JOURNAL OF OCULAR PHARMACOLOGY AND THERAPEUTICS Volume 28, Number 2, 2012 ª Mary Ann Liebert, Inc. DOI: 10.1089/jop.2011.0117

Intraoperative Versus Intraoperative 5- for Trabeculectomy: A Systematic Review and Meta-Analysis

Zhong-Jie Lin,1,* You Li, M.D.,1,2,* Jin-Wei Cheng, M.D.,1 and Xiao-He Lu, M.D.2

Abstract

Purpose: The aim of this study was to assess the intraoperative application of mitomycin C (MMC) compared to 5-fluorouracil (5-FU) on the outcome of trabeculectomy and to examine the balance of risk and benefit. Methods: Pertinent studies were selected through systematic searches of major literature databases, including the Cochrane Library, PubMed, Embase, and Chinese Biomedicine Database. Internet searches of search engines, the professional associations’ websites, and the manufacturers’ databases were also performed. Clinical con- trolled trials comparing 5-FU with MMC in trabeculectomy were selected. The primary efficacy measure was the weighted mean difference (WMD) in percentage intraocular pressure reduction (IOPR%) at follow-up end point. The secondary efficacy measure was the relative risk (RR) for ‘‘qualified’’ (with or without medical therapy) success of trabeculectomy at follow-up end point. The third efficacy measure was RR for ‘‘complete’’ (without medical therapy) success of trabeculectomy at follow-up end point. The fourth efficacy measure was RR for adverse events, including wound leak, hypotony, endophalmitis, and shallow anterior chamber (AC). The pooled effects were calculated using the random effects model by RevMan version 5.0 software. Results: Eight studies enrolling a total of 536 patients were included in the meta-analysis. MMC was associated with significantly more IOPR% compared with 5-FU, with a WMD of 7.09 [95% confidence interval (CI) 1.47– 12.70] at follow-up end point (P = 0.01). MMC was comparable with 5-FU in qualified success rate, with a RR of 1.09 (0.99–1.20) at follow-up end point (P = 0.09). MMC was comparable with 5-FU in complete success rate, with a RR of 1.17 (0.79– 1.75) at follow-up end point (P = 0.43). Rates of adverse events did not differ significantly between 5-FU and MMC, with an RR of 0.71 (0.22–2.28) for bleb leakage, 1.40 (0.72–2.72) for hypotony, 1.63 (0.27–9.75) for endophthalmitis, and 0.95 (0.41–2.21) for shallow AC. Conclusions: Intraoperative MMC is more effective in lowering IOP in trabeculectomy compared with in- traoperative 5-FU, but is comparable with intraoperative 5-FU in both qualified and complete success rate. Intraoperative use of both agents may contribute equally to adverse events.

Introduction surgery to prevent the conjunctiva healing onto the sclera. The alkylating properties of MMC inhibit DNA replication, n trabeculectomy, the problem to be resolved is the which led to its use first as an anticancer drug.2 5-FU is a Ihealing response, which leads to failure of some operated DNA topoisomerase I inhibitor and works through irre- eyes after a period of time. To resolve this problem and fa- versible inhibition of thymidylate synthase.3 cilitate intraocular pressure (IOP)-lowering efficacy, various MMChasbeenreportedinrandomizedcontrolledtrials4–12 antifibrosis agents have been used in trabeculectomy. Mito- (RCTs) to reduce mean IOP at 12 months. Postoperative 5-FU mycin C (MMC) and 5-fluorouracil (5-FU) were first used to implications have been assessed with RCTs.13–16 A system- modify the wound healing response in the early 1980s.1 atic review has proved that MMC reduces mean IOP at 12 MMC is a drug used during the initial stages of months17 and RCTs for 5-FU have proved IOP reduction

1Department of Ophthalmology, Shanghai Changzheng Hospital, Second Military Medical University, Shanghai, China. 2Department of Ophthalmology, Zhujiang Hospital, Southern Medical University, Guangzhou, China. *These authors contributed equally to this work.

166 MMC VERSUS 5-FU FOR TRABECULECTOMY 167 at 1 year compared to placebo.18 Khaw suggested that a single intraoperative application of 5-FU might be sufficient to control postoperative proliferation of scar tissue at the Baseline

