Eplerenone versus placebo for chronic central serous chorioretinopathy: the VICI RCT

Article history

The research reported in this issue of the journal was funded by the EME programme as project number 13/94/15. The contractual start date was in April 2016. The final report began editorial review in December 2019 and was accepted for publication in June 2020. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The EME editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

© Queen’s Printer and Controller of HMSO 2021. This work was produced by Lotery et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2021 Queen’s Printer and Controller of HMSO

Background

Central serous chorioretinopathy (CSCR), predominantly affecting men of working age, is the fourth most common vision-threatening retinopathy. 1–4 Fluid spontaneously gathers under the retina causing a neurosensory retinal detachment, which leads to permanent vision loss in up to one-third of cases. 5 The disease course is variable; some cases spontaneously resolve, up to 50% develop chronic CSCR [subretinal fluid (SRF) for > 3 months] and others resolve then recur or affect the fellow eye. 2,4

Characterised by widespread retinal abnormalities, including localised retinal detachments and a thickened choroid with dilated vessels, chronic CSCR leads to a permanent reduction in vision. 1 Abnormalities of choroidal circulation are thought to play a pivotal role in disease pathogenesis. Distinct choroidal abnormalities, including excess dilatation and hyperpermeability, can be seen in patients, and primate studies demonstrate choroidal endothelial cell defects. 2,6 Corticosteroids, both endogenous and exogenous, are the most significant risk factor for CSCR, with mineralocorticoid receptor overactivation in rodents producing a CSCR phenotype. 2,7 The cause of CSCR is unknown, although occurrence within families suggests heritable factors and Schubert et al. 8 identified the first genetic determinants when they identified variation in the cadherin gene in male patients with CSCR. A recent genome-wide association study identified mutations in complement factor H to be significantly associated with chronic CSCR. 9,10

Treatments with some evidence of benefit include half-dose verteporfin photodynamic laser therapy (PDT) and subthreshold micropulse laser treatment. 11–13 However, these treatments are not available in most NHS hospitals and verteporfin for PDT is not licensed for this indication. Furthermore, PDT can have side effects that reduce vision, including retinal scarring, which is permanent, and choroidal ischaemia, which can be temporary. 14 Anti-vascular endothelial growth factor (VEGF) agents used to treat neovascular macular degeneration have not shown efficacy for treating CSCR. 15 Therefore, standard care for CSCR is often observation only; there is no gold standard treatment.

Rationale

Research into the aetiology of CSCR has opened new avenues for potential treatments. In a rat model of CSCR, Zhao et al. 16 showed that one of the features of the condition, choroidal vasodilation, was induced by aldosterone acting via an endothelial vasodilatory potassium channel KCa2.3. Aldosterone is a mineralocorticoid receptor (MR) activator. Blockade of this pathway prevented aldosterone-induced choroidal thickening. This work was translated to a clinical setting. We treated two patients with chronic CSCR with the oral MR antagonist eplerenone for 5 weeks and observed an improvement in visual acuity as well as resolution of retinal detachment and choroidal vasodilation. The observed benefit was maintained for 5 months after the patients stopped taking eplerenone. 16 These results identified MR signalling as a pathway controlling choroidal vascular bed relaxation and provide a pathogenic link with CSCR. This research supports the hypothesis that blockade of MRs could be used as a treatment for CSCR. A subsequent pilot study of 13 patients with chronic CSCR found that eplerenone was associated with a significant reduction in central macular thickness (CMT) and SRF and an improvement in visual acuity in some patients. 17

A small number of double-masked randomised controlled trials (RCTs) investigating eplerenone to treat CSCR have shown some benefit in reducing SRF and/or improving visual acuity. 18–20 However, the studies were limited by being conducted at single centres and with too few patients to draw meaningful conclusions.

In summary, there are several small studies suggesting efficacy of eplerenone in treating CSCR but there have been no adequately powered placebo-controlled RCTs to conclusively demonstrate whether or not eplerenone is a safe and effective treatment. The VICI trial provides evidence for whether or not eplerenone is an effective treatment and contributes to the knowledge gap on the natural history of the disease. This trial is especially important because ophthalmologists are prescribing eplerenone off licence for CSCR without robust evidence of its efficacy.

Aim and objectives

The aim of this study was to compare the efficacy and safety of eplerenone with usual care versus placebo with usual care for chronic CSCR for 12 months in a Phase III randomised placebo-controlled clinical trial.

Primary objective:

1. to evaluate whether best corrected visual acuity (BCVA) following eplerenone therapy with usual care is superior to placebo with usual care in eyes with chronic CSCR.

Secondary objectives:

1. to evaluate whether or not eplerenone treatment with usual care is better than placebo with usual care for resolution of SRF

2. to describe the safety profile of eplerenone treatment with usual care (compared with placebo with usual care)

3. to evaluate whether or not participant-reported visual function improves with eplerenone treatment with usual care compared with placebo with usual care

4. to describe how the choroid responds to treatment in CSCR

5. to describe how retinal pigment epithelium (RPE) function changes over 1 year in CSCR as measured by autofluorescence (AF)

6. to evaluate how low-luminance visual acuity (LLA) changes with eplerenone treatment.

Mechanistic substudy objectives:

1. to explore treatment response by conducting imaging studies of retina and choroid

2. to generate a biobank of DNA, serum and plasma for future mechanistic studies.

Study design

The VICI trial was a Phase III, parallel, randomised (1 : 1 ratio), multicentre, double-masked trial, stratified by site (i.e. participating hospital) and BCVA score, comparing the superiority of oral eplerenone plus usual care with placebo plus usual care in patients with CSCR.

The trial was registered with the International Standard Randomised Controlled Trial Number (ISRCTN) registry on 6 May 2016 (identifier: ISRCTN92746680). Ethics approval was granted by the Wales Research Ethics Committee (REC) 1 on 30 April 2016 (identifier: 16/WA/0069). Health Research Authority (HRA) approval was granted on 21 July 2016. A Clinical Trials Authorisation was granted by the Medicines and Healthcare products Regulatory Agency (MHRA) on 31 March 2016 [European Union Drug Regulating Authorities Clinical Trials Database (EudraCT) 2016–000113–70]. Recruitment opened on 14 December 2016 and closed on 28 February 2018.

The trial was managed by a trial management group, coordinated and monitored by the Bristol Trials Centre Clinical Trials and Evaluation Unit (BTC-CTEU) and sponsored by the University Hospital Southampton NHS Foundation Trust. Trial conduct was overseen by an independent Trial Steering Committee (TSC) that received recommendations from an independent Data Monitoring and Steering Committee (DMSC) (see Appendix 2).

The trial protocol is available online and has been published. 21,22

Participants

Setting

Patients were identified and recruited from outpatient ophthalmology clinics at 22 secondary and tertiary care NHS trusts in the UK between December 2016 and February 2018. Sites were required to have SPECTRALIS^® (Heidelberg Engineering Ltd, Hemel Hempstead, UK) imaging equipment to minimise heterogeneity in retinal imaging between sites. Broad coverage of England and Northern Ireland was achieved, making the trial accessible to a wide population of patients with CSCR. The following NHS trusts participated in the trial:

• University Hospital Southampton NHS Foundation Trust

• Belfast Health and Social Care Trust

• Bradford Teaching Hospitals NHS Foundation Trust

• Brighton and Sussex University Hospitals NHS Trust

• City Hospitals Sunderland NHS Foundation Trust

• East Lancashire Hospitals NHS Trust

• Frimley Health NHS Foundation Trust

• Guy’s and St Thomas’ NHS Foundation Trust

• King’s College Hospital NHS Foundation Trust

• Leeds Teaching Hospitals NHS Trust

• Manchester University NHS Foundation Trust

• Moorfields Eye Hospital NHS Foundation Trust

• Newcastle upon Tyne Hospitals NHS Foundation Trust

• Oxford University Hospitals NHS Foundation Trust

• Royal Liverpool and Broadgreen University Hospitals NHS Trust

• Royal Wolverhampton NHS Trust

• Sheffield Teaching Hospitals NHS Foundation Trust

• Southend University Hospital NHS Foundation Trust

• Torbay and South Devon NHS Foundation Trust

• University Hospitals Bristol NHS Foundation Trust

• University Hospitals Coventry and Warwickshire NHS Trust

• York Teaching Hospital NHS Foundation Trust.

