Article Text

Download PDFPDF

Predictors of treatment response to intravitreal anti-vascular endothelial growth factor (anti-VEGF) therapy for choroidal neovascularisation secondary to chronic central serous chorioretinopathy
  1. Khaled Romdhane1,
  2. Marta Zola1,
  3. Alexandre Matet1,
  4. Alejandra Daruich1,
  5. Martine Elalouf1,
  6. Francine Behar-Cohen2,
  7. Irmela Mantel1
  1. 1 Department of Ophthalmology, University of Lausanne, Jules Gonin Eye Hospital, Lausanne, Switzerland
  2. 2 Medical Retina Unit, INSERM UMR1138, Paris, France
  1. Correspondence to Dr Irmela Mantel, Department of Ophthalmology, Jules Gonin Eye Hospital, Fondation Asile des Aveugles, University of Lausanne, 1015 Lausanne, Switzerland; irmela.mantel{at}fa2.ch

Abstract

Purpose The aim of this study was to evaluate the effect of anti-vascular endothelial growth factor (VEGF) therapy on choroidal neovascularisation (CNV) complicating central serous chorioretinopathy (CSC) using multimodal imaging, and to identify possible predictive factors of the treatment response.

Design Retrospective study.

Methods Data of 27 eyes with CNV complicating CSC treated with anti-VEGF therapy (either ranibizumab or aflibercept) were reviewed. Response to anti-VEGF treatment was evaluated by change in visual acuity, intra/subretinal fluid modifications and CNV changes on optical coherence tomography angiography (OCTA). Univariate and multivariate analyses were performed to identify predictive factors for central retinal thickness (CRT) change and for the relative degree of treatment response (complete, incomplete or absent fluid reduction).

Results CRT was significantly reduced at 32±15 days after 2.8±1.3 injections (p=0.0004) as was the subretinal fluid (p=0002). Complete fluid resorption was observed in 45% of cases. Best corrected visual acuity did not significantly improve (p=0.18). CNV area (p=0.09) and CNV flow area (p=0.07) did not significantly decrease. No changes in CNV pattern were noted. Univariate analysis identified greater CRT at baseline (p<0.0001), greater amount of subretinal fluid (p<0.0001), a shorter period of retinal fluid (p=0.04) and female gender (p=0.04) as predictors for CRT reduction. After multivariate analysis the factor of greater CRT at baseline (p<0.0001) proved independent. The degree of treatment response was dependent on the size of CNV surface (p=0.05) and flow area (p=0.05) on OCTA in the univariate analysis, and the latter independent after multivariate analysis. In addition, a shorter time period of retinal fluid appeared to play a role (p=0.01 multivariate, p=0.19 univariate).

Conclusion The anti-VEGF response was highly variable and often incomplete, suggesting that CNV was not solely responsible for the fluid accumulation. Predictive factors may guide indication for anti-VEGF in CNV associated with CSC.

  • neovascularisation
  • retina
  • choroid
  • treatment other

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Central serous chorioretinopathy (CSC) is characterised by serous retinal detachment (SRD) of the neurosensory retina that is usually self-resolving over 3–6 months.1 2 Choroidal vessel hyperpermeability and congestion leading to abnormal fluid leakage from the choroid to the subretinal space through retinal pigment epithelial defects are believed to be the cause of the disease.3 4

Chronic CSC is usually defined as persistence of SRD for at least 6 months or widespread retinal pigment epithelium (RPE) decompensation with or without SRD, which may or may not be associated with active leakage sites.2 5 Chronic CSC has a worse prognosis and is more likely to be complicated by choroidal neovascularisation (CNV) than the acute form, leading to more severe visual impairment.4 6 7 It is part of a recently proposed pachychoroid spectrum, which classifies CSC with CNV as pachychoroid neovasculopathy.8

CNV typically requires treatment with intravitreal anti-vascular endothelial growth factor (anti-VEGF) agents. Their effectiveness has been shown for neovascular age-related macular degeneration (AMD)9 and for CNV of any other origin.10

However, the treatment indication for CSC complicated by CNV is often difficult to establish. The mere presence of CNV in chronic CSC may not necessarily require treatment. However, the presence of exudative activity might be attributable to the underlying CSC and/or to the CNV. To the best of our knowledge, no study has reported differentiating criteria. These would be of clinical importance as the treatment options for CSC itself differ from anti-VEGF therapy, including options such as laser,11 photodynamic therapy (PDT)12 or oral mineralocorticoid receptor antagonists.13–15

The aim of this study was to evaluate chronic CSC with CNV treated with anti-VEGF agents to quantify their effect on fluid reduction and CNV structure, and to identify possible predictive factors for the treatment response using multimodal imaging.

