The Influence of Ultra-Wide field Fluorescein Angiography on the Diagnosis and Management of Diabetic Retinopathy

Abstract

Purpose: To study the influence of ultra-widefield fluorescein angiography (FA) on the diagnosis and management of diabetic retinopathy (DR). Methods: Ten experts in DR completed an online survey in which they were asked to diagnose and manage DR cases using different imaging modalities. Experts independently reviewed 20 cases of DR and provided a diagnosis and management plan for each case, first based on ultra-widefield color-free and red-free images alone and again with the corresponding ultra-widefield FA images. Experts were polled on their diagnostic confidence, use of FA in clinical practice, and opinions on the value of ultra-widefield FA. Based on the reference standard diagnosis, primary outcomes included diagnostic sensitivity and specificity with and without ultra-widefield FA. Secondary outcomes included intergrader agreement, expert confidence, management outcomes, and an analysis of experts' opinions on the clinical use of ultra-widefield FA. Results: Diagnostic sensitivity (95% CI) increased from 36% (29%-43%) to 69% (62%-75%) (P < .05) with ultra-widefield FA. Intergrader agreement (Fleiss kappa statistic, 0.29 [95% CI, 0.21-0.27] vs 0.44 [95% CI, 0.40-0.47]; P < .05) and expert confidence (38% vs 65%) also improved. In 39% of responses, management was changed from observation to treatment. Although 40% of experts did not request FA with the initial ultra-widefield color-free/red-free images, 80% found ultra-widefield FA clinically useful when provided. Conclusions: Diagnosis, treatment, and expert opinions on the use of FA all changed when a corresponding ultra-widefield FA was available. Incorporating FA into routine clinical practice may facilitate more accurate clinical decision-making.

Keywords: diabetic retinopathy; fluorescein angiography; ultra-wide field imaging.

Figure 1.

Experts’ responses to the questions: (A) Rate your comfort in diagnosing diabetic retinopathy (DR) based on ultra-widefield color fundus photography and ultra-widefield fluorescein angiography (FA). (B) Rate how useful you find ultra-widefield color fundus and ultra-widefield FA for evaluating DR. (C) How often do you use ultra-widefield FA in your clinical practice to evaluate DR?

Figure 2.

Survey study design presented to diabetic retinopathy experts. In Part I of the study, experts are presented with 20 clinical cases (color fundus and red-free images only) and are asked to provide the diagnosis, management plan, and recommended clinical follow-up. In Part II, experts are presented with the same cases in a new, random order with the addition of the corresponding fluorescein angiography. The experts are prompted again to provide the diagnosis, management, and recommended clinical follow-up. Experts are not able to see their previous responses from Part I.

Figure 3.

The 10 expert responses to each of the 20 presented cases. Cases are subdivided into different graphs based on their consensus reference standard diagnosis. The frequency with which experts selected each category of diabetic retinopathy (DR) based on the ultra-widefield color fundus and red-free images alone is compared with the addition of the corresponding ultra-widefield fluorescein angiography (FA). Increased diagnostic accuracy was more frequently associated with the use of ultra-widefield FA.

Figure 4.

Differences in diagnostic sensitivity and specificity of experts when grading diabetic retinopathy cases with and without the corresponding ultra-widefield fluorescein angiography.

Figure 5.

(A) A case of high-risk proliferative diabetic retinopathy ( PDR) and (B) a case of non-high-risk PDR are shown. In (A), with ultra-widefield color fundus and red-free images alone, most experts (80%) selected a diagnosis of nonproliferative diabetic retinopathy (NPDR, 50% severe NPDR and 30% very severe NPDR). With the addition of ultra-widefield fluorescein angiography (FA), 90% of experts selected a diagnosis of PDR (40% non-high-risk PDR and 50% high-risk PDR). In (B), 80% of experts diagnosed the case as moderate NPDR with ultra-widefield color fundus/red-free images alone. None of the experts selected a diagnosis of proliferative disease. With the addition of FA, all experts changed their diagnosis to PDR (80% non-high-risk PDR and 20% high-risk PDR). With this change in diagnosis, management also changed in both cases. In (A), 70% of experts elected to observe the patient based on ultra-widefield color fundus/red-free images alone. With the addition of the ultra-widefield FA, 100% of experts treated the patient with either panretinal photocoagulation (PRP) (50%), antivascular endothelial growth factor (anti-VEGF) (20%), or both (30%). In (B), 100% of experts elected to observe based on color fundus/red-free photos alone. With the addition of ultra-widefield FA, 90% elected to treat the patient with PRP (56%), anti-VEGF (11%), or both (33%). When polling the experts, 100% found ultra-widefield FA useful for diagnosis and management in both cases.

Figure 6.

Intergrader agreement measured as the unweighted kappa statistic. Each expert was compared with all other experts. This strip-plot displays the mean unweighted kappa statistic for a single expert (open circle) for each category. The average unweighted kappa statistic across all experts is displayed as a closed circle. The whiskers denote the 95% CI. The nonparametric sign test was used to determine statistical significance when comparing mean unweighted kappa statistics for paired ultra-widefield color fundus and red-free images alone vs the paired color fundus and red-free images with the corresponding fluorescein angiogram. P < .05 is considered significant and is denoted with an asterisk.