Conduct

Screening was a two-stage process. Patients were initially screened against criteria that could be assessed from medical records. Eligible patients were provided with a patient information leaflet (PIL) and where possible were given at least 24 hours to consider participating in the trial. Written informed consent was obtained from interested patients to undergo trial-specific screening assessments required to assess full eligibility. If participants were eligible following assessments and did not withdraw consent, they were randomised into the trial. Eligibility was determined by experienced ophthalmologists at participating sites. Consent was taken by experienced research nurses or ophthalmologists at the sites.

Interventions

Randomised participants received a dose of 25 mg/day of oral eplerenone or matching placebo for 1 week. If tolerated (blood serum potassium level of ≤ 5.0 mmol/l), the dose was increased to 50 mg/day of oral eplerenone for up to 12 months or until complete resolution of SRF or a blood serum potassium level of > 5.0 mmol/l. Because the interventions were masked, participants in both treatment arms followed the same dose escalation schedule.

Eplerenone was administered as 25-mg or 50-mg tablets that were overencapsulated to match the placebo capsules. The placebo capsules were backfilled with lactose, which was chosen because lactose is a constituent of eplerenone tablets. No restrictions were placed on taking the eplerenone or placebo capsules; capsules could be taken at any time of day, with or without food.

Participants took the capsules home in the form of one bottle of 10 25-mg capsules (1 week’s supply) or one, two or three bottles of 36 50-mg capsules (1, 2 or 3 months’ supply, respectively) depending on the length of time between follow-up visits.

Treatment decisions

Outcomes

Assessments

Adverse events

Serious adverse events (SAEs) and non-serious AEs were recorded throughout the 12-month follow-up period. Participants’ general practitioners were notified of their participation, with a request to inform the local research team about any suspected AEs or reactions.

At each follow-up visit participants were asked to report any AEs experienced since the previous visit. All events were recorded on a case report form (CRF) that coded them by Medical Dictionary for Regulatory Activities (MedDRA) version 14.1 categories. SAEs were reviewed by the local treating clinician, who was masked to the allocation. The treating clinician made the decision on whether or not the SAE was related to the intervention.

Adverse events were reviewed at least monthly by the Trial Management Group and twice per year by the DMSC, which gave recommendations of continuation of the trial to an independent TSC. Unexpected SAEs that were causally related to the intervention were subject to expedited reporting to the REC, MHRA and DMSC.

Sample size

We calculated that 45 patients in both groups would be sufficient to detect a difference of five or more letters in BCVA score at 12 months between the eplerenone and placebo groups with 90% power and 5% significance. This assumed the following:

• a standard deviation (SD) of change in BCVA score of nine letters11

• the correlation between baseline and any follow-up BCVA score was 0.5 letters27

• data were collected from a minimum of two follow-up assessments per participant28

• the correlation between BCVA score on follow-up visits was 0.8 letters.

The target sample size was 104, which allowed for < 15% drop-out over the 12-month period. There were no changes to the sample size during the trial.

Randomisation

Participants were randomly allocated to either the eplerenone with usual care group or the placebo with usual care group in a 1 : 1 ratio. The randomisation was stratified by centre and visual acuity level (low, ETDRS BCVA score of 54–67 letters; high, ETDRS BCVA score of 68–85 letters) and used block randomisation with block sizes of two and four. Allocations were generated by computer, in advance of starting the trial, by a statistician at the co-ordinating centre and supplied to the Newcastle Specials Pharmacy Production Unit (Newcastle Upon Tyne, UK). Randomisation occurred only once information used to identify a participant uniquely and confirm eligibility had been entered into a secure internet-based randomisation system, accessible only to approved trial personnel. Randomisation was performed by local ophthalmologists or delegated members of the local trial team (e.g. research nurses or trial co-ordinators) within 4 weeks of screening.

Masking

Participants, clinicians, research nurses, outcome assessors, pharmacists, the trial management team and statisticians were masked to the allocations. Masking was implemented by overencapsulation of eplerenone/placebo to produce identical capsules. Capsules were supplied in bottles with labelling that was identical except for a unique bottle number. Bottle numbers were assigned against an allocation list. Only personnel at the manufacturing pharmacy and the trial database manager had access to the allocation list for the purposes of producing and managing the investigational medicinal product (IMP). The trial manager and site pharmacists were able to unmask the allocation if requested, for example in the event of a SAE where knowledge of the allocation would affect the patient’s care. The final decision to unmask lay with the chief investigator or co-lead investigator.

Data collection and follow-up schedule

Participants were followed up for 12 months (except where participants withdrew from follow-up). Data were collected at baseline and 1 week, 4 weeks and 3, 6, 9 and 12 months post randomisation (Table 2) and at 1 week and 4 weeks post restarting the IMP, if applicable. All prospective data were collected face to face during outpatient clinics. Images captured by OCT, AF, FFA, ICGA and colour fundus photography (CFP) taken during research visits were sent in pseudonymised format to the Network of Ophthalmic Reading Centres UK for independent grading by trained assessors masked to the allocations (see Multimodal image grading for central serous chorioretinopathy for details). Retrospective data (pre-screening eligibility and medical history) were collected by local research teams from medical notes and by asking participants for information. Data were recorded on paper CRFs and entered into a secure internet-based database. 26

Colour fundus photography

The presence of bullous collections of SRF, foci of hypo- and hyperpigmentation and exudate, which are common features of CSCR, was recorded. The colour image in Figure 1 shows an example of a region of SRF with a surrounding area of exudate seen as streaks. The colour images were used ad hoc to help adjudicate on retinal features seen with other imaging modalities if these were not clear from a single imaging modality.

FIGURE 1.

Colour fundus imaging. The colour image shows a region of SRF (blue arrow) with a surrounding area of exudate seen as streaks (black arrow). For scale, the optic disc is approximately 1.5 mm in diameter.

Fundus fluorescein angiography

The transit of fluorescein through retinal vessels can be seen clearly, and under normal conditions the dye remains in the retinal circulation. The choroidal circulation is leaky and, as a result, fluorescein enters the extravascular space and is seen as a low-intensity diffuse background level of fluorescence similar to the appearance of ground glass. In CSCR, leakage of fluorescein occurs and results in the appearance of a fluorescein leak during the angiographic frame, which was recorded. The appearance (phenotype) of the leak can vary. A single circular region of leak is termed ‘ink-blot’. Leaks can sometimes track vertically; these have the appearance of a ‘smoke stack’. A third variety, termed ‘chronic epitheliopathy’, appears as speckled diffuse hyperfluorescence, representing a very chronic disease state with no definite morphological features (see Figure 2 for examples). FFA phenotypes (ink-blot, smoke stack, chronic epitheliopathy or no leakage visible) were recorded at baseline and 12 months to address secondary outcome 12. In Figure 2a, two areas of early hyperfluorescence superotemporal and nasal to the optic disc are seen. In the later frame of the angiogram (Figure 2b), these areas of hyperfluorescence increase in size and hyperfluorescence while the fluorescence in the normal retinal blood vessels fades. This is described as a ‘smoke stack’ form of leakage. In a different patient, early (Figure 2c) and late (Figure 2d) frames of a fluorescein angiogram show some pinpoint leakage superior to the fovea, which increases slightly in size in Figure 2d, and a central area of increased hyperfluorescence in the fovea appears in Figure 2d. This is described as an ‘ink blot’ form of leakage. Finally, in a different patient, diffuse areas of chronic retinal pigment epitheliopathy are seen in both early (Figure 2e) and late (Figure 2f) frames of the fluorescein angiogram. These areas of punctate hyperfluorescence are seen superotemporal to the optic disc and temporal and inferotemporal to the fovea.

