Automated Detection of Inflammatory Activity in Sarcoid Uveitis and Birdshot Chorioretinopathy Patients Using Optical Coherence Tomography and Fluorescein Angiography

Abstract

Purpose : Automated quantification of ocular inflammation provides continuous measurements of inflammation and improves the ability to differentiate activity in uveitis management. This subgroup analysis evaluates the utility of automated imaging-based measures of ocular inflammation in sarcoid uveitis (SU) and birdshot chorioretinopathy (BC).

Methods : This is a subgroup analysis of the prospective imaging of inflammation study (IQI). Recently active uveitis patients were prospectively imaged using ultra-wide field fluorescein angiography (UWFA), spectral domain optical coherence tomography (SD-OCT), and anterior segment optical coherence tomography (AS-OCT) at baseline and follow-up visits. Patients were identified as active based on clinical exam with physician interpretation of imaging. A masked grader performed automated software analysis to measure total retinal leakage (TRL) measured on UWFA, OCT central subfield thickness (CST) and AS-OCT identification of cells. Activity was met based on one or more of the following criteria: TRL > 4%, CST > 350 microns, and AS-OCT > 8 cells/mm³. In this analysis, patients with SU and BC are analyzed. Central 3-disc diameter (3DD) leakage on UWFA and intraretinal fluid percentage on OCT were also analyzed.

Results : A total of 40 eyes from 20 patients (13 SU, 7 BC) with a total of 164 visits were evaluated. BC patients met activity threshold based on AS-OCT, SD-OCT, AND UWFA in, 19.5%, 10.7%, and 2%, respectively. In SU patients, activity was noted in 20% of UWFA, 16.2% of AS-OCT, and 5.8% of SD-OCT images. In SU patients, when determined active by any imaging modality, TRL, 3DD, CST, AS-OCT were increased compared to inactive visits. In BC, when FA and AS-OCT identified activity, there was an increase in TRL and AS-OCT compared to inactive visits. When SD-OCT determined activity, CST and 3DD were higher compared to inactive visits.

Conclusions : Patients with BC and SU are identified active more often based on AS-OCT and TRL on FA, respectively. An increase in all TRL, 3DD, CST, and AS-OCT in patients with active SU validates our automated imaging software in detecting disease activity objectively. Based on our limited number of patients with BC, further investigation will be conducted to improve our ability to differentiate imaging activity in BC patients.

This is a 2021 ARVO Annual Meeting abstract.