intervention.36 Surgical intervention has also been proposed for a ‘‘taut’’ hyaloid membrane.37 Diagnosis of a ‘‘taut’’ hyaloid membrane is clinically based on a ‘‘glistening’’ reflex from the surface of the macula in conjunction with fluorescein angiographic evidence of diffuse oozing from tissue deep in the retina. The OCT demonstration of an associated epiretinal membrane conceptually supports the potential benefit of surgical intervention for this ‘‘taut’’ hyaloid membrane. Although the morphologic information from OCT is very helpful in diagnosing both of these conditions, vitrectomy for diabetic macular edema is still not currently considered to be standard-of-care as evidenced by the current National Eye Institute-sponsored trial to evaluate vitrectomy for diabetic macular edema.38

6.2Heidelberg Retinal Tomograph (HRT)

The Heidelberg Retinal Tomograph (HRT) (Heidelberg Engineering, Heidelberg, Germany) is a confocal laser scanning tomograph that sequentially acquires images in the x,y plane along the z-axis. The HRT II image contains

384 384 pixels in the central 4.5 4.5 mm square of the retina for a transverse resolution of approximately 12 mm. Given the sampling strategy of the HRT the transverse resolution is equivalent throughout the scanned area. As compared to the sampling strategy of OCT the HRT will provide more ‘‘edema’’ data points in the paracentral macula outside of the central 500 mm radius central subfield macular thickness (CSMT). However, it appears that global measures of diabetic macular edema consistently correlate with CSMT and provide no apparent differential information to modify treatment or prognosis.39 Therefore the theoretical advantage of HRT sampling strategy may not be a clinically relevant advantage.

Unlike OCT in which retinal thickness is measured by computing the distance from the RPE reflection and the internal limiting membrane (ILM) reflection, the HRT measures a change in the distribution of the reflected light which is associated with retinal edema. An edema index can be generated for each scanned point of the macula. This edema index measures the optical effect of edema but does not quantify the distance between the RPE and ILM. When this edema index is compared to a clinically derived map drawn by retina specialists examining patients with diabetic macular edema (clinically

significant macular edema), the HRT was 92% sensitive and 68% specific in detecting areas with edema40 and as a consequence the HRT has not gained wide acceptance. In the absence of clinical trials it remains to be determined whether HRT adds value to the patient above clinical examination.

6.3 Retinal Thickness Analyzer (RTA)

The Retinal Thickness Analyzer (RTA; Talia Technology Ltd., Neve-Ilan, Israel) works on a principle of slit lamp biomicroscopy. Sixteen sequential vertical scans are rapidly generated across the central 3 3 mm of the macula to generate a topographical map of retinal thickness. The axial resolution is up to 50 mm and the transverse resolution is 187 mm. RTA software generates a foveal average thickness measurement. When this foveal average thickness measurement is compared to a clinically derived map drawn by retina specialists examining patients with diabetic macular edema (clinically significant macular edema) the RTA was 57% sensitive and 71% specific in detecting areas with edema.40 This poor sensitivity and specificity translates into a reduced reliability in measuring macular edema as compared to stereo fundus photography41 and OCT technology,42 and as a consequence the RTA has not gained widespread use in clinical practice nor in clinical trials in managing diabetic macular edema. In the absence of clinical trials it remains to be determined whether RTA adds value to the patient above clinical examination.

6.4 Microperimetry

Retinal thickness measured by OCT correlates only modestly with visual acuity. Alternative explanations for visual loss (such as duration and severity of systemic diabetes and duration of edema) other than just the quantity of edema logically play a role. Fundus microperimetry is a scanning laser ophthalmoscope technology which maps fixation characteristics and sensitivity within the macula. Whereas visual acuity measures foveolar function,

microperimetry can measure parafoveal visual function. In diabetic patients eccentric fixation and unstable fixation correlate with poorer visual acuity, mean OCT thickness, and a cystic OCT morphology.43,44 The additional information provided by microperimetry over and above visual acuity data may logically allow diabetic macular edema to be subclassified into categories demonstrating differential response to therapeutic interventions. Such an investigation, however, has not yet been performed and it remains to be determined whether microperimetry adds value to the patient above clinical examination.

6.5 Color Fundus Photography

The large data set of the ETDRS demonstrates that stereoscopic color fundus photographs provide one with great predictive power to determine the risk of progression to proliferative diabetic retinopathy45 (see Chapter 5). Dependent on the level of experience, ophthalmoscopy is often not as sensitive as slide film in detecting diabetic retinopathy.46 Sevenfield stereoscopic digital imaging is equally sensitive to slide film in detecting neovascularization of the disc, neovascularization elsewhere, and clinically significant macular edema.47 Evaluation of stereoscopic digital images of fields 1 and 2 only with JPEG compression demonstrates a high correlation with seven-field stereoscopic slide film in detecting diabetic retinopathy.48 Non-stereoscopic digital imaging has poor sensitivity in detecting clinically significant macular edema.49 In summary, the acquisition of stereo color fundus photographs (either with slide film or digitally acquired) may add value to the evaluation of the diabetic patient dependent upon the experience of the examiner. These results may allow teleophthalmology strategies to effectively screen diabetic patients for appropriate referrals.

