Interventions for central serous chorioretinopathy: a network meta-analysis

Abstract

Background: Central serous chorioretinopathy (CSC) is characterized by a thickened and dysfunctional choroid which is accompanied by a serous detachment of the neural retina. The effects on the retina are usually self-limiting, although some people are left with irreversible vision loss due to progressive and permanent photoreceptor damage or atrophy of the retinal pigment epithelium (RPE). There has been a variety of interventions used in CSC, including, but not limited to, laser treatment, photodynamic therapy (PDT), and drug therapy with mineralocorticoid receptor antagonists or intravitreal injection of anti-vascular endothelial growth factor (anti-VEGF) agents. However, it is not known whether these treatments offer significant long-term advantages over observation or each other. At present, there is no evidence-based consensus on the management of CSC.

Objectives: This is an update of a Cochrane review first published in 2015 where 25 studies with 1098 participants were included. Since then, many trials have been conducted and reports published. Our primary objective was to assess the comparative effectiveness of multiple interventions for CSC. The secondary objective was to provide the relative ranking of the interventions for CSC using network meta-analysis.

Search methods: We searched CENTRAL, MEDLINE, Embase, and three trial registries in 29 March 2024, together with reference checking.

Selection criteria: Randomized controlled trials (RCTs) that compared any intervention for CSC with any other intervention for CSC or control.

Data collection and analysis: Two review authors (CL, LP) independently selected studies and extracted data. Our outcomes of interest were best corrected visual acuity (BCVA), recurrence of CSC, persistent CSC, contrast sensitivity, central retinal subfield thickness, quality of life, and adverse events. We used standard methodological procedures expected by Cochrane. We used Cochrane's statistical software, Review Manager, to perform pairwise analyses and Stata to perform network meta-analysis (NMA). For pairwise comparisons, we pooled data from studies using fixed or random-effects models if there were fewer or more than three studies, respectively. We conducted NMAs using a multivariate meta-analyses approach and ranked interventions using the surface under the cumulative ranking (SUCRA). We used the Confidence in Network Meta-Analysis (CINeMA) approach to assess and present the certainty of evidence for NMA results.

Main results: This review includes 4015 participants from 67 RCTs in total. Additionally, we identified 31 ongoing clinical trials. Trials compared aflibercept, crocin, lutein, eplerenone, spironolactone, prednisolone eye drops, PDT, subthreshold micropulse laser (SML) (577 nm) between each other or respective control groups (e.g. observation, sham injection, or placebo). Studies were conducted in Europe, North and South America, the Middle East, and Asia. Most of the trials were small, enrolling fewer than 50 participants, and poorly reported. A substantial proportion of trials were not masked, and it remained unclear whether key aspects of the trial, such as allocation concealment, had been done. Eight (13%) studies were funded by industry and 21 (31%) by non-industry sources. Overall, 23 (34%), 34 (51%), and 10 (15%) studies were rated at high, moderate, and low risk of bias. Twenty-two studies were included in the pairwise meta-analyses, contributing data to at least one prespecified outcome (change in best corrected visual acuity, recurrence or persistence of CSC, change in contrast sensitivity or central retinal thickness, quality of life, or adverse events) with a follow-up of six to 18 months. These RCTs assessed the effect of oral medication treatments (such as antioxidants, beta-blockers, carbonic anhydrase inhibitors and mineralocorticoid receptor antagonists), intravitreal anti-VEGF injections, laser-assisted treatments (such as pulsed and non-pulsed laser approaches), PDT, and meditation. Most studies had a moderate risk of bias. Pairwise meta-analyses mostly failed to find evidence of differences in effect. We did not have any comparisons with more than 10 studies per analysis to assess the risk of publication biases. Regarding harms, most studies did not report harms in a standardized way and reported no treatment-related harms. Specific harms reported included significant RPE damage among those receiving conventional SML and Grade 1 choroidal ischemia in 3/51 eyes receiving PDT, but the evidence is very uncertain. We were unable to conduct an NMA of recurrence or harms due to sparse data. To enable fuller data for our network of change in BCVA, we classified the interventions into seven unique groups by the types of pharmacologics, laser treatments, and levels of PDT. We excluded interventions for which the assumption of transitivity was not met (i.e. focal unpulsed laser treatment, H. pylori eradication therapy), and performed a NMA with 17 trials of the seven treatment groups (21 comparisons). The NMA did not find any evidence of differences between the treatments that were analyzed. The SUCRA analysis for BCVA suggested the following order for the highest ranking treatments: < 50% PDT (SUCRA = 81.1), supplement (59.0), eplenerone (57.7), anti-VEGF (50.3), control (47.9), ≥ 50% PDT (36.5), and pulsed laser (17.5). SUCRA also suggested low-dose PDT, eplenerone, and supplement had the highest probabilities of being the best (≥ 19.6%), compared to the others (≤ 6.3%). However, the reliability of these SUCRA estimates is limited due to poor overall connectivity in the network, leading to an increased risk of inconsistency between direct and indirect comparisons and increased influence of individual studies. We judged most comparisons as being at moderate (13/21) or low (7/21) confidence, mostly because of imprecision and within-study bias. No comparisons had high certainty.

Authors' conclusions: CSC remains an enigmatic condition, in large part due to a natural history of spontaneous improvement in a high proportion of people and also because no single treatment has provided overwhelming evidence of efficacy in published RCTs. While a number of interventions have been proposed as potentially efficacious, the risks of biases and the relatively small number of participants enrolled and successfully followed limit the utility of existing data. Our results did not show the superiority of any treatment option over another. Low-dose photodynamic therapy, supplements, and eplenerone had the greatest SUCRA values and probabilities of being the best treatments for improving visual acuity, although our confidence in the evidence for these interventions is very low to moderate. Larger and high-quality RCTs comparing these treatments are warranted.

Trial registration: ClinicalTrials.gov NCT01325181 NCT00987077 NCT01990677 NCT01710332 NCT00817245 NCT00963131 NCT01327170 NCT01574430 NCT01256580 NCT01585441 NCT01982383 NCT01019668 NCT01552044 NCT01971190 NCT02153125 NCT02215330 NCT02354170.