Ultrawidefield Imaging in Diabetic Retinal Disease

With enhanced ability to visualize the retinal periphery, screening and treatment are more effective.

COLIN S. TAN, MBBS, FRCSEd (Ophth), MMed (Ophth) • SEENU M. HARIPRASAD, MD

Diabetic retinopathy affects up to 93 million people worldwide and causes up to 17% of total blindness.^1,2 In the United States, diabetic retinopathy is the leading cause of new-onset vision loss among working-age adults. It is projected that, by 2050, the number of Americans with diabetic retinopathy will increase to 16 million. Of these, approximately 3.4 million will have sight-threatening diabetic retinopathy.³

The frequency of diabetic macular edema increases with the severity of diabetic retinopathy, being 3% among those with mild, nonproliferative diabetic retinopathy and increasing to 71% among those with proliferative diabetic retinopathy.⁴ Its frequency also increases with the duration of disease, with 30% of adults who have had diabetes for more than 20 years suffering from DME.⁵

Diabetes, and especially DME, has been shown to have considerable impact on patients’ quality of life. A study by Hariprasad et al⁶ used the 25-item National Eye Institute Visual Function Questionnaire (NEI VFQ-25) to evaluate vision-specific quality of life among patients with diabetes. The authors reported lower scores among type 2 diabetes patients with DME, compared to control groups with diabetic retinopathy only, glaucoma, age-related macular degeneration, and cataracts.

To reduce the morbidity caused by diabetic retinopathy and DME, early detection and treatment are required. Accurate evaluation of the presence and severity of diabetic retinopathy is essential for allowing timely interventions to be initiated.⁷

Screening for diabetic retinopathy is essential for detection of early changes of diabetic retinopathy and DME. While screening programs around the world vary, many employ some form of fundus photography to detect diabetic eye disease. In addition, ophthalmologists also rely heavily on fundus imaging to diagnose and evaluate the complications of diabetes and to monitor treatment outcomes.

FUNDUS IMAGING

Standard fundus cameras capture 30º to 55º views of the posterior pole. Using the seven standard fields from the Early Treatment Diabetic Retinopathy Study, up to 75º of the posterior pole can be imaged using a montage of photos. If the patient is cooperative, a contact lens system, such as the Ocular Staurenghi 230 SLO Retina Lens (Ocular Instruments, Bellevue, WA), can obtain images covering 120º of the retina.

Each of these options has specific disadvantages. A single field image covers only a very small segment of the posterior pole. While the field of view is larger using the seven standard ETRDS fields, it has the disadvantage of image artifacts at the borders, where the individual photos overlap.

In addition, when obtaining steered images during fluorescein angiography, each of the images is captured at a different phase of the angiogram. Because the transit of fluorescein occurs within a relatively short period of time (10 to 30 seconds), important information can be lost as a result.

ULTRAWIDEFIELD IMAGING

In recent years, several ultrawidefield (UWF) imaging devices have been introduced and rapidly adopted by ophthalmologists, especially retinal specialists (Figure 1). Widefield imaging refers to the ability to capture 102º of the posterior pole with one image. In contrast, UWF can capture 200º with one image.

Figure 1. A) Ultrawidefield (UWF) pseudocolor image of the posterior pole in a patient with proliferative diabetic retinopathy. Extensive panretinal photocoagulation scars are seen. B) UWF fluorescein angiogram of the same eye illustrating areas of leakage from retinal neovascularization. There are several areas of capillary dropout.

A recent noncontact lens, used on the Heidelberg Spectralis device (Heidelberg Engineering, Heidelberg, Germany), can image 102º of the posterior pole for both fluorescein and indocyanine green angiography. This device, however, is not able to capture color fundus photographs.

Another common range of UWF devices is marketed by Optos (Dunfermline, United Kingdom). The Optos 200Tx can image up to 200º of the retina in a single image. Using steered images, even larger regions of the retina can be imaged.

In addition to pseudocolor photography, fundus autofluorescence (FAF) and fluorescein angiography (FA) are available on some UWF devices and indocyanine green angiography (ICGA) will soon be available. UWF imaging has been shown to be of relevance in many common retinal diseases, including diabetic retinopathy,^8-10 AMD,^11,12 retinal vein occlusion,^13,14 uveitis, and retinal dystrophies. In a paper evaluating FAF among patients with neovascular AMD, those with non-neovascular AMD, and normal controls, several types of abnormal peripheral FAF patterns were described.¹¹ More interesting is that the frequencies of these FAF abnormalities varied with the severity of AMD, being most frequent in patients with neovascular AMD.

DETECTION OF DIABETIC RETINOPATHY

Fundus photographs have often been employed to screen for diabetic retinopathy or detect its progression or as an objective record for the documentation of the patient’s status. Single field 50º fundus photographs are often used, with some other centers obtaining two or more overlapping fields. In clinical studies, the seven standard ETDRS fields are often used, but these are time-consuming to acquire and not logistically feasible for routine screening programs.

Because the lesions associated with diabetic retinopathy may occur anywhere in the retina, relying only on the central 50º to 75º of the posterior pole can miss lesions that occur more peripherally. Hence, UWF fundus photography has the potential to play an important role in the screening and documentation of diabetic retinopathy.