19 (mm Hg) drainage site. Clinicians now appear to prefer the in- mean IOP traoperative application of agents for the modification of wound healing, and routine postoperative injections of 5-FU Sex are rarely used. Therefore, we focused on intraoperative ra- (M/F) ther than perioperative use of 5-FU in this review. The results of two reviews comparing MMC and 5-FU to age Mean placebo have suggested a similar effect for the two agents in (years) inhibiting scarring after trabeculectomy.17,18 Some clinicians retrospective; Pro, prospective

consider MMC more powerful and some prefer 5-FU. A no. eyes previous review suggested that MMC results in a greater Total decrease in IOP than 5-FU.20 However, a recent meta-analysis

found no statistically significant difference between the in- no. 21 Total traoperative application of these two antifibrosis agents. patients Clinically, these two agents seem to have different results. Therefore, to evaluate the efficacy and safety of these two antifibrosis agents, we undertook a systematic review and meta-analysis focusing on intraoperative MMC and in- traoperative 5-FU use in all relevant controlled clinical trials. End point length (month) Materials and Methods Time This meta-analysis was performed according to a pre- (min) determined protocol described in the following paragraph, and standard systematic review techniques, as outlined by MMC Con

the Cochrane Reviewers’ Handbook, were followed at all Intraoperative stages of the process.22 (mg/mL)

Outcome measures

Time (min) Theprimaryoutcomewasthepercentagereductionfrom preoperative to postoperative in IOP (IOPR%). When authors 5-FU

reported mean and standard deviation (SD) of IOP and IOPR, Con Intraoperative we used them directly. When not available, we computed them (mg/mL) according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions: IOPR = IOPbaseline - 2 2 IOPend point and SDIOPR = (SD baseline + SD end point –SDbaseline · operation Combined Baseline Characteristics of Eligible Controlled Clinical Trials 1. Table Random SingleRetroRandom USA MultipleRetro Single USARetroPro West Single AfricaPro Single No NoRandom USA No Single Single USA Single USA India Korea No 50 50 PHACO 50 No No 5 No 5 50 5 50 0.5 0.2 0.4 5 50 50 5 50 0.2–0.5 3 2 0.2 1 2 5 2/5 5 0.2/0.4 0.5 84 0.2 3/5 58 18 12 1 3.5 36 5 103 68 108 57 16 115 12 68 108 57 65.5 63 6 65.5 65.7 64/51 75 34/34 16 73 NA 81 23.1 NA 30.3 62.5 32 40 24.8 28/35 81 47.1 19.3 50.4 55 24.5 9/7 52.3 NA 29.6 NA 31.1 30.5 24 28 29 27 26 31 30 25

FIG. 1. The selection flowchart of clinical trials included in F-FU, 5-fluorouracil; MMC, mitomycin C; Con, concentration; M/F, male/female; IOP, intraocular pressure; Random, prospective randomized; Retro, nonrandomized; NA, not applicable; PHACO, phacotraceculectomy. Kim et al. 2008 Singh et al. 2000 Smith et al. 1997 Budenz et al. 1999 Vijaya et al. 2000 SeiY et al. 1994 Singh et al. 1997 Palanca-Capistrano 2009 the present meta-analysis. Trial (reference) Design Center Location 168 LIN ET AL.

Table 2. Quality Scoring Components for 8 Cinical Trials Included

Quality score component Score

First Author (year)Ref. I II III IV V Over all Percentage (%)

Palanca-Capistrano (2009)24 11 2 7 4 4 28 87.50 Kim (2008)25 11 2 4 1 3 21 65.63 Singh (2000)26 11 2 7 5 4 29 90.63 Smith (1997)27 11 2 4 1 3 21 65.63 Budenz (1999)28 10 2 4 1 2 19 59.38 Vijaya (2000)29 11 2 4 1 1 19 59.38 SeiY (1994)30 7 0 4 0 2 13 40.63 Singh (1997)31 11 2 6 3 4 26 81.25

1/2 SDend point) .IOPR%= IOPR/IOPbaseline and SDIOPR% = undergone trabeculectomy or phacotrabeculectomy; (3) SDIOPR/IOPbaseline. intervention—intraoperative MMC was administered at any When the difference in means (MD) and its t-value [t- concentration and dose, and was compared with in- value also can be obtained from a computer by entering = traoperative 5-FU of any concentration and dose; (4) outcome tinv (P value, Ntreat + Ncontrol - 2) into any cell in a Microsoft variables—at least one of the following outcome variables, Excel spreadsheet] were reported, SD = MD - t - 1 · IOPR%, complete success rate, and qualified success rates. - 1 - 1 - 1/2 (N treat + N control) . Studies involving other types of glaucoma surgery, such as For efficacy, the proportion of qualified success and nonpenetrating glaucoma surgery, were excluded. Other ex- complete success was also used. Qualified success was de- clusion criteria included repeated glaucoma surgery. fined as target end point IOP with or without medications. After completion of the searches, two review authors Compete success was defined as target end point IOP (Z.J.L., Y.L.) working independently assessed the titles and without medications. The fourth outcomes were adverse abstracts of all obtained reports for a rough judgment of an event rates in either group, including wound leaks, hypot- article’s eligibility. The full-text copies of possibly and defi- ony, late endophthalmitis, and shallow anterior chamber. nitely relevant trials were obtained and assessed by the two authors independently according to the definitions in the Search strategy criteria, which were checked by J.W.C. Only trials meeting

