A, Log-rank test, P = .26. B, Log-rank test, P = .002. C, Log-rank test, P = .002.
A, Log-rank test, P = .001. B, Log-rank test, P = .001. C, Log-rank test, P < .001.
eTable 1. Criteria used for definition of success after initial trabeculectomy with adjunctive mitomycin C.
eTable 2. Additional procedures and glaucoma surgeries performed after initial trabeculectomy with mitomycin C in the 2 cohorts.
eFigure 1. Bar graph of course of intraocular pressure after trabeculectomy with mitomycin C in cohorts of patients of African vs European descent.
eFigure 2. Bar graph of number of glaucoma medication at baseline through 5 years of follow-up in cohorts of patients of African vs European descent.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Nguyen AH, Fatehi N, Romero P, et al. Observational Outcomes of Initial Trabeculectomy With Mitomycin C in Patients of African Descent vs Patients of European Descent: Five-Year Results. JAMA Ophthalmol. 2018;136(10):1106–1113. doi:10.1001/jamaophthalmol.2018.2897
What are the outcomes of initial trabeculectomy with mitomycin C in patients of African descent compared with those in patients of European descent?
In this matched cohort study of 222 patients, success rates of initial trabeculectomy with mitomycin C were worse in patients of African descent compared with patients of European descent when more stringent criteria for success were used.
These findings suggest that new approaches for controlling wound healing after trabeculectomy are needed and that the role of nonfiltering glaucoma surgeries in patients of African descent needs to be investigated.
There is evidence that patients of African descent (AD) experience higher surgical failure rate after trabeculectomy without antimetabolites.
To compare outcomes of initial trabeculectomy with mitomycin C in AD patients with those of patients of European descent (ED) and to identify prognostic factors for failure.
Design, Setting, and Participants
In this retrospective matched cohort study, 135 eyes of 105 AD patients were matched with 135 eyes of 117 ED patients by age (within 5 years), surgeon, lens status, and follow-up time (within 1 year) from a single tertiary academic center.
Initial trabeculectomy with mitomycin C.
Main Outcomes and Measures
Criteria A, B, and C defined qualified success rates as final intraocular pressure of 18 mm Hg or less, 15 mm Hg or less, and 12 mm Hg or less, respectively, in addition to 20% or more, 25% or more, and 30% or more reduction of intraocular pressure or reduction of 2 or more medications. Kaplan-Meier survival curves were compared with log-rank test in AD and ED patients, and Cox proportional hazard models were used to estimate the influence of race/ethnicity on surgical success accounting for confounding variables.
Of the 105 AD patients, 56 (53.3%) were female, and the mean (SD) age was 67.5 (10.4) years; of the 117 ED patients, 64 (54.7%) were female, and the mean (SD) age was 68.2 (10.0) years. For AD patients compared with ED patients, the qualified success rates at 5 years for criteria A were 61% and 67%, respectively (difference, 7.3%; 95% CI, 4.4-10.4); for criteria B, 43% and 60% (difference, 17.6%; 95% CI, 15.2-20.0); and for criteria C, 25% and 40% (difference, 15.8%; 95% CI, 11.1-20.5). On multivariable Cox regression analyses, AD was associated with higher failure rate with criteria B and C for qualified success and with all criteria for complete success (ie, no need for medications). Incidence of bleb leaks was higher in the AD group (29 vs 11 eyes; P = .002). Additionally, AD patients required additional glaucoma surgeries more often than ED patients (47 vs 26 eyes; P = .004).
Conclusions and Relevance
African descent was associated with higher failure rates and higher incidence of bleb leaks after initial trabeculectomy with mitomycin C compared with European descent. If this is subsequently shown to be a cause and effect, the findings need to be considered when surgical treatment of glaucoma is contemplated in AD patients.
