Purchase this article with an account.
Melina Cavichini Cordeiro, Kunny Dans, Mahima Jhingan, Manuel Amador, Dirk-Uwe G Bartsch, Eric Nudleman, William R Freeman; Confusion Matrix Evaluation of the Clinical Utility of OCT Angiography in AMD-associated Choroidal Neovascularization. Invest. Ophthalmol. Vis. Sci. 2019;60(9):3004. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate the ability of OCTA in identifying the presence or absence of CNV and its clinical utility in determining neovascular disease activity in eyes with age-related macular degeneration (AMD) using the confusion matrix (2x2 table).
Clinical parameters (symptoms and change in visual acuity) along with retinal imaging, including fluorescein angiography (FA) and optical coherence tomography (OCT) using the Heidelberg Spectralis were used as gold standard to determine disease activity. OCT angiography (OCTA) using the Optovue RTVue was performed on the same day. Images were graded by 2 masked retina specialists based on the presence or absence of CNV on OCTA, and clinical activity on gold standard tests based on the presence of leakage on FA and retinal fluid on OCT. A pro re nata (PRN) regimen of intravitreal anti-VEGF injections was given to patients with neovascular activity.
One hundred and forty-six eyes of 103 patients with AMD were retrospectively reviewed. Of these, 41 eyes had clinically active wet CNV, 21 eyes had inactive CNV lesions, and 84 had dry AMD with high risk drusen. OCTA was 85% accurate in determining the presence or absence of CNV in AMD. There was a 21.0% and 4.8% false positive and negative detection rate by OCTA, respectively. In a substudy, eyes with reactivated or well treated inactive CNV were studied to evaluate OCTA’s utility in demonstrating clinical activity. OCTA was 60% accurate in determining clinical activity and need for treatment. False positives and negatives were 30.0% and 10.0%, respectively.
Compared to gold standard, OCTA has a high sensitivity and negative predictive value in determining the presence or absence of CNV in AMD. However, in eyes with wet AMD receiving PRN anti-VEGF therapy, the low specificity rate in determining disease activity suggests that its accuracy in determining need for treatment based by OCTA alone is limited. Hence, caution must be exercised in interpreting results to avoid unnecessary treatment. Further studies need to be undertaken to determine if differences occur in OCTA imaging of active and inactive CNV in wet AMD.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.
Table 1. Determination of presence or absence of CNV in eyes with active wet AMD and dry AMD.
Table 2. Determination of neovascular disease activity in eyes with wet AMD receiving PRN anti-VEGF therapy.
This PDF is available to Subscribers Only