Mitomycin C for the prevention of corneal haze in photorefractive keratectomy: a meta-analysis and trial sequential analysis

Author: Chang YM, Liang CM, Weng TH, Chien KH, Lee CH.

Geographical coverage: Worldwide

Sector: Biomedical

Sub-sector: Treatment

Equity focus: None

Study population: Myopia and myopic astigmatism patients that underwent PRK.

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis and trail sequential analysis.

Qualitative synthesis method: Not applicable

Background: Photorefractive keratectomy (PRK) has been widely used in correction of refractive error for more than 20 years. Due to its painful sensation and long recovery time, refractive procedures have increasingly developed, such as transepithelial PRK, laser epithelial keratomileusis, epipolis laser in-situ keratomileusis (Epi-LASIK), LASIK and small incision lenticule extraction. However, these procedures still have both advantages and disadvantages. Therefore, PRK remains a popular refractive surgery due to the absence of flap-related complications, ability to maintain an acceptable corneal architecture and applicability in thinner corneas. Although PRK has many advantages, it also has some disadvantages including corneal haze. The effects of MMC on corneal haze formation after PRK have been investigated in the past 20 years; however, the results are contradictory.

Objectives: To evaluate the anti-haze effect and visual outcome after intraoperative mitomycin C (MMC) use during photorefractive keratectomy (PRK) in myopia or myopic astigmatism patients.

Main findings:

Overall, authors found MMC an important anti-haze agent in PRK for reducing both early- and late-onset haze and can also help improving predictability of refraction and subjective postoperative visual acuity.

A total of 17 articles were included according to the criteria. Eleven RCTs, five cohort studies and one case-control study were identified as 16 full-texts and one conference abstract.

Authors found no significant difference in the pre-SE of the study participants; however, two studies did not provide pre-SE values. All included studies used 0.02% MMC in surgeries, except for one study that used 0.01% MMC, and the administration time was below 120 seconds in all studies.

Five RCTs had low risk in random sequence generation and allocation concealment, while the others just revealed randomisation without allocation disclosure. There were six RCTs that described blinding of outcome assessment, while others presented unclear methods. As for selective reporting, outcomes were not found in all RCTs. In non-randomised studies, the Newcastle-Ottawa Scale (NOS) was used to evaluate the risk of bias. In cohort studies, no trials demonstrated less than five stars, and one trial showed nine stars being identified as having the lowest risk of bias. Only one case-control study with four stars was involved, representing a high risk of bias.

Of the included 11 randomised controlled trials, five cohort and one case-control studies, 3,536 eyes (2,232 and 1,304 in the MMC and control groups, respectively) were enrolled for meta-analysis. The TSA disclosed strong evidence of decline in corneal haze rate in the MMC group compared with that of the control group. In the subgroup analysis of duration, MMC seemed to reduce corneal haze rate in early-onset and late-onset haze. Predictability of refraction and visual acuity were greater in the MMC groups, but not significantly. The proportion of patients losing at least two lines of best corrected visual acuity postoperatively in the MMC groups was lower than that in the control groups. The corneal postoperative ECD showed no significant difference between the MMC and control groups.

Publication year, number of the eyes, follow-up duration, MMC administration time and pre-op SE could not fully explain the heterogeneity of the corneal haze rate. No significant small study bias was observed by the authors for the primary or secondary outcomes.

Authors emphasise the need to investigate the effect of MMC on corneal ECD loss in larger RCTs.

Methodology:

Study evaluated myopia and myopic astigmatism patients that underwent PRK. Studies published before February 2020 that satisfied inclusion criteria included. Inclusion criteria are: (1) study design: RCTs, comparative cohort studies, and case-control studies; (2) population: myopia or myopic astigmatism patients; (3) intervention: compared MMC with placebo or no intervention during PRK; and (4) outcomes: evaluation one of the clinical outcomes: corneal haze, visual acuity, spherical equivalent (SE) and endothelial cell density (ECD). Conference abstracts that provided useful information were also considered. Exclusion criteria were: (1) studies not involving the desirable clinical outcomes; (2) animal or laboratory studies; (3) follow-up duration of less than three months; and (4) studies including hyperopia patients.

Authors searched for literature in the PubMed, EMBASE, Cochrane Library and Google Scholar, without language restriction to obtain studies published through February 2020. A manual search for conference abstracts and references cited in previous studies was performed.

Two authors independently conducted abstract screening and data extraction. Discrepancies between the two reviewers were resolved through a discussion. Additionally, two authors performed quality assessment of RCTs according to the Cochrane Collaboration Reviewers’ Handbook for Systematic Reviews of Interventions.

Patient data from each included study was pooled and presented as mean standard deviation or numbers. Authors used weighted mean difference continuous variables, and the odds ratio for dichotomous variables. Analyses were conducted using the Mantel-Haenszel random-effects model and Review Manager, with a p-value less than 0.05 considered statistically significant. Publication bias was tested using Egger’s regression and funnel plot. Heterogeneity was assessed using I2 statistics, sensitivity tests, and meta-regression analyses. Possible moderators were tested to explore heterogeneity. Subgroup analyses were performed based on follow-up duration and study designs, with six months as a cut-off point. Studies were categorised into two groups based on study design. Finally, trial sequential analysis was performed to quantify the statistical reliability of data in the cumulative meta-analysis.

Applicability/external validity: Authors identify a number of limitations which may make their findings less applicable, for example, inclusion of non-RCTs with higher risk of bias; and the fact that treatment was not always necessarily operationalised in the same way.

Geographic focus: Search was unrestricted in terms of language and geography and a number of studies produced in low- and middle-income countries were included. However, authors did not consider how findings may vary for different geographic settings.

Summary of quality assessment:

While the approaches used to identify, include and critically appraise studies were generally robust, no attempts were made to include unpublished material or to contact relevant experts to identify additional material, meaning that the search results may be limited. However, the data analysis undertaken was very thorough and robust, with a range of possible reasons for the identified results thoroughly examined. For these reasons, we have medium confidence in the findings of this review.