Trabeculectomy remains the current standard for surgical treatment of uncontrolled glaucoma, especially in patients who need a substantial reduction in intraocular pressure (IOP) or who have very low IOPs.1 Patients of African descent (AD) develop glaucoma earlier and are susceptible to faster progression of glaucomatous optic neuropathy compared with patients of European descent (ED).2-5 Additionally, AD patients experience a higher surgical failure rate and a greater number of postoperative complications after glaucoma filtering surgery compared with ED patients, presumably owing to increased scarring after surgery.6-14 As a result, African race is considered to be an important risk factor for failure of filtration surgery, warranting careful consideration of alternative options for such patients.6,15-17 Adjunctive use of mitomycin C (MMC) improves the success rate of trabeculectomy, but it increases the risk of late bleb leak and bleb-related infection.18,19 To our knowledge, few studies have addressed outcomes of trabeculectomy with adjunctive MMC in AD patients, and long-term data are lacking on this important subject.20 Also, the rates and types of complications after trabeculectomy with MMC have not been well defined in AD patients.
Quiz Ref IDThe current study examines long-term surgical outcomes in AD patients who underwent initial trabeculectomy with adjunctive MMC according to predefined criteria and compares them with a cohort of matched ED patients for age, surgeon, lens status, and length of follow-up. We also explored prognostic factors for failure in the 2 cohorts of patients.
We retrospectively reviewed the clinical database at the Glaucoma Division of the Stein Eye Institute, David Geffen School of Medicine at UCLA, University of California, Los Angeles, to find all AD patients with glaucoma who had an initial trabeculectomy with adjunctive MMC by 1 of the 5 attending glaucoma surgeons between March 1997 and July 2015. A total of 135 eyes from 105 AD patients met the inclusion criteria and were matched to a parallel cohort of 135 eyes from 117 ED patients. Matching was based on age (within 5 years), surgeon, lens status (phakic vs pseudophakic), and follow-up time (within 1 year). The study was approved by the human subject protection committee and the institutional review board at the University of California, Los Angeles, and was carried out in accordance with the tenets of the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. Informed consent was waived because the study was retrospective and low risk.
Patients were selected based on the following inclusion criteria: trabeculectomy performed as the initial incisional glaucoma surgery; use of adjunctive MMC; a diagnosis of primary open-angle glaucoma, chronic angle-closure glaucoma, pseudoexfoliative glaucoma, pigmentary glaucoma, or normal-tension glaucoma; and a minimum follow-up of 3 months. Eyes with diagnoses other than those described above, history of conjunctival surgery, or concomitant cataract extraction were excluded. Demographic data, baseline ocular characteristics, intraoperative and postoperative findings, and complications until last available follow-up or repeated glaucoma surgery were extracted from medical records. Baseline preoperative IOP and number of medications were determined by averaging IOPs and number of medications from the visits in the 3-month time period prior to the surgery date. Snellen visual acuity measurements were converted into logMAR equivalents.21 Leaks were defined as early if a leak occurred 3 months or more before the date of surgery and as late if a leak occurred 3 months or more after the date of surgery.
Surgeries were performed after retrobulbar or peribulbar anesthesia. For limbus-based surgery, a 4-0 silk bridle suture was placed under the superior rectus muscle tendon. A limbus-based conjunctival flap was prepared superiorly, with an incision 10 to 12 mm posterior to the limbus. The conjunctiva and Tenon capsule were dissected anteriorly to the limbus. A large cellulose sponge or multiple sponge pieces soaked with MMC (0.2-0.3 mg/mL) were placed over the intended site of the scleral flap for 1 to 3 minutes. The MMC concentration was based on the routinely used concentration at the time of surgery. Duration of MMC application was determined by each attending surgeon; most eyes (116 of 135 eyes [85.9%] in each cohort) had a 1-minute exposure. A one-half thickness scleral flap (approximately 3 × 2.5 mm) was dissected forward into clear cornea. After resection of an anterior trabecular block (approximately 2 × 1 mm), iridectomy was performed. The scleral flap was closed with two to six 10-0 nylon sutures. Conjunctiva and Tenon capsule were closed in a single layer with running 9-0 Vicryl sutures (Ethicon Inc). For fornix-based surgery, a 6-0 silk traction suture was placed in the superior cornea adjacent to the limbus, and a fornix-based conjunctival flap centered at the 12-o’clock position was prepared. Conjunctiva and Tenon capsule were closed with two to four 9-0 Vicryl sutures anchored at the limbus.