Materials and methods

This retrospective study was conducted in Jules-Gonin Eye Hospital, Medical Retina Unit, Lausanne, Switzerland. Data were collected from a consecutive series of patients treated with anti-VEGF drugs for CSC complicated by CNV between July 2015 and May 2018.

Study population

This study included a consecutive series of patients with CSC, treated with anti-VEGF therapy for CNV, identified to have flat irregular pigment epithelial detachment (FIPED) on spectral domain optical coherence tomography (SD-OCT) B-scans, and CNV signs on multimodal imaging including SD-OCT, OCT angiography (OCTA), and dye angiography with fluorescein (FA) and indocyanine green (ICG). All types of CSC were included, and the categories are defined in table 1.

Table 1

Definitions for the categorisation used in the study

Exclusion criteria were insufficient image quality, confounding retinal pathologies, or PDT or focal laser treatment performed within 6 months before evaluation. Concomitant treatment with oral mineralocorticoid receptor antagonist (spironolactone or eplerenone) was allowed.

For each patient, a baseline time point was determined. This corresponded to the visit when the first anti-VEGF treatment was injected in the study eye. However, multimodal imaging was not always immediately available. Thus, a later time point was allowed on the condition that the patient did not receive any treatment for at least 3 months before baseline visit.

Data collection

The collected data included age, sex, ocular history, medication, the laterality of the study eye, the type of CSC and the best available visual acuity measured on ETDRS charts then converted into LogMar. Routine retinal imaging, which included a 6×6 mm cube (98 B-scans) on OCT (Heidelberg Spectralis, Heidelberg, Germany), was performed during each follow-up visit. OCTA was acquired every 3 months using the RTVue XR Avanti (AngioVue system, Optovue, Fremont, CA). This included a 3×3 mm volume scan (304×304 A-scans) and a 6×6 mm volume scan (400×400 A-scans). FA (Heidelberg retina angiograph (HRA), Heidelberg, Germany) and ICG angiography (HRA) were performed at referral and repeated according to the ophthalmologist’s recommendation, particularly before starting anti-VEGF treatment.

Treatment

Intravitreal injections of 0.5 mg/0.05 mL of ranibizumab (Lucentis; Genentech, South San Francisco, CA), or 2 mg/0.05 mL of aflibercept (Eylea; Regeneron, Tarrytown, NY) were performed on a monthly as-needed regimen. The choice for ranibizumab or aflibercept was at the clinician’s discretion.

Image analysis

SD-OCT B-scans were analysed for the type of fluid (subretinal/intraretinal) and the quantity of subfoveal fluid (manual measurement of the vertical distance between the neuroretina and the RPE). Subfoveal choroidal thickness was measured manually on enhanced depth imaging, from the RPE to the internal limit of the sclera.

FA was analysed for the presence of CNV according to the usual CNV criteria (table 1).

On OCTA, CNV was identified as an abnormal vascular branching on en face slab between the RPE and Bruch’s layer and/or by the presence of a decorrelation signal within the FIPED on the OCTA B-scans. The automated layer segmentation on AngioVue OCTA was manually corrected for the line on the Bruch membrane, RPE and internal limiting membrane using the V.2017.1.2.150 of the software. En face visualisation of the CNV was performed on a custom slab, putting the upper layer offset at 0 µm from the RPE and the lower layer offset at −15 µm from Bruch’s membrane, thus including mainly the FIPED (preferentially on the 3×3 mm images if capturing the entire CNV).

For qualitative CNV description on OCTA, we used the classification used in previous studies (‘medusa’, ‘seafan’ or ‘indistinct’ patterns).16 17 The presence of anastomoses in the periphery of the CNV and any dense microvasculature were also noted (figures 1 and 2). The CNV area was manually delineated with the AngioVue inbuilt drawing tool, and the flow of the custom CNV layer was measured within this area (figures 1 and 2).

Figure 1

Chronic central serous chorioretinopathy in a right eye complicated by choroidal neovascularisation (CNV): left column before treatment. (A) Represents a well-defined CNV with medusa pattern visualised on optical coherence tomography angiography (OCTA; en face custom slab after manual segmentation correction), peripheral anastomosis and dense microvasculature are present. (B) The inbuilt drawing tool of the RTVue XR Avanti device allows delineating the CNV for calculating its area and flow. (C) The B-scan OCT shows subfoveal retinal detachment with hyper-reflective flat irregular pigment epithelial detachment. Right column: after three intravitreal injections of aflibercept. (D) Shows a smaller perfused area of the CNV with reduction of the peripheral anastomosis and dense microvasculature. (E) Shows CNV area and flow reduction. (F) Resolution of the subfoveal fluid on spectral domain OCT (SD-OCT).