FIGURE 2.

Fundus fluorescein angiogram phenotypes. (a) Early-frame smoke stack; (b) late-frame smoke stack; (c) early-frame ink-blot; (d) late-frame ink-blot; (e) early-frame chronic epitheliopathy; and (f) late-frame chronic epitheliopathy.

Other FFA variables were graded but were not specified as secondary outcomes, including the timing, morphology and spatial distribution of the leaks. Leaks can occur during any stage of the angiogram. Leaks that occurred within the first 30 to 60 seconds were classified as ‘early’ and those that occurred after 60 seconds were classified as ‘late’. The morphology or pattern of leakage was also determined. Leaks can be at multiple sites of the fundus and, consequently, grading required classification as unifocal or multifocal. The spatial distribution of the leak was also determined as central or peripheral (central within the macular arcades and peripheral beyond macular arcades).

Indocyanine green angiography

The dye indocyanine green binds tightly to large plasma proteins and remains within the choroidal vessels. Consequently, the morphology of the choroidal circulation is better observed with this dye. Because CSCR is primarily a choroidal disease, considerable attention was given to grading of ICGA images. The inclusion in the angiographic run of the arterial phase of the indocyanine green is important and graders were asked to state whether or not appropriate frames were captured. ICGA images were used to identify reduced choroidal permeability at 12 months (secondary outcome 8), which was recorded as ‘yes’ or ‘no’. Figure 3 shows examples of ICGA images taken early and late in the run. In the early frame of the ICGA the retinal vessels are seen extending from the optic nerve (oval dark area at the right of the frame) sweeping across superior and inferior to the macula forming arcades (Figure 3a and c). The choroidal vessels are seen throughout the frame as linear bands of hypercyanescence. These vessels are broader and the cyanescence is of lesser intensity than in the retinal vessels. There is a focal area of punctate cyanescence in the central macula. The late frame of the ICGA shows multiple foci of hypercyanescence at the fovea, in the macular arcades and in the superior part outside the macular arcade, representing leakage from choroidal vessels.

FIGURE 3.

Early- and late-frame indocyanine green angiograms. (a) Early frame showing abnormally dilated choroidal vasculature with flow voids; (b) late frame showing multiple foci of leakage with staining of the RPE (punctate hypercyanescence); (c) early frame showing normal-looking choroidal vasculature; and (d) late frame showing leakage from choroidal vessels with areas of hypercyanescence and punctate staining of the RPE.

Other key features of interest that were graded but did not contribute to the secondary outcomes were spots (circular regions of hypercyanescence) and plaques (irregular regions of hypercyanescence), the number of plaques and the area of the largest plaque. We also recorded whether or not the position of the largest plaque was subfoveal. Temporal characterisation was also important and abnormal hypercyanescence was classified as appearing ‘early’ or ‘late’ and with a spatial distribution ‘macular’ or ‘peripheral’. The visibility of the individual choroidal vessels in the early and late ICGA frames and the presence of dilated vessels anywhere in the angiographic frame was noted.

Optical coherence tomography

Optical coherence tomography allows the assessment of the relationships between the retina, RPE and choroid in unprecedented detail. The phenotypic characterisation of the vascular choroid, including details of the vascular profiles and retinochoroidal interface, has been markedly enhanced, permitting assessments of the presence of choroidal varices, engorged dilated choroidal vessels and diffuse thickening of the overall choroidal structure. In addition, changes in the outer retinal layers – namely, presence of SRF, disorganisation and/or shedding of the photoreceptors into the subretinal space, and intactness of the external limiting membrane and ellipsoid zone – could be ascertained. Changes at the level of the RPE/Bruch’s membrane, such as thinning or thickening of this structure or presence of small foci of pigment epithelial detachments, were also identified. Among other descriptive measures, the presence of outer retinal atrophy resulting in increased transmission of the OCT signal was captured.

Optical coherence tomography images can be used for measurements such as of the vertical height of pockets of fluid as well as of the thickness of tissue compartments. In addition, automated outputs of central macular volume are available with the current generation of OCT capture systems. For the VICI trial, measurements were made on the foveal image and included the height of SRF and choroidal thickness at the fovea and the thickest point.

Optical coherence tomography images were used to measure CSRT (µm) at baseline, 4 weeks and 3, 6, 9 and 12 months (secondary outcome 2); the change in SRFT (µm) at 12 months, adjusted for baseline (secondary outcome 3); and the thickness of the choroid (µm; measured subfoveally) at 12 months (secondary outcome 7). In addition, OCT was used to assess the resolution or reduction of SRF (secondary outcomes 9, 10) and the time taken for reduction or resolution of SRF to occur (secondary outcome 13).

Example OCT images are shown in Figure 4. The OCT image in Figure 4b, a cross-section of the fovea (bold line of the raster shown in the reflectance image), displays the neurosensory retina with the multiple layers of neural cells. The innermost fine hyper-reflective band is the internal limiting membrane. The outer two bright bands represent the outer regions of the photoreceptors and the RPE. The choroid is the crescent-shaped area located on the outside of the RPE. Vascular profiles, surrounded by intervascular pillars, are visible within the choroid. In the central region of the OCT image in Figure 4b there is a region of dark hyporeflectivity separating the neurosensory retina from the pigment epithelium, which represents the collection of SRF.

FIGURE 4.

Optical coherence tomography images. (a) Baseline OCT image of an eye with CSCR. The dark area lying between the retina and the pigment epithelium (blue arrow) represents a region with SRF. (b) OCT image of the same eye at 12 months. The area of SRF has reduced (blue arrow). The white arrow shows an area of residual disturbance of the RPE and outer retinal layers, seen as an irregular region of hyper-reflectivity. (c) OCT image at 12 months from another participant showing complete resolution of SRF (blue arrow). The retina is thinned and the outer nuclear band (white arrow) has almost disappeared in the region where SRF was originally present.

Biobank samples

Pre randomisation, consenting participants donated approximately two 10-ml samples of blood in ethylenediaminetetraacetic acid vacutainers and one 10-ml sample of blood in clot-activating gel vacutainers. Samples were posted at room temperature to the University of Southampton and processed into DNA, plasma and serum components, which were stored at –80 °C for use in future ethics-approved studies. Donating blood samples to the biobank was an optional aspect of study participation.

Adherence monitoring

Participants’ adherence to the IMP regimen was monitored throughout the trial. Participants were asked to return used bottles of capsules from their previous follow-up visit at their next follow-up visit (if applicable; see Safety criterion for dose escalation and continuation of study drug and Central serous chorioretinopathy resolution and recurrence treatment criteria). Returned capsules were counted by clinical trials pharmacists and recorded in a secure web-based IMP management system. 22 The number of expected capsules consumed was calculated as the number of days from the date the participant was informed to continue on the IMP to the date of the follow-up visit at which the capsules were returned, or the date of the follow-up visit at which the capsules should have been returned if the follow-up visit was missed. (If a visit was missed, no IMP could be prescribed for the subsequent period and monitoring adherence did not apply.) If participants took > 70% of the expected number of capsules in a follow-up period, they were categorised as adherent. During protocol development, the adherence threshold was discussed by the trial management group and a consensus was reached, defining satisfactory adherence as > 70% of pills that should have been taken in accordance with the protocol. The threshold was subsequently discussed and agreed at an investigators’ meeting.

