Outcomes of the Shunt Tube Exposure Prevention Study (STEPS), A Randomized Clinical Trial

Abstract

Purpose:

To compare the long-term safety and efficacy of amniotic membrane-umbilical cord (AM-UC) and pericardium patch grafts in reducing glaucoma shunt tube exposure.

Design:

Multicenter, prospective, randomized clinical trial.

Subjects:

Adults with uncontrolled glaucoma undergoing glaucoma drainage device (GDD) implantation.

Methods:

Subjects were randomized to receive GDD with either AM-UC (Amnioguard™, BioTissue, Miami, FL) or pericardium (Tutoplast®, IOP, Costa Mesa, CA) to cover GDD tubes. Subjects were followed clinically with anterior segment optical coherence tomography (AS-OCT) to prospectively assess patch graft stability and host-tissue integration.

Main Outcome Measures:

Tube exposure, graft thinning and graft-related complications.

Results:

A total of 81 eyes of 81 patients (50 females, 31 males) aged 67±13 years underwent GGD implantation using Baerveldt (n=72) or Ahmed valve (n=9). Tubes were inserted in the anterior chamber (n=71), Sulcus (n=6) or pars plana (n=4). Tube ligation was performed with Baerveldt GDD along with fenestration (n=51) or orphan trabeculectomy (n=21). Tubes were covered with AM-UC (n=41) or pericardium (n=40). The mean follow-up time was 29±8 months (range, 13-40 months). Tube exposure occurred in 1 eye (2%) in the AM-UC group at 3 months and 2 eyes (5%) in the pericardium group at 2 and 6 months (p=0.54). Sequential AS-OCT showed better host-tissue integration and significantly less graft thinning in the AM-UC group. Early graft thinning (≤ 3 months) occurred in 5 eyes (12%) in the AM-UC group and 17 eyes (43%) in the pericardium group (p=0.002). Late thinning occurred in 2 eyes (5%) and 11 eyes (28%) in the AM-UC and pericardium group, respectively (p=0.007). Graft translucency and cosmetic appearance of AM-UC were superior to pericardium. No evidence of graft rejection or infection was associated with the patch grafts in either group.

Conclusion:

AM-UC grafts are well-tolerated and offer an alternative to pericardium for safe and stable tube shunt coverage. Its high tensile strength, low immunogenicity, and excellent host-tissue integration significantly reduced graft thinning. AS-OCT is a useful tool in evaluating patch grafts following GDD surgery.

Précis

This is the first randomized clinical trial to compare long-term safety and efficacy of amniotic membrane-umbilical cord (AM-UC) versus pericardium patch grafts in reducing glaucoma shunt tube exposure and graft thinning.

Glaucoma drainage devices (GDDs) are being used more frequently in the treatment of uncontrolled glaucoma.¹ While effective at reducing disease progression, tube exposure is a serious postoperative complication that increases the risk of sight-threatening intraocular infection.^2-5 Several surgical approaches have been described to minimize tube exposure, including long scleral tunneling,⁶ scleral flaps,⁷ Tenon advancement,⁸ and allogenic patch grafts such as sclera, pericardium, and cornea.^9-13 Although these grafts have reduced the rate of tube exposure, they are prone to progressive thinning and silent melt, possibly due to immune-related processes.^5,14,15 We speculated that graft thinning might also result from poor host-tissue integration. We further speculated that a thicker version of amniotic membrane (AM), with high tensile strength and low immunogenicity, may provide good tectonic support and avert immune-related graft thinning and melting.¹⁶ We also assumed that AM might have an unknown biological activity to promote host-tissue integration and increase graft survival.

Our previous studies have supported the above hypotheses and demonstrated the short-term safety and efficacy of a 300μ-thick AM (Figure 1B), harvested from the amniotic sac near the umbilical cord (UC). Tube exposure rate was reduced to 2.3%, and anterior segment optical coherent tomography (AS-OCT) demonstrated a well-integrated conjunctiva-AM bilayer, which maintained its thickness over time.¹⁷ These findings suggested that the biologically active AM may work differently from other patch graft materials. However, further increase of tissue thickness and rigidity was required as commonly accepted criteria for grafts used in covering GDDs. Consequently, we hypothesized that a patch graft harvested from the UC with its compact structure and overlying AM (Figure 1C) would be ideal for that purpose, including increased tensile strength. Therefore, in this randomized clinical trial, we have evaluated AM-UC patch graft as a primary GDD tube covering compared to the well-established pericardium patch graft, with regards to graft thinning, tube exposure, and host-tissue integration.

