Rami HE, Barham R, Gupta A, Fleming A, van Hemert J, Sun JK, Silva PS, Aiello LP. Correlation of hemorrhages and/or microaneurysms (H/Ma) severity and counts between ultrawide field (UWF) retinal imaging and standard ETDRS color stereoscopic retinal photography (ETDRS Photos). Poster presented at the 2016 ARVO Annual Meeting; May 1, 2016. Seattle, WA. [abstract] Invest Ophth Vis Sci. 2016 Sep 1; 57(12):1660.

PURPOSE: To evaluate the agreement in H/Ma identification and severity between UWF images and ETDRS photos.

METHODS: Eyes with no to moderate nonproliferative diabetic retinopathy (NPDR) by ETDRS photos were identified. UWF imaging (stereoscopic 200° pairs of each eye, Optos P200Tx) and ETDRS photos (stereoscopic 35 mm color 7-field 30° photography, Zeiss FF4) were both acquired by standardized protocol at the same visit. H/MA severity was evaluated according to the ETDRS detailed grading scale. H/Ma were counted in ETDRS fields 1-7 and in the UWF areas not covered by ETDRS photos. For localization, UWF images used standardized digitally defined ETDRS photo fields. All images were evaluated at a centralized reading center.

RESULTS: A total of 133 eyes [25 (18.8%) no DR, 47 (35.3%) mild NPDR and 61 (45.9%) moderate] were evaluated and 727/798 (91.1%) fields were gradable for H/Ma in both UWF and ETDRS photos. Simple/weighted kappa agreement for H/MA severity overall and for fields 2-7 were: All=0.61/0.69, F2=0.70/0.77, F3=0.62/0.73, F4=0.50/0.62, F5=0.54/0.65, F6=0.64/0.70, and F7=0.58/0.63. Overall, exact severity agreement was 74.6% and within 1-step in 99.2%. ETDRS photos identified more severe H/Ma in 12% and UWF images in 13%. A mean of 43.4 H/Ma were identified in the ETDRS photo field area compared to 47.8 identified within the same ETDRS area using UWF images (p<0.0001). A total of 15.5 additional H/Ma (35.7% increase) were identified in the peripheral UWF fields. Peripheral H/Ma were identified in 4 (16%) eyes with no DR observed on ETDRS photos.

CONCLUSIONS: There is good to excellent agreement between UWF images and ETDRS photos in determining H/Ma severity. There is excellent correlation in identifying H/Ma within ETDRS photo fields, similar to historical ETDRS H/Ma agreement rates. UWF peripheral fields identified 36% more H/Ma, suggesting a more severe H/Ma severity in 13%. Given the UWF:ETDRS H/Ma correlation, additional lesions detected in peripheral fields and known risks associated with peripheral H/Mas, future detailed evaluation of H/Ma counts using UWF images may provide a useful tool for evaluation of disease activity and progression.