these criteria were assessed for methodological quality. For Clinical trials comparing intraoperative MMC use with in- the publications reporting on the same study population, the traoperative 5-FU use were identified through a systematic article reporting the results of the last end point was in- search of the Cochrane Library, PubMed, and EMbase. A broad cluded, and data that could not be obtained from this pub- search strategy combined terms related to MMC (including lication were obtained from others. MeSH search using mitomycins and keyword search using The observers were blinded to the names of the authors words mmc, mutamycin, m?tom?cin, mitoc?n and amet?cin?), and their institutions, the names of the journals, sources of terms related to 5-FU (including MeSH search using fluoro- funding, and acknowledgments, as well as the financer of the uracil and keyword search using words fluorouracil, flur- study. acilium, fluracil, fluoruracil and flu), and terms related to surgery (including MeSH search using exp Filtering-Surgery Data extraction and Glaucoma-Surgery, and keyword search using words trabeculectom, sclerostom, filter near surg, filtrat near surg and Data extraction was performed by two reviewers (Z.J.L., glaucoma near surg), with a filter to restrict results to clinical Y.L.) independently. Any disagreement was resolved by dis- trial, meta-analysis and randomized controlled trial. The orig- cussion. For each study and each type of treatment, the fol- inal search was performed in January, 2011. Regular alerts lowing data were extracted: The authors of the study, the year every 3 months were established on PubMed and Embase of publication, information on study design, location of trial, databases to capture new studies until June, 2011, and updated length of study, number of subjects, patient age, sex, and IOP searches on the Cochrane databases regularly. measurements. Numeric discrepancies between the two in- The internet was searched using the Google and Yahoo! dependent data extractions were resolved after discussion. search engines and using the terms 5-fluorouracil and mi- tomycin C. A manual search was performed by checking the Qualitative assessment reference lists of original reports and review articles, re- trieved through the electronic searches, to identify studies The qualities of clinical trials included were assessed by not yet included in the computerized databases. two independent observers (Z.J.L. and Y.L.) using a previ- ously reported quality assessment system that was for both 23 Trials selection randomized and nonrandomized studies. The system contained 27 items distributed between five subscales about Published and unpublished trials fulfilling the following reporting (10 items), external validity (3 items), bias (7 items), selection criteria were included in the present meta-analysis: confounding (6 items), and power (1 item). Any discrepancy (1) Study design—controlled clinical studies, including co- in the qualitative assessment between the two observers was hort studies or randomized trials; (2) population—patients discussed and a consensus was reached. The total score of with glaucoma, including primary and secondary, who were each trial was expressed as a percentage of the maximum MMC VERSUS 5-FU FOR TRABECULECTOMY 169

FIG. 2. Funnel plot of stud- ies comparing intraoperative mitomycin C (MMC) and intraoperative 5-fluorouracil (5-FU) on qualified success rate (left) and percentage in- traocular pressure reduction (IOPR%, right). RR, relative rate; MD, mean difference (both calculated in a random- effects model); SE, standard error.