Figure 1: Thickness and Structure of Amniotic Membrane and Umbilical Cord Patch Grafts.

Surgical microscope view (Top) and light microscope with Hematoxylin and Eosin stain (Bottom) showing: Amniotic membrane (AM) graft consists of a devitalized epithelium, basement membrane, and stroma and has an average thickness of 100μm (A). A 300μm-thick AM, with a thicker spongy stroma, can be harvested from the amniotic sac near the umbilical cord (UC) (B). A 400-500μm-thick patch graft can be harvested from the wall of UC, it also consists of devitalized amniotic epithelium, basement membrane, and compact stroma. (Bar = 500)

METHODS

Design and Participants:

This is a multicenter randomized clinical trial to compare the long-term safety and efficacy of AM-UC derived patch grafts and pericardium patch grafts in reducing glaucoma shunt tube exposure. The study was approved by the Institutional Review Board (IRB) of each of the participating sites: Bascom Palmer Eye Institute, University of Miami (Miami, FL); New York Eye and Ear Infirmary (New York, NY); Manhattan Eye, Ear and Throat Hospital, Northwell Health (New York, NY); and Columbia University, Harkness Eye Institute (New York, NY). The study was conducted in accordance with the Health Insurance Portability and Accountability Act (HIPAA) and Declaration of Helsinki. The study is registered at clinicaltrials.gov with trial identifier: NCT01551550.

Prior to participating in the study, written informed consent was obtained from all subjects. A detailed medical and ocular history were then obtained including medical conditions and current and previous treatments of glaucoma. All subjects also underwent a complete ophthalmic examination at baseline to determine eligibility for the study. Inclusion criteria included: subjects 21 years and older of both genders and all ethnic groups, who had uncontrolled glaucoma with or without previous conjunctival cutting surgery including prior failed trabeculectomy. Exclusion criteria included: previous GDD surgery, chronic ocular inflammation, recent ocular infection within 14 days prior to study entry, previous cyclodestructive procedure, no light perception vision, inability or unwillingness to give written informed consent or non-adherence to study requirements.

After meeting the eligibility criteria, study subjects acquired a unique patient identification number (PIN) and were randomly assigned to receive either GDD with AM-UC (study group) or GDD with pericardium (control group). To ensure that both Study and Control groups are equally represented, block randomization (n=4) was performed via a secure online service (randomization.com) to generate a random allocation sequence for each clinical site.

Surgical Technique:

Surgery was performed for both groups at each of the participating sites as previously described.¹² In brief, under local anesthesia with sedation, a fornix-based conjunctival incision was made in the planned implantation quadrant and blunt dissection was carried out to identify the recti muscles. A Baerveldt 350 mm² implant (Abbott Medical Optics, Abbott Park, IL) or Ahmed valve PF7 (New World Medical, Inc., Rancho Cucamonga, CA) was primed and placed beneath or between the muscles, respectively. The plate was secured to the sclera using two 9-0 nylon sutures (Ethicon Inc., Johnson & Johnson, Somerville, NJ, USA), approximately 10 mm posterior to the surgical limbus. Baerveldt tubes were occluded with an external 8-0 Vicryl suture (Ethicon Inc., Johnson & Johnson, Somerville, NJ, USA). On a case-to-case basis, the tube was trimmed to an appropriate length, and inserted into the anterior chamber or sulcus through a scleral fistula created using a 23-gauge needle, or into the vitreous cavity through a sclerostomy after a complete skirt to skirt pars plana vitrectomy. The tube was secured to the underlying sclera with a single pass 8-0 Vicryl stay suture close to the insertion site. Based on randomization, the study group received a single layer of AM-UC (Amnioguard™, BioTissue, Miami, FL) and the control group received a single layer of pericardium (Tutoplast®, IOP, Costa Mesa, CA) patch grafts. According to the manufacturers, the thickness of both AM-UC and pericardium ranged from 400μm to 500μm. The patch grafts were reconstituted (thawed or rehydrated), rinsed with balanced salt solution with antibiotics, cut to a rectangular shape of approximately 5×7 mm, and secured over the anterior portion of the extraocular segment of the shunt tube using 2 interrupted 8-0 Vicryl sutures near the limbus. The anterior edge of the AM-UC and pericardium were beveled to avoid dellen formation. The conjunctiva was then re-apposed with 8-0 Vicryl sutures and subconjunctival antibiotics and steroids were administered in all patients.

