Early Vitrectomy for Severe Proliferative Diabetic Retinopathy in Eyes with Useful Vision

Results of a Randomized Trial-Diabetic Retinopathy Vitrectomy Study Report 3

THE DIABETIC RETINOPATHY VITRECTOMY STUDY RESEARCH GROUP

Abstract: Three hundred seventy eyes with advanced, active, proliferative di­abetic retinopathy (PDR) and visual acuity of 10/200 or better were randomly assigned to either early vitrectomy or conventional management. After 4 years of follow-up, the percentage of eyes with a visual acuity of 10/20 or better was 44% in the early vitrectomy group and 28% in the conventional management group. The proportion with very poor visual outcome was similar in the two groups. The advantage of early vitrectomy tended to increase with increasing severity of new vessels. In the group with the least severe new vessels, no advantage of early vitrectomy was apparent. [Key words: Diabetic Retinopathy Vitrectomy Study, early vitrectomy, proliferative diabetic retinopathy, randoniized clinical trial.] Ophthalmology 95: 1307 -1320, 1988

The Diabetic Retinopathy Vitrectomy Study (DR VS) was designed to evaluate the risks and benefits of per­forming early pars plana vitrectomy in eyes with advanced proliferative diabetic retinopathy (PDR). The DRVS comprises three studies. Two of these are randomized clinical trials involving early vitrectomy, and the third is a natural history study. The 2-year results in the natural history study and in the randomized trial for eyes with recent severe vitreous hemorrhage have been published. 1,2

The purpose of this article is to report results of the second

clinical trial, which compares early vitrectomy with con­ventional management in eyes with extensive, active, neovascular or fibrovascular proliferations, and useful vi­sion.

Originally received: June 14, 1988. Manuscript accepted: July 26. 1988.

Supported by contracts from the National Eye Institute. National Institutes of Health.

Reprint requests to Office of Biometry, The National Eye Institute, Bldg 31, Room 6A·10, 9000 Rockville Pike, Bethesda, MD 20892.

METHODS

ELIGIBILITY CRITERIA

Patients who had a history of diabetes, were 18 years of age or older, had no history of pituitary ablation, and gave written consent after receiving a detailed explanation of the study were eligible for entry if at least one eye ex­hibited extensive, active, neovascular or fibrovascular proliferations, and had a best-corrected visual acuity of 10/200 or better. The definition of extensive, active, neo­vascular or fibrovascular proliferations was developed

1307

OPHTHALMOLOGY • OCTOBER 1988 • VOLUME 95 • NUMBER 10

from clinical impression and a review of 160 eyes followed for 1 year or more in the natural history portion of the DRVS. To fulfill this definition it was necessary for an eye to have either (1) severe new vessels and severe fibrous proliferations; (2) severe new vessels and red vitreous hemorrhage; or (3) moderate new vessels, severe fibrous proliferations, and red vitreous hemorrhage. These se­verity levels are defined in the Eligibility Grading section of the Appendix. Stereoscopic fundus photographs of seven standard fields3 taken at the first baseline visit un­derwent a preliminary grading for eligibility at the Reading Center. Eyes in which the photographs did not show the required minimum amounts of new vessels and/or fibrous proliferations, but in which the baseline ophthalmic ex­amination had documented their presence, were declared "eligible by clinical examination."

Principal exclusion criteria were: (1) previous vitrec­tomy; (2) extensive neovascularization of the iris or new vessels involving more than 3 clock hours of the trabecular meshwork, neovascular glaucoma, or intraocular pressure greater than 30 mmHg despite treatment; (3) retinal de­tachment involving the center of the macula with visual acuity less than 10/50; (4) visual acuity less than 10/200; and (5) renal disease severe enough to require renal trans­plantation or chronic dialysis, or other coexistent disease that would make survival or availability for regular follow­up visits for at least 2 years unlikely.

BASELINE INFORMATION COLLECfED

At the baseline visit, a brief history and a detailed ophthalmic examination were carried out and recorded on a standard form. The examination included slit-lamp biomicroscopy and gonioscopy to detect new vessels on the iris or in the anterior chamber angle, binocular indirect ophthalmoscopy, and fundus examination with slit lamp and contact lens. Best-corrected visual acuity was mea­sured with the Diabetic Retinopathy Study protocol and charts, but at a distance of 10 feet.4 The fundus photo­graphs used to assess eligibility were subsequently graded in detail using the Modified Airlie House Classification,3

in which some lesions, such as new vessels on or within 1 disc diameter of the optic disc (NVD), are graded in only one of the seven photographic fields, while others, such as new vessels elsewhere (NVE), are graded in mul­tiple fields. These gradings were used to provide measures of the severity of each of more than 20 individual lesions, as described in the Appendix.

In addition, gradings of NVD and NVE were used to define four combined new vessel severity (NVC) subgroups: NVC-l (least severe), new vessel severity no greater than moderate and this severity level present in no more than one photographic field; NVC-2 (moderately severe), new vessels moderate in two or more fields, but not severe in any field; NVC-3 (severe), NVD severe or NVE severe in at least one field; NVC-4 (very severe), NVD severe and NVE severe in at least one field. Appen­dix Figure A5 defines these categories by the NVD and NVE grading scales.

