Table 1 summarizes the baseline characteristics of the study sample. Their mean patient age at the time of surgery was 57.11 ± 19.04 years (range: 18 to 92 years).

Demographic characteristics of patients in the stent, ligature and combined groups were compared (Table 2). There was no significant difference among the three groups in terms of diagnosis, phakic status, number of previous surgeries, pre-operative visual acuity, pre-operative IOP, implant location and location of tube fixated. However, there were significant differences among the three groups in terms of the number of pre-operative medications (stent, 4.48 ± 0.77; ligature, 4.07 ± 0.63 and combined groups, 3.67 ± 0.52; P = 0.002) and BGI model (Baerveldt 350: 18 ± 60.00 vs 116 ± 92.06 vs 2 ± 33.33 and Baervedlt 250: 12 ± 40.00 vs 10 ± 7.94 vs 4 ± 66.67 for the stent, ligature and combined groups, respectively; P < 0.001; Table 2). Further, 53 (32.52%) eyes were diagnosed with primary open-angle glaucoma, 121 (74.23%) had previously undergone trabeculectomy and 132 (81.48%) had previously undergone cataract surgery. The median number of previous intraocular surgeries was 3, 3 and 2 in the stent, ligature and combined groups, respectively. The majority of eyes’ phakic status was pseudophakia (64.52%, 67.20% and 50.00% in the stent, ligature and combined groups, respectively). The median pre-operative best-corrected VA (BCVA) in logMAR unit was 0.88 (0.30–1.30), 0.88 (0.40–2.12) and 1.3 (0.40–2.30) and mean pre-operative IOP was 23.19 ± 11.84, 24.91 ± 9.04 and 22.67 ± 6.41 in the stent, ligature and combined groups, respectively. Considering quadrant, devices were implanted superotemporally (113, 69.33%), superonasally (21, 12.88%), inferotemporally (15, 9.20%) and inferonasally.

Table 2 Characteristics of the Three Groups

The median (IQR) follow-up period was 60.8 (21.23–66.03), 17.63 (10.23–32.33) and 13.1 (12.33–14.33) months for the stent, ligature and combined groups, respectively (Table 3).

Table 3 Follow-Up Period and Treatment Outcomes

The propensity score was calculated based on age, sex, eye side, diagnosis, number of previous surgery, lens status, pre-operative BCVA (logMAR), pre-operative IOP, implant location, location of tube fixation and BGI model.

There were no significant differences between stent and ligature groups in terms of post-operative IOP changes [mean difference (95% confidence interval) = 0.53 (−0.49, +1.55) vs −0.02 (−0.84, +0.81); P = 0.411], analyzed by repeated measurements with mixed models stratified by propensity score and eye side (Figure 1A).

Figure 1 Postoperative follow-up time of IOP (A) and BCVA (logMAR) (B) between two groups.

There were no significant differences between stent and ligature groups in terms of post-operative BCVA (logMAR) change [mean difference (95% confidence interval) = 0.02 (−0.13, +0.18) vs –0.05 (−0.18, +0.07); P = 0.465], analyzed by repeated measurements with mixed models stratified by propensity score and eye side (Figure 1B).

The hazard ratio and 95% confidence interval of treatment failure for the ligature group were 2.57 (0.72–9.19) compared with the stent group, analyzed using multilevel parametric survival model, stratified by eye side and propensity score, and adjusted by age, gender, previous surgery and pre-operative IOP (Figure 2). Repeated measurement analysis with excluded missing data at baseline and last follow up is available in the Supplementary material (Tables S1–S3).

Figure 2 Adjusted parametric cumulative hazard function illustrating treatment failure comparing between the two groups and analysed by mixed-effects exponential proportional hazard regression (P = 0.147).

Discussion

Trabeculectomy with mitomycin C can be effective in lowering IOP, but extensive dissection, ostomy formation and suturing can make treatment outcomes unpredictable and lead to complications. Common trabeculectomy complications include hemorrhage, scarring, cataract formation, hypotony, bleb-related infection, functional vision loss and failure.^2,18 Among all options, BGI is more resistant to conjunctival scarring and is an increasingly used intervention in refractory glaucoma. This device has previously shown promising implications as the first-line treatment of glaucoma in patients who have undergone intraocular surgery prior to filtering surgery. However, BGI, which is a non-valved implant, may have dreadful complications during both the early and late post-operative periods, which include post-operative hypotony/hypertony, flat anterior chamber with or without tube-endothelial touch and choroidal detachment.¹⁹ Many complications result from inappropriate aqueous flow restriction during the early post-operative period. Bleb resistance surrounding the BGI endplate tends to have minimal resistance to aqueous outflow until approximately post-operative 6-weeks.⁶ Various techniques have been used intra-operatively to restrict the aqueous flow in this non-valved device. External ligation, internal stenting or their combined technique has been used to achieve the optimum aqueous outflow and to allow encapsulation to form above the BGI endplate. This study found that the ability to control IOP and BCVA (logMAR) during the post-operative period were similar in both techniques. The hazard ratio and 95% confidence interval of treatment failure for the ligature group were 2.57 (95% CI, 0.72–9.19) compared with that of the stent group at the 5-year follow-up. There was no statistical significant difference between the two groups; however, there was a trend toward better surgical result in the stent group compared with the ligature group. One possible explanation for this is the differences in the types of absorbable sutures used. Polyglactin sutures hold their tensile strength for approximately 2–3 weeks in tissues and are completely absorbed by hydrolysis within 56–70 days [Coated VICRYL^® (polyglactin 910) Violet braided absorbable suture; Ethicon Inc.]. In contrast, chromic catgut holds its tensile strength for 18–21 days and is degraded by proteolytic enzymes and phagocytic cells within 90 days. Hydrolysis produces less tissue reaction than enzymatic degradation.²⁰ Chromic catgut is a virtually monofilament suture that theoretically produces few tissue reactions than multifilament polyglactin and generates a milder inflammatory reaction at the capsular endplate level. This may lead to low drainage resistance of encapsulation. However, the in vivo inflammatory reaction between the chromic catgut and polyglactin sutures remains unclear.^21,22

Despite the trend toward the better result for the stent group, chromic catgut sutures must be handled carefully because it can easily weaken or break when crushed by certain instruments.

The second possible explanation for the discrepancy in the results is the difference in the timing of suture removal between the two groups. In the stent group, the chromic catgut sutures were removed at post-operative 4–6 weeks. In contrast, polyglactin suture in the ligature group was allowed to spontaneously resolve, which may vary upon individual tissue reactions.

There are some limitations to this study. First, the retrospective nature of the study may have led to significant biases. Second, the relatively small number of patients in the stent group limited the statistical power of this study to identify the surgical risk factors. Future study with larger sample size may be warranted. Third, this study included eyes that underwent multiple previous ocular surgeries. Patients were enrolled from Ramathibodi Hospital, a referral center, representing the severe end of the glaucoma spectrum. Extensive conjunctival scarring may compromise the surgical results compared with primary surgery. Finally, this study included data pertaining to drainage devices implanted by multiple surgeons using a standardized surgical technique.

To the best of our knowledge, this is the first study to compare two tube-modification techniques. This study found a trend toward a better result for the stent group than for the ligature group. Further research with large sample size and randomized control trial is warranted to confirm results and identify the risk factors associated with surgical failure.