Validation of 3-dimensional Analysis of Choroidal Neovascularization (CNV) as a surrogate marker for response to therapy

Abstract

Purpose : To assess changes in 3-dimensional (Volumetric) analysis of CNV in eyes treated with ranibizumab using a novel algorithm and evaluate correlation between changes in volume of CNV, changes in area of CNV using Fluorescein angiography (FA) and changes in central retinal thickness (CRT)

Methods : Algorithm validation was performed using 10 eyes with 30° macular OCT scans (25 B-scans). Each B-scan was segmented at internal limiting membrane (ILM) and Bruch’s membrane (BM) by Heidelberg Eye Explorer software (Heyex). Volume of the central 3-mm region in Heyex software was noted (1A). All B-scans for each subject were exported and a custom generated algorithm was used to project a 3-mm circle on the en-face OCT image (1B). A boundary identification tool was used to manually identify points on B-scans corresponding to 3-mm circle and its volume was measured (1B). Results from the algorithm were compared with results from Heyex. After validation, OCT scans of 8 eyes with CNV from two visits 6 months apart were processed using our algorithm to obtain volumes of the CNV lesion (1C).Fluorescein angiography (FA) images of the same subjects from the same visits were evaluated for Area of CNV was also calculated for these eyes using FA images from the corresponding visits using Image J. Changes in central retinal thickness (CRT) on OCT, CNV area on FA and volume on custom algorithm were compared using t-test and linear regression

Results : The mean volume (n=10 eyes) calculated for validation analysis using the custom algorithm was within 0.05 mm³ (p>0.05). In the diseased eyes, the percent change in the area of CNV lesion on FA, volumetric change on MATLAB and change in CRT demonstrated a 4.11%, 19.31%, 10.07% reduction respectively. Mean change in area measured on FA, and volume on MATLAB from baseline to month 7 was not statistically significant (-0.44mm² and -0.50 mm³, respectively; p>0.05). Linear Regression demonstrated a statistically significant linear relationship between change volume with the change in CRT (r:0.01; p=0.03). A non-significant relationship was seen between the change in area and change in CRT (r: 0.005; p=0.60) (Fig. 2)

Conclusions : CNV lesions are 3-dimensional and volumetric analysis of such lesions using OCT may be more sensitive to detecting smaller changes with response to therapy

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.