achievable score. Good quality refers to a quality score not teristics of the eligible studies are summarized in Table 1. A lower than 50%. total of 589 eyes of 536 patients were enrolled. The duration of follow-up ranged from 6 months to 84 months. Mean age Statistical analysis ranged from 47 to 75 years. Among the 262 patients whose sex were available,24,25,27,29 135 were male and 127 were fe- Not all of the trials reported on all the outcomes of inter- male. Trabeculectomy with intraoperative MMC or 5-FU was est. For each comparison and outcome, we undertook sepa- undergone in 7 trials, and phacotrabeculectomy was under- rate meta-analyses. Outcome measure was assessed on an gone in 1 trial. Three trials had a prospective, parallel, ran- intent-to-treat (ITT) basis. Considering the different clinical domized design; 2 had a prospective, nonrandomized characteristics among study groups and the variation of design; and 3 had a retrospective, nonrandomized design. sample sizes, we assumed that heterogeneity was present, The quality assessment is summarized in Table 2. Funnel even when no statistical significance was identified, and we plots on qualified success rate was asymmetric while funnel decided to combine data by using a random-effects model to plots on IOPR was symmetric (Fig. 2), suggesting probable achieve more conservative estimates. publication bias. For dichotomous outcomes, relative risk (RR) was esti- mated. Weighted mean difference (WMD) was calculated for continuous outcomes. Analyses were carried out in RevMan Percentage of IOPR version 5.0 software (Cochrane Collaboration, Oxford, United The percentage of IOPR comparing MMC with 5-FU was Kingdom). The results were reported with 95% confidence reported. Five studies reported data related to IOPR% at < intervals (CIs). A P value 0.05 was considered statistically various time. We chose the IOPR at follow-up end point (for 1 significant on the test for overall effect. To detect publication trial we chose the figure of the 12-month to reduce bias, the biases, asymmetry in funnel plots was visually examined. others of follow-up end point). MMC was found more effec- tive than 5-FU in lowering IOP. The differences in IOPR% Sensitivity analysis comparing MMC with 5-FU were statistically significant, with Sensitivity analysis was undertaken to evaluate the effect of WMD 7.09 (95% CI, 1.47–12.70). For the subgroup including methodological characteristics of controlled clinical trials in RCTs, the differences in IOPR% were also statistically signif- term of trial design, which was differentiated as retrospective, icant (WMD 3.70, 1.23–6.17). For the subgroup including pro- prospective nonrandomized, and randomized. Another sensi- nonrandomized clinical trials, the differences in IOPR% were tivity analysis was undertaken to evaluate the effect of baseline also statistically significant (WMD 9.61, 2.94–16.28). For the of controlled clinical trials in term of operation, which was subgroup including retrospective clinical trials, the differences differentiated as phocotrabeculectomy and trabeculectomy. in IOPR% were not statistically significant (WMD 7.53, - 1.39–16.46). For the phacotrabeculectomy subgroup, the differences in IOPR% were not statistically significant (WMD Results 2.00, - 1.41–5.41). When excluding the study that phaco- The selection flow of controlled clinical trials is shown in trabeculectomy was undergone, the differences in IOPR% Fig. 1. We reviewed the full text of 12 articles from 118 were statistically significant (WMD 8.46, 1.99–14.93) (Table 3). studies identified from our initial literature search and hand search. A total of 9 studies met criteria for inclusion, and Qualified success rate there was one duplicate publication. Therefore, 8 clinical trials involving 536 patients were included in the final meta- Eight studies reported the proportions of patients achiev- analysis.24–31 ing target end point IOP with or without medications at follow-up end point; the difference in qualified success rate Trials characteristics, populations, and quality between the MMC group and 5-FU group was not statisti- cally significant (pooled RR 1.09, 0.99–1.20). For the subgroup The trials were conducted in various countries, including including RCTs, the difference in qualified success rate be- West Africa, United States, India, and Korea. The charac- tween the MMC group and 5-FU group was not statistically 170 LIN ET AL.

Table 3. IOPR% Comparing the MMC Group with the 5-FU Group

MMC 5-FU

Mean (SD) Mean (SD) WMD WMD (random) Weight (random) Trial No. (%) No. (%) 95% CI (%) 95% CI (%)

All Palanca-Capistrano et al.24 58 55.96(5.92) 57 52.26(7.50) 22.10 3.70 (1.23 to 6.17) Kim et al.25 30 49.20(14.48) 38 34.11(14.48) 17.10 6.09 ( - 0.84 to 13.02) Smith et al.27 36 63.79(6.59) 37 61.79(8.20) 21.30 2.00 ( - 1.41 to 5.41) Budenz et al.28 29 27.57(4.52) 28 13.43(4.81) 22.10 14.14 (11.72 to 16.56) Vijaya et al.29 16 64.65(11.25) 16 55.04(7.65) 17.40 9.61 (2.94 to 16.28) Subtotal 169 176 100.00 Heterogeneity: Tau2 = 35.59, chi2 = 48.40, df = 4 P < 0.00001, I2 = 92% Test for overall effect: 7.09 (1.47 to 12.70) Z = 2.47, P = 0.01 Randomized Palanca-Capistrano et al.24 58 55.96(5.92) 57 52.26(7.50) 100.00 3.70 (1.23 to 6.17) Heterogeneity: not applicable Test for overall effect: Z = 2.38, P = 0.02 Pro-nonrandomized Vijaya et al.29 16 64.65(11.25) 16 55.04(7.65) 100.00 9.61 (2.94 to 16.28) Heterogeneity: not applicable Test for overall effect: Z = 2.83, P = 0.005 Retrospective Kim et al.25 30 40.20(14.48) 38 34.11(14.48) 29.90 6.09 ( - 0.84 to 13.02) Smith et al.27 36 63.79(6.59) 37 61.79(8.20) 34.60 2.00 ( - 1.41 to 5.41) Budenz et al.28 29 27.57(4.52) 28 13.43(4.81) 35.50 14.14 (11.72 to 16.56) Subtotal 95 103 100.00 Heterogeneity: Tau2 = 56.90, chi2 = 33.52, df = 2 P < 0.00001, I2 = 94% Test for overall effect: 7.53 ( - 1.39 to 16.46) Z = 1.65, P = 0.10 Phaco-exluded Palanca-Capistrano et al.24 58 55.96(5.92) 57 52.26(7.50) 27.90 3.70 (1.23 to 6.17) Kim et al.25 30 40.20(14.48) 38 34.11(14.48) 21.80 6.09 ( - 0.84 to 13.02) Vijaya et al.29 16 64.65(11.25) 16 55.04(7.65) 22.20 9.61 (2.94 to 16.28) Budenz et al.28 29 27.57(4.52) 28 13.43(4.81) 28.00 14.14 (11.72 to 16.56) Subtotal 133 139 100.00 Heterogeneity: Tau2 = 37.43, chi2 = 35.61, df = 3, P < 0.00001, I2 = 92% Test for overall effect: 8.46 (1.99 to 14.93) Z = 2.56, P = 0.01 Phaco Smith et al.27 36 63.79(6.59) 37 61.79(8.20) 100.00 2.00 ( - 1.41 to 5.41) Heterogenity: Not applicable Test for overall effect: Z = 1.15, P = 0.25