Quiz Ref IDThe primary outcome measure was qualified success rate (ie, success rate with or without medications), and secondary outcome measures were complete success rates (with no need for medications), IOP level, number of medications, postoperative complications, and need for additional glaucoma surgery. Qualified success was defined as follows: criteria A, final IOP of 18 mm Hg or less with either 20% or greater reduction in IOP or reduction of at least 2 medications compared with the preoperative period; criteria B, final IOP of 15 mm Hg or less and either 25% or greater reduction in IOP or reduction of at least 2 medications; and criteria C, final IOP of 12 mm Hg or less and either 30% or greater reduction in IOP or reduction of at least 2 medications (eTable 1 in the Supplement).22 Complete surgical success was similarly defined with the additional requirement of no need for glaucoma medications.
Unpaired t test and Wilcoxon signed rank test were used to compare continuous variables with Gaussian and non-Gaussian distributions, respectively. Categorical variables were compared with χ2 test. Kaplan-Meier survival analyses and log-rank test were used to compare long-term outcomes based on the 3 success criteria between AD and ED cohorts. In a retrospective study, the time intervals between consecutive visits are variable; therefore, the accuracy of the time to failure somewhat depends on length of the time between visits. To address this issue, once the visit at which the failure was established was determined, failure was considered to have occurred in the middle of the preceding time interval. We required confirmation of failure except when failure was identified on the last available visit. Cox proportional hazard regression analysis was used to determine predictive factors for failure. The following potential predictors were evaluated in univariate analyses modeling surgical failure: sex, age, race/ethnicity, family history of glaucoma, history of diabetes, history of hypertension, lens status, baseline IOP, baseline number of medications, central corneal thickness, baseline visual field mean deviation, fornix-based vs limbus-based conjunctival flap, duration of MMC application, number of scleral flap sutures, and early or late bleb leaks. Variables with a 2-sided P value less than .20 or variables clinically known to affect failure after trabeculectomy were included in multivariable analyses. A shared frailty adjustment was used to account for correlation between eyes of the same patient. The proportional hazards assumption was verified by reviewing smoothed plots of scaled Schoenfeld residuals after fitting the final multivariate model and by reviewing log-log plots. All P values were 2-tailed, and a P value less than .05 was considered statistically significant. Analyses were not adjusted for multiple comparisons. Statistical analyses were performed using R version 3.4.1 (The R Foundation) and Stata version 14.0 (StataCorp).
A total of 135 eyes of 105 AD patients and 135 eyes of 117 ED patients were eligible. Table 1 describes demographic and clinical characteristics of the enrolled patients. Of the 105 AD patients, 56 (53.3%) were female, and the mean (SD) age was 67.5 (10.4) years; of the 117 ED patients, 64 (54.7%) were female, and the mean (SD) age was 68.2 (10.0) years. The baseline mean (SD) IOP was 19.4 (7.1) mm Hg in eyes of AD patients and 18.1 (7.2) mm Hg in eyes of ED patients (P = .16). The mean (SD) number of baseline glaucoma medications was 3.3 (1.0) for AD patients and 3.0 (1.1) for ED patients (P = .02). The mean (range) follow-up time for AD patients was 4.8 (0.3-13.5) years and for ED patients was 5.2 (0.3-13.0) years. Surgery was carried out with a fornix-based conjunctival flap in 114 eyes (84.4%) in the AD cohort and in 120 eyes (88.9%) in the ED cohort. Mitomycin C was used in 3 different concentrations (0.2, 0.25, and 0.3 mg/mL). Duration of MMC application in the 2 cohorts is summarized in Table 1. A total of 94 eyes (69.6%) in the AD cohort and 123 eyes (91.1%) in the ED cohort had 2 scleral flap sutures placed. No intraoperative complications occurred in either cohort.
