neuronal cell loss in conditions without diabetes. Capillary microangiopathy secondary to neurosensory damage has rarely been described in a

laboratory model of glaucoma and in retinitis pigmentosa.102,108 These data suggest a contributory

role of neuroretinal cell loss in causing retinal capillary microangiopathy.

In the final stages of diabetic macular ischemia, neurosensory cellular apoptosis is seen in association with retinal capillary occlusion. Glutamate excitotoxicity appears to play a role in ischemia-associated neuroretinal cell death.109 Plasminogen activators and

oxidative damage contribute to the accumulation of glutamate.110,111 Clinically, this is manifested by thin-

ning of the macula, which is readily demonstrated on optical coherence tomography (Fig. 8.4).

8.3 Natural History

The natural history of diabetic macular ischemia is not well described. One explanation for the lack of information is the relative rarity of clinically identifiable macular ischemia in large prospective clinical trials using standard fluorescein angiography.112 Video-fluorescein angiography with the scanning laser ophthalmoscope reveals macular capillary dropout prior to the development of microaneurysms.113 The Early Treatment Diabetic Study provided some degree of information on the short-term natural history of diabetic macular ischemia. DMI was graded in 1,243 eyes over a 5-year period. However, the study was limited by attrition (36%)

and by the inclusion of only a few cases of severe macular ischemia. Analysis of the data demonstrated no change in the distribution of severity of diabetic macular ischemia over the 5-year study period.114 Over the long term, however, diabetic macular ischemia is known to progress. The perifoveal intercapillary area and size of foveal

avascular zone increase with advancing stages of diabetic retinopathy.13,113,115,116 Retinal vascular

remodeling may result in limited revascularization

at the border of ischemic retina and may result in visual improvement.68,117 The presumed mechan-

ism appears to be intraretinal neovascularization.117,118 The potential for improvement in

DMI is unknown, but is probably very low. There are no definitive data on the bilaterality or symmetry of DMI. Macular nonperfusion has been reported to occur sooner after the diagnosis of type 2 diabetes than type 1 diabetes.119 The higher prevalence of coexisting systemic hyperten-

sion in patients with type 2 diabetes may offer an explanation of this finding.10,120–122 However, the

true date of onset in type 2 diabetes is more difficult to assess than in type 1 diabetes. The time between the true onset and the diagnosis of type 2 diabetes may be less in young patients than in older patients. In addition, the prevalence of diabetic retinopathy was reportedly higher in type 1 than in type 2 diabetes mellitus in a study of an adolescent cohort.122 Thus, it appears likely that diabetic macular ischemia develops and progresses as an integral part of diabetic retinopathy related to severity and duration of hyperglycemia.

8.4 Clinical Evaluation

Various modes of clinical study provide information on the structural and functional effects of diabetic macular ischemia. Fundoscopy and fundus photography demonstrate ‘‘featureless’’ areas of ischemic retina in which no microaneurysms, blot hemorrhages, or exudates are present (Fig. 8.5). Surrounding the area of ischemia is the hypoxic penumbra where there are dilated ectatic capillaries and other typical findings of diabetic retinopathy.68 Larger caliber vessels traversing ischemic areas are

attenuated, sheathed, or appear as ghost vessels (Fig. 8.6).7,123,124 As seen in other ischemic retino-

pathies, focal depressions in the macula are due to ischemic infarcts.125

Fluorescein angiography (FA) is the gold standard in the clinical diagnosis of diabetic macular ischemia. FA displays relative hypofluorescent areas of retina where there is absence of blood flow through the macular capillaries. At the edges of ischemic retina, associated angiographic findings include arteriolar changes and capillary dilatation. The hallmark findings include enlargement and irregularity of the foveal avascular zone (FAZ) and widened, non-uniform spaces between macular

capillaries indicative of intervening capillary loss (Fig. 8.7).7,8,13,115,126 In the late phase of the fluor-

escein angiogram, the relative hypofluorescent patches of ischemia are surrounded by hyperfluorescent leakage (Fig. 8.8).8 The normal FAZ as determined by trypsin digestion studies is reported to have an average longest diameter of 0.65 mm with a considerable range from 0.12 to 1.2 mm