If a participant failed to return used bottles of capsules, the bottles were categorised as ‘lost’. It was assumed that ‘lost’ bottles contained no capsules because the participant had taken them all and discarded the bottle. A sensitivity analysis to test the effect of this assumption on the primary outcome was performed. This was performed by using multiple imputation to impute the number of pills remaining in bottles confirmed as ‘lost’.

Emergency unmasking

In the event of a medical emergency, research personnel followed working instructions to facilitate unmasking of a participants’ allocation by the trial manager at the BTC-CTEU (during working hours) and local site pharmacists (outside working hours).

1. The internet-based IMP management system that was used by sites and the BTC-CTEU to allocate, track and account for bottles of IMP throughout the trial had an unmasking function. On completion of a minimum data set (name of person requesting unmasking; reason for unmasking; person doing the unmasking) the system would unmask a participants’ allocation. The unmasking function was role-restricted to users who were delegated pharmacy or trial administrator permissions.

2. To enable emergency unmasking by out-of-hours on-call pharmacists who were not registered on the IMP management system, a generic username and password specific to each site were created and disclosed only to site pharmacists. This ensured that, in a medical emergency, local pharmacy personnel would be able to unmask a participant without needing to register and await approval from the trial manager.

3. To enable unmasking in the event of internet failure, local pharmacies were provided with sealed code-break envelopes for the 25-mg-dose bottles of IMP. Every participant, irrespective of overall time on IMP, was allocated an initial bottle of 25-mg capsules at randomisation but would not be guaranteed to receive subsequent bottles of IMP. Therefore, it was necessary to produce code-break envelopes for the 25-mg bottles only. Pharmacists were instructed to look up the first bottle number allocated to the participant in the event of an unmasking event. Envelopes were manufactured by Newcastle Specials Pharmacy Production Unit. Code-break envelopes were retrieved by the BTC-CTEU at the end of the trial and inspected for evidence of tampering.

Unmasking events were monitored by the trial manager. If a participant’s allocation was unmasked via the above options 1 or 2, an automated e-mail alert to the trial e-mail address was triggered. In the event of unmasking via option 3, local pharmacists were required to report having unmasked allocation to the trial manager at the earliest opportunity. Code-break envelopes were returned to the co-ordinating centre at the end of the trial.

Statistical methods

The analysis and safety populations comprised all participants randomised into the study, excluding those who withdrew and were unwilling for their data already collected to be used. All analyses were directed by a pre-specified SAP (see the NIHR project web page)21 and performed on an intention-to-treat basis. Continuous variables are summarised using the mean and SD or, if the distribution was skewed, the median and interquartile range (IQR). Categorical variables are summarised as number and percentage.

The intention was to adjust all models for the two stratification factors included in the randomisation: visual acuity level (low, ETDRS BCVA score of 54–67 letters; high, ETDRS BCVA score of 68–85 letters) as a fixed effect and centre as a random effect. Owing to small numbers of patients at some centres, it was not always possible to estimate a random effect for each centre. If this was the case, centres with a small number of patients were combined (e.g. centres with one or two patients were grouped and centres with three or four patients were grouped). For time-to-event outcomes it was not possible to fit centre as a random effect, so this was accounted for by using a clustered sandwich estimator to adjust the standard errors of the estimates for clustering within centre.

The primary outcome and other longitudinal outcomes (LLA and CSRT) were analysed using linear mixed-effects methods, with treatment effects presented as mean differences (MDs) with 95% confidence intervals (CIs). Both time and time by treatment interactions were added to each of the models as fixed effects to allow the treatment effect at 12 months to be estimated. Baseline values measured before treatment had started were also added to each model as a fixed effect. Difference variance/covariance structures were assessed and the structure that provided the best fit was used to model within-patient errors. Likelihood ratio tests determined that an unstructured covariance structure provided the best fit in all longitudinal models. Time-to-event outcomes were analysed using proportional hazards parametric survival models for interval-censored data, and treatment estimates presented as hazard ratios (HRs) and 95% CIs. Complete resolution of SRF and complete or partial resolution of SRF were censored at the time of last follow-up if SRF did not resolve completely or partially. Disease recurrence was censored at time of last follow-up if disease did not recur. Continuous outcomes measured at 12 months [SRFT, patient-reported visual function (NEI-VFQ-25) and choroidal thickness] were compared using linear regression, adjusted for baseline values, with treatment estimates presented as MDs. For all methods outlined, underlying assumptions were checked using standard methods (e.g. tests for normality, residual plots) and if assumptions were not valid then alternative methods of analysis were sought.

A pre-specified exploratory analysis was performed for the primary outcome, BCVA score, to assess the effect of adherence and treatment. Two indicators, one for treatment (on treatment vs. not on treatment) and one for adherence (adhered vs. did not adhere), were generated for each post-randomisation time point at which BCVA was assessed. If a patient was not on treatment owing to disease resolution, they were considered adherent; if a patient was not on treatment for any other reason, they were deemed non-adherent. Interactions between treatment group and these indicators were added to the model and indicator status-specific treatment effects were estimated. A sensitivity analysis reassessing the effect of adherence and treatment on BCVA score after imputing pill counts for lost bottles was also performed.

A sensitivity analysis pre-specified in the SAP but not in the protocol was performed for time-to-event outcomes – complete resolution of SRF and complete or partial resolution of SRF – because an initial review of baseline data found imbalances between treatment groups in SRF. This was performed by adding baseline SRF level to the models.

A post hoc sensitivity analysis of BCVA, CSRT, SRFT and choroidal thickness was performed, which involved adjusting for whether or not the participant had received PDT. These were performed as the proportion of patients receiving PDT throughout the trial differed between the treatment groups. For outcomes measured at multiple time points (BCVA and CSRT), an indicator was added to the model that was updated to ‘yes’ at the point at which the participant received PDT. For outcomes measured at 12 months only (SRFT and choroidal thickness), an indicator for whether or not the participant had received PDT in the study eye at any point throughout the trial was added to the model.

For hypothesis tests, two-tailed p-values of < 0.05 were considered statistically significant. Likelihood ratio tests were used in preference to Wald tests. No formal adjustment was made for multiple testing. However, formal statistical comparisons were not made for outcomes with low event rates and consideration has been given to the number of tests performed when interpreting the results.

All analyses were performed in Stata^® version 15.1 (StataCorp LP, College Station, TX, USA).

In all tables, missing data are described in footnotes. Rules for imputing missing data were outlined in the SAP and depended on the level of missing data. Multiple imputation was required for NEI-VFQ-25 score, choroidal thickness and SRFT analyses because missing data at 12 months accounted for 10%, 9% and 8% of observations, respectively. Imputation by chained equations was used to generate multiple complete data sets (using the Stata -ice- command) and results were combined using Rubin’s rules.

To set our findings in context, we did a systematic review and meta-analysis, searching for other trials comparing eplerenone and placebo [identified using medical subject headings (MESH) by searching for the following: ‘central serous choroidal retinopathy’ AND (‘eplerenone’ OR ‘mineralocorticoid receptor antagonists’) AND ‘publication type = randomized controlled trial’]. We included trials that investigated eplerenone treatment in patients with chronic CSCR. Trials were excluded if they did not investigate eplerenone or included only patients with acute CSCR. We assessed the risk of bias29 and re-analysed data reported by the trials to obtain treatment effects for BCVA score and SRF thickness using the same metric (i.e. the difference between groups in the change in BCVA score and SRF thickness from baseline to the end of treatment). If it was not possible to calculate the SD of the mean change from baseline for a group, an average of the SDs from the other studies was used. The treatment effects were synthesised in a fixed-effects meta-analysis to estimate weighted MDs between eplerenone and placebo groups according to the inverse variance method.

Patient and public involvement

Before the trial commenced, a representative (Geraldine Hoad) from the Macular Society (URL: www.macularsociety.org; accessed 2 September 2020) provided input on the PIL to ensure that the information provided to patients was acceptable and clear.