The most common complication in AD patients was bleb leak (29 eyes [21.5%]), of which 21 were early leaks and 11 were late leaks; bleb leaks occurred less frequently in ED patients (11 eyes [8.1%]) (Table 2). Cataract extraction was performed postoperatively in 76 eyes (56.3%) in the AD cohort and 64 eyes (47.4%) in the ED cohort. The most common additional glaucoma surgery in AD patients was repeated trabeculectomy (23 AD vs 10 ED eyes) (eTable 2 in the Supplement). The total number of additional procedures was greater in AD patients than in ED patients (47 of 135 [34.8%] vs 26 of 135 [19.3%]) (eTable 2 in the Supplement). For both cohorts, the reductions in IOP and glaucoma medications after surgery were significant at all visits compared with baseline (eFigures 1 and 2 in the Supplement).
Kaplan-Meier survival curves demonstrate the cumulative probabilities of qualified and complete surgical success over the first 5 years of follow-up (eFigures 1 and 2 in the Supplement), as the number of remaining eyes in each group became small at that point. Based on criteria A, mean (SD) qualified success rates at the first, third, and fifth years of follow-up were 82.0% (3.3), 66.0% (3.8), and 61.0% (4.7), respectively, for AD patients and 90.0% (2.6), 75.7% (3.9), and 67.0% (4.5) for ED patients (difference at 5 years, 7.4%; 95% CI, −5.6 to 20.3; P = .26) (Figure 1). The mean (SD) complete surgical success rates at the first, third, and fifth years of follow-up based on criteria A were 62.7% (4.2), 39.0% (4.6), and 30.1% (4.6), respectively, for AD patients and 78.0% (3.6), 62.2% (4.4), and 51.0% (4.9) for ED patients (difference at 5 years, 20.9%; 95% CI, 7.8-34.0; P = .001) (Figure 2).
Based on criteria B, mean (SD) qualified success rates at the first, third, and fifth years of follow-up were 71.0% (3.9), 52.0% (4.5), and 43.0% (4.8), respectively, for AD patients and 83.0% (3.2), 65.0% (4.3), and 60.0% (4.7) for ED patients (difference at 5 years, 17.6%; 95% CI, 4.0-30.2; P = .002). The mean (SD) complete surgical success rates at the first, third, and fifth years of follow-up based on criteria B were 59.7% (4.2), 36.1% (4.4), and 27.7% (4.4), respectively, for AD patients and 74.8% (3.7), 56.8% (4.5), and 50.6% (4.7) for ED patients (difference at 5 years, 22.9%; 95% CI, 10.3-35.6; P < .001).
Based on criteria C, mean (SD) qualified success rates at the first, third, and fifth years of follow-up were 57.0% (4.2), 38.0% (4.3), and 25.0% (4.1), respectively, for AD patients and 69.0% (4.0), 56.0% (4.4), and 40.0% (4.8) for ED patients (difference at 5 years, 15.8%; 95% CI, 3.5-28.2; P = .002). The mean (SD) complete surgical success rates at the first, third, and fifth years of follow-up based on criteria C were 50.8% (4.3), 27.0% (4.0), and 17.1% (3.6), respectively, for AD patients and 65.9% (4.1), 50.1% (4.5), and 38.3% (4.7) for ED patients (difference at 5 years, 20.2%; 95% CI, 8.6-31.7; P < .001).
Cox proportional hazard regression analysis was used to determine prognostic factors for failure according to criteria A, B, and C (Table 3). For qualified success based on the final multivariable model, AD was a risk factor for failure with criteria B (hazard ratio [HR], 1.73; 95% CI, 1.20-2.48) and criteria C (HR, 1.87; 95% CI, 1.36-2.58). Early wound leak was a failure risk factor for criteria B (HR, 1.72; 95% CI, 1.02-2.89). Male sex (HR, 1.56; 95% CI, 1.13-2.14) and glaucoma type other than primary open-angle glaucoma (HR, 1.90; 95% CI, 1.34-2.68) were additional risk factors for failure according to criteria C (Table 3). For complete surgical success, AD (HR range, 1.77-2.07) and glaucoma type other than primary open-angle glaucoma (HR range, 1.55-1.86) were failure risk factors with all 3 sets of criteria. Additionally, early wound leak (HR, 1.81; 95% CI, 1.11-2.95) was a risk factor with criteria B and male sex (HR, 1.53; 95% CI, 1.12-2.08) was a risk factor with criteria C.