Fig. 8.6 Severely ischemic retina in which sclerotic major retinal vessels appear as white ‘‘ghost vessels’’

Fig. 8.8 Late-phase fluorescein angiogram demonstrating diabetic macular ischemia as areas of relative hypofluorescence surrounded by diffuse intraretinal fluorescein leakage

Fig. 8.7 Enlarged, irregular foveal avascular zone with widened spaces between macular capillaries resulting in increased perifoveal intercapillary area (PIA)

area of the FAZ.127 When measured by fluorescein angiography, the mean diameter (the average of

two perpendicular measurements) is 0.53– 0.73 mm.126,128 In diabetic retinopathy the average

longest diameter of the FAZ was reported to be 0.94 mm with a range from 0.74 to 1.02 mm.126 The mean diameter in diabetic eyes was 0.79 with a range of 0.66–0.91 mm.126 Given the irregularity

of the FAZ in diabetic retinopathy126, the FAZ area may be a more reliable measurement of ischemia than FAZ diameter. The normal FAZ area is

0.205–0.405 mm2, similar to that reported in mild-

to-moderate NPDR.2,13,113,115,116,126,129,130 How-

ever, the high end of the normal range of FAZ measurements exceeds 2 mm2 in area.115,126,131 In

severe NPDR and PDR, the reported range of the FAZ is 0.42–0.96 mm2.2,13,115,126,129 Certainly, there

are cases of severe macular ischemia that exceed these reported ranges (Fig. 8.4). For comparison, one disk areais equal to 1.77 mm2, assuming one disk diameter equals 1,500 mm.

Grading the degree of diabetic macular ischemia by fluorescein angiography is useful for clinical studies, but is limited by media opacity and lacks uniformity among published reports. The proportion of ungradable angiograms in the ETDRS was low (2%), but selection bias might suggest that media opacity may prevent detailed evaluation in the clinical setting.114 In a subsequent study in which the ETDRS scheme of grading DMI was used by Goebel et al., fluorescein angiograms were ungradable in 11% of eyes despite the exclusion of eyes with advanced cataract and corneal opacities.132 Assessment of macular ischemia by evaluating the perifoveal capillaries is appropriate as they appear particularly susceptible to occlusion from diabetes and

FAZ dimensions are strongly positively correlated with the severity of capillary nonperfusion in the posterior retina.7,126 Various measures of DMI include presence/absence of enlarged or irregular FAZ, nonperfusion within 1,500 mm from foveal center, gradations of areas of ischemia based on

disk area, and computerized calculations of the area of ischemia.13,81,133–135 The measured peri-

meter of the foveal avascular zone and the FAZ area may be a reasonable substitute for the ETDRS method of assessment, although it does not take into account the impact of ischemia outside the FAZ.13 The ETDRS approached assessment of macular ischemia by the use of reference photographs that demonstrated the boundaries of the grading groups. For example, standard photograph 1A represented the lower boundary of medium grade capillary loss (Fig. 8.9).114 Ischemic changes that were graded by the ETDRS included the degree of capillary loss, the size of the FAZ, capillary dilatation, and abnormalities of the arterioles. Capillary loss within 1,500 mm from the fovea was evaluated by dividing the area into five subfields. Each subfield was graded by severity: grade 0 (absence of capillary loss), grade 1 (questionable capillary loss), grade 2 (definite capillary loss, but less than

standard photo 1A), grade 3 (moderate capillary loss, equal to or greater than standard photo 1A, but less than standard photo 2), and grade 4 (severe capillary loss, equal to or greater than standard photo 2).8 The reader is directed to Fig. 8.10 for ETDRS standard photograph 2. The size of the foveal avascular zone was quantified by linear dimension: grade 0 (less than 300 mm), grade 1 (equal to 300 mm), grade 2 (300–500 mm), grade 3 (greater than 500 mm), and grade 8 (cannot grade, e.g., severely irregular FAZ). The outline of the FAZ was also graded by the degree of destruction of the normally smooth round to oval contour: grade 0 (normal), grade 1 (questionable), grade 2 (less than one-half is destroyed), grade 3 (more than one-half of FAZ destroyed, but some remnants remain), grade 4 (FAZ completely destroyed), and grade 8 (cannot grade). Capillary dilatation was graded by standard photographs with the same five-step grading scale as that used for capillary loss. Arteriolar abnormalities associated with ischemia were graded by severity and included focal narrowing, pruning, staining, broadening, and blurred contour. Focal arteriolar narrowing of perpendicular side branches is usually seen in smaller, terminal, or near-terminal branches (Fig. 8.11). Pruning is a short arteriolar stump