Postoperative Treatment and Follow-up

All eyes received post-operative topical fluroquinolone antibiotics (Zymar, Allergan, Inc., California) 4 times daily for one week and Prednisolone 1% (Omnipred, Alcon Laboratories, Inc., Texas) every 2 hours for one week with tapering in 6-8 weeks. Anti-glaucoma medications were added as needed and recorded. Subjects returned for mandatory follow-up visits with contiguous windows at 1 day, 1 week, 1 month, 3, 6, 12, 18, and 24 months. Visits outside the study schedule were also recorded. Subjects were evaluated both clinically and with the high-resolution spectral-domain AS-OCT (RTVue®, OptoVue, Inc., Fremont, CA or Cirrus, Zeiss Meditec, Dublin, CA) to assess patch graft thickness, stability and host-tissue integration. Whenever good optical penetration could be achieved with clear visualization of the upper edge of the tube underneath the conjunctiva/patch graft complex (Figures 2A, 2B), a series of four horizontal cut sections showing the round cut section of the tube were obtained, to ensure the cut sections were as perpendicular to the tube plane as possible. A line was then drawn from the apex of the tube to the surface of the conjunctiva (Figure 2B). The lowest of the 4 values was recorded as the thickness. If good optical penetration could not be achieved (Figure 2C, 2D), such as in early imaging of pericardium patch grafts, an estimation of the conjunctiva/patch graft complex thickness was achieved by extending a curved line along the natural curvature of the tube/sclera utilizing the upper edge of the part of the tube shown on the image that is not covered by a patch graft on median longitudinal OCT cut sections along the length of the tube (Figure 2E, 2F). In addition to the measurements, adequate visualization of tube margins indicative of enough optical penetration observed in a visit after such visualization had not been possible in a previous visit was considered as a sign of patch graft thinning after ruling out major subconjunctival bleeding, although not conclusively indicative of such (Figure 3F).

Figure 2: Measurement of Graft Thickness using Anterior Segment Optic Coherence Tomography (AS-OCT).

A cross section AS-OCT is taken and the conjunctiva/graft complex thickness is measured by drawing a line from the apex of the tube to the surface of the conjunctiva (A, B). If the tube is not visible (C,D), a longtudinal section AS-OCT is used to project the apex of the tube by draw a line extending from the visible non-coverd part (E,F).

Figure 3: Host-Tissue Integration in Amniotic membrane-Umbilical Cord (A) vs. Pericardium (B).

Sequential anterior segment optical cohence tomograpgy (AS-OCT) shows better host-tissue integration in the amniotic membrane-Umbilical cord (AM-UC) compared to pericardium. At 1month, there is a potential space between the conjuncticva and both grafts (C,D arrows). The translucent AM-UC allows good visulaization of the tube (C), whereas the opaque precardium obscures the tube (D). Additionally, the conjunctiva appears thicker over the pericardium (B). At 12 months, the AM-UC graft merges well in the subconjunctival space with maintained thickness and increased reflectivity (E), which indicates good host-tissue integration. On the contrary, the potential space persists over the pericardium indicating poor integration (F). The upper edge of the tube is also visible (F, Star) indicating thinning of the pericardium (F). Clinically, AM-UC covered area (G) was aesthetically more appealing than the opaque pericardium-covered area (H).

Outcome Measures:

Outcome measures were assessed in a masked fashion by reviewing all preoperative and postoperative data including slit lamp examination, color slit lamp photography and AS-OCT to monitor: (1) tube exposure; (2) tectonic integrity of the graft and the development of asymptomatic graft thinning; (3) other complications such as wound leak, leaking bleb, shallow or flat anterior chamber, hyphema, hypotony, choroidal effusion, motility restriction, uncontrolled intraocular pressure (IOP) and reoperation.

