pharmacologic (i.e., anti-VEGF) treatment early on. However, as the Þbrovascular membrane contracts, extensive peripheral anterior synechiae (PAS) form and angle closure develops. This type of glaucoma is a difÞcult disease to treat pharmacologically, and usually glaucoma surgical intervention is needed to manage the IOP.

In summary, the NVG is managed by treating the inciting factor that is causing neovascularization in the angle, which is ischemia-induced increased VEGF levels, and, secondly, by treating the increased IOP.

5.4Tractional Retinal Detachment

As PDR progresses, contraction of the Þbroblasts in the preretinal neovascular membranes causes traction retinal detachment (TRD), macular dragging, and, sometimes, a combined traction-rhegmatogenous RD. Additional Þndings of anterior hyaloidal Þbrovascular proliferation with cyclitic membranes occur in severe cases with anterior extraretinal neovascularization that results in tractional detachment of peripheral retina and ciliary body. Hypotony may ensue in many such cases [18].

Eyes with nonclearing VH, bilateral VH, and/or TRD threatening or involving the macula frequently undergo PPV. The presence of combined tractionrhegmatogenous retinal detachment, regardless of macular status, is another indication for PPV. Surgical management and the outcome depend on the extent of posterior vitreous separation and the extent (e.g., focal vs. broad), location (e.g., anterior to the equator vs. posterior to the equator), and severity of vitreoretinal adhesion. In some severely ischemic eyes, antiangiogenic adjuvant pharmacologic therapy is supplemented to treat diabetes-related VH, retinal neovascularization, rubeosis iridis, and NVG.

5.5Treatment

5.5.1Panretinal Laser Photocoagulation

Panretinal laser photocoagulation (PRP) is the standard of care for the treatment of severe nonproliferative and proliferative DR [19]. The risk of severe visual loss is decreased by at least 50 % with PRP [19, 20]. The Diabetic Retinopathy Study (DRS) identiÞed four risk factors for severe visual loss: (1) any retinal neovascularization (NV), (2) location of NV at or within 1-disc diameter of the optic disc,

(3) severity of NV (NV at the optic disc > standard photograph 10A (1/4Ð1/3-disc area), NV elsewhere ≥1/2-disc diameter), and (4) vitreous or subhyaloid hemorrhage. Eyes with three or four risk factors have high-risk PDR since the risk of severe visual loss (visual acuity <5/200 on two consecutive visits) is 50 % at the 5-year follow-up without treatment. The beneÞts of PRP are most pronounced in eyes with high-risk PDR.

Both the DRS and the Early Treatment Diabetic Retinopathy Study (ETDRS) showed that PRP reduces the risk of severe visual loss in the setting of severe

NPDR. One should consider PRP in patients with severe NPDR or non-high- risk PDR especially for type 2 diabetes mellitus patients without macular edema [21] as well as in patients who have signiÞcant visual loss due to PDR in the fellow eye or who cannot be followed reliably. PRP before high-risk PDR characteristics develop is associated with an increased risk of moderate visual loss during the Þrst year after PRP, but it is associated with a decreased risk by 2Ð5 years after treatment.

The PRP technique involves delivering discrete, moderately white laser burns to the entire retina outside of the anterior to the temporal arcades (bringing the treatment no closer than three-disc diameters to the macular center) and optic nerve (bringing the treatment no closer than 500 μm to the optic nerve head). Regression of NV typically occurs within 3 weeks of treatment. One proposed mechanism of action is that PRP ablates the retina and thereby markedly reduces the source of upregulated VEGF and other promoters of angiogenesis (Another, not mutually exclusive, hypothesis is that PRP improves oxygenation to the remaining retina.). The increased ratio of proangiogenic factors (e.g., VEGF and erythropoietin) to antiangiogenic factors (e.g., pigment-epithelium- derived factor) promotes neovascularization. This ratio is decreased after laser treatment [12].