MMC, mitomycin C; 5-FU, 5-fluorouracil; SD, standard deviation; WMD, weighted mean difference (random-effect model); CI, confidence interval. WMDs more than zero denoted an advantage for mitomycin C, and those less than zero denoted an advantage for 5-fluorouracil; 95%CIs of WMDs not including 0 denoted a statistically significant advantage. significant (RR 1.07, 0.94–1.21). For the subgroup including the phacotrabeculectomy subgroup, the difference in quali- retrospective clinical trials, the difference in qualified success fied success rate between the MMC group and 5-FU group rate between the MMC group and 5-FU group was not sta- was not statistically significant (RR 2.00, - 1.41–5.41). When tistically significant (RR 1.22, 0.68–2.19). For the subgroup excluding the study in which phacotrabeculectomy occurred, including pro-nonrandomized clinical trials, the difference in the difference in qualified success rate between the MMC qualified success rate between the MMC group and 5-FU group and 5-FU group was not statistically significant, either group was not statistically significant (RR 1.06, 0.86–1.30). For (pooled RR 1.10, 1.00–1.22) (Table 4). MMC VERSUS 5-FU FOR TRABECULECTOMY 171

Table 4. Qualified Success Rate Comparing MMC Group with 5-FU Group

MMC 5-FU

No. No. Qualified No. Qualified RR Test for Test for of trials of eyes success rate of eyes success rate 95% CI heterogeneity overall effect

All 8 296 0.91 293 0.82 1.09 (0.99, 1.20) Q = 26.71, P = 0.0004 Z = 1.67, P = 0.09 Randomized 3 156 0.92 148 0.86 1.07 (0.94, 1.21) Q = 4.72, P = 0.09 Z = 1.07, P = 0.29 Retrospective 3 95 0.88 103 0.74 1.22 (0.68, 2.19) Q = 38.55, P < 0.00001 Z = 0.65, P = 0.51 Prospective 2 45 0.93 42 0.86 1.06 (0.86, 1.30) Q = 2.48, P = 0.12 Z = 0.54, P = 0.59 Phaco-ex 7 260 0.90 256 0.79 1.10 (1.00, 1.22) Q = 14.30, P = 0.03 Z = 1.91, P = 0.06 Phaco 1 36 1.00 37 1.00 2.00 (-1.41, 5.41) Not applicable Z = 1.15, P = 0.25

MMC, mitomycin C; 5-FU, 5-fluorouracil; RR, relative risk (calculated using a random-effect model); CI, confidence interval; Phaco, phacotrabeculectomy.