Sample Size Calculation:

Based on previous publications, tube erosion rate ranged from 2.6% to 16% for pericardium, which also has a progressive thinning rate of approximately 25%.¹⁵ On the other hand, tube erosion rate for AM was 2.3%, with no detected graft thinning.¹⁷ Therefore, we assumed an average of 34.5% combined rate of tube exposure and graft thinning in the pericardium and a rate of 2.3% in the AM-UC. To achieve 90% power and an alpha error of 0.05, 34 subjects per group were enough to establish a statistically significant difference. We anticipated a 20% dropout rate and enrolled an additional 14 subjects to account for dropouts. Therefore, the total sample size was adjusted to 82 subjects.

Statistical Analysis:

Evaluation of continuous variables was achieved using a non-paired, two-tailed Student t-test. Categorical variables were evaluated with chi-square test, Fisher’s exact test, or Spearman correlation as appropriate. The tube exposure and graft thinning rates were compared between the treatment groups using 2-sided Student t-test, Mann-Whitney U-test, and X² tests, and the risk factors were also assessed to determine any possible correlation. For qualitative parameters, inter-observer agreement was quantified by calculating the average measure interclass correlation coefficients. Data was expressed as the mean ± standard deviation. The results were analyzed with the statistical software Sigma Stat 2.03 and SPSS 20.0 (SPSS, Chicago, IL). P<0.05 was considered statistically significant for all analyses.

RESULTS

A total of 82 eyes of 82 patients (51 females, 31 males, aged 67±13 years) received GGD implantation with Baerveldt 350 mm2 (Abbott Medical Optics, Abbott Park, IL) in 72 eyes, and Ahmed valve FP7 (New World Medical, INC, Rancho Cucamonga, CA) in 10 eyes. One female patient that received Ahmed valve was lost to follow up and excluded from analysis, leaving a total of 81 eyes from 81 patients. GDD were placed in the superior-temporal quadrant (n=76), superior nasal (n=1), and inferior nasal (n=4). Tubes were inserted in the anterior chamber (n=71), Sulcus (n=6) or pars plana without the use of pars plana clip (n=4). Tube ligation was performed with Baerveldt GDD along with fenestration (n=51) or orphan trabeculectomy (n=21). Tubes were covered with AM-UC (n=41) or pericardium (n=40). In addition to the orphan trabeculectomy, GDD was combined with corneal transplant (n=2), phacoemulsification (n=11), and pars plana vitrectomy (n=6). Baseline clinical characteristics and demographic data are summarized in Table 1. There was no statistically significant difference between groups regarding age, race, ethnicity, preoperative diagnosis, and number of previous surgeries. The mean follow-up was 29±8 months (range, 13-40 months). Despite the impressive adherence of the subjects to the follow-up schedule, the contiguous time windows eliminated few missed visits i.e. 4 subjects with one or more lost follow up visits were included in the statistical analysis.

Table 1.

Subjects’ Demographics and Preoperative Characteristics

	AM-UC Group (n = 41)	Pericardium Group (n = 40)	p-value
Age (years)	65.4 ± 13.6	68.9 ± 12.3	0.23
Range	23, 92	36, 88
Gender			0.17
Male	19 (46%)	12 (30%)
Female	22 (54%)	28 (70%)
Ethnicity			0.82
Hispanic	15 (37%)	16 (40%)
Non-Hispanic	26 (63%)	24 (60%)
Race			0.04
Caucasian/White	23 (56%)	26 (65%)
African-American/Black	18 (44%)	10 (25%)
Asian	0 (0%)	4 (10%)
Eye			0.92
Right	18 (44%)	18 (45%)
Left	23 (56%)	22 (55%)
Lens Status			0.61
Phakic	15 (39%)	12 (31%)
Aphakic	2 (5%)	1 (3%)
Pseudophakic	22 (56%)	26 (67%)
Diagnosis			0.93
Open-angle glaucoma	32 (78%)	30 (75%)
Angle-closure glaucoma	3 (7%)	4 (10%)
Other	6 (15%)	6 (15%)
Number of Previous Surgeries	1.5 ± 1.0	1.7 ± 1.5	0.55
Range	0, 4	0, 6
Implant Type			0.74
Baerveldt	37 (90%)	35 (88%)
Ahmed	4 (10%)	5 (12%)

^*Values are reported as mean ± SD, median (min, max) or number (percent) as indicated.

