is entering the stage of regression, with few or no new vessels and extensive fibrous proliferations.

Extensive neovascularization in the anterior chamber angle is a strong indication for scatter photocoagulation, if it is feasible, regardless of the presence of high-risk characteristics. If this treatment is carried out before extensive closure of the angle has occurred, full-blown neovascular glaucoma can be prevented. When opacities of the media preclude retinal photocoagulation, cryoapplications or vitrectomy with endophotocoagulation may be used.

The presence of extensive retinal hemorrhages, IRMAs, venous beading, and opaque small arteriolar branches, often accompanied by prominent soft exudates, suggests rapidly progressive closure of the retinal capillary bed and severe retinal ischemia. New vessels usually are present in such eyes but may be relatively unimpressive. Severe retinal ischemia increases the urgency to initiate scatter photocoagulation, whether or not DRS high-risk characteristics are present, since eyes so affected appear to be at greater risk of anterior segment neovascularization (4). Patients should be aware of this risk and the risk of a sudden decrease in central vision, which may occur with occlusion of the remaining arterioles supplying the macula.

Systemic factors also should be considered in deciding whether to initiate treatment in patients with very severe NPDR or moderate PDR. Clinical impression suggests that progression of retinopathy may accelerate during pregnancy (96–100) or with the development of renal failure (101–105). If photocoagulation is deferred until high-risk characteristics develop, and this occurs in the later stages of pregnancy or when renal transplantation or dialysis is required, these more pressing problems may make it difficult to complete photocoagulation according to a schedule considered optimal from the ophthalmologic point of view. If measures to improve long-standing poor glycemic control are planned when retinopathy is already at this stage, photocoagulation of at least one eye should be considered because of the phenomenon of early worsening, described earlier.

PRP and Macular Edema

Macular edema sometimes increases, at least temporarily, after scatter photocoagulation, and this may be followed by transient or persistent reduction of visual acuity (106, 107). The ETDRS documented small early harmful effects of scatter photocoagulation, particularly full scatter, in eyes with macular edema, as well as in those without. The DRS also found early harmful effects, which were greater in the xenon group. At the 6-week posttreatment visit, 21% of argon-treated and 46% of xenon-treated eyes that had macular edema and were free of high-risk characteristics at baseline had a decrease in visual acuity of two or more lines, compared with 9% of untreated eyes. Comparable percentages for eyes with neither macular edema nor high-risk characteristics were 9%, 18%, and 3%, respectively. After 1 year of followup, the greater progression of retinopathy in untreated eyes had led them to catch up with treated eyes; in the group having macular edema without high-risk characteristics at baseline, the percentages with a decrease in visual acuity of two or more lines were 32, 33, and 34%, respectively, in the argon-treated, xenon-treated, and control groups (108). A multivariable analysis confirmed the independent effects of macular edema and treatment and provided no evidence of any interaction (109). Both the

ETDRS and the DRS support the clinical impression that eyes with macular edema requiring scatter treatment are at less risk of visual acuity loss when focal or grid treatment to reduce the macular edema precedes scatter photocoagulation. If a delay of scatter treatment seems undesirable, the ETDRS protocol can be used, combining focal/grid treatment for macular edema with scatter treatment in the nasal quadrants at the first episode of photocoagulation and adding scatter in the temporal quadrants at one or more subsequent episodes (110). Certainly, scatter treatment should not be delayed when the risks of vitreous hemorrhage or neovascular glaucoma seem high, regardless of the status of the macula.

PRP Treatment Techniques

It is important to realize that the size of the burn produced depends not only on the spot-size setting used, but also on power and duration; so it is difficult to compare techniques, even those using the same wavelength and spot-size setting, on the basis of number and theoretical size of burns. It is also difficult to describe burn strength; power level is not very helpful, since the required power for a burn of given strength depends on the clarity of the media and the pigmentation of the fundus, even if spot-size setting and duration are kept constant. One useful measure of burn strength is the need for retrobulbar anesthesia. What we consider to be optimal burn strength with the argon laser is just below the level at which treatment under topical anesthesia with 300–500-m, 0.1-s burns becomes painful for most patients. With topical anesthesia only, it usually is difficult or impossible to obtain burns of adequate strength when their duration is longer than 0.1 or 0.15 s or their size is >500 m. The ETDRS protocol for full scatter treatment provides useful guidelines for initial treatment, calling for a total of 1,200– 1,600, 500- m, 0.1-s argon laser burns of moderate intensity placed one-half to one burn apart and divided between two or more episodes (at least 2 weeks apart, if two episodes; at least 4 days apart, if three or more episodes). Burns usually appear to enlarge slightly within several minutes after their application, resulting in the closer spacing of the scatter burns.

Blankenship (111) suggested that keeping the posterior limit of scatter treatment farther from the posterior pole may reduce harmful treatment effects. More peripheral treatment protocol may provide a useful alternative for initial treatment of eyes with macular edema in which the urgency of treatment for severe PDR is thought to preclude division of scatter treatment between two or more episodes.

