Aims To quantify the changes of the choroidal structure in the enhanced depth imaging optical coherence tomographic (EDI-OCT) images after intravitreal aflibercept (IVA) injections for polypoidal choroidal vasculopathy (PCV).
Methods Retrospective, observational case series. Forty eyes of 40 treatment-naive patients who underwent IVA for PCV were examined by EDI-OCT before, and 3 months and 12 months after IVA. The EDI-OCT images were binarised by ImageJ software. The cross-sectional luminal and stromal areas of the inner and outer subfoveal choroid of 1500 µm width were quantified.
Results The stromal but not the luminal area of the inner choroid was significantly decreased at 3 months and 12 months after the IVA (stromal area, both p<0.001; luminal area, both p>0.050). On the other hand, the luminal but not the stromal area of the outer choroid was significantly decreased at 3 months and 12 months (luminal area, both p<0.001; stromal area, both p>0.050). The Pachychoroid Index, ratio of luminal/stromal area (L/S ratio) of the outer choroid divided by the L/S ratio of the inner choroid, was significantly decreased at 3 months and 12 months (both p<0.050). The Pachychoroid Index was increased and returned almost to the baseline level after recurrences and decreased again after successful re-treatment. The baseline Pachychoroid Index was significantly correlated with the presence of a dry macula, thinner fovea and better visual acuity at 12 months (all p<0.050).
Conclusion The binarisation of the EDI-OCT images can be used to quantify the activity of PCV and to predict the prognosis after IVA.
- Treatment Medical
Statistics from Altmetric.com
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.
Intravitreal aflibercept (IVA) injections can improve the visual acuity and macular morphology in eyes with polypoidal choroidal vasculopathy (PCV).1 Earlier studies using enhanced depth imaging optical coherence tomography (EDI-OCT) showed that the central choroid was thicker in eyes with PCV, and the central choroidal thickness (CCT) was reduced after antivascular endothelial growth factor (anti-VEGF) therapy.2–4 As best as we know, there have been no reports quantifying the changes of the luminal and stromal areas of the choroid in eyes with PCV after anti-VEGF therapy.
We have reported that the EDI-OCT images can be converted to binary images which can then be used to quantify the luminal and stromal areas of the choroid.5–10 This technique has been widely used,11–13 and we have used this technique to investigate the choroidal structure in eyes with PCV.
Recently, Pang et al14 proposed a new clinical entity called, ‘pachychoroid’ which is a spectrum of diseases with a thickened choroid including PCV and central serous chorioretinopathy (CSC). PCV and CSC have similar characteristics including choroidal hyperpermeability detected by indocyanine green angiography (ICGA),15 increased choroidal thickness16 and dilated, thin-walled choroidal vessels determined histopathologicallly.17 Most recently, we proposed a CSC index, renamed here as the ‘Pachychoroid Index’, which is an index derived from an analysis of the different choroidal layers in the binarised EDI-OCT images.10 ,18 This technique was sensitive enough to detect stromal swelling in the inner choroid and congestion in the outer choroid in CSC.10 ,18 We demonstrated that the Pachychoroid Index was significantly higher in CSC than in controls,10 ,18 and it was significantly reduced after photodynamic therapy (PDT) for CSC.18 However, the Pachychoroid Index in PCV and its change after anti-VEGF therapy has not been determined.
Thus, the purpose of this study was to quantify the changes of the luminal and stromal areas of the inner and outer choroid after IVA injections for PCV.
Materials and methods
This was a retrospective, observational case series of 40 eyes of 40 consecutive treatment-naive patients who underwent IVA injections for PCV at the Tokushima University Hospital between October 2013 and November 2015. The demographic data of the patients are presented in table 1. The patients were examined by EDI-OCT before, and 3 months and 12 months after the initial IVA.
The diagnosis of PCV was based on the ICGA findings that showed polypoidal lesions at the border of branching choroidal vascular networks. The exclusion criteria included a myopic refractive error greater than −6.0 dioptres and other ocular diseases that might affect the clinical findings. Twenty-one untreated healthy fellow eyes and age-matched, sex-matched and refractive error-matched 38 normal controls were also examined.