Fig. 8.9 ETDRS standard photograph 1A. Lower boundary of macular ischemia for medium grade capillary loss. (Reproduced with permission from the Early Treatment Diabetic Retinopathy Study Research Group)

Fig. 8.10 ETDRS standard photograph 2. Lower boundary of macular ischemia for severe grade capillary loss. (Reproduced with permission from the Early Treatment Diabetic Retinopathy Study Research Group)

directed to an area of capillary loss (Fig. 8.12). Staining is described as narrow fluorescent lines visible on each side of the arteriolar blood column. Staining may also be seen as an arteriolar segment appearing more fluorescent than neighboring segments (Fig. 8.13). Broadening of an arteriolar segment is due to staining of the arteriolar wall and is frequently preceded by a side branch or by focal narrowing of the arteriole. Broadening of an arteriolar segment and blurring of arteriolar contour are due to abnormal fluorescein leakage due to increased permeability. Despite the use of standard photographs, clear definitions, highly trained angiographers, and expert

clinicians, there was only moderate inter-observer agreement on grading the degree of DMI in the Early Treatment Diabetic Retinopathy Study (weighted kappa = 0.41–0.60).8 In the clinical setting, the ETDRS method of grading diabetic macular ischemia may be impractical and of limited clinical use. A recently proposed International Classification of Diabetic Retinopathy did not include grading of macular ischemia.136

Optical coherence tomography can accurately and reliably quantify macular thickness in diabetic retinopathy.137 In diabetes before clinical detection of retinopathy, pericentral retinal thickness may be

Fig. 8.13 Arteriolar staining/broadening (see arrow). (Standard photograph 4, reproduced with permission from the Early Treatment Diabetic Retinopathy Study Research Group)

reduced due to ischemic neurosensory tissue loss.138,139 Retinal nerve fiber layer thickness is sig-

nificantly reduced in preproliferative diabetic retinopathy.140 In specific areas of retinal ischemia there is a disturbance of the inner retina and highreflective deposit between the outer segments and the retinal pigment epithelium (Fig. 8.14).141 In the absence of edema, the ischemic areas in the macula appear thin on OCT (Fig. 8.4). Coexisting macular pathology is common in diabetic retinopathy and may complicate the interpretation of OCT.142 Idiopathic preretinal membranes may cause macular thickening and small retinal cysts.143 Although the quality and quantity of cystic changes may be helpful in differentiating between diabetic macular edema from preretinal membrane, both cause

retinal thickening that may offset neuroretinal thinning from DMI.141,143–145 This may explain the

finding that macular ischemia confounds the corre-

lation of visual acuity with macular thickness in diabetic macular edema (Fig. 8.15).132,146

Perimetry provides useful information on functional loss of vision in diabetic retinopathy beyond that of visual acuity.147 Various methods of perimetry show significant reduction of retinal sensitivity in diabetic eyes prior to the development of clinical findings of retinopathy and these findings are pre-

dictive of future development of diabetic retinopa- thy.97,148–150 Short-wave automated perimetry

(SWAP) isolates the blue–yellow neural pathway

and appears more sensitive than standard white- on-white automated perimetry (SAP).147,151,152

