The Potential Value of Ultra-Widefield Imaging: JAMA Ophthalmology

In the context of an increasing global prevalence of diabetes, accurate risk stratification of individuals with diabetic retinopathy (DR) may help to optimize the use of health care resources, both in the community and in the eye clinic. The classification of DR has traditionally been centered on, and, until recently, confined by the limited field of traditional retinal imaging techniques. The Diabetic Retinopathy Severity Scale (DRSS) is based on the Early Treatment Diabetic Retinopathy Severity (ETDRS) standard 7-field photography, which although it captures only about 30% to 35% of the retinal area provides substantial information for risk stratification in DR. Noncontact, ultra-widefield (UWF) retinal imaging technology enables data capture from an additional 50% of the retinal area; evaluation of its relevance in the prediction of disease progression in DR was investigated as part of the DRCR Retina Network, Protocol AA.

Marcus and coauthors, for the DRCR Retina Network identified that predominantly peripheral lesions (PPL) as detected on UWF fluorescein angiography (UWF-FA)—but not with ETDRS standard 7-field photography or UWF color imaging alone—was predictive of disease worsening in patients with DR (1.7-fold increased risk over 4 years) as determined using the DRSS. In a prespecified secondary analysis of Protocol AA data, Silva and coauthors identified that greater retinal nonperfusion as determined using UWF-FA at baseline was associated with a higher risk of disease worsening and receipt of DR treatment following adjustment for baseline DRSS and known systemic risk factors. Taken together, these studies strongly support the predictive value of UWF-FA in patients with DR, with PPL appearing to contribute a greater risk than nonperfusion. These complementary findings demonstrate how additional data from the retinal periphery may refine existing systems of disease classification and have the potential to enhance the care of patients with DR through the earlier detection of treatable disease by identifying patients at greater risk of requiring intervention.

Although these studies may have immediate clinical impact in well-resourced health care settings, their relevance may be tempered somewhat by the significant resources needed to administer fluorescein intravenously to a large group of patients with DR in resource-limited health care settings. This may offset any benefit to health care economies from the more accurate risk stratification of patients with DR. Further barriers to UWF-FA assessment of patients with DR include access to UWF imaging technology and image interpretation (grading of PPL and nonperfusion on UWF-FA imaging) by relevant clinicians. Indeed, the relevance of any novel disease severity scale for DR that incorporates UWF-FA findings will depend on the extent to which these data can be captured, interpreted, and applied in different health care contexts.

UWF-FA may not be feasible as a screening tool. However, oral administration of fluorescein to patients with DR may increase accessibility to UWF-FA by circumventing the need for staff trained in intravenous cannulation and injection. The development of UWF optical coherence tomography angiography—currently capable of imaging to the insertion of the vortex veins at the equator —may also permit the evaluation of peripheral retinal perfusion in the future without the need for the systemic administration of an exogenous fluorophore and its practical challenges.

Deep learning techniques are capable of predicting sex, age, smoking status, and the risk of developing type 2 diabetes and chronic kidney disease based on retinal imaging studies alone. These technologies may yet identify features on UWF color images in patients with DR that are not apparent to human observers in this study (ie, look beyond classic DR lesions) and that may similarly add predictive value for ocular and systemic disease outcomes in patients with DR with or without adjunctive clinical metadata (ie, systolic blood pressure). However, the accurate identification of all peripheral DR lesions and nonperfusion may be dependent on the administration of fluorescein.

The authors' observations draw attention to the extent to which the pathophysiology of DR may be driven by, or manifest in, the retinal periphery. An analysis of UWF-FA data in study participants with DR identified that peripheral arterioles of narrow caliber were associated with disease severity and nonperfusion, an effect that was not identified with more central arterioles. Moreover, peripheral arteriolar narrowing was more common in eyes with predominantly peripheral lesions. Taken together, these 3 studies underline the value of UWF imaging in demonstrating its ability to predict disease severity, the risk of peripheral nonperfusion, and the risk of disease progression in DR. Further investigation may evaluate the choriocapillaris and choroid as well as the retinal circulation in patients with diabetes, comparing biomarkers in the periphery vs those observed at the posterior pole and their predictive value for disease progression.

More broadly, the findings of this study draw attention to the value of peripheral retinal assessment through UWF imaging in the phenotyping of other ocular disorders. For retinal vascular diseases (eg, retinal vein occlusion, sickle cell retinopathy, retinopathy of prematurity), UWF imaging may similarly offer value in predicting risk of progression through the inclusion of additional data from peripheral retina, which may be more vulnerable to ischemia. More intriguingly, UWF imaging may identify peripheral retinal features for disorders with clinical signs at the posterior pole, whichmay offer added diagnostic or prognostic value. For example, temporal linear drusen like deposits are regularly seen in the peripheral fundus in North Carolina macula dystrophy, a feature identified on review of UWF images.8 For individuals with age-related macular degeneration, UWF-FA and UWF color imaging identified an association between peripheral and midperipheral pigmentary changes, but not peripheral drusen, and delayed dark adaptation. It is likely that further studies using UWF imaging techniques may refine systems of disease classification and aid in the prediction of disease progression for a range of retinal disorders that may impact clinical management.

Ultimately, the demands placed on global resources by DR necessitate a more refined method of risk stratification for individuals with DR—beyond the DRSS alone—in order to safely lengthen follow-up intervals for lower-risk patients. This will enable the redirection of resources to individuals at higher risk of sight-threatening disease. For the purposes of screening individuals with diabetes in the community, further studies are required to compare UWF color imaging with standard retinal imaging protocols. However, for patients with proven retinopathy, the findings of these 2 studies increase the likelihood that UWF imaging, in some form, will contribute to risk stratification. Further investigation may determine whether the rate of change of nonperfusion, or PPL, as identified on serial UWF-FA assessments adds additional predictive value.

In the future, it is likely that aggregated data from retinal imaging studies, together with other ocular and systemic parameters (eg, diabetic control, blood pressure, serum lipids) shown to influence the risk of disease progression will personalize follow-up intervals for patients with DR. Moreover, UWF imaging could be incorporated into a "virtual"model of eye care whereby retinal imaging studies are undertaken in the community and interpreted by reading centers, in part alleviating the burden of monitoring DR in the eye clinic.

Source: Imran H. Yusuf; Andrew J. Lotery; JAMA Ophthalmology

doi:10.1001/jamaophthalmol.2022.3166