In the early postoperative period, eyes with AM-UC showed less postoperative conjunctival injection compared to eyes with pericardium, which was managed with the routine postoperative treatment. The AM-UC covered area was aesthetically more appealing than the opaque pericardium- covered area. The translucency of AM-UC allowed direct visualization of the underlying tube and permitted easy laser suture-lysis when needed to control early postoperative ocular hypertension in AM-UC group (n=7), while this was not possible through the opaque graft in the pericardium group (n=6).

Tube exposure occurred in 1 eye (2.4%) in AM-UC group within 3 months after receiving Ahmed valve implant, and in 2 eyes (5%) in the pericardium group at 2, and 6 months after receiving Baerveldt glaucoma implant. No statistically significant difference in the tube exposure rates between groups was observed (p=0.54). The 3 tubes were inserted into the AC in the superior temporal quadrant, and the Baerveldt tubes were fenestrated. Meibomian gland dysfunction (MGD) was present in all 3 cases, conjunctival recession was observed in one eye which received combined phacoemulsification cataract surgery, and conjunctival dehiscence in the other eye which received combined corneal transplantation. All subjects with tube exposure underwent surgical revision to cover the tube with a corneal patch graft (VisionGraft, Tissue Banks International Inc., Baltimore, MD). After revision, one eye in the pericardium group had a second exposure of the tube and the plate. The GDD was removed, and the vision has markedly deteriorated. However, no signs of infection were observed in any of the studied eyes.

Sequential AS-OCT showed better host-tissue integration and significantly lower graft thinning in AM-UC group (Figure 3). Graft thickness and density at one-month was recorded and compared to the following visits and between groups (Figure 4). Images were also stored for masked comparison using subject ID, however, since AM-UC was translucent, the type of treatment was obvious to the examiner. In addition, the opaque pericardium precluded deeper penetration of light and hence affected the precise OCT determination of thickness or tube visualization (Figure 4B). The average graft thickness of AM-UC at one month was 340 ± 194 μm (range 200-570 μm), which was lower than the average one-month measurable thickness of pericardium 424±139 μm (range 370-540 μm), (p=0.03), probably due to more compression with the tube. Early graft thinning, defined as >20% reduction of the graft thickness at 3-month, occurred in 5 eyes (12%) in AM-UC with an average thickness of 255±78 μm (range 182-324 μm), and 17 eyes (42.5%) in the pericardium group with an average thickness of 318±71 μm (range 243-3740μm), (p=0.002). Late thinning at 12-month continued in 2 eyes (4.9%) in AM-UC with an average thickness of 206±27 μm (range 179-233 μm), and 11 eyes (27.5%) in the pericardium group with an average thickness of 176 ± 66 μm (range 116-238 μm), (p=0.007). Graft thinning occurred within the first 12 months in both groups, with insignificant changes in the AM-UC thereafter. In cases with progressive thinning/melting, the conjunctiva remained stable over the tube without erosion (Figure 4F). Graft thinning was observed more frequently in subjects who received Baerveldt implant with tube fenestration (3 out of 5 in AM-UC group, and 13 out of 17 in the pericardium group). One case in the pericardium group showed partial thinning adjacent to localized bleb from possible leaking around the tube. The incidence of graft thinning was significantly related to 3 variables: gender, number of previous surgeries, and treatment group. More specifically, graft thinning was detected more in the females than males (36% vs 12.9%, p= 0.023), in subjects who had more previous ocular surgeries (p= 0.027), and subjects who received pericardium vs. AM-UC (42.5% vs 12.2%, p= 0.002) (Table 2). Multiple regression confirmed only treatment group (p=0.009) and the number of previous surgeries (p=0.008) were significant factors for thinning. In addition, there was no significant correlation between graft thinning and the remaining variables such as age, race, ethnicity, eye (right vs. left), lens status, type of previous surgeries, type of GDD implant, or the follow-up time.

Figure 4: Graft Thinning in Amniotic membrane-Umbilical cord vs. Pericardium.

Sequential AS-OCT shows less graft thinning in the AM-UC vs. pericardium. Baseline graft thickness and density at one month was easily detected in AM-UC (A) than Pericardium (B). At 6 months, insignificant graft thinning (< 20%) occurs in AM-UC (c) whereas pericardial thinning was accompanied with tube visualization after being obscured (D). At 12 months, significant graft thinning (>20%) occurs in AM-UC (E), whereas total graft melt is seen in pericardium (F).

Table 2.