A patient and public involvement (PPI) group comprising VICI trial participants was formed during the trial after all participants had been recruited. Participants from nine sites, selected by geographical proximity to London where the meetings would be held, were invited to join the PPI group by local research teams. Two PPI meetings were held: one during follow-up and one when the results were available. An agenda was prepared for the first meeting, with input from Caroline Barker and Francesca Lambert (PPI facilitators, University Hospital Southampton NHS Trust).

Topics covered at the first meeting were as follows:

• How did you find out about the VICI trial?

• Decision-making processes behind taking part in research.

• Getting to your trial visits. (We aimed to find out what made attending regular follow-up visits possible and whether or not there were barriers to attending visits.)

• Taking the study tablets regularly. (We aimed to find out what methods participants used to remind themselves to take the study drug daily.)

• Introduction to the biobank.

• Taking part in future events as a public representative.

A second PPI meeting was held after the analysis stage to develop a results leaflet that was understandable and acceptable to trial participants.

Screened patients

In total, 402 patients were screened for the trial (Figure 5). During the initial eligibility screening stage from patient medical records, 97 patients were found to be ineligible. Eighty-two initially eligible patients were not approached for consent. Among patients who were eligible and given a PIL, 44 declined to consent to trial-specific screening assessments and participation in the trial. A total of 164 initially eligible patients consented to trial-specific screening assessments, of whom 64 were ineligible. One patient who was eligible and initially gave consent withdrew shortly before randomisation. See Figure 5 for reasons why patients were excluded from the trial at each screening stage. There was variation in the ratio of screened to randomised participants per site because of the different methods of screening employed by sites. Therefore, the proportions of screened patients classified as eligible and consenting should be interpreted with caution.

FIGURE 5.

The VICI trial Consolidated Standards of Reporting Trials (CONSORT) flow diagram. CTIMP, clinical trial of investigational medicinal product. Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Randomised patients

The first site started to recruit (i.e. to screen patients) 14 December 2016. Recruitment closed 28 February 2018. A total of 114 participants were randomised between 11 January 2017 and 22 February 2018 (Figure 6). Follow-up (i.e. last patient, last visit) was completed on 28 February 2019.

FIGURE 6.

Predicted and actual recruitment.

The trial over-recruited by 10 participants for ethical and logistical, but not methodological, reasons. At the point of reaching the recruitment target of 104 participants, participants who had already consented and were undergoing screening continued to enter the trial if eligible up to the point of recruitment closing.

Representativeness of the participants

Baseline characteristics of patients are presented in Table 3 (separately for those who were ineligible at initial screening, eligible at initial screening but who did not consent, consented and underwent trial-specific screening assessments but who were not randomised because they were found to be ineligible, and randomised). On average, patients who were ineligible at initial screening were older than all other patients, who were of similar age. A lower proportion of men were ineligible at screening (65%) and a higher proportion were eligible but did not consent (85%) compared with those who were consented (75%).

Participant withdrawals

Twenty-six participants permanently ceased taking the study drug during the 12-month follow-up period; 18 out of 26 participants agreed to continue with follow-up to 12 months and 8 out of 26 participants withdrew from follow-up before 12 months. The reasons for stopping the study drug and subsequent withdrawal from follow-up, if applicable, are detailed in Table 4. Time to stopping the study drug is shown in Appendix 4, Figure 21. Follow-up time classified as ‘on’ or ‘off’ IMP for each participant is shown in Figure 7.

FIGURE 7.

Time participants were on and off the IMP during the trial. (a) Placebo; and (b) eplerenone. Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Protocol deviations

Few protocol deviations occurred during the trial (Table 5). All participants were dispensed the correct drug and followed the correct dosage regimen at all time points. Two participants, one in the eplerenone group and one in the placebo group, were informed to continue taking the study drug despite having a serum potassium level > 5.0 mmol/l. In both cases the deviation was detected by the trial manager through remote central monitoring and the participants were informed to stop taking the study drug 3 weeks after the blood test result. The most common protocol deviation was visits attended outside of the specified visit window. This deviation occurred equally between groups. Distributions of follow-up visits around the visit windows are shown in Appendix 4, Figure 22.

Numbers analysed

The analysis population consisted of all 114 randomised participants; no participant who withdrew did not consent for their data to be used. The number of participants included in the analysis of each outcome is presented in Table 7.

Adherence to the investigational medicinal product regimen

All participants in the eplerenone and placebo groups adhered to the IMP regimen between baseline and week 1 according to the definition of taking > 70% of the expected number of capsules in that time period, as described in Chapter 2, Adherence monitoring. This is important as decisions to increase the dose from 25 mg to 50 mg were made on the assumption that the participant had been adherent and potassium levels remained at ≤ 5.0 mmol/l. At subsequent time periods, at least 80% of participants were adherent (according to the same definition) and adherence was well balanced between the groups (Table 8). Out of 497 bottles dispensed to participants in the placebo group and 559 bottles dispensed to participants in the eplerenone group, 46 and 57 were not returned, respectively (see Appendix 4, Table 34). It was assumed that these bottles were discarded empty (i.e. the participant had taken all the pills and discarded the bottle), so the capsule counts were recorded as zero.

Additional treatments administered as usual care

Clinicians were permitted to administer usual care for CSCR during the 12-month follow-up period at their discretion (Table 9). No participants received thermal laser therapy. One participant in the placebo group received subthreshold laser therapy in their study eye while prescribed the study drug. In the placebo group six patients received nine PDT treatments (i.e. some patients received PDT more than once) in their study eye; two of these six patients were prescribed the study drug at the time of administration. In the eplerenone group three participants each received one PDT treatment; one of these participants received PDT in their study eye and was prescribed the study drug at the time of administration.

Participant characteristics and data at baseline

The mean participant age at baseline was 48.7 ± 7.6 years (placebo, 49.9 ± 7.9 years; eplerenone, 47.4 ± 7.1 years) and approximately three-quarters were male. The median duration of CSCR before randomisation was 9 months overall and was well balanced between groups [placebo, 9 months (IQR 6–18 months); eplerenone, 8 months (IQR 6–22 months)]. See Table 10 for baseline demography, blood test results, medical history and ocular examination data by group and overall, and Table 11 for baseline outcome data by group and overall.

Biobank sample collection

A total of 109 participants donated DNA, plasma and serum samples to the biobank at the University of Southampton for use in future ethically approved studies.

Best corrected visual acuity

Median BCVA score increased by about four letters in both groups during follow-up (Figure 8). Modelled mean BCVA score at 12 months was 80.4 letters (SD 4.6 letters) in the eplerenone group and 79.5 letters (SD 4.5 letters) in the placebo group. Mean BCVA score at 12 months did not differ between the groups (eplerenone minus placebo: 1.73 letters, 95% CI –1.12 to 4.57 letters; p = 0.24; Table 12 and Figure 8). In the exploratory analysis, adherence to treatment did not change the effect of eplerenone on BCVA score, irrespective of whether or not pill counts were imputed for lost IMP bottles (Tables 13 and 14). In the post hoc analysis, which adjusted for photodynamic therapy administered during follow-up, the effect of eplerenone on BCVA score was unchanged (see Appendix 4, Table 35).

FIGURE 8.

Primary outcome: BCVA score over time. Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Exploratory analyses of the effects of age and confluent/granular hyperautofluorescence on BCVA score in the trial cohort are shown in Appendix 4, Figure 23. Neither factor influenced BCVA score (age: –0.16 letters per year of age, 95% CI –0.35 to 0.04 letters per year of age; presence of confluent/granular hyperautofluorescence: –1.75 letters, 95% CI –4.78 to 1.28 letters).