6.6 Fluorescein Angiography

Historically the development of fluorescein angiography (and fluorescein angiography conferences) defined the onset of medical retina as a distinct

discipline.50,51 Since its original descriptions in 1960s52–54 fluorescein angiography remained the dominant imaging technique for managing retinal and choroidal vascular diseases until OCT became widely available at the start of the 21st century. By illustrating the competency of the retinal vascular and choroidal circulation the pathophysiology of diabetic retinopathy became much better understood. The landmark clinical trials of the Diabetic Retinopathy Study55 and Early Treatment Diabetic Retinopathy Study used fluorescein angiography in classifying disease severity,56 guiding laser therapy,57 and evaluating the response to therapy.35 With this historical background one may ask the question that in the 21st century does fluorescein angiography still add clinical value to the patient?

Regarding proliferative diabetic retinopathy the Diabetic Retinopathy Study (DRS) defined indications for benefiting from scatter photocoagulation.58 Findings which placed the patient at high risk for losing vision were defined. These high-risk factors are (1) new vessels present, (2) new vessels located on or within 1 disc diameter of the disc, (3) new vessels moderate to severe (NVD standard photograph 10A or, for eyes without NVD, NVE 1/2 disc area), and (4) vitreous or preretinal

hemorrhage present.59 Determination and diagnosis of these features are based on clinical examination and not fluorescein angiography. In the DRS the cumulative 2-year rates of severe visual loss was as high as 36.9% in controls (see Table 6.1).

Groups F, H, I, and J have three or four high-risk factors and in those groups scatter photocoagulation significantly reduces the 2-year risk of severe visual loss. In groups with less than three high-risk factors, the difference between treatment and control groups was not significant and therefore prompt scatter photocoagulation was not advised. One can see that in a diabetic patient with vitreous hemorrhage but without clinically visible neovacularization (Groups B and D) the identification of neovascular tissue within the standard seven fields would change one’s recommendation for prompt scatter photocoagulation. Therefore, in diabetic patients with vitreous hemorrhage but without identifiable neovascular tissue fluorescein angiography to attempt to identify leakage associated with occult NVE or NVD would change one’s therapeutic recommendation.

In patients with nonproliferative diabetic retinopathy the indication to benefit from focal photocoagulation is clinically significant macular edema. As

reviewed in Chapter 5 (Defining Diabetic Retinopathy Severity) this definition is a clinical diagnosis based on slit lamp biomicroscopy and does not rely on fluorescein angiography. Therefore the information obtained from fluorescein angiography should not change one’s recommendation as to whether or not the patient should undergo laser therapy. However, the Early Treatment Diabetic Retinopathy Study utilized a combination of two laser strategies to treat diabetic macular edema. A focal treatment was applied directly to all leaking microaneurysms between 500 and 3,000 mm from the center of the fovea. (This pattern of leakage arising from microaneurysms was termed ‘‘focal’’ leakage in the ETDRS.) A grid treatment was subsequently applied to all areas of diffuse leakage and for areas of capillary nonperfusion. (This pattern of leakage was termed ‘‘diffuse’’ leakage in the ETDRS). Identification of microaneurysms, diffuse leakage, and areas of capillary nonperfusion are all features identified by fluorescein angiography. Fluorescein angiography is therefore indicated based on ETDRS data to help guide initial laser therapy.

The terms focal and diffuse fluorescein leakage are often used imprecisely.60 Focal fluorescein leakage presumably arising from microaneurysms was differentiated in the ETDRS from diffuse fluorescein leakage presumably arising from leaking capillaries. The value of distinguishing between diffuse and focal diabetic macular edema is uncertain (see Sidebar in Chapter 7 Diabetic Macular Edema).35

Although ETDRS protocol utilized fluorescein angiography to guide therapy it is not clear whether such guidance is necessary to obtain beneficial results. Since the results of the ETDRS were released, interpretation of ‘‘ETDRS laser technique’’ in clinical practice varies. The Diabetic Retinopathy Clinical Research Network (DRCR.net) is a collaborative network involving 150 sites (both community and university based) with over 500 investigators.61 Based on a survey of its investigators the initial protocol compared two prevalent laser treatment techniques for management of diabetic macular edema.62 Microaneurysms were specifically treated in one laser treatment protocol (so-called modified ETDRS focal/ gird [mETDRS]). The visual acuity results at 12 months were not statistically different from the protocol in which microaneurysms were specifically not treated (so-called modified macular grid treatment

