Efficacy and Safety of Bevacizumab in the Treatment of Pterygium: An Updated Meta-Analysis of Randomized Controlled Trials

Purpose Studies investigating efficacy and safety of bevacizumab in pterygium have increased and reported controversial results. Thus, we updated this meta-analysis to clarify the issue. Methods Studies were selected through search of the databases Embase, PubMed, Web of Science, and the Cochrane Central Register of Controlled Trials (CENTRAL) from their inception up until June 2017. The pooled risk ratio (RR) and 95% confidence interval (CI) were calculated for recurrence and complication rates by using random effects model. Results 1045 eyes in 18 randomized controlled trials (RCTs) enrolled. Overall, the pooled estimate showed a statistically significant effect of bevacizumab on the reduction of recurrence (RR 0.74, 95% CI 0.56–0.97, P=0.03). Subgroup analyses presented significant results beneficial to bevacizumab (primary pterygium group, RR 0.53, 95% CI 0.33–0.83, P=0.006; conjunctival autograft group, RR 0.48, 95% CI 0.25–0.91, P=0.02; and follow-up longer than 12 months group, RR 0.36, 95% CI 0.13–0.99, P=0.05). No statistically significant difference was observed in complication rates. Conclusions Application of bevacizumab showed a statistically significant decrease in recurrence rate following removal of primary pterygia, or in cases with conjunctival autograft, or with follow-up longer than 12 months, while complications were not increased.


Introduction
Pterygium is one of the most common ocular surface diseases, which is characterized by the fibrovascular conjunctiva tissue proceeding from the bulbar conjunctiva towards the cornea. It limits eye movements and causes dry eye, irritation, foreign body sensation, and even decrease of visual acuity [1]. e primary treatment for pterygium is surgery, and the major problem of the treatment is the high recurrence rate, varying between 38% and 88% in bare sclera, 5%-30% in conjunctival autograft, and 0%-15% in limbal conjunctival autograft [2]. Many adjuvant therapies have been developed to reduce recurrence including mitomycin C [3,4], 5-FU [5][6][7], and radiotherapy [8,9].

Search Strategy.
e databases of Embase, PubMed, Web of Science, and the Cochrane Central Register of Controlled Trials (CENTRAL) were searched from their inception up until June 2017. Details of the search strategies were described in the Search Strategy file. Endnote software was used to exclude the duplications. Titles and abstracts were scrutinized to deduct apparently irrelevant studies. Full texts were retrieved and assessed for qualification. A manual search was executed by checking the reference lists of all retrieved studies and reviewing articles to distinguish studies not found by the electronic searches. Language was not restricted.

Inclusion and Exclusion Criteria.
e articles were considered qualified if the studies fulfilled the following inclusion criteria: (1) participants: pterygium patients (including primary pterygium, impending recurrent pterygium, and recurrent pterygium); (2) intervention: topical or subconjunctival bevacizumab, regardless of operation or not. e dose of bevacizumab, follow-up periods, or length of fibrovascular growth passing the corneal limbus were not confined; (3) comparison: bevacizumab and control; (4) outcomes: recurrence and/or complication rates; and (5) publication type: RCT. RCTs without exact raw data available for extraction were excluded.

Outcome Measurements.
e primary outcome measurements were recurrence and complication rates. Recurrence was diagnosed when any fibrovascular growth crossed the limbus and extended over the cornea to any distance by slit-lamp examination. e number of recurrences was estimated at the endpoint of the follow-up in each study. Complications such as lacrimation, inflammation, photophobia, conjunctival erythema, conjunctival flap edema, conjunctival graft loss, subconjunctival hemorrhage, corneal dellen, severe conjunctival or corneal scarring, and systemic complications were counted. e number of complications at the last documenting time during the follow-up in each study was calculated.

Data Extraction.
e data were extracted by two reviewers (Yi Sun and Bowen Zhang) independently. Discrepancies were resolved by discussion to reach a consensus between the investigators. e information collected from each study included the first author's last name, year of publication, study design, location and duration of the study, sample size including sex, age, and diagnoses, type of treatment and control, route of administration, and dose of bevacizumab.

Risk of Bias Assessment. Two reviewers (Yi Sun and
Bowen Zhang) separately evaluated the risk of bias in each study according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions 5.3. e authors reviewed the studies and assigned a value of "high," "low," or "unclear" to the following items: (1) selection bias (Was there sufficient generation of the randomization sequence and allocation concealment?); (2) performance and detection bias (Was there blinding of participants, personnel, and outcome assessors?); (3) attrition bias (Were there incomplete outcome data and how to deal with this?); (4) reporting bias (Was there evidence of reporting outcome selectively?); and (5) other sources of bias (Were there any other potential threats to validity?). Any disagreement was discussed until a consensus was reached.

