Relative efficacy and safety of preservative-free latanoprost (T2345) for the treatment of open angle glaucoma and ocular hypertension: an adjusted indirect comparison meta-analysis of randomised clinical trials
Methodological quality of the review: Low confidence
Author: Cucherat M, Stalmans I, Rouland JF
Geographical coverage: United States of America (USA), United Kingdom (UK), Canada, China, Turkey, Italy and Egypt
Sector: Glaucoma
Sub-sector: Treatment of open angle glaucoma and ocular hypertension
Equity focus: None specified
Review type: Effectiveness review
Quantitative synthesis method: Meta-analysis
Qualitative synthesis method: Not applicable
Background: The preferred treatment for ocular hypertension (OH) and open angle glaucoma (OAG) is prostaglandin analogues (PGA). Currently, there is no available data on the relative safety and efficacy of preservative-free latanoprost and other PGA (with and without a preservative). There has been increasing interest in undertaking adjusted indirect comparison meta-analyses to optimally estimate the relative effectiveness and safety of alternative healthcare interventions when available data does not allow direct comparisons.
Objectives: To assess the relative efficacy and tolerability of preservative-free latanoprost (T2345) compared with other PGA for the treatment of OH and OAG by adjusted indirect comparison meta-analysis.
Main findings: 21 randomised controlled trials (RCTs) were included in the meta-analysis. Nine trials were conducted in North America, two in Europe and ten in other regions or in >1 continent. Overall, there was some variation in the quality of the trials included in the analysis data set. Seven trials were double-masked, 11 were single-masked (investigator or evaluator) and one was open. Information on masking was not available for two trials. All trials had a parallel-group design. Analysis of the inclusion criteria of the selected studies, and the characteristics of the patients enrolled, indicated similarity of the study populations in terms of age, sex ratio and baseline IOP.
No statistically significant differences in IOP at three months were seen between T2345 and travoprost (preserved with benzalkonium chloride [BAK], polyquaternium-1 or sofzia), and other BAK-preserved PGA: bimatoprost 0.03%, bimatoprost 0.01% or latanoprost. T2345 was statistically significantly superior to BAK-tafluprost (mean difference: 0.47 mm Hg, 95% confidence interval, [−1.52;−0.28]). The risk of hyperemia was statistically significantly lower with T2345 than with all the other PGA.
Authors conclude that indirect comparisons never found preservative-free latanoprost (T2345) to be statistically significantly inferior to the other PGA in terms of efficacy on IOP and showed statistically significant superiority over BAK-tafluprost. The risk of hyperemia was statistically significantly lower with T2345 than with all the other PGA. The results of indirect comparisons may be of great value to clinical and resource decision-makers in the absence of large randomised trials permitting multiple direct head-to-head comparisons.
Authors also note that evidence generated by indirect comparisons is not as strong as that which would be generated by large, double-blind randomised trials directly comparing a large number of treatment options.
Methodology: MEDLINE, EMBASE, and Cochrane Controlled Trials Register (up to December 2011) databases were used to identify RCTs evaluating PGA for the treatment of OAG and OH. No language or publication restrictions were applied.
The following studies were excluded: comparisons involving drugs that had subsequently been withdrawn from the market; study participants had to have OAG or OH (excluding secondary glaucoma) with a baseline IOP exceeding 20 mm Hg and have undergone a wash-out period before randomisation if previously treated for these conditions; studies that had enrolled patients having undergone intra-ocular laser treatment or surgery within the previous three months or having received during the study concomitant systemic or ocular treatments that could affect IOP. Pharmacokinetic, animal and laboratory studies, and studies in healthy volunteers, were also excluded.
Risk of bias on studies included were assessed by using the following evidence-based criteria: method of allocation concealment, randomisation technique, double-blinding and description of withdrawals and dropouts. Authors did not mention if this was conducted by two reviewers independently. One reviewer extracted data from the selected trials and a second reviewer checked this data for accuracy, using a standard data extraction sheet. However, it’s not clear if this was conducted independently.
The authors conducted a direct meta-analysis employing a random-effect model, using odds ratio to summarise dichotomous results and the weighted mean difference (WMD) to summarise continuous results. Statistical heterogeneity was quantified by using the I2 statistics, 19 of which approximates the percentage of the total variation (within and between studies) that is due to between-study variation. Publication bias was not assessed due to the small number of trials corresponding to each particular comparison. Adjusted indirect comparisons were performed using the Bucher method. The main endpoints were IOP measured at three months and at incidences of hyperemia.
Applicability/external validity: The authors did not discuss the applicability/external validity of the results.
Geographic focus: Authors did not restrict the search to high- or low-income countries. This study included studies from USA, UK, Canada, China, Turkey, Italy and Egypt
Summary of quality assessment: Overall, there is low confidence in the conclusions about the effects of this study, as important limitations were identified. Authors conducted a partially comprehensive search for relevant studies. However, reference lists in included articles were not checked and authors/experts were not contacted as part of the search strategy. Review authors did not mention in the review if selection of studies, quality assessment and data extraction of included studies were conducted by two reviewers independently. In addition, the review does not make clear which evidence is subject to low risk of bias in assessing causality, and which is likely to be biased.