Low-luminance visual acuity

Median LLA score appeared to be lower in the eplerenone group at baseline but the IQRs for the groups clearly overlapped (see Appendix 4, Table 36). Therefore, median LLA score appeared to improve more in the eplerenone group (9 letters) than in the placebo group (1 letter) (Figure 9). The modelled LLA score mean at 12 months in the eplerenone group was 63.9 letters (SD 4.8 letters) and in the placebo group was 65.3 letters (SD 3.7 letters). Mean LLA score at 12 months did not differ between the groups (eplerenone minus placebo: 0.69 letters, 95% CI –3.79 to 5.02 letters, p = 0.79).

FIGURE 9.

Low-luminance visual acuity over time. Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Patient-reported visual function using the NEI-VFQ-25 questionnaire

Higher NEI-VFQ-25 scores represent better visual function. Median NEI-VFQ-25 scores and IQRs at baseline and 12 months for eplerenone and placebo groups, overall and for each questionnaire domain, are shown in Appendix 4, Table 37. There were no differences between groups [modelled means were 86.5 letters (SD 5.3 letters) for the eplerenone group and 89.1 letters (SD 4.4 letters) for the placebo group]. Mean NEI-VFQ-25 score at 12 months, adjusted for baseline score, did not differ between the groups (eplerenone minus placebo:–2.39 letters, 95% CI –5.45 to 0.68 letters; p = 0.13) (Figure 10).

FIGURE 10.

Effect size for patient-reported visual function.

Resolution and recurrence of central serous chorioretinopathy

Classification of CSCR in study eyes as having resolved completely, resolved partially or not having responded (based on resolution of SRF) to IMP at scheduled time points is shown in Table 15, by eplerenone or placebo group. CSCR could resolve completely at one time point but then recur. Figure 11 shows follow-up time by group for each participant, classified as not completely resolved or completely resolved. Table 16 shows the number of participants’ fellow eyes with any, and new, CSCR.

FIGURE 11.

Time participants did or did not have complete resolution of SRF during the trial. Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Table 16 shows the number of participants in each group with complete resolution of SRF at any time during follow-up, with recurrence of SRF after complete resolution and with complete or partial resolution of SRF at any time during follow-up. The table also describes estimates by group of median times to complete resolution of SRF, recurrence of SRF and complete or partial resolution of SRF, and hazard ratios (HRs) for the effects of eplerenone versus placebo on time to complete resolution (HR 0.78, 95% CI 0.41 to 1.51; p = 0.463), time to recurrence of SRF (after complete resolution) (HR 1.10, 95% CI 0.45 to 2.66; p = 0.836) and time to complete or partial resolution (HR 1.23, 95% CI 0.75 to 2.01; p = 0.418). Note that estimated median times to complete resolution of SRF are outside the period of follow-up (i.e. > 1 year) because < 50% of participants experienced complete resolution of SRF during the study. See Appendix 4, Table 38, for data contrasting the results of the primary and sensitivity analyses, adjusting for baseline SRF, for time to complete, and complete or partial, resolution of SRF.

Findings from optical coherence tomograms

Median SRFT and choroidal thickness at baseline and 12 months in each group are shown in Table 17. In addition, these tables (see also Figure 13) show the treatment effects of eplerenone versus placebo, both of which favoured the placebo group (eplerenone minus placebo: SRFT –48.08 µm, 95% CI –82.73 to –13.43 µm, p = 0.01; choroidal thickness –38.53 µm, 95% CI –64.74 to –12.31 µm, p = 0.004).

Median CSRT in each group at each time point is shown in Appendix 4, Table 39, and Figure 12. There were no differences between the groups at 12 months, as evidenced by no difference in the estimated treatment effect of eplerenone versus placebo on CSRT, adjusted for baseline CSRT (eplerenone minus placebo: 24.4 µm, 95% CI –7.9 to 56.6 µm; p = 0.142) (Figure 13).

FIGURE 12.

Central subfield retinal thickness over time. Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

FIGURE 13.

Effect sizes of CSRT, SRFT and choroidal thickness.

The choroid of fellow eyes was usually graded as thickened, despite clinical judgements that CSCR was mainly absent in fellow eyes (see Table 16). A post hoc analysis of choroidal thickness in fellow eyes, carried out to investigate the consistency of the effect favouring placebo in study eyes (because eplerenone is a systemic treatment), is shown in Table 18. The effects of eplerenone versus placebo on choroidal thickness for study and fellow eyes are contrasted in Figure 14. The post hoc analysis shows that the fellow eye effect was consistent with the study eye effect.

FIGURE 14.

Choroidal thickness in study eye and fellow eye. Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Reproduced with permission from Lotery et al. 30 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license, which permits others to copy and redistribute the material in any medium or format, for non-commercial use, with no derivatives, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc-nd/4.0/. The figure includes formatting changes to the original.

Post hoc analyses of CSRT and choroidal thickness were also carried out to take account of having had PDT administered to the study eye during follow-up. See Appendix 4, Table 40, for data contrasting the results of the primary analyses for these outcomes and the post hoc analyses, which were very similar.

Fluorescein and indocyanine green angiographic and retinal pigment epithelium autofluorescence outcomes

Study eyes with FFA images available at both baseline and 12 months (n = 104) were included in the FFA analysis. All study eyes showed leakage at baseline. The most common baseline phenotype in both groups was ink-blot, followed by chronic epitheliopathy and then smoke stack. Table 19 shows the FFA phenotype of study eyes at baseline cross-tabulated by phenotype at 12 months in the eplerenone and placebo groups. In both groups at 12 months, approximately half of phenotypes remained the same as at baseline, approximately one-third converted to a different leakage phenotype and just under one-fifth showed no visible leakage.

In both the eplerenone and placebo groups the proportion of participants with reduced choroidal permeability at 12 months was low (Table 20). Statistical comparisons of treatment effects were not performed as fewer than 10 events occurred in total.

Table 21 shows the area of macular RPE hypoautofluorescence (mm²) at baseline and 12 months and the area change at 12 months compared with baseline in the eplerenone and placebo groups. Only eight patients in total had macular RPE hypoautofluorescence at baseline and/or 12 months, so no statistical comparisons between groups were made.

Amendments to the patient information leaflet

The PIL was amended following review by a representative from the Macular Society. Minor clarifications were made to the wording to make it more concise and understandable to patients. The full list of possible side effects was moved from the main body of the text to an appendix at the end as it was suggested that the list could be overwhelming for patients. The main potential side effect, hyperkalaemia, was retained in the main body of text.

Patient and public involvement group

Eight participants (five men and three women) from six sites opted to join the PPI group and attended the first meeting on 21 June 2018. Five of the original eight PPI group members (two men and three women) from three sites attended the second meeting on 22 June 2019.

Main findings

Patient and public involvement

Input from representatives of the Macular Society was invaluable for producing a PIL that was understandable and acceptable to participants. Moving the list of potential drug side effects from the main body of the text to an appendix made the PIL less overwhelming and improved the flow of text while ensuring that this important information was still available to potential participants to aid their decision-making. This PPI-guided PIL format is being applied in new studies coordinated by the BTC-CTEU. However, despite the relocation of the side effects, the PPI group still reported the listed side effects as off-putting when considering whether or not to participate in the trial.

An unusual aspect of the PPI group formed for the VICI trial was the recruitment of active trial participants during the trial follow-up period. We assembled the group in this way because attendance at follow-up visits and adherence to the intervention was high and loss to follow-up and withdrawal rates were low. This presented an opportunity to explore why a RCT was going well, something that is rarely investigated or documented. There was mild concern that (1) we risked unmasking if participants compared their symptoms or disease resolution status during the meeting and (2) participants would withdraw from the trial if they became upset by discussions at the meeting. To address point (1), participants were reminded at the beginning of the meeting of the masked design of the trial and why it is important, and they were asked not to discuss and compare their CSCR symptoms. The PPI group were happy with this request and respected it. The risk of point (2) occurring was considered low. However, careful planning ahead of the meeting ensured that agenda items were appropriate and non-inflammatory. Two experienced PPI facilitators were consulted to ensure that the language used in the meeting was appropriate and constructive.