Regression Analysis of the Risk factors for graft thinning

	B	S.E.	Wald	df	P Value	Exp(B)	95% C.I. for EXP(B)
							Lower	Upper
Gender	1.334	.611	4.761	1	.029*	3.797	1.145	12.586
Constant	−1.910	.536	12.703	1	.000	.148
Previous Surgeries	.694	.248	7.831	1	.005*	2.001	1.231	3.253
Constant	−2.114	.545	15.062	1	.000	.121
Patch Graft	1.672	.575	8.467	1	.004*	5.322	1.726	16.409
Constant	−1.974	.477	17.109	1	.000	.139

B: beta; S.E.: Standard error; df: Degree of freedom;

^*: significance; EXP(B): Odd ratio

There was no evidence of graft rejection, severe inflammation or infection associated with the patch grafts in either group. Other postoperative complications, not related to the grafts material, occurred at similar rates in the study and control groups. Hyphema developed in 3 study eyes and 2 control eyes, and spontaneously resolved within 2 weeks. Persistent hypotony occurred in 2 eyes in each group, along with choroidal effusion in the study group which required further tube occlusion, repositioning and AM-UC replacement with a corneal patch graft. Corneal edema occurred in 4 study and 3 control eyes with combined phacoemulsification, shallow AC and iridocorneal or tube touch. One patient had corneal graft failure and received corneal transplantation. Diplopia also developed in 2 eyes in each group.

Lastly, the IOP and the number of glaucoma medications were also comparable between groups; the mean preoperative IOP was 26.8±3.4 mmHg (range 18-31) in the AM-UC group and 26.4±4.9 (range 17-33) in the pericardium group (p=0.67). Mean postoperative IOP at 3 months was 14.4±5.7mmHg (range 5-32) in AM-UC group and 15.2±6.5mmHg (range 6-30) in the pericardium group (p=0.56). Mean IOP at 1 year postoperatively was 15.8±6.3 mmHg (range 6-26) in AM-UC group and 15.3±6.1 mmHg (range 6-25) in the pericardium group (p=0.72). At 2 years postoperatively, mean IOP was 14.9±4.9 mmHg (range 7-21) in the AM-UC group and 15.0±5.2 mmHg (range 6-28) in the pericardium group (P=0.93). The mean number of preoperative glaucoma medications was 3.3±0.5 in AM-UC group and 3.4± 0.5 in the pericardium group (p=0.37). At 3-month the number of glaucoma medications reduced to 0.9±0.8 in AM-UC group and 1.2±0.9 in the pericardium group (p=0.12). At 1 year postoperatively, mean number of glaucoma medications was 1.2±0.8 in the AM-UC group and 1.5±0.7 in the pericardium group (p=0.08). The difference remained statistically insignificant at 2 years with a mean number of medications of 1.5±1.0 in the AM-UC group, and 1.8±0.8 in the pericardium group (p=0.15).

DISCUSSION:

To the best of our knowledge, this is the first randomized clinical trial to evaluate AM-UC, versus pericardium, as a primary covering of GDD tubes. Our results demonstrated the long-term safety and efficacy of AM-UC in reducing tube exposure and graft thinning. The AS-OCT findings further support our hypotheses that AM-UC may not only provide strong tectonic support and avert the possible immune-related graft melting but also could promote improved host-tissue integration to prolong graft survival. In addition, the cosmetic appearance of AM-UC was superior to pericardium due to graft translucency, particularly when the GDD was placed inferiorly.

In this study, tube exposure rate was similar in the AM-UC group (2.4%) and pericardium group (5%), (p=0.54). The incidence and timing of tube exposure were also comparable to our previous study and the literature using different materials and techniques.^13-15,17,18 A large meta-analysis of 38 studies also demonstrated a tube exposure rate of 2.0±2.6% with an average follow-up of 26±3 months.⁵ While we have relatively longer follow-up of 29±8 months, tube exposure occurred within 2-6 months and was not preceded by progressive graft thinning in either group. However, a longer follow-up for this cohort may have resulted in different results. All 3 eyes had MGD and tubes were inserted in the AC at the superior temporal quadrant, compatible with a mechanical rubbing mechanism as previously suggested.^15,19,20 Other risk factors include previous and or combined surgeries, which might induce conjunctival dehiscence or scarring.²⁰ We suggest that inadequate healthy conjunctival covering was a direct cause of early patch graft exposure and melting.¹⁷ Complete conjunctival covering is also crucial as a source of native cells to achieve host-tissue integration until the patch graft achieves full tectonic strength thereby circumventing thinning.¹⁷

In the current study, the incidence of early graft thinning was significantly lower in AM-UC (12%) compared to pericardium (42.5%) (p=0.007). The incidence of late progressive graft thinning was also significantly lower in AM-UC (4.9%) compared to pericardium (27.5%) (p=0.002). In addition, graft thinning was insignificant after 12 months in AM-UC without further incidence of tube exposure.