Side effects of PRP include visual Þeld constriction, night blindness, color vision changes, serous retinal and choroidal detachment, cycloplegia, and worsening of macular edema. In one study [22], ~8 % of patients had decreased central vision >2 lines due to persistent macular edema. Biweekly treatment may reduce this risk [23, 24]. There is also the possibility of visual loss immediately following PRP. The DRS reported vision loss of two to four lines within 6 weeks of PRP in 10Ð23 % of patients vs. 6 % of controls [25]. Severe visual loss was typically associated with the use of xenon photocoagulation, but this modality is no longer used, in part due to the unwieldy delivery system and also due to the visual Þeld defects induced by the larger spots.

In cases where full PRP treatment must be administered quickly (e.g., 1Ð2 sessions), one can administer topical corticosteroid and cycloplegia therapy (e.g., prednisolone acetate 1 % q.i.d. and atropine 1 % b.i.d.) to prevent the development of choroidal detachment and secondary angle closure glaucoma. In the setting of combined CSME and high-risk PDR, one can prophylax against the exacerbation of DME by providing sub-Tenon triamcinolone injection prior to PRP [26] or by administering intravitreal ranibizumab or bevacizumab just before or shortly after PRP is applied [27]. Intravitreal anti-VEGF agent injection may precipitate development of or progression of RD if signiÞcant NV is present [28, 29]. In contrast, the DRS showed that PRP does not exacerbate or precipitate traction retinal detachment in PDR.

In early NVG cases, PRP is the mainstay of treatment; the treatment may work by reducing oxygen demand and decreasing ischemia-related VEGF levels, resulting in regression of neovascularization. However, if synechial angle closure has already occurred, PRP may not have signiÞcant effect on the intraocular pressure (IOP). Surgical intervention with either Þltration surgery or a shunt procedure is then indicated to control the IOP.

5.5.2Pars Plana Vitrectomy and Endophotocoagulation

Pars plana vitrectomy (PPV) is performed for treatment of eyes with advanced DR, including PDR with nonclearing VH and TRD, especially if involving or threatening the macula. Panretinal photocoagulation is provided intraoperatively. For eyes with rubeosis iridis and neovascularization of the anterior chamber angle, the laser should be performed extensively, from outside the arcades up to the anterior edge of ora serrata. Temporally, treatment should extend no closer than 3-disc diameters temporal to the macular center.

The Diabetic Retinopathy Vitrectomy Study (DRVS) clearly showed that in type 1 diabetics, eyes that undergo PPV early for VH have a better chance of regaining VA of ≥20/40 (25 % in the early PPV group vs. 15 % in deferred group at the 2-year follow-up; p=0.01) [30]. The DRVS also demonstrated a beneÞt of early vitrectomy in eyes with severe active Þbrovascular proliferation. No PRP was performed during surgery. This study reported a Þnal visual acuity of ≥20/40 at 4-year follow-up in 44 % of eyes in the early vitrectomy group and in 28 % of the conventional management group (p<0.05) [31]. The conventional management included observation, PRP, or vitrectomy for RD involving the macula or for severe persistent VH of 6 months or more. Eyes with the most severe Þbrovascular proliferation showed the greatest beneÞt with early vitrectomy [32].

TRD involving or threatening the macula, a combined tractional-rhegmatogenous RD, and/or non-resolving premacular hemorrhage are other indications for early vitrectomy. Extramacular TRD typically progresses very slowly and is usually observed [33]. Recent advances in surgical instrumentation and vitrectomy machines have improved anatomic success of surgery for diabetic TRD due to more effective membrane dissection. The visual outcome after diabetic vitrectomy can be difÞcult to predict since macular ischemia can, on occasion, lead to unexpected visual results after surgery. Diabetic patients can retain relatively good vision despite substantial enlargement of the foveal avascular zone. Currently, vision improves ≥2 lines in ~75 % of patients and worsens in ~10 % of patients after diabetic vitrectomy [34]. Predictive factors for functional success include preoperative visual acuity, anatomic status of macula, complex Þbrovascular membrane dissection (iatrogenic retinal breaks, use of long-acting intraocular tamponade), and iris neovascularization [34].

