For OCT-A Trial Endpoints, Quality, Technology Must Improve

Although optical coherence tomography angiography (OCT-A) images may one day be an important endpoint in multicenter clinical research, scan quality, and scan analysis must significantly improve before being used as an outcome in research. This is according to research results published in Translational Vision Science & Technology. 

Often, multicenter clinical trials utilize centralized reading centers to determine whether images are of a sufficient quality to grade. In the current study, researchers evaluated the quality of OCT-A images as assessed via a central reading center 3 multicenter clinical studies for diabetic retinopathy. 

All images were obtained between August 2016 and October 2019, with grading between January 2018 and November 2019. Image quality determination was performed using an empiric SSI threshold of 55 for AngioVue (Optovue) images and the manufacturer recommended SSI threshold of 7 for the AngioPlex (Carl Zeiss Meditec). 

Analyses included 7539 scans from 787 eyes of 461 participants (mean age, 62±12 years; 68% White, 56% men, 83% with type 2 diabetes). Patients were subdivided via study protocol: the first, protocol AA, included 5473 scans from 579 eyes; the second, protocol W, included 1929 scans from 171 eyes; and the third, from protocol AE, included 137 scans from 37 eyes. The average number of scans per participant was 16 (±10; range, 2-59).

Overall, 61% of scans were of good quality: 62% in the 3×3 mm scans, 63% in the 6×6 mm scans, and 59% in the optic nerve scans. Investigators found a moderate correlation in scan quality among scans from the same eye (intraclass correlation=0.52). Participants with good quality scans were more likely to be young men with better visual acuity (60±12 years, 64% men, Early Treatment Diabetic Retinopathy Study letter score 81.6±9.7). 

Of the 6216 scans captured via AngioVue, 61% passed quality control. The study determined reasons for poor image quality included low SSI (67%) and excess motion (24%). An amended grading protocol was applied to 21% of these scans, 40% of which failed quality control and 22% of which had an artifact present in addition to low SSI. 

Of the 1323 scans captured via AngioPlex, 63% were of good quality. Scan failures due to excess motion accounted for 47% of images; media opacity and low SSI accounted for only 21% and 19% of images, respectively. An amended grading protocol was applied to 46% of these images, of which 32% failed quality control and 5% had an artifact present in addition to low SSI. 

There were 662 images that were captured via both AngioVue and AngioPlex. Of these, 47% were good quality for both modalities, 16% were good quality for the AngioPlex only, and 16% were good quality for the AngioVue only. A total of 20% of images were of poor quality for both modalities. 

Study limitations include a lack of manufacturer recommendations for SSI limits for the AngioVue, the potential learning curve of graders, and an inability to identify additional ways to improve scan quality. 

“OCT-A provides the ability to identify and quantify microvascular changes resulting from diabetic retinopathy on a more precise scale than previous standard techniques such as fundus photography or fluorescein angiography,” the researchers concluded. “OCT-A has the potential to become an important endpoint in future multicenter clinical trials; however, the consistency of obtaining high-quality OCT-A scans must improve dramatically through hardware and software innovation.” 

Disclosure: Several study authors declared affiliations with the pharmaceutical industry. Please see the original reference for a full list of authors’ disclosures. 

Reference

Lujan BJ, Calhoun CT, Glassman AR, et al; for the DRCR Retina Network. Optical coherence tomography angiography quality across three multicenter clinical studies of diabetic retinopathy. Published online March 2, 2021. Transl Vis Sci Technol. doi:10.1167/tvst.10.3.2