Complete success rate Discussion

Three studies reported the proportions of patients The present systematic review suggested that in- achieving target end point IOP without medications at fol- traoperative MMC is more statistically effective in IOP low- low-up end point, the difference in complete success rate ering in trabeculectomy compared with intraoperative 5-FU, between the MMC group and 5-FU group was not statisti- but comparable with intraoperative 5-FU in both qualified cally significant (pooled RR 1.17, 0.79–1.75). For the sub- and complete success rate. Both agents contribute equally to group including retrospective clinical trials, the difference in adverse events. complete success rate between MMC group and 5-FU group Abdu et al. analyzed data from 9 clinical trials and re- was not statistically significant, either (RR 1.46, 0.50–4.25). ported that intraoperative MMC is comparable with perio- For the subgroup including pro-nonrandomized clinical tri- perative 5-FU in mean IOP and success rate.21 In contrast to als, the difference in complete success rate between the MMC that analysis, we focused on intraoperative MMC and in- group and 5-FU group was not statistically significant, either traoperative 5-FU instead of perioperative 5-FU and reviewed (RR 1.00, 0.89–1.12) (Table 5). 8 controlled clinical studies in trabeculectomy, comparing intraoperative MMC with intraoperative 5-FU, and found Side effects that intraoperative MMC application was associated with greater IOP-lowering efficacy, with statistically significant No significant differences comparing between MMC and differences in IOP reduction compared with intraoperative 5- 5-FU were found in the incidence of bleb leakage, hypotony, FU at follow-up end point. However, both groups achieved endophthalmitis, and shallow anterior chamber, with the comparable qualified success rate at follow-up end point. pooled RRs being 0.71 (0.22–2.28), 1.40 (0.72–2.72), 1.63 The first strength of the present analysis is that we focused (0.27–9.75), and 0.95 (0.41–2.21), respectively (Table 6). on direct comparison between intraoperative MMC and in- traoperative 5-FU rather than indirect comparison between Sensitivity analysis perioperative MMC and perioperative 5-FU. Furthermore, in Trials designed as retrospective, prospective non- the quality assessment of trials, we used a system for both randomized, and randomized were stratified, and the differ- randomized and nonrandomized trials that was definitely ences of qualified success rate and IOPR% were compared, better than the Jadad scoring system only for randomized respectively. For the subgroup including retrospective clinical trials. Five (62.5%) of all including studies were non- 32 trials, the differences in IOPR% were not statistically signifi- randomized design, which may fail to detect actual results. cant (WMD 7.53, - 1.39–16.46). Results of the rest were sta- The third strength is that we undertook subgroup meta-an- tistically significant. Trials that included trabeculectomy and alyses by only including randomized clinical trials and phocotrabeculectomy were stratified and the differences of subgroup meta-analyses by excluding trials in which pha- qualified success rate and IOPR% were compared, respec- cotrabeculectomy occurred. Two independent co-authors tively. For the phacotrabeculectomy subgroup, the differ- judged the eligibility of articles and extracted data from the ences in IOPR% were not statistically significant (WMD 2.00, eligible articles, with discrepancies resolved after discussion - 1.41–5.41). Results of the rest were statistically significant. by all of the authors. Only the series of the same patient

Table 5. Complete Success Rate Comparing the MMC Group with 5-FU Group

MMC 5-FU

No. of No. of Complete No. Complete RR Test for Test for trials eyes success rate of eyes success rate 95% CI heterogeneity overall effect

All 3 82 0.76 91 0.62 1.17 (0.79, 1.75) Q = 15.69, P = 0.0004 Z = 0.79, P = 0.43 Retro 2 66 0.70 75 0.53 1.46 (0.50, 4.25) Q = 9.76, P < 0.002 Z = 0.70, P = 0.49 Pro 1 16 1.00 16 1.00 1.00 (0.89, 1.12) Not applicable Z = 0.00, P = 1.00

MMC, mitomycin C; 5-FU, 5-fluorouracil; RR, relative risk (calculated using a random-effect model); Retro, retrospective; Pro, prospective. 172 LIN ET AL.

Table 6. Side Effect Comparing MMC Group with 5-FU Group

MMC 5-FU

No. No. No. RR Test for Test for of trials of eyes Rate of eyes Rate 95% CI heterogeneity overall effect

Bleb-leak 4 177 0.0904 176 0.1023 0.71 (0.22, 2.28) Q = 5.59, P = 0.13 Z = 0.58, P = 0.57 Hypotony 5 214 0.0935 202 0.0644 1.40 (0.72, 2.72) Q = 1.71, P = 0.79 Z = 0.98, P = 0.33 Endophthalmitis 2 88 0.0341 95 0.0211 1.63 (0.27, 9.75) Q = 0.06, P < 0.81 Z = 0.53, P = 0.59 Shallow AC 5 180 0.0556 182 0.0604 0.95 (0.41, 2.21) Q = 1.95, P = 0.75 Z = 0.12, P = 0.90