Statistical Analysis.
e recurrence and complication rates were handled as dichotomous variables measured as the risk ratio (RR) with a 95% confidence interval (CI). Due to the diversity in sample size and the differences in clinical characteristics among the studies, it was presumed that heterogeneity existed even when no statistical significance was observed. erefore, the data were pooled using a random effects model. Statistical heterogeneity among the studies was assessed by calculating a Cochran Q statistic and an I 2 statistic. Subgroup analysis and sensitivity analysis for the recurrence rate were carried out to evaluate the impact of the following factors on the results: (a) participants: primary pterygium, impending recurrent pterygium, and recurrent pterygium; (b) intervention: topical use or subconjunctival injection of bevacizumab; type of operation or not; and (c) follow-up periods: ≤6 months, 6～12 months, and ≥12 months. We explored asymmetry in funnel plots to detect publication biases. e analysis was performed using RevMan 5.3 ( e Cochrane Collaboration, Copenhagen, Denmark).

Literature Search.
Literature search and selection process are summarized in Figure 1. A total of 99 articles were initially enrolled. After removing duplications, the abstracts of the remaining studies were inspected, and 29 articles with possibly relevant trials were further identified in full texts. Eighteen randomized controlled trials (RCTs) were deemed eligible after a full text screening and were finally included in this meta-analysis.

Meta-Analysis.
15 studies reported recurrences. Definitions of pterygium recurrence of the included randomized clinical trials are shown in Table 2. Overall recurrence rate of this meta-analysis was summarized in supplementary data file.
Publication bias for recurrence rates and complications was checked by evaluating funnel plots (supplementary data file).

Discussion
is meta-analysis, updated with 1045 eyes in 18 RCTs showed that bevacizumab would significantly reduce pterygium recurrence rate after surgery in either case of primary pterygium or use of conjunctival autograft or follow-up longer than 12 months. Complications of bevacizumab were not increased compared with the control.
An earlier meta-analysis performed by Hu indicated that bevacizumab had no statistically significant effect on preventing pterygium recurrence [35]. Hu included 9 RCTs, of which 7 reported recurrence and 8 reported complications, whereas in our current meta-analysis, we report raw data on recurrences in 15 and complications in 17 studies. e inclusion of more trials and more cases renders our analysis more statistically significant.
According to Prabhasawat [36], corneal recurrence with fibrovascular tissue covering the excision area and invading the cornea (grade 4) was the true recurrence. However, the    definition of recurrence adopted in literatures varied. e inconsistent definition of pterygium recurrence in the included studies ( Table 2) implied that the conclusion of the meta-analysis should be interpreted prudently. Study by Razeghinejad defined recurrence as any fibrovascular growth of conjunctival tissue extending more than 1.5 mm across the limbus [37]. In addition, data of recurrence in table 3 of the literature were found incorrect. us, the study was excluded. Moreover, the significant effect of bevacizumab on decreasing recurrence in the follow-up longer than 12 months group would suggest that longer follow-up in the future studies could further favor the effect. RR for the overall recurrence rate was 0.74, with 95% CI [0.56, 0.97]. After removal of the study by Motarjemizadeh [31], I 2 decreased to 0% and RR was 0.98, with 95% CI [0.92, 1.05], but it did not affect the conclusive result in subgroup analysis on the pterygium type or administration route of bevacizumab. erefore, sensitivity analysis was unstable and the heterogeneity was mainly caused by this study. However, there was no reason to exclude the study after comprehensive reading of the full text.
ere was no statistically significant difference in overall complications and subconjunctival hemorrhage between bevacizumab group and control group, showing the safety of bevacizumab. e sensitivity analysis for the complication was stable. It is different from the previous meta-analysis by Hu [35], who reported the bevacizumab group was associated with a higher risk of developing subconjunctival hemorrhage. e funnel plot for the recurrence and complication rates displayed asymmetry. is could be due to factors other than publication bias, including poor methodological quality, true heterogeneity, artefactual variation, and chance.

Journal of Ophthalmology
Our study had several potential limitations. First, the heterogeneity may result from different administration route of bevacizumab, different type of pterygium, surgeon's experience, and follow-up duration. Second, sensitivity analyses of the recurrence rate were not stable. erefore, caution is required in their interpretation and more research is still needed.
Despite these limitations, the evidence from the updated meta-analysis shows that bevacizumab application following pterygium surgery provides a statistically significant decrease in recurrence rate in cases of primary pterygium, or use of conjunctival autograft, or follow-up longer than 12 months without an increase in complications. Further study of the long-term efficacy of bevacizumab on reducing pterygium recurrence based on the definition of true recurrence (grade 4) will be needed. Disclosure e funding organization played no role in the design or conduct of this study.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.