Figure 2

Chronic central serous chorioretinopathy in a right eye complicated by choroidal neovascularisation (CNV): left column before treatment. (A) Represents an ill-defined CNV with seafan pattern visualised on optical coherence tomography angiography (OCTA; en face custom slab), peripheral anastomosis and dense microvasculature are present. (B) The inbuilt drawing tool of the device allows delineating the CNV calculating its area size and flow. (C) Corresponding spectral domain OCT (SD-OCT) showing subfoveal and intraretinal fluid with hyper-reflective flat irregular pigment epithelial detachment. Right column: after four intravitreal injections of ranibizumab. (D) Shows proliferation of the CNV with denser microvasculature. (E) CNV area and flow increase. (F) Resolution of the subfoveal fluid on SD-OCT but persistence of the intraretinal fluid.

Outcomes that were measured criteria included visual acuity change and intra/subretinal fluid changes (change in central retinal thickness (CRT) in microns and relative responses in the categories presented in table 1). CNV changes were assessed on OCTA both quantitatively and qualitatively (figures 1 and 2).

Statistical analyses

Descriptive statistical analyses were performed. The changes after treatment were analysed using paired Wilcoxon and χ2 tests for continuous and categorical variables, respectively. Univariate and multivariate analyses served to identify risk factors associated with the response. In the multivariate analysis, the logistic model was used for the categorical outcome measure (responder categories), and the linear regression model was used for the continuous outcome measure (CRT change). Factors included in the multivariate model needed a p value <0.2 in the univariate analysis. For data analysis, a Microsoft Excel 2010 spreadsheet and JMP software for Windows (V.8.0.1, SAS Institute, Cary, NC) were used. A two-tailed probability of 0.05 or less was considered statistically significant.

Results

Baseline characteristics

The study included 27 eyes (14 right eyes (44%)) of 25 patients (20 male, 5 female) with a mean age of 64 years (8 SD). Further baseline characteristics are summarised in table 2.

Table 2

Baseline characteristics of all patients included in the study

CNV identification was more successful on OCTA (26 eyes/96%) (figure 3), in particular on B-scan using the decorrelation signs within the FIPED (25 eyes/93%), compared with en face OCTA (19 eyes/70%) (figure 4).

Figure 3

Optical coherence tomography angiography (OCTA; en face slab) (A) showing a well-defined choroidal neovascularisation (CNV) with medusa pattern and corresponding OCT B-scan (C) showing decorrelation signal of the CNV inside the flat irregular pigment epithelium detachment. Mid-phase fluorescein angiography (B) of the same eye taken the same day showing signs of chronic central serous chorioretinopathy with hyperfluorescent atrophic descending tract masking the visualisation of the CNV and corresponding mid-phase indocyanine green angiography (D) showing choroidal hyperpermeability without clear signs of CNV.

Figure 4

Optical coherence tomography angiography (OCTA; en face slab) (A) showing an ill-defined medusa-like high decorrelation signal. OCTA B-scan (C) showing decorrelation signal inside the flat irregular pigment epithelium detachment and thus confirming the choroidal neovascularisation (CNV). Mid-phase fluorescein angiography of the same eye obtained the same day (B) showing a leakage point without clear CNV sign. Mid-phase indocyanine green angiography (D) showing an abnormal hyperfluorescent vascular pattern evoking CNV.

The anti-VEGF agents administered were aflibercept in 22 eyes (81%) and ranibizumab in 5 eyes (19%). A mean number of 2.8 injections (1.3 SD) were administered before the eyes were re-evaluated with OCTA after a mean of 3.4 months (1.9 SD) and a mean of 32 days (15 SD) after the last injection was administered.

Treatment responses

Treatment responses, including visual acuity, CRT, subfoveal choroidal thickness and quantitative measurements of CNV on OCTA, are summarised in table 3.

Table 3

Anti-VEGF treatment response in eyes with central serous chorioretinopathy and choroidal neovascularisation

There was a significant reduction in CRT, which was consistent with a significant reduction of subretinal fluid height. This response was even more significant in previously treated eyes (CRT reduction: p=0.002 in pretreated eyes vs p=0.06 in treatment-naïve eyes); subretinal fluid height was p=0.003 in pretreated versus p=0.10 in treatment-naïve eyes. Forty-five per cent of eyes responded to treatment completely with resolution of SRD. A majority of them (67%) were documented for fluid resolution within a time period less than 3 months after anti-VEGF therapy initiation. Cases with complete or partial treatment response showed fluid reduction in spatial correlation with the CNV location in all but one case. There was no significant change in best corrected visual acuity, choroidal thickness, CNV area, flow area within the CNV and flow density on OCTA.