All 40 eyes underwent three consecutive monthly IVA injections (2.0 mg) as a loading dose, and received additional IVAs given as needed during the follow-up period. The additional IVA injections were given if the vision decreased to >0.2 logarithm of the minimal angle of resolution (logMAR) units or if fluid or oedema involving the macula in the spectral-domain OCT (SD-OCT) images remained for at least 1 month following the previous IVA injection.
Spectral-domain optical coherence tomography
SD-OCT was performed with the Heidelberg Spectralis instrument (Heidelberg Engineering, Heidelberg, Germany). The central foveal thickness (CFT) and CCT were measured manually using the calliper function. All EDI-OCT images were obtained between 11:00 hours and 13:00 hours to minimise the effect of the diurnal variations of the choroidal structures.7
A dry macula was defined as one with an absence of subretinal or intraretinal spaces in the SD-OCT images. A recurrence was defined as the reappearance of subretinal or intraretinal spaces in the SD-OCT images after the presence of dry macula over 3 months without IVA injections.
Quantification of luminal, stromal and overall choroidal areas
The binarisation of the horizontal EDI-OCT images was done by a modified Niblack's method as described in detail (figure 1).5 The EDI-OCT images were analysed with the ImageJ software (ImageJ, V.1.47, NIH, Bethesda, Maryland, USA). The examined choroidal area was 1500 µm wide, was centred on the fovea, and extended vertically from Bruch's membrane to the chorioscleral border. The luminal and stromal areas were automatically calculated in the binarised EDI-OCT images using ImageJ software.
The binarisation of the EDI-OCT image was done three times for each image, and the three measurements were averaged. Eyes with PCV often have haemorrhages beneath the retinal pigment epithelium (RPE) and retina, and these haemorrhages can block the visibility of the choroid in the EDI-OCT images.19 In 10 eyes in which the horizontal images could not be analysed because of the shadowing effect due to high haemorrhagic RPE detachments, the vertical scans were analysed at all time points. Five eyes in which both horizontal and vertical scans could not be analysed were excluded.
Evaluations of the choroidal areas were also performed for the inner and outer choroid. The inner choroid included the choriocapillaris and medium choroidal vessel layer, and the outer choroid included the larger choroidal vessel layer. The margin of the inner and outer choroid was determined by the method proposed by Branchini et al20 with some modifications. The segmentation was done on the binarised OCT images which enhanced the appearance of the margin of the larger choroidal vessels (figure 1).10 ,18 As reported,21–23 the outer choroid on the EDI-OCT images was defined as areas consisting of large hyporeflective spaces representing large luminal spaces. The inner choroid was defined as areas consisting of small-sized to medium-sized hyporeflective spaces, surrounded by hyper-reflective stroma, giving a mottled appearance. Sim et al22 reported that the thickness of the inner and outer choroidal sublayers can be quantified with good reliability, repeatability and reproducibility.
We have proposed an index called the Pachychoroid Index that represents the relative values of the luminal and stromal areas of the inner and outer choroid.10 ,18 The formula to determine the Pachychoroid Index was:
This Pachychoroid Index can quantify the structural differences of the inner and outer choroid. A Pachychoroid Index of 1.0 indicates that the ratio of luminal/stromal area (L/S ratio) of the inner choroid is equal to the L/S ratio of the outer choroid. If both a swelling of the stroma in the inner choroid and a vascular dilatation in the outer choroid are present, the index becomes higher.
Paired t-tests were used to determine the significance of the differences between paired samples, and unpaired t-tests for the differences between two independent groups. Pearson's correlation tests were used to determine the significance of the correlations. The significance of the fluctuations in the EDI-OCT parameters was determined by repeated-measures analysis of variance with Greenhouse-Geisser corrections. The Bonferroni test was used for post hoc analyses. The intrarater correlation coefficients were calculated by one-way random effects model for measurements of agreement. A p value of <0.050 was considered statistically significant.
The mean number of IVAs during the 12 months was 4.3±1.8 times (range 3–9). The intrarater agreement was high with an intraclass correlation coefficient >0.980 for the measurements of the baseline luminal and stromal areas of the inner and outer choroid.