Quiz Ref IDWe found that AD patients demonstrated a higher propensity for failure with more stringent criteria, which amounted to 77% to 107% increased risk on multivariate analyses. Other risk factors consisted of non–primary open-angle glaucoma diagnosis (55% to 86% higher failure rate with criteria A, B, and C for complete surgical success and 90% higher failure rate with criteria C for qualified success) and early bleb leak (81% higher failure rate with criteria B for complete surgical success and 72% higher failure rate with criteria B for qualified success). With the most lenient criteria (qualified success with criteria A), initial trabeculectomy had similar 5-year efficacy in AD and ED patients (66% vs 67%). We also found a higher incidence of early and late bleb leaks in AD patients.
The earlier onset, higher incidence, greater severity, and faster progression of glaucoma have been well documented in AD patients.23-37 Studies have shown that trabeculectomy is an effective treatment for uncontrolled glaucoma in AD patients.3,17,38,39 One of the main findings of the Advanced Glaucoma Intervention Study16 was that AD patients fared better when the initial surgical approach was argon laser trabeculoplasty. This could be attributed to the higher failure rate of trabeculectomy without adjunctive MMC in AD patients (18%) than white patients (13%).
Use of intraoperative and postoperative antimetabolites has enhanced efficacy of trabeculectomy and improved long-term bleb survival.40 However, preliminary evidence suggests that AD patients remain at higher risk for failure following filtration surgery.41 In a 2012 series of 47 African Caribbean patients, Shah et al20 showed a survival rate of 86% at 36 months when an IOP of 21 mm Hg or less was used as success criterion. Three independent risk factors for failure were preoperative use of acetazolamide, pseudophakia, and higher preoperative IOP.42 Two small studies comparing outcomes of placement of the Ex-PRESS mini shunt with MMC in AD and ED patients disagreed on whether AD patients had worse outcomes compared with ED patients.43,44
Use of multilevel criteria for success after glaucoma surgery in this study reflects the goals of contemporary treatment strategies. This especially holds true for patients whose glaucoma has progressed at relatively low IOPs and for patients with severe glaucomatous optic neuropathy. Our results confirmed that trabeculectomy has a higher failure rate in AD patients compared with ED patients. The qualified success rates in AD patients were worse than in ED patients through at least 5 years of follow-up with more stringent criteria that are consistent with current practices of glaucoma treatment (ie, criteria B and C). Patients of African descent had higher failure rates with all complete success criteria. These patients also had higher IOPs and number of medications along with a higher need for additional glaucoma surgeries. This is consistent with the relevant literature on outcomes of filtering surgery without antimetabolites in AD patients. Broadway et al44 found a higher number of fibroblasts and macrophages in the superficial and deeper layers of conjunctival substantia propria of AD patients than in ED patients. This partly explains the higher propensity for scar formation and failure after filtration surgery in AD patients.
Early postoperative bleb leak is a relatively common complication after trabeculectomy, and there is wide variation in the reported incidence (0% to 30%).45 It is more common with fornix-based than limbus-based flaps.45,46 One study found that early postoperative leak in eyes with fornix-based conjunctival flaps did not affect outcomes of trabeculectomy.46 In our study, early bleb leak occurred more frequently in eyes with a fornix-based flap. However, a higher incidence of any leaks was observed in the AD cohort regardless of conjunctival flap type. Presence of early wound leak was associated with higher failure rates with criteria A (P = .03) but not criteria B (P = .056). Both early and late bleb leaks occurred more frequently in the AD cohort. We considered the possibility that a higher concentration or longer duration of MMC application could have been used in AD patients, which might have accounted for an increased risk of leaks; however, we found no difference between these parameters in the 2 cohorts. Alternatively, the clinicians might have been more aggressive postoperatively in AD patients aiming for lower pressures, thus causing higher early and late bleb leaks. Another possibility is that faster healing or scarring in AD patients could happen in a fornix-to-limbus direction. Therefore, the aqueous could not permeate more posteriorly and would be forced toward the limbal wound, resulting in more frequent leakage.