The CNV pattern did remain identical in all but one case, which changed classification from a medusa type to an indistinct pattern. The peripheral anastomosis changed only mildly, with some relative reduction seen in 5 out of 13 eyes. However, microvascularisation within the CNV did not show any changes.

Predictive factors

Univariate analysis for predictors was performed for the response category (complete, partial or absent response) and for CRT change. The results are summarised in table 4.

Table 4

Univariate analysis for predictors of treatment response

A complete response to anti-VEGF therapy was more likely with larger sizes of CNV and larger areas of neovascular flow on en face OCTA. CRT reduction was greater for females, fluid duration less than 3 months, greater CRT or greater subretinal fluid at baseline. The few cases with CNV identified on fundus fluorescein angiography (FFA) presented a near-significant higher CRT reduction after treatment.

A multivariate analysis was performed for both outcomes (response category and CRT change) and included all variables with a p value <0.2 in the univariate analysis. After stepwise linear regression analysis, the final model for the CRT change was significant (p<0.0001) and revealed as an independent factor for greater CRT reduction a higher baseline CRT (p<0.0001). In terms of response category, stepwise logistic regression led to a final significant model (p=0.03), identifying as marginally independent factors the shorter period of retinal fluid accumulation (<3 months, p=0.01) and greater CNV flow area (p=0.01).

Discussion

This investigation of a consecutive series of eyes with CSC complicated by CNV and their responses to anti-VEGF treatment revealed varying responses in terms of fluid reduction and allowed for identification of predictive factors. It is challenging to determine the degree to which the exudation in CSC complicated by CNV is attributable to the neovascular process or to the underlying CSC condition. Our study indicates observational elements that might help differentiate the fluid origin in these complex cases and help in the selection of the proper treatment strategy.

The response profile to intravitreal anti-VEGF therapy revealed that slightly less than half (45%) of the patients had complete fluid resorption, while a greater proportion (55%) showed an incomplete or absent response. In addition, visual acuity did not improve.

This response profile is clearly different from the response in neovascular AMD in which most patients (approximately 80%) reach a dry macula after three anti-VEGF agent injections,18 and are associated with significant visual acuity improvement. In our cohort, we included eyes with CNV in CSC that had previously been treated with anti-VEGF drugs (more than 3 months before baseline). As this might have influenced the response profile, we investigated both subgroups: pretreated and treatment naïve. However, treatment-naïve eyes did not show a better response. On the contrary, it was the pretreated eyes that showed a stronger benefit in terms of fluid absorption.

Therefore, it appears that CNV may be present in CSC without being responsible for the exudative fluid, as evidenced by its non-responsiveness to anti-VEGF treatment. This is not infrequent, as our results showed. In addition, some of the observed responsiveness to anti-VEGF treatment might in reality be self-resolving CSC, as this may happen in natural course of the disorder. This would even further increase the proportion of CNV which is actually non-responsive to anti-VEGF treatment, and probably not exudative by itself.

The subretinal and/or intraretinal fluid in CSC may originate from active CNV or from the exudative activity of CSC itself. Differentiating the origin of the pathological fluid with its potential visual threat is of great importance for appropriate treatment choice. Anti-VEGF therapy is the most appropriate treatment option in case of exudation from the CNV.10 On the other hand, CSC-related exudation is usually treated with either laser, PDT or oral mineralocorticoid receptor antagonists.14 19 However, to date, no clear evidence exists that might help the clinician determine the origin of the exudative activity and the choice of appropriate treatment. Our findings that pretreated eyes showed a significant reduction in CRT and subretinal fluid, while treatment-naïve eyes did not, might simply reflect the absence of clear indicators for good anti-VEGF therapy responders. A history of favourable response to prior anti-VEGF treatment is currently the best indicator for future favourable response to anti-VEGF treatment.

The CNV pattern on OCTA was dominated by a diffuse indistinct pattern (68%). The observation of a frequent unorganised indistinct pattern might be an indication of slow growth of the CNV, which is responding to a chronic low-grade imbalance of vascular mediators, probably with less aggressive neovascular growth than in other disorders. This hypothesis corresponds well with the following: little structural change in the CNV after anti-VEGF treatment; moderate prevalence of the signs of activity of peripheral anastomosis or dense internal microvasculature, again with little change; and limited effect of anti-VEGF treatment on the exudative activity, which might originate from the CSC rather than CNV. Using the same logic, we can understand the non-significant change of the neovascular flow on OCTA after anti-VEGF treatment. Apparently, the neovessels in CSC are relatively mature, and therefore not very responsive to anti-VEGF therapy (figure 5).