Changes of retina and whole choroid after treatment
The CCT and CFT were significantly decreased from the baseline at 3 months and 12 months (all, p<0.050; figure 2). Both the luminal and stromal areas of the whole choroid were significantly decreased at 3 months and 12 months (all p<0.005), but the L/S ratio was not significantly changed at 3 months and 12 months (p=1.000, p=0.925, respectively).
Analyses of inner and outer choroidal areas
In the inner choroid, the stromal but not the luminal area was significantly decreased at 3 months and 12 months (both p<0.001, stromal area; p=1.000, p=0.188, luminal area; figure 2). The L/S ratio of the inner choroid was significantly increased at 3 months and 12 months (p<0.001, p=0.031, respectively). The total inner choroidal area was significantly decreased at 3 months and 12 months (p=0.021, p=0.003, respectively).
In the outer choroid, the luminal but not the stromal area was significantly decreased at 3 months and 12 months (both p<0.001, luminal area; p=0.063, p=0.132, stromal area). Thus, the L/S ratio was significantly decreased at 12 months (p=0.033), but not at 3 months (p=0.092). The total outer choroidal area was significantly decreased at 3 months and 12 months (both p<0.001).
Changes of Pachychoroid Index
The Pachychoroid Index of all eyes was significantly decreased at 3 months and 12 months (p<0.001, p=0.011, respectively; figure 3A). The Pachychoroid Index was significantly higher at the baseline (1.832±0.692), 3 months (1.476±0.707) and 12 months (1.554±0.796) than that of the healthy fellow eyes (0.988±0.465) and normal control eyes (1.029±0.411, all p<0.005).
In 10 eyes with a recurrence, there were significant fluctuations in the Pachychoroid Index during the follow-up period (p<0.001, figure 3B). The high Pachychoroid Index decreased after the initial IVA, increased and returned almost to the baseline level at the time of recurrence and decreased again after successful re-treatment.
Correlation between Pachychoroid Index and clinical findings
The baseline Pachychoroid Index in eyes without a dry macula at 12 months (2.195±0.685, 17 eyes) was significantly higher than that in eyes with a dry macula (1.563±0.576, 23 eyes; p=0.003). The baseline Pachychoroid Index was significantly correlated with the decreased CFT and improved best-corrected visual acuity (BCVA) at 12 months (p=0.024, p=0.002, respectively; figure 4). The BCVA improvement in logMAR units and CFT decrease were defined as the values at 12 months subtracted from the baseline values.
The baseline CCT and number of IVA were not significantly correlated with the presence of a dry macula, decreased CFT, improved BCVA and Pachychoroid Index at 12 months (all p>0.100).
As best as we know, the changes of the choroidal structure after anti-VEGF therapy have not been reported in eyes with PCV. Our results showed that the stromal but not the luminal area of the inner choroid was significantly decreased after IVA. On the other hand, the luminal but not the stromal area of the outer choroid was significantly decreased after IVA. In eyes with CSC, we recently reported that the stromal but not the luminal area of the inner choroid was also significantly decreased after PDT.18 In the outer choroid, the luminal but not the stromal area was significantly decreased after PDT for CSC. These results suggest that the decreased CCT after IVA for PCV or PDT for CSC may be mainly attributed to a decrease in the exudative changes in the inner choroidal stroma and the reduction of the dilation of the outer choroidal vessels.18
In eyes with CSC, the Pachychoroid Index was significantly higher than that in controls, and it was significantly reduced after PDT.10 ,18 In the present study, the Pachychoroid Index was significantly decreased after IVA for PCV, and the Pachychoroid Index increased at the time of a recurrence and decreased after successful re-treatment. The baseline Pachychoroid Index was significantly higher than that of the fellow eyes and normal controls. These results indicate that the Pachychoroid Index may be associated with the activity of PCV and could serve as a marker for recurrences. In addition, the baseline Pachychoroid Index was significantly correlated with the presence of a dry macula, decreased CFT and improved BCVA at 12 months which suggests that the Pachychoroid Index may be able to predict the changes in these factors after IVA injections.