Compared to the literature, AM-UC represents one of the lowest rates of graft thinning among all patch grafts that have been used for primary tube coverage and measured both clinically and with AS-OCT. Smith et al.¹⁵ retrospectively compared pericardium (n=23), to sclera (n=23) and dura mater (n=18), and reported patch graft thinning/melting of 26% in all grafts over 2 years. Raviv et al.¹² reported an 11.4% rate of progressive thinning/melting in pericardium within 10 months. In these studies, progressive thinning and or melting were determined clinically by the disappearance of the opaque graft materials leaving a clearly visible tube beneath intact conjunctiva. Although, gamma-irradiated sterile cornea patch grafts have become a popular choice for tube covering, De Luna et al.²¹ have recently reported a surprisingly high rate of progressive corneal graft thinning over an average period of 1.7 years. They used AS-OCT and found complete melting of the corneal patch graft in 16.6% of cases. It has to be noted that one of the 3 cases with tube exposure in this study had a second exposure with a complete melting of the subsequent corneal patch graft as well.

Consistent with the literature, our results confirmed two main risk factors for graft thinning: the type of patch graft and the number of previous surgeries.²² In contrast, some studies found no significant differences between graft materials nor previous surgeries.^23,24 We also found an association between tube fenestration and graft thinning in both groups. This suggests an effect of early contact with the aqueous humor which contains proinflammatory cytokines and needs further investigation. No significant correlation between graft thinning and the remaining variables such as age, race, ethnicity, eye (right vs. left), lens status, type of previous surgeries, or the follow-up time was observed.

The composite outcome used for sample size determination in this study is consistent with prior studies in this field,^25-27 and based on the endpoint recommendations of the World Glaucoma Association Guidelines for Glaucoma Surgical Trials.²⁸ While there are concerning issues with use of composite outcomes in randomized trials, sometimes they are the only practical way to proceed. A randomized trial to detect a decrease in the rate of tube exposure from 5% to 2.5% with 90% power would require randomization of nearly 2600 patients.²⁹

Limitations of this multicenter study include the variation in surgeons’ techniques and the potential bias due to the opaque nature of pericardium in the control group. Inter-surgeon analysis showed no significant effect on the study results in either group. In addition, patch graft thickness could not be precisely measured using AS-OCT in the pericardium group. However, we overcame this limitation by carefully estimating the average thickness in the pericardium group using the longitudinal views. While ultrasound biomicroscope (UBM) was considered as it may penetrate deeper, the images were less clear and distorted by echoes from the implant’s material. Furthermore, the available UBM, at the time of planning the study, required an eyecup to hold a coupling medium in a supine position, which puts the eye under the risk of post-operative infection and pressure-induced tube erosion owing to the use of eyecups. Therefore, AS-OCT was selected for this study due to several advantages, including the higher resolution and non-contact scanning in the setting position. In other words, AS-OCT was better in terms of patient safety, comfort and compliance in the clinic setting. Another limitation is the comparison tissue used, i.e. pericardium. This study only allows conclusions between AM-UC and pericardium and not other commonly used patch materials such as sclera and cornea. As stated above, previous studies compared pericardium and sclera and showed similar rates of tube exposure and graft thinning. The cornea also carries similar risk of graft thinning. As stated above, we selected the pericardium given its comparable thickness, consistency, availability, and cost.

In conclusion, AM-UC grafts are well-tolerated and offer an excellent alternative to pericardium for safe and stable shunt tube coverage. AM-UC’s high tensile strength, low immunogenicity and host-tissue integration properties improved long-term graft survival and significantly reduced graft thinning, however, we did not find a difference in rates of exposure. AS-OCT is a useful tool in evaluating translucent patch grafts following GDD surgery.