Areas of decreased retinal sensitivity on perimetry

reliably correlate well with angiographic areas of capillary dropout.153–155 With regard to central

macular ischemia, mean thresholds of SWAP show abnormalities before loss of visual acuity and the findings correlate well with increasing size of foveal

avascular zone and perifoveal intercapillary area.147,155 Worsening of functional defects on SAT

and SWAP correlates well with progression of diabetic retinopathy.152 Furthermore, visual field defects detected by SWAP are more common in eyes with macular edema, but may reflect ischemic damage rather than macular edema itself.156 If confirmed, SWAP may be helpful in stratifying patients with diabetic macular edema for potential visual return with treatment. Microperimetry offers the option to test retinal sensitivity while directly observing the fundus and demonstrates loss of retinal

sensitivity in areas of ischemia and reduced sensitivity at the border of ischemia.141,157 Although perime-

try may provide helpful information on the functional effect of DMI, it is primarily employed in research at this time.

Fig. 8.14 Comparison of normal (top) fundus and spectraldomain optical coherence tomography (SD OCT) and severe nonproliferative diabetic retinopathy (middle). The white arrow in the fundus photos shows the direction of SD OCT scans. Top right SD OCT scan shows all ten layers of the normal retina: innermost hyperreflective retinal nerve fiber layer (RNFL), the hyporeflective ganglion cell layer (GCL), the hyperreflective inner plexiform layer (IPL), the hyporeflective inner nuclear layer (INL), the hyperreflective outer plexiform layer (OPL), the hyporeflective outer nuclear layer (ONL), the hyperreflective

external limiting membrane (ELM), the hyporeflective photoreceptor cell layer (PR), the hyperreflective photoreceptor inner/ outer segment junction (IS/OS), and the hyperreflective retinal pigment epithelium/choriocapillaris (RPE/CC). The middle right image demonstrates thinning of inner retinal layers and abnormal high OCT reflectivity between the IS/OS and the RPE/CC on the SD OCT scan in the nonperfused area (NPA, double-sided arrow) of the retina. The bottom image shows the NPA on the corresponding fluorescein angiogram. (Reprinted from Unoki et al.141 Copyright # 2007, with permission from Elsevier)

The electroretinogram (ERG) provides an objective measurement of retinal function in diabetes. Prior to the development of clinical or angiographic findings of diabetic retinopathy, ERG most consistently shows an increase in the oscillatory potential implicit time (OP1, suggesting dysfunction among

bipolar, amacrine, and ganglion cells possibly related to preclinical circulatory insufficiency), and other functional inner retinal abnormalities are variably described.90 Progressive retinal dysfunction on ERG correlates with and may be predictive of progressive diabetic microangiopathy seen

clinically.90 Macular dysfunction studied by multifocal ERG (mfERG) is inconsistently localized to specific areas of apparent pathologic fundus lesions,

perhaps due to the presence of occult macular ischemia.161,162 Foveal cone ERG parameters may be

abnormal in diabetic retinopathy with or without edema, but severity of dysfunction on mfERG

seems to correlate with degree of retinal thickness.163,164 The wealth of information on ERG

and diabetic retinopathy includes, but does not clearly separate, the effects of diabetic macular ischemia and edema.90 Electroretinography is not used in the routine clinical management of PDR.

Color perception testing reveals a variety of

defects in diabetic retinopathy beginning before significant visual loss.96,165 Approximately 50% of

patients enrolled in the ETDRS had abnormal color vision on the Farnsworth-Munsell 100 hue test.166 Tritan color defects are the most common color deficiency in diabetic retinopathy and may

relate to an S-cone (blue cone) pathway dysfunction or S-cone cell loss.167–169 In the ETDRS, factors

associated with impaired hue discrimination included age, diabetic macular edema, and neovascularization.170 The severity of macular ischemia

and altered retinal blood flow has also been shown to correlate with a tritan defect.171,172 At this time,

little useful clinical information is gleaned by color vision testing.

Contrast sensitivity testing reveals abnormal function before visual acuity testing and may be more sensitive and specific for degree of retinopathy than color vision testing.94 Contrast sensitivity loss correlates well with fluorescein angiographic

evidence of capillary dropout and macular edema.173,174 Impaired contrast sensitivity may be

reversed with improvement in metabolic control in non-ischemic nonproliferative diabetic retinopathy.175 Oxygen supplementation partially reverses abnormalities in contrast sensitivity and hue