Figure 5

Optical coherence tomography angiography (OCTA; en face custom slab) and corresponding spectral domain OCT (SD-OCT) of choroidal neovascularisation (CNV) complicating central serous chorioretinopathy with indistinct pattern and no clear anastomosis. The images at the left (A, B,C) are taken at baseline and the images on the right (D, E, F) are from the follow-up visit. The first line (A and D) represents the OCTA en face slab after manual correction of the segmentation layer, and the second line (B and E) shows the same image with manual delineation of the CNV area. No modifications of the CNV pattern were noted on OCTA with increase of the exudative activity on SD-OCT despite anti-vascular endothelial growth factor (VEGF) treatment.

One might speculate to what degree the phenomenon of a relatively mature neovascular structure or the CNV and its relatively low exudative activity is expression of slow growth due to low-level VEGF expression, and to what degree it could be a useful pathophysiological response compensating for altered choriocapillaris. The precise pathological process is so far unclear.

However, if the CNV shows exudative activity, the CNV-related visual threat might be greater due to the anatomical disruption of the visual cycle and the retinal integrity.

Our explorative analysis of possible predictors for greater response to anti-VEGF treatment, measured as complete or incomplete responders (in the absence of reliable measurements for the remaining pathological fluid), and as absolute CRT reduction, respectively, showed some interesting results. It appeared that a greater size of CNV and a greater flow area as measured on en face OCTA were correlated with an increased chance to respond completely to the anti-VEGF treatment. On the other hand, the CRT reduction was dependent on the baseline CRT and the baseline amount of subretinal fluid, as measured at its highest point. This is not surprising, as less fluid and lower CRT quickly reaches the ceiling effect for possible improvement. More interestingly, eyes with more recent appearance of intraretinal and/or subretinal fluid (less than 3 months) had a better chance of responding to anti-VEGF treatment. Thus, on the background of chronic CSC with its low-grade exudative activity, a recent appearance of fluid with visualisation of CNV may be indicative of CNV-related fluid rather than acute CSC exudation. In addition, there was a finding that eyes of females showed more CRT reduction. The reason for this observation is unclear. However, there is some evidence that CSC with and without CNV may be somewhat different in women.20 Interestingly, identification of CNV on FFA was near statistical significance for greater CRT improvement. This is biologically plausible as these correspond with the rare classic CNV type that is usually more aggressive and exudative.

Our study has some limitations, besides the inherent weaknesses of a retrospective study. It is an exploratory study with a limited number of eyes. Part of the imaging interpretation relies on the investigators’ appreciation rather than objective measurements. The manual correction of the OCTA segmentation and delimitation of CNV area, although indispensable, carries a risk of interpretation bias. Indeed, no pre-existing segmentation could be used to visualise specifically the area of the flat irregular pigment epithelium detachment.

In conclusion, anti-VEGF therapy is part of the treatment options for CSC with CNV. However, the anti-VEGF treatment response was highly variable and often incomplete, suggesting that CNV was not solely responsible for the fluid accumulation. CNV itself might indeed sometimes be quiescent, even in the presence of fluid, which could be attributable to CSC exudation. Predictors for good treatment response to anti-VEGF therapy were female sex, greater CRT at baseline, recent fluid appearance, and large size and flow area of CNV measured on OCTA before treatment. In addition, a history of previous response to anti-VEGF treatment may be useful for predicting future outcomes. Keeping in mind that the exudative activity is not always due to the CNV, these identified factors may help the clinicians taking the decision to treat or not with anti-VEGF agents. Further prospective studies in larger groups are needed to better differentiate vision-threatening CNVs in CSC from quiescent CNVs. Moreover, these results raise the issue of whether quiescent CNVs might be potentially beneficial for the patient developing in response to the critical metabolic situation.

Acknowledgments

We thank Editage (www.editage.com) for English language editing.

References

Footnotes

  • Contributors KR and IM contributed to the design of the study, to the acquisition, analysis and interpretation of data, to writing, revising and final approval of the work. FBC, AM and AD contributed to the acquisition of data, to the obtainment of the Ethics Committee of the Swiss Federal Department of Health approval, to the writing, revising and final approval of the work. MZ and ME contributed to the acquisition of data, to the writing, revising and final approval of the work.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval Swiss Federal Department of Health (No CER-VD 19/15 and 20/15). A waiver for consent was obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost B Jonas