An important point raised by the PPI group was to ensure that any substudies, such as the substudy in this trial in which participants donated samples to a biobank for future research, are clearly explained to patients prior to obtaining consent. Although information about the biobank samples was included in the PIL and was separated from the main trial information in the consent form, it was advised that more could be done to make this type of information apparent. The suggestions made by the group have been taken onboard and will be implemented in future studies. The outcome of the PPI meeting was shared with teams at participating sites.

The feedback received from some members of the PPI group that the NEI-VFQ-25 was, in their view, not suitable for assessing visual-function-related quality of life in patients with CSCR was taken on board. At present, a quality of life assessment tool specifically for patients with CSCR is an unmet need. Further research is required to develop and validate an appropriate assessment tool.

The PPI group partly attributed the ease of taking the study medication to the lack of restrictions on taking the capsules with regard to time of day and food intake. This is an important consideration for future studies as flexibility could minimise burden on participants and lead to improved adherence. However, such flexibility is not always possible depending on the guidelines for the study drug. Where restrictions are mandated by guidelines it could be useful to identify alternative strategies that make adherence easier, such as sending regular text message reminders to participants or implementing online or paper diaries.

Gaining PPI input on the content of the results leaflets from a subset of trial participants was highly beneficial because the perceived priority of different results differed between them and the professional stakeholders. We highly recommend engaging with trial participants or members of the public when preparing dissemination documents for a lay audience. We also learned from the participants that the NEI-VFQ-25 was, in their opinion, not an appropriate patient-reported outcome measure (PROM) for assessing their quality of life. It would be beneficial in future studies of CSCR to use a PROM that is sensitive enough to detect improvements in quality of life in patients with CSCR.

Strengths and limitations

The VICI trial had several strengths. All but three participants contributed to the primary analysis and participants attended almost all scheduled visits. There were no protocol deviations compromising the treatment comparisons. The trial was powered to detect a clinically important difference in BCVA targeted by several large multicentre trials of treatments for retinal conditions and,31–33 in fact, had more power than anticipated owing to participants attending almost all scheduled visits.

Limitations included the need to discontinue treatment if CSCR completely resolved during follow-up or an elevated serum potassium level was detected. These issues may have reduced the observed treatment effect but were required to safeguard participants. Hyperkalaemia was detected in 14% of participants using the potassium threshold adopted for the study of 5.0 mmol/l but, in all but one of these patients, the absolute potassium level that triggered discontinuation of IMP was below the threshold for reducing the dose or withdrawing the treatment as specified in the SmPC for eplerenone for use in its licensed indication (5.5–5.9 mmol/l: reduce dose or withdraw treatment if on lowest dose; ≥ 6.0 mmol/l: withdraw treatment). We adopted the lower threshold for the trial because we wanted to ensure the safety of participants who did not have heart failure and therefore had no opportunity to benefit from eplerenone for this indication. However, participants in the trial were younger and fitter than patients who are usually prescribed eplerenone for heart failure and, with hindsight, we think that the established potassium threshold would have been more appropriate. We discontinued IMP when CSCR completely resolved because, when this happened, participants had no opportunity to benefit from the IMP but would have been at risk from possible side effects. Discontinuing IMP when CSCR completely resolved allowed us to compare recurrence of CSCR after complete resolution. We observed no difference in time to recurrence between groups.

The main protocol deviation was study visits taking place outside the visit window, which occurred equally between the two groups. The main reasons for this were participants taking holidays and having work commitments. Occasionally, study visits were delayed owing to staffing issues at sites. Owing to the controlled supply of the study drug (i.e. dispensing only enough to last between follow-up visits) and the safety monitoring required, we were not able to expand the visit windows. Such issues are likely to occur in future studies in this population. We will investigate strategies to make follow-up more efficient to reduce the amount of time participants spend in clinics and the resources research teams need to attribute to follow-ups.

Our inability to control the use of co-treatments could have introduced bias if these were used differentially by group. We observed that more participants in the placebo group were treated with PDT than in the eplerenone group. However, co-interventions (including PDT) were used infrequently. A post hoc analysis that adjusted for the imbalance between groups in the proportion of study eyes treated with PDT produced results that did not differ from the results of the primary analysis. We believe that our results are generalisable to a wider population of patients with CSCR. The visual acuity eligibility criterion for the trial meant that some patients, who would otherwise have been eligible, were excluded because their BCVA score was too good (see Main findings, Study conduct). However, we have no reason to believe that the effects of eplerenone in these patients would have been any different. This view is supported by the consistent effect of eplerenone on choroidal thickness in the study and fellow eyes of participants, as fellow eyes could be considered to represent similarly less severe CSCR (i.e. those patients whose BCVA score was too good). In this study, fellow eyes were judged by ophthalmologists not to have CSCR, but grading of OCTs showed considerable choroidal thickening. The same logic would apply to patients who did not meet the trial criterion for chronicity. Requiring eligible participants to have SRF tended to shift the threshold towards chronicity. We required the duration of the current CSCR episode to be ≥ 4 months but did not consider whether or not a participant had had a previous episode of CSCR, and if so how many, or how quickly a previous episode or episodes had resolved. In other respects, the characteristics of the trial participants were similar to the wider CSCR population.

Neovascular age-related macular degeneration (nAMD) is a common phenocopy of CSCR. To avoid nAMD cases being inadvertently recruited into the trial, we did not recruit any patients aged > 60 years because nAMD usually presents at a later age than this. In addition, we checked for nAMD with fluorescein angiography and ICGA at baseline. Presence of CNV at baseline was an exclusion criterion, as was a history of previous anti-VEGF therapy. These criteria, we believe, stopped any patients with nAMD phenocopies entering the study.

We reduced the risk of patients with dome-shaped maculopathy being recruited with our exclusion criterion of myopia of > –6 dioptres. Patients on MEK inhibitors can develop SRF similar to CSCR. However, no patients on such drugs were recruited.

Lessons for the future

There are two lessons for the future in relation to manufacture of an IMP for trials.

The first lesson is that triallists should obtain quotes for IMP manufacture from several pharmacies. We obtained three quotes, which varied by ≈ £100,000 for the same specification. Triallists are rarely experts with respect to preparation/formulation of the drug of interest as an IMP or issues relating to manufacture of an effective placebo and may have to choose a supplier with little information about the track record of specific pharmacies. At the outset, it would have been helpful to have a checklist of issues to consider and a list of ‘accredited’ pharmacies. From its oversight of a wide range of studies, the NIHR may be well placed to help triallists in this respect. In this trial, budgeting for the IMP was made unnecessarily complicated because the manufacturing pharmacy did not distinguish the cost of the active drug from the cost of manufacturing the placebo and over-encapsulation of both drug and placebo. This prevented the sponsor from attributing the cost of the IMP to treatment versus research cost categories at the time of preparing the grant application. Triallists require manufacturing pharmacies to itemise research quotes accordingly.

Given the problems we experienced with low stock and expiry times, the second lesson relates to the amount of IMP (drug and placebo) ordered from the manufacturing pharmacy. We recommend that triallists order substantially more IMP than required simply on the basis of the number of scheduled doses. This recommendation is tempered by the need to weigh up the cost of wastage against the risks to the trial of running out or the expense of moving stock around. Triallists also need to consider potential threats to the supply of IMP owing to issues such as a delayed start to recruitment or slower than projected recruitment.

There is also a lesson for future trials on treatments for CSCR that relates to visual acuity. The fact that visual acuity in eyes with CSCR can be improved by prescribing additional positive optical power raises the question of how a visual acuity eligibility criterion should be defined, or if eligibility needs to defined in terms of visual acuity at all. One possibility would be to specify the visual acuity criterion with the most up-to-date refraction prescription preceding the onset of CSCR, but this would be difficult to implement. An alternative would be to not define eligibility in terms of visual acuity. This is attractive because visual acuity does not capture well the visual problems that patients experience.