Koizumi et al3 reported that the baseline CCT was not significantly different between PCV eyes with and without a dry macula at 3 months after IVA injections. In the present study, the baseline CCT was not significantly correlated with the presence of a dry macula, decreased CFT, and improved BCVA at 12 months. These results suggest that the CCT is not a good indicator of the visual prognosis after IVA injections for PCV which supports the usefulness of a choroidal analysis using the binarisation of the EDI-OCT images.
The choroidal hyperfluorescence in the ICGA images of eyes with PCV was located in the area of the dilated choroidal veins, and hyperpermeability appeared to be present in the choriocapillaris and in the dilated choroidal veins.19 ,24 In our study, the stromal areas of the whole choroid were significantly reduced at 3 months and 12 months after IVA which indicates a decrease of the stromal swelling in the whole choroid after IVA. A recent study on monkeys found a reduction of the number of choriocapillaris endothelial cell fenestrations after IVA.25 The reason for the decreased stromal swelling after IVA may be related to the reduction of the leakage from the fenestrated vascular walls.2
In eyes with PCV, histopathological studies showed dilated vessels under Bruch's membrane.17 Yang et al19 reported that the diameter of the outer choroidal vessels in the EDI-OCT images was significantly larger in eyes with PCV than in controls. The choroidal thickness was significantly correlated with the largest choroidal vessel diameter suggesting that the choroidal thickening was mainly associated with dilatation of the outer choroidal vessels. Our results showed that this outer choroidal vascular dilatation may be reduced after IVA injections.
In an animal model, the area of the choriocapillaris was significantly reduced after IVA.25 This is consistent with the decrease in the total inner choroidal area after IVA in our study. Julien et al25 reported that the lumens of the choriocapillaris were smaller in monkey eyes after IVA than in controls. However, the luminal area of the inner choroid was not significantly changed after IVA in our patients. The reasons for this discrepancy may be that immunohistochemistry was performed 1 day and 7 days after IVA in the monkeys, and that the monkey eyes were healthy without PCV. In eyes with pachychoroid neovasculopathy, Pang et al14 reported that a dilation of the larger choroidal vessel layer may often be associated with the obliteration of the inner choroidal vessels. Our results showed that the L/S ratio of the inner choroid was significantly smaller in eyes with PCV than in controls (1.408±0.535 vs 2.142±0.839). This inner choroidal vascular obliteration in PCV eyes before IVA may contribute to the different responses to IVAs from that of healthy monkey eyes.
This study has limitations. First, the sample size was small. Second, a retrospective study can have sampling biases. Third, the follow-up period of 1 year may not be long enough. Lastly, manual segmentation of the inner and outer choroid is not completely objective. However, to the best of our knowledge, this is the first study that evaluated the changes in the choroidal structures in eyes undergoing IVA for PCV.
In conclusion, the significant correlations between the baseline Pachychoroid Index and the presence of dry macula, reduced CFT and better BCVA at 12 months after the IVAs indicate that the index can serve both as a marker for activity of PCV and also as a predictor of the effect of IVA injections. Binarisation of the EDI-OCT images is a useful and non-invasive method to quantify the choroidal structures.
All authors became members of Japan Clinical Retina Study (CREST) group and researched. The authors thank Professor Emeritus Duco Hamasaki of the Bascom Palmer Eye Institute of the University of Miami for providing critical discussions and suggestions for the study and revision of the final manuscript.
Contributors All authors have given final approval of this version to be published. Design of the study (YM, SS and TS), conduct of the study (ED, HS, KA, MN, CY, TK and ME), management of the data (ED, HS, KA, MN, CY and ME), analysis of the data (ED, YM, HS, KA, TK, ME and SS), interpretation of the data (ED, YM, TK, SS and TS), preparation of the manuscript (ED and YM) and overall coordination (TS).
Funding This work was supported in part by grant-in-aid 16K11288 (to YM) from the Ministry of Education, Science, Sports and Culture, Japan, as well as by the grant from Bayer Japan to the Tokushima University.
Competing interests None.
Patient consent Obtained.
Ethics approval The Institutional Review Board of Tokushima University Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data are available from the corresponding author on request.