MMC, mitomycin C; 5-FU, 5-fluorouracil; RR, relative risk (calculated using a random-effect model); CI, confidence interval; AC, anterior chamber. group at the last end point were included in the present and post operative treatment with 5-fluorouracil and mito- analysis. Asymmetry was showed by funnel plots, indicating mycin-c: Long term effects in vivo on subconjunctival and less possibility of publication biases. scleral fibroblasts. Int. Ophthalmol. 16:381–385, 1992. One major limitation of this analysis was that patients 2. Loon, S.C., and Chew, P.T. A major review of were not stratified into high, medium, and low risk of tra- in glaucoma therapy. Ophthalmologica 213:234–245, 1999. beculectomy failure subgroups, which may possibly produce 3. Longley, D.B., Harkin, D.P., and Johnston, P.G. 5-fluoro- more interesting results.33 A second limitation is that our uracil: Mechanisms of action and clinical strategies. Nat. Rev. analyses of IOPR, success rate, and adverse events were Cancer 3:330–338, 2003. based on data pooled from trials of different durations. It 4. Andreanos, D., Georgopoulos, G.T., Vergados, J., Papacon- was due to lack of data reported in all phases of follow-up. It stantinou, D., Liokis, N., and Theodossiadis, P. Clinical eval- uation of the effect of mitomycin-C in re-operation for primary was a compromise proposal to choose the data of follow-up open angle glaucoma. Eur. J. Ophthalmol. 7:49–54, 1997. end point. Another limitation is that only published studies 5. Carlson, D.W., Alward, W.L., Barad, J.P., Zimmerman, M.B., were included. To avoid publication bias, we conducted not and Carney, B.L. A randomized study of mitomycin aug- only an electronic search but also a manual search to identify mentation in combined phacoemulsification and trabecu- all potentially relevant articles, including published and lectomy. Ophthalmology 104:719–724, 1997. nonpublished ones. Unfortunately, it is possible that we may 6. Cohen, J.S., Greff, L.J., Novack, G.D., and Wind, B.E. A have failed to include some papers, especially those pub- placebo-controlled, double-masked evaluation of mitomycin lished in other languages. The other limitations are the fac- C in combined glaucoma and cataract procedures. Ophthal- tors such as small sample size, inadequate allocation mology 103:1934–1942, 1996. concealment, or inadequate or no double blinding, which 7. Costa, V.P., Comegno, P.E., Vasconcelos, J.P., Malta, R.F., and may greatly affect the interpretation of the results. The in- Jose´, N.K. Low-dose mitomycin C trabeculectomy in patients clusion in meta-analysis of studies with very small sample with advanced glaucoma. J. Glaucoma 5:193–199, 1996. sizes may have a paradoxical effect of decreasing the power 8. Shin, D.H., Kim, Y.Y., Sheth, N., Ren, J., Shah, M., Kim, C., of random-effects tests.34 Therefore, in future, rigorous ran- and Yang, K.J. The role of adjunctive mitomycin C in sec- domized controlled trials with long enough follow-up and ondary glaucoma triple procedure as compared to primary big enough sample size are strongly recommended to further glaucoma triple procedure. Ophthalmology 105:740–745, 1998. evaluate the real IOP-lowering effect of intraoperative MMC, 9. Martini, E., Laffi, G.L., Sprovieri, C., and Scorolli, L. Low- compared with intraoperative 5-FU in trabeculectomy. dosage mitomycin C as an adjunct to trabeculectomy. A In conclusion, the results of this meta-analysis of 8 ran- prospective controlled study. Eur. J. Ophthalmol. 7:40–48, 1997. domized controlled trials suggested that intraoperative 10. Robin, A.L., Ramakrishnan, R., Krishnadas, R., Smith, S.D., MMC is more effective in IOP lowering in trabeculectomy Katz, J.D., Selvaraj, S., Skuta, G.L., and Bhatnagar, R. A long- term dose-response study of mitomycin in glaucoma filtra- compared with intraoperative 5-FU but comparable with tion surgery. Arch. Ophthalmol. 115:969–974, 1997. intraoperative 5-FU in both qualified and complete success 11. Shin, D.H., Simone, P.A., Song, M.S., Reed, S.Y., Juzych, rate. Intraoperative use of both agents may contribute M.S., Kim, C., and Hughes, B.A. Adjunctive subconjunctival equally to adverse events. mitomycin C in glaucoma triple procedure. Ophthalmology 102:1550–1558, 1995. Acknowledgment 12. Wu, L., and Yin, J. The effect of mitomycin C on filtration This paper was supported by Shanghai Municipal Natural surgery of glaucoma with poor prognosis. Zhonghua Yan Ke Za Zhi Science Foundation, and National Natural Science Founda- 32:32–34, 1996. 13. The Fluorouracil Filtering Surgery Study Group. Fluorour- tion of China. acil Filtering Surgery Study one-year follow-up. Am. J. Ophthalmol. 108:625–635, 1989. Author Disclosure Statement 14. Goldenfeld, M., Krupin, T., Ruderman, J.M., Wong, P.C., The authors have no competing financial interests. Rosenberg, L.F., Ritch, R., Liebmann, J.M., and Gieser, D.K. 5- Fluorouracil in initial trabeculectomy. A prospective, random- References ized, multicenter study. Ophthalmology 101:1024–1029, 1994. 15. Ophir, A., and Ticho, U. A randomized study of trabecu- 1. Khaw, P.T., Sherwood, M.B, Doyle, J.W., Smith, M.F., lectomy and subconjunctival administration of fluorouracil in Grierson, I., McGorray, S., and Schultz, G.S. Intraoperative primary . Arch. Ophthalmol. 110:1072–1075, 1992. MMC VERSUS 5-FU FOR TRABECULECTOMY 173