The number of scleral flap sutures influences postoperative filtration and IOP variation. A greater number of flap sutures allows for greater control of IOP reduction postoperatively by allowing smaller incremental drops in IOP after suture lysis. However, well-constrained flaps restrict aqueous flow that might result in increased scarring if timely suture lysis is not done.47 A larger proportion of eyes from ED patients had the scleral flap sutured with only 2 nylon sutures (91% vs 70%); nonmeasurable variations in tightness of flap sutures may have influenced the amount of flow resistance.
Quiz Ref IDOur study had limitations. We addressed the limitations of a retrospective study by matching the 2 cohorts on important confounding variables for failure, including age, surgeon, and lens status. Other sources of bias in a retrospective setting cannot be completely ruled out. A shortcoming of our study is that we could not verify whether cataract surgery prior to trabeculectomy had been done through a small clear-cornea incision in pseudophakic eyes. However, it is likely that most, if not all, of the enrolled patients with pseudophakic eyes underwent phacoemulsification through a small clear-cornea incision, as was customary in this part of the country. Another limitation of this study is that race/ethnicity was self-identified by the study participants, as recorded in the medical record. In the cohorts included, no patients identified themselves as being of mixed race. It has been proposed that rather than race/ethnicity itself, some associated characteristics are more likely related to outcomes of interest.35,47 While we were able to explore some clinical factors, such as central corneal thickness, socioeconomic and behavioral factors (including education, socioeconomic status, and adherence to follow-up/visits) could not be adjusted for. We were also not able to study bleb appearance and presence and extent of bleb pallor in a retrospective study.
In summary, in this retrospective longitudinal observational study, AD patients demonstrated a higher failure rate than ED patients after initial trabeculectomy with MMC with more stringent success criteria. They also demonstrated more frequent bleb leaks. These findings have important implications for management of glaucoma in AD patients. Nonfiltering glaucoma surgeries or new approaches to controlling wound healing after trabeculectomy should be considered in AD patients.
Accepted for Publication: May 23, 2018.
Corresponding Author: Kouros Nouri-Mahdavi, MD, MSc, Stein Eye Institute, David Geffen School of Medicine at UCLA, University of California, Los Angeles, 100 Stein Plaza, Los Angeles, CA 90095 (firstname.lastname@example.org).
Published Online: July 19, 2018. doi:10.1001/jamaophthalmol.2018.2897
Author Contributions: Mr Nguyen and Dr Nouri-Mahdavi had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Law, Caprioli, Nouri-Mahdavi.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Nguyen, Fatehi, Miraftabi, Caprioli, Nouri-Mahdavi.
Critical revision of the manuscript for important intellectual content: Nguyen, Fatehi, Romero, Kim, Morales, Giaconi, Coleman, Law, Caprioli, Nouri-Mahdavi.
Statistical analysis: Nguyen, Fatehi, Morales, Law, Caprioli, Nouri-Mahdavi.
Administrative, technical, or material support: Nguyen, Fatehi, Romero, Miraftabi, Law, Nouri-Mahdavi.
Study supervision: Nguyen, Law, Caprioli, Nouri-Mahdavi.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Coleman has received personal fees from Alcon and Glaukos. Dr Nouri-Mahdavi has received research support and nonfinancial support from Heidelberg Engineering. No other disclosures were reported.
Funding/Support: This study was supported by an unrestricted departmental grant from Research to Prevent Blindness.
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.