If visual acuity does not reflect the visual problems that patients experience, there is also a question about whether or not it should be the primary outcome in trials of CSCR. This question is particularly relevant when considering a trial of PDT because PDT might be expected to reduce BCVA scores by a few letters but still be effective in treating the condition (i.e. avoiding the distortion symptoms about which patients complain and avoiding future costs to the NHS of managing the condition). There needs to be a debate involving all stakeholders about the pros and cons of choosing different outcomes (e.g. BCVA, retinal fluid, health-related quality of life, psychometric test of visual distortion) as the primary outcome. Other functional measures of vision such as microperimetry also need to be considered.

The disadvantage of not using visual acuity as an eligibility criterion or as the primary outcome is the difficulty in putting CSCR in context alongside other visually disabling conditions. The challenge in CSCR trials is that the range of visual acuity loss varies widely. Some patients have severe visual loss, but these patients may be excluded from entering trials owing to irreversible structural damage. Patients more representative of the majority of CSCR patients may have refracted visual acuities in the normal visual range. Other functional measures such as microperimetry may be a better end point for future clinical trials. However, microperimetry is not used routinely in clinical practice and so many eye departments do not have access to such equipment. A future study correlating structural changes (e.g. SRF) with microperimetry would be useful. If they were strongly correlated, then presence of SRF could be considered as a trial end point because SRF is easily measured with OCT equipment, which is readily available in eye departments.

Future research

We believe that there would be considerable value in following the VICI trial cohort for longer. As described in Strengths and limitations, participants could have had previous episodes of chronic CSCR before the one that qualified them for the trial; some had complete resolution during the trial and some of those had a recurrence of CSCR during the trial. The natural history of resolution and recurrence of CSCR has not been well characterised. Notwithstanding the fact that we did not characterise episodes of CSCR before recruitment, the trial cohort represents the most comprehensively documented CSCR cohort, and information about the subsequent CSCR experience of participants (e.g. via an annual follow-up) would be valuable to both ophthalmologists and future patients.

With respect to other potential treatments, the PLACE trial raised the possibility that half-fluence PDT could be effective. 12 We do not consider this single trial to be sufficient evidence to conclude that half-fluence PDT is effective because of flaws in the trial, notably the lack of masking. Nevertheless, we believe that the trial should be replicated. It is difficult to know what capital investment would be required to do such a trial in the UK. Most UK centres have PDT equipment (it was previously used to treat nAMD), but it may have been decommissioned or be obsolete. 34 Verteporfin is expensive but the patent expires in October 2020. Moreover, if multiple patients attend one clinic, one vial of verteporfin can be used to treat three or four patients with half-fluence PDT. If such a trial were to be designed, we recommend that microperimetry be considered as the primary outcome. However, investment in microperimetry equipment would probably be needed.

Contribution of authors

Professor Andrew Lotery (https://orcid.org/0000-0001-5541-4305) (Professor of Ophthalmology and Chief Investigator) conceived the trial, obtained funding, managed the trial with the Trial Management Group, developed the retinal image grading protocols and managed the grading process, interpreted the data and co-authored the first draft of the report.

Professor Sobha Sivaprasad (https://orcid.org/0000-0001-8952-0659) (Consultant Ophthalmologist) obtained funding, designed the trial, was a member of the Trial Management Group, developed the retinal image grading protocols and managed the grading process, interpreted the data and co-authored the first draft of the report.

Dr Abby O’Connell (https://orcid.org/0000-0001-7598-927X) (Trial Manager) managed all aspects of the trial including monitoring and co-authored the first draft of the report.

Miss Rosie A Harris (https://orcid.org/0000-0002-2405-667X) (Medical Statistician) was responsible for data management, analysed the data and co-authored the first draft of the report.

Dr Lucy Culliford (https://orcid.org/0000-0002-9255-6617) (Research Fellow) obtained funding, designed the trial, oversaw trial management and co-authored the first draft of the report.

Ms Angela Cree (https://orcid.org/0000-0002-1987-8900) (Senior Research Manager) obtained funding, oversaw the biobank, was a member of the Trial Management Group and co-authored the first draft of the report.

Ms Savita Madhusudhan (https://orcid.org/0000-0002-8203-0929) (Consultant Ophthalmologist) developed the retinal image grading protocols and managed the grading process.

Ms Helen Griffiths (https://orcid.org/0000-0003-3908-884X) (Research Technician) processed the biobank samples.

Miss Lucy Ellis (https://orcid.org/0000-0001-8179-5172) (Trial Manager) was responsible for trial management during trial set-up.

Professor Usha Chakravarthy (https://orcid.org/0000-0002-2606-3734) (Consultant Ophthalmologist) provided expert input on the trial, developed the retinal image grading protocols and managed the grading process, interpreted the data and co-authored the first draft of the manuscript.

Professor Tunde Peto (https://orcid.org/0000-0001-6265-0381) (Consultant Ophthalmologist) provided expert input on the trial, developed the retinal image grading protocols and managed the grading process and interpreted the data.

Professor Chris A Rogers (https://orcid.org/0000-0002-9624-2615) (Professor of Medical Statistics) obtained funding, designed the trial and interpreted and analysed the data.

Professor Barnaby C Reeves (https://orcid.org/0000-0002-5101-9487) (Professor of Health Services Research) obtained funding, designed the trial, managed the trial, interpreted the data and co-authored the first draft of the report.

Publications

Willcox A, Culliford L, Ellis L, Rogers CA, Cree A, Chakravarthy U, et al. Clinical efficacy of eplerenone versus placebo for central serous chorioretinopathy: study protocol for the VICI randomised controlled trial. Eye 2019;33:295–303.

Lotery A, Sivaprasad S, O’Connell A, Harris RA, Culliford L, Ellis L, et al. Eplerenone for chronic central serous chorioretinopathy in patients with active, previously untreated disease for more than 4 months (VICI): a randomised, double-blind, placebo-controlled trial. Lancet 2020;395:294–303.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research. The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the MRC, NETSCC, the EME programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the EME programme or the Department of Health and Social Care.

Trial Steering Committee

(Chairperson: September 2016 to April 2017) Professor Susan Downes, Consultant Ophthalmologist, Oxford Eye Hospital, Oxford, UK.

(Chairperson: April 2017 to trial end) Mr Ben Burton, Clinical Director of Research, James Paget University Hospital, Great Yarmouth, UK.

Dr Gabriella Czanner, Lecturer in Ophthalmic Statistics, University of Liverpool, Liverpool, UK.

Dr Dolores Conroy, Director of Research, Fight for Sight, London, UK.

Professor Theresa McDonagh, Independent Professor of Heart Failure and Consultant Cardiologist, King’s College Hospital, London, UK.

Mrs Blanche Morrissey, Patient Representative, Buckinghamshire, UK.

Miss Veronica Rossi, Patient Representative, London, UK.

Mr Mandeep Bindra, Consultant Ophthalmologist and Vitreo-retinal Surgeon, Stoke Mandeville Hospital, Aylesbury, UK.

Mr Quresh Mohammed, Consultant Ophthalmologist, Cheltenham General Hospital, Cheltenham, UK.

Data Monitoring and Safety Committee

(Chairperson: September 2016 to September 2018) Professor Toby Prevost, Professor of Medical Statistics, Imperial College London, London, UK.

(Chairperson: September 2018 to end of trial) Mr Mark Dayer, Consultant Cardiologist, Musgrove Park Hospital, Taunton, UK.

Professor Simon Taylor, Professor of Ophthalmology, Royal Surrey County Hospital, Guildford, UK.

Professor Baljean Dhillon, Professor of Clinical Ophthalmology, University of Edinburgh, Edinburgh, UK.