Among the techniques currently in use, the number of episodes in which initial scatter treatment is carried out varies from one to four or more. Those techniques using a smaller number of larger burns tend toward a single episode with retrobulbar anesthesia, whereas those using a larger number of smaller burns nearly always divide treatment into two or more episodes. Multiple episodes make it easier to avoid retrobulbar anesthesia and its occasional complications, but they may cause delays and inconvenience for patients who must travel long distances for treatment. Angle-closure glaucoma secondary to serous detachment of the peripheral choroid and ciliary body is less common when scatter treatment is carried out in two or more sessions over a period of 1 or 2 weeks (112, 113), and some observers believe that small losses in visual acuity also may be less common.

Substantial regression of new vessels usually occurs within days or weeks after the initial application of scatter photocoagulation, and eyes in which new vessels continue to grow despite initial treatment, or recur after partial or complete regression, usually respond well to additional treatment (114–117). Because techniques, assessments, and inclusion criteria between studies vary, it is difficult to directly compare them to report the response rate to an initial course of full PRP. It appears that, on average, about two thirds of eyes have a satisfactory response to initial scatter treatment. As mentioned earlier, this ratio tends to be more favorable in patients with type II diabetes. Patients with severe intraretinal lesions and actively growing new vessels, who typically have type I diabetes, often need multiple treatments. In most cases re-treatment gives positive results (36, 117).

The ETDRS protocol contains guidelines for follow-up treatment that seem suitable for general use. Six factors are considered: (1) change in new vessels since the last visit or last photocoagulation treatment, (2) appearance of the new vessels (caliber, degree of network formation, extent of accompanying fibrous tissue), (3) frequency and extent of vitreous hemorrhage since the last visit or last photocoagulation treatment, (4) status of vitreous detachment, (5) extent of photocoagulation scars, and (6) extent of traction retinal detachment and fibrous proliferations (118). If new vessels appear to be active, as suggested by formation of tight networks, paucity of accompanying fibrous tissue, and increase in extent in comparison to the previous visit, additional photocoagulation is considered. A single episode of vitreous hemorrhage coincident with the occurrence of extensive posterior vitreous detachment, particularly if the only vitreoretinal adhesion remaining is at the disc, argues less for additional photocoagulation than do recurrent hemorrhages unrelated to such an occurrence. The extent and location of photocoagulation scars also may influence the decision regarding additional photocoagulation treatment. If the previous scatter burns appear widely spaced, or if there are areas where scatter was omitted, additional photocoagulation is considered more seriously.

Vitrectomy for PDR

When vitrectomy was initially introduced in 1970 by Machemer et al. (119), the major indications in eyes with PDR were severe vitreous hemorrhage that had failed to clear spontaneously after a year and traction retinal detachment involving the center of the macula. As this procedure came into widespread use, it was recognized that it might be of value earlier in the course of very severe PDR (120). A clinical trial, the Diabetic Retinopathy Vitrectomy Study (DRVS), was established by the National Eye Institute to explore this possibility. In one part of the DRVS, eyes with recent severe vitreous hemorrhage (hemorrhage sufficient to completely obscure the posterior pole and to reduce visual acuity to 5/200 or less for at least 1 month) were randomly assigned to either early vitrectomy or conventional management (i.e., follow-up without vitrectomy unless retinal detachment involving the center of the macula occurred or the hemorrhage failed to clear during a 1-year waiting period) (25). After 2 years of follow-up, recovery of good vision (visual acuity of 10/20 or better) was observed more frequently in the early vitrectomy group, but loss of light perception tended also to occur more frequently in the early vitrectomy group (Table 6). Early vitrectomy appeared to be clearly advantageous only in patients with T1DM. These results suggest that early vitrectomy

Table 6

Percentages of eyes with visual acuities of 10/20 or better and no light perception (NLP) at the 2-year follow-up visit, by type and duration of diabetes and treatment group

E early vitrectomy group, D deferral group

From (25), copyright American Medical Association

aD minus E rather than E minus D (as for visual acuity ≥10/20), so that a positive value is a difference in favor of early vitrectomy, as is the case for visual acuity ≥10/20

should be considered in eyes with recent severe diabetic vitreous hemorrhage when it is known from prior examination that fibrovascular proliferations are severe, particularly if it appears that macular potential is good. Older patients with severe vitreous hemorrhage sometimes have surprisingly mild PDR, and in such patients it usually is preferable to allow more time for spontaneous clearing of vitreous hemorrhage before considering vitrectomy, particularly if vision in the fellow eye is good.

As vitrectomy techniques have evolved and improved and the frequency of serious complications decreased, additional indications have been suggested. These included traction on the disc, peripapillary retina, or macula that distorts these structures and leads to substantial reduction in visual acuity; opaque fibrous proliferations in front of the macula; and extensive preretinal hemorrhage (120–122) (Fig. 5b, c).

In a second study the DRVS compared early vitrectomy vs. conventional management in eyes that had extensive active neovascular or fibrovascular proliferations and useful vision, 65% of which had had previous photocoagulation (Table 7) (14, 123). In eyes with the most severe new vessels (the severe and very severe categories), early vitrectomy appeared to provide a greater chance of good vision with no increase in risk