[MMG]). In a subsequent DRCR.net study comparing this ‘‘modified ETDRS’’ laser technique to intravitreal triamcinolone, fluorescein angiography was not required and performed only at investigator discretion.63 This ‘‘modified ETDRS’’ laser treatment has not been compared to protocol ETDRS laser treatment in a clinical trial. However, there is support in the literature64 and it appears that current clinical practice based on DRCR.net protocols demonstrates a positive treatment response when laser is not guided by fluorescein angiography. The ETDRS also found no fluorescein angiographic variable of prognostic importance for outcome.35

Retinal vascular diseases characterized by ischemia are associated with an upregulation of vascular endothelial growth factor65 with a subsequent increased risk of retinal and iris neovascularization. Multiple clinical research protocols are currently investigating the benefit of anti-VEGF intravitreal injections in the management of proliferative diabetic retinopathy and diabetic macular edema.66–68 As a means to document retinal circulation, fluorescein angiography is the best ancillary study to investigate ischemia. In contrast to ischemia of the macula, ischemia of the peripheral retina has a greater association with an increased production of VEGF and other vasoproliferative factors. If one could categorize peripheral retinal ischemia in diabetic retinopathy would this add value for the patient?

The technology currently exists to give us this information. Conventional fluorescein angiography images the retinal circulation within the posterior pole. Ultra wide-field fluorescein angiography is a technique which can visualize the peripheral retina far beyond the seven standard photographic fields of the Modified Airlie House Classification (see Fig. 6.8).

The results of the ETDRS are typically viewed with regard to its conclusions regarding the management of macular edema. One of the principal questions asked by the Early Treatment Diabetic Retinopathy Study (and perhaps why it was named ‘‘early treatment’’) was ‘‘when in the course of diabetic retinopathy is it most effective to initiate photocoagulation therapy?’’ Patients were eligible if they had mild, moderate, or severe nonproliferative retinopathy or early proliferative retinopathy. These patients were then randomized to early photocoagulation or deferral of photocoagulation until high-risk characteristics developed. In patients who received

Fig. 6.8 Example of ultra wide field fluorescein in angiography taken with a scanning laser ophthalmoscope

early photocoagulation the pattern of scatter laser photocoagulation was randomized to either ‘‘mild’’ scatter or ‘‘full’’ scatter. About 1843 patients received a full-scatter treatment composed of 1,200–1,600 burns and 1,868 patients received a mild-scatter treatment composed of 400–650 burns.27 At 5 years the rate of developing high-risk proliferative retinopathy was 18.4% in the full-scatter group and 28.7% in the mild-scatter group. Therefore, mild-scatter photocoagulation is not as effective as full-scatter photocoagulation in preventing progression to high-risk proliferative retinopathy. However, mildscatter photocoagulation was better than observation in that 40.7% of patients in the observation group developed high-risk proliferative retinopathy at 5 years. The ETDRS therefore provides evidence that mild-scatter photocoagulation provides a reduction in the rate of progression to high-risk characteristics but not as great a rate as full scatter.

It is intuitive that if one were to have angiographic information from the retinal periphery which would subclassify diabetic retinopathy with regard to the extent of peripheral ischemia then this subclassification might allow the mild-scatter treatment protocol to achieve the degree of benefit demonstrated by full scatter without as great a risk of suffering moderate visual loss (9.7% at 4 months in full scatter vs. 6.4% at 4 months in mild scatter).27 At this time a study using a subclassification of peripheral retinal ischemia to direct the extent of laser treatment has not been performed.

Table 6.2 Occurrence of moderate visual loss at 5 years

Moderate vision loss

Adapted with permission from ETDRS Research Group27.

If one believes that a greater extent of peripheral retinal ischemia results in a greater upregulation of VEGF and a greater chance of diabetic macular edema then one might expect panretinal photocoagulation to provide some benefit for patients with diabetic macular edema. ETDRS data, however, do not support this hypothesis. The rate of moderate vision loss at 5 years in patients with macular edema who received scatter photocoagulation as a sole therapy was not much different from controls (see Table 6.2).

This data would therefore argue against the logic that wide-field fluorescein angiography would alter one’s therapy. Therefore, although it makes intuitive sense, the information gained from wide-field fluorescein angiography currently has not been demonstrated to add value beyond classifying retinopathy based on fundus examination alone.

In retinal vascular diseases ischemia of the macula and an enlarged foveal avascular zone may be a mechanism associated with visual loss exclusive of macular edema.69 The ETDRS categorized macular ischemia by grading the size of the foveal avascular zone, the features of the outline of the foveal avascular zone, and the areas of capillary loss <1,000 mm from the foveal center. There was only moderate agreement among the ETDRS professional graders in classifying capillary loss and abnormalities of the foveal avascular zone.70 Schemes to classify severity of macular ischemia are inconsistent.71,72 These inconsistencies between graders and between study centers make it difficult to know whether the additional information gained from fluorescein angiography of the macula adds therapeutic or prognostic value to patients. Since the classification schemes of the ETDRS and Global Diabetic Retinopathy Project (see Chapter 5) do