Adjunctive pharmacotherapy with anti-VEGF drugs in combination with PRP and PPV has shown promise in some studies [27]. Ocriplasmin, a new drug approved for vitreomacular adhesion (VMA) [35], may also be a promising adjunct to relieve some vitreoretinal traction in diabetic TRD. Studies are underway to evaluate its effectiveness in diabetic TRD.

5.5.3Anti-VEGF Treatment

Drug-induced regression of retinal neovascularization in PDR was Þrst shown by the Macugen Diabetic Retinopathy Study Group [36] and Gonzalez and co-workers [37]. Since then, many case series have described temporary but robust regression of retinal and iris neovascularization with anti-VEGF treatment [38Ð40].

It may take 2Ð3 weeks to see the effect of PRP laser on iris and retinal neovascularization. In contrast, rapid and robust regression of iris neovascularization occurs within days after anti-VEGF intravitreal injections.

In a retrospective study, Avery et al. [27] showed complete or partial reduction in leakage of disc neovascularization by ßuorescein angiography within 1 week after intravitreal bevacizumab (IVB) in 44 eyes with PDR. Seventy-three percent of eyes had complete resolution of neovascularization [27]. The effect of these injections may be temporary but may be extremely helpful in NVG eyes to promptly initiate regression of neovascular tissue. As noted earlier, PRP-induced NV regression may not occur for a couple of weeks.

5.5.3.1Anti-VEGF Therapy as Adjunctive Treatment to PRP

In a prospective study of 80 eyes with PDR, Mirshahi et al. [41] administered 1.25 mg of IVB with the Þrst session of PRP laser treatment and found 87.5 % of the bevacizumab-treated eyes and 25 % of the sham eyes showed complete regression of neovascularization at 6 weeks (p < 0.005). The regression rate was identical, however, at week 16. Cho et al. [42] reported a lower incidence of VH in the cohort that was treated with 1.25 mg bevacizumab plus PRP compared to the cohort that received only PRP.

The response to anti-VEGF therapy is extremely rapid. Sometimes regression is seen within a couple of days [41, 43, 44]. As noted earlier, anti-VEGF treatment is being used as an adjuvant to PRP where prompt regression of neovascularization may be needed (e.g., neovascular glaucoma) or in an eye with PDR where the view of the fundus is compromised and precludes adequate PRP [40, 45, 46].

Anti-VEGF injections have been used in diabetic patients with VH to induce regression of PDR, promote faster resolution of VH, and possibly prevent vitrectomy [47]. The Diabetic Retinopathy Clinical Research Network (DRCR.net) compared intravitreal ranibizumab (0.5 mg) injections (n = 125) with intravitreal saline injections (n = 136) in eyes with VH and PDR. No clinically signiÞcant difference was noted in the rate of vitrectomy by 16 weeks in the two groups. The cumulative probability of PPV was 12 and 17 %, respectively. The rate of recurrent VH in the Þrst 16 weeks was, however, lower in the ranibizumab group [48].

Some studies report a lower incidence of VH after PPV in diabetics with adjuvant use of bevacizumab [49, 50].

5.5.3.2Perioperative Use of Anti-VEGF Therapy for PDR

Intravitreal bevacizumab is also being used as a preoperative adjunct before PPV in selected severe cases of PDR with VH, with or without TRD, to reduce the complication of intraoperative bleeding [43, 51]. Huang et al. [52] injected IVB in 40 eyes with PDR and VH, followed by PRP; a second IVB was given 4Ð6 weeks later for persistent VH. Eyes underwent PPV if VH persisted >12 weeks. The average time for clearance of VH in the bevacizumab-treated group was 11.9 weeks compared to 18.1 weeks in the control group (p = 0.02). Only 10 % of IVB-treated eyes underwent PPV compared to 45 % of the control group (p = 0.01). Eyes with TRD were not included in this study.