In a prospective study of 103 diabetic patients, Silva et al¹⁰ compared the sensitivity and specificity of UWF images with dilated ETDRS photography and dilated fundal examination by a retinal specialist. The authors reported exact agreement between UWF images and ETDRS photographs in 84% of eyes (weighted kappa 0.85), with 91% of comparisons occurring within one step.

Analyzing by disease severity, there was 99% perfect agreement between eyes with no DR and those with any DR present. The agreement was 92% when differentiating eyes with very severe nonproliferative diabetic retinopathy or better from proliferative retinopathy.¹⁰

Silva et al¹⁵ also reported that one-third of lesions in diabetic retinopathy were located outside the seven standard ETDRS fields. Some of these lesions included intraretinal microvascular abnormalities and new vessels elsewhere. More importantly, peripheral lesions located outside of the seven standard fields would have suggested a more severe level of diabetic retinopathy in 10% of eyes.¹⁵ Because these are relatively recent findings, the implications of peripheral lesions on the risk of progression of diabetic retinopathy are unknown.

The sensitivity and specificity of UWF images for detecting DR on ETDRS photographs were 99% and 100%, respectively, in the study by Silva et al.¹⁰ Another paper, by Wilson et al,¹⁶ reported that UWF images achieved sensitivity of 83.6% compared to 82.9% for digital photography in identifying referable disease.

One other significant advantage of UWF imaging is that the acquisition time is less than half that of dilated ETDRS photography. Silva et al reported that the time taken to acquire nonmydriatic UWF images was 170±80 seconds, compared to 370±130 seconds for ETDRS photography.¹⁰

PERIPHERAL NONPERFUSION AND ITS ROLE IN DME

One of the most interesting observations from the ability to perform UWF FA is the realization that there exists considerable nonperfusion in the peripheral retina among eyes with retinal vascular diseases, including diabetic retinopathy and RVO (Figure 2).

Figure 2. Ultrawidefield fluorescein angiogram of a patient with proliferative diabetic retinopathy. Several areas of leakage from new vessels are seen. There is also extensive peripheral retinal nonperfusion.

Because UWF images exhibit significant distortion of the image at the peripheries, it is currently not possible to quantify the area of nonperfusion in actual anatomical units (eg, mm²). In the future, software will be available to correct the peripheral image distortion so that actual areas of nonperfusion can be measured.

The current method of assessing amounts of retinal nonperfusion uses the ischemic index, which is the number of pixels within the region of nonperfusion expressed as a percentage of the total visible retina.¹⁷

Investigators have reported a broad range of ischemic indices for retinal vascular diseases. Patel et al⁸ reported that, in a series of patients with diabetic retinopathy, the mean ischemic index was 47%, and it ranged from 0% to 99%. Similarly, in RVO, the ischemic indices have also varied from 0% to 100%, with means of 14.8% and 25% described by various authors.^13,17

In addition to the risk of causing retinal neovascularization, extensive retinal nonperfusion also has an impact on macular edema. This is believed to be caused by the production of VEGF by the ischemic areas. VEGF is a potent vasodilator that weakens the walls of the capillaries at the macula, which enhances permeability and causes leakage of lipids and fluid, resulting in edema.

A paper by Wessel et al¹⁸ reported that 26.3% of eyes with retinal nonperfusion had DME, compared to only 8.7% of eyes without ischemia. The severity of macular edema is influenced by the amount of retinal nonperfusion.

A study by Patel et al⁸ grouped 148 eyes into four cohorts, based on the severity of diabetic retinopathy. The authors reported that patients with mild nonproliferative diabetic retinopathy had a mean ischemic index of 0%, whereas those with PDR varied from 53% to 65%.

Those with larger areas of nonperfusion had more recalcitrant DME and needed more treatment with macular photocoagulation. Among patients with active PDR, the decrease in central macular thickness was the smallest (7.2%), and these patients required a mean of 5.7 sessions of macular photocoagulation.

An Important Point

It is important to point out that not all studies have reported a relationship between the presence of retinal nonperfusion and DME. A paper by Sim et al¹⁹ found no relationship between DME and peripheral ischemia or peripheral leakage.

Similarly, Wessel et al¹⁸ did not find a correlation between retinal nonperfusion and central macular thickness measured using OCT. These findings may indicate that not all regions of peripheral nonperfusion are of similar relevance. Some regions may contribute more significantly to macular edema formation than others, and it will be interesting to learn whether these can be determined using the current imaging modalities.

As a result of these findings, ophthalmologists are now considering novel ways to manage DME. Whereas panretinal photocoagulation was previously thought to worsen DME, it is now believed that patients with recalcitrant DME may benefit from targeted retinal photocoagulation to regions of retinal nonperfusion. Treating these zones could reduce VEGF production, causing a decrease in the edema at the macula.

Targeted retinal photocoagulation has the advantage of reducing the amount of treatment required, while sparing areas of perfused (and presumably viable) retina.⁸ Other authors have suggested combination therapy: using anti-VEGF drugs to inhibit existing levels of VEGF, while performing targeted retinal photocoagulation to reduce further VEGF production. This treatment could potentially enhance the durability of anti-VEGF treatment.

CONCLUSION

While imaging technology continues to advance, there are occasions when a new modality or feature has the potential to change the paradigms and fundamentally alter the way we view a disease. UWF imaging has given us insights into diabetic retinopathy and DME that enhance our understanding of their pathophysiologies and provide new ideas regarding their management. RP

REFERENCES