16. Ruderman, J.M., Welch, D.B., Smith, M.F., and Shoch, D.E. A 28. Budenz, D.L., Pyfer, M., Singh, K., Gordon, J., Piltz-Seymour, prospective, randomized study of 5-fluorouracil and filtra- J., and Keates, E.U. Comparison of phacotrabeculectomy tion surgery. Trans. Am. Ophthalmol. Soc. 85:238–253, 1987. with 5-fluorouracil, mitomycin-C, and without antifibrotic 17. Wilkins, M., Inda, A., and Wormald, R. Intra-operative mi- agents. Ophthalmic Surg. Lasers. 30:367–374, 1999. tomycin C for glaucoma surgery. Cochrane Database Syst. 29. Vijaya L, Mukhesh BN, Shantha B, Ramalingam S, Sathi Rev. 2005;(4):CD002897. Devi AV. Comparison of low-dose intraoperative mitomy- 18. Wormald, R.,Wilkins, M.E,, and Bunce, C. Post-operative 5- cin-C vs 5-fluorouracil in primary glaucoma surgery: A pilot fluorouracil for glaucoma surgery. Cochrane Datebase Sys. study. Ophthalmic Surg. Lasers 31:24–30, 2000. Rev. 2001;(3):CD001132. 30. Oh, S.Y., Youn, D.H., Kim D.M., and Hong C. The effects of 19. Khaw, P.T., Sherwood, M.B., MacKay, S.L., Rossi, M.J., and intraoperative mitomycin c or5-fluorouracil on glaucoma Schultz, G. Five-minute treatments with fluorouracil, floxur- filtering surgery. Korean J Ophthalmol. 8:6–13, 1994. idine, and mitomycin have long-term effects on human Te- 31. Singh, K., Egbert, P.R., Byrd, S., Budenz, D.L., Williams, non’s capsule fibroblasts. Arch Ophthalmol. 110:1150-4, 1992. A.S., Decker, J.H., and Dadzie, P. Trabeculectomy with in- 20. Akarsu C, Onol M, and Hasanreisoglu B. Postoperative 5- traoperative 5-fluorouracil vs mitomycin C. Am. J. Ophthal- fluorouracil versus intraoperative mitomycin C in high-risk mol. 123:48–53, 1997. glaucoma filtering surgery: Extended follow up. Clin. Exp. 32. Javitt JC. When does the failure to find a difference mean Ophthalmol. 31:199–205, 2003. that there is none? Arch. Ophthalmol. 107:1034–1040, 1989. 21. Abdu, M., Chen, X.-Y., Kadir, J., and Arkin, A. Mytomycin C 33. Clarke, J.C.K., and Schlottmann, P.G. Mitomycin C versus 5- versus 5-fluorouracil for trabeculectomy: A systematic re- Fluorouracil for wound healing in glaucoma surgery (Pro- view. Chin. J. Evidence-Based Med. 10:730–739, 2010. tocol). Cochrane Datebase Sys. Rev. 2009;(3):CD006259. 22. The Cochrane Collaboration. Cochrane Handbook for Sys- 34. Downs, S.H., and Black, N. The feasibility of creating a tematic Reviews of Interventions 5.1.0. Updated March 2011. checklist for the assessment of the methodological quality both Available at www.cochrane-handbook.org. Accessed June of randomised and non-randomised studies of health care in- 15, 2011. terventions. J. Epidemiol. Commun. Health 52:377–384, 1998. 23. Downs, S.H., and Black, N. The feasibility of creating a checklist for the assessment of the methodological quality Received: July 7, 2011 both of randomised and non-randomised studies of health Accepted: September 9, 2011 care interventions. J. Epidemiol. Commun. Health 52:377–384, 1998. Address correspondence to: 24. Palanca-Capistrano, A.M., Hall, J., Cantor, L.B., Morgan, L., Jin-Wei Cheng, M.D. Hoop, J., and WuDunn, D. Long-term outcomes of in- Department of Ophthalmology, Shanghai Changzheng Hospital traoperative 5-fluorouracil versus intraoperative mitomycin 415 Fengyang Road C in primary trabeculectomy surgery. Ophthalmology 116: Shanghai 200003 185–190, 2009. China 25. Kim, H.Y., Egbert, P.R., and Singh, K. Long-term compari- son of primary trabeculectomy with 5-fluorouracil versus E-mail: [email protected] mitomycin C in West Africa. J. Glaucoma 17:578–583, 2008. and 26. Singh, K., Mehta, K., Shaikh, N.M., Tsai, J.C., Moster, M.R., Budenz, D.L., Greenfield, D.S., Chen, P.P., Cohen, J.S., Baer- Xiao-He Lu, M.D. veldt, G.S., and Shaikh, S. Trabeculectomy with intraoperative Department of Ophthalmology mitomycin C versus 5-fluorouracil. Prospective randomized Zhujiang Hospital . Ophthalmology 107:2305–2309, 2000. Southern Medical University 27. Smith, M.F., Doyle, J.W., Nguyen, Q.H., and Sherwood, M.B. Guangzhou 515282 Results of intraoperative 5-fluorouracil or lower dose mito- China mycin-C administration on initial trabeculectomy surgery. J. Glaucoma 6:104–110, 1997. E-mail: [email protected]