1308

Patients were classified as having type I diabetes if di­abetes was diagnosed at 20 years of age or younger and insulin was being used at the time of entry into the study. Patients 40 years of age or older at diagnosis were classified as having type II diabetes (regardless of insulin use), as were patients with diabetes diagnosed at a younger age, ifthey were not receiving insulin at the time of entry into the study. Those patients receiving insulin whose diabetes was diagnosed between 21 and 39 years of age, inclusive, were assigned to an intermediate group, which presumably was made up of a mixture of patients with type I and type II diabetes.

RANDOMIZATION AND FOLLOW-UP VISITS

Patients eligible at the first baseline visit returned be­tween 2 weeks and 3 months later for a second visit, at which random assignment to early vitrectomy or conven­tional management was made, if results of clinical ex­amination confirmed continuing eligibility. An eye in which visual acuity had fallen below 10/200 could still be enrolled, if the cause of the decrease was considered to be vitreous hemorrhage and ophthalmoscopy (or ultrasound examination if the fundus was obscured) showed the cen­ter of the macula to be attached. Between July 1979 and May 1983, 381 eyes of 345 patients were randomized to early vitrectomy or conventional management in 15 clin­ics. Because of slow patient recruitment, four clinics dis­continued participation in the trial and no follow-up be­yond 2 years was available for their 11 patients. All data from the 11 eyes of these patients have been excluded, leaving a total of 370 eyes. Regular follow-up visits, at which the baseline ophthalmic examination (including measurement of best-corrected visual acuity) was repeated, were scheduled 3, 6, 12, 18,24,36, and 48 months after randomization. Additional visits were carried out as needed for clinical management.

EARLY VITRECfOMY AND CONVENTIONAL MANAGEMENT PROTOCOLS

In both treatment groups, the goals of vitrectomy in­cluded: (1) the removal of all vitreous opacities, except those in the peripheral vitreous anterior to the equator; (2) the division of all membranes and strands extending between the vitreous base and the optic disc or other areas ofvitreoretinal attachment; and (3) the division of elevated retino-vitreo-retinal sheets or bands and separation of epiretinal membranes from the retina, with removal of these structures to the extent feasible, and segmentation of those remaining into separate islands. Lensectomy was permitted only if it would significantly increase the chance of successful surgery; it was performed in 6.2% of the 177 eyes that were phakic at the time of vitrectomy in the early group. Scatter (panretinal) endophotocoagulation was not used during surgery. Washout of vitreous hem­orrhage occurring after vitrectomy and reducing visual acuity to less than 10/200 were permissible if hemorrhage

DRVS • DRVS REPORT 3

Table 1. Baseline Characteristics by Treatment Group

Total Total No. of No. of

Characteristics Early Conventional* Eyes Characteristics Early Conventional * Eyes

No. of eyes observedt 181 189 370 Fibrous proliferations Age at baseline (yrs) elsewhere"

Mean 39.8 40.9 40.4 ~M/1 27.6 30.2 28.9 SD 13.7 14.5 14.1 M/2345 34.8 27.5 31.1 Maximum 74 78 78 ;?S/1 37.6 42.3 40.0 Minimum 18 18 18 Retinal elevation II

Age at diagnosis (yrs) Absent 33.1 25.9 29.5 Mean 20.8 22.2 21.5 O-D/23 31.5 37.0 34.3 SD 14.5 15.9 15.2 ;?D/45 35.4 37.0 36.2 Maximum 60 66 66 Preretinal hemorrhage" Minimum 1 1 1 Absent 36.5 31.7 34.1

Duration of diabetest (Years) 0-D/23 32.6 43.9 38.4 Mean 19.0 18.8 18.8 ;?D/45 30.9 24.3 27.6 SD 8.0 7.4 7.7 Vitreous hemorrhage" Maximum 46 42 46 Absent 19.3 19.6 19.5 Minimum 1 <1 <1 0-D/23 41.4 48.1 44.9

Sex ;?D/45 39.2 32.3 35.7 F (%) 48.6 43.9 46.2 Hemorrhages and/or

Race microaneurysms" Black (%) 13.8 10.1 11.9 <M/1 19.9 22.8 21.4

Type of diabetes (%)§ M/123 48.6 45.0 46.8 Type I (%) 58.0 53.4 55.7 ;?M/45 31.5 32.3 31.9 Mixed (%) 23.8 26.5 25.1 Venous caliber Type II (%) 18.2 20.1 19.2 abnormalitiesll

Baseline visual acuity Absent 32.0 33.9 33.0 ;?10/20 (%) 31.5 31.7 31.6 0-D/1 48.1 42.3 45.1 10/30-10/50 (%) 26.5 34.4 30.5 ;?D/23 19.9 23.8 21.9 ~10/70 (%) 42.0 33.9 37.8 Intraretinal microvascular

History of photocoagulation abnormalities" Photocoagulation (%) 64.6 66.1 65.4 Absent 62.4 63.5 63.0

New vessels on or within 1 0-D/1 21.0 21.7 21.4 DD of the optic disc" ;?D/23 16.6 14.8 15.7

~D 36.5 36.0 36.2 Hard exudate II DM-M 36.5 37.0 36.8 Absent 34.8 31.2 33.0 ~MS 27.1 27.0 27.0 0-D/45 36.5 35.4 35.9

New vessels elsewhere" >M/1 28.7 33.3 31.1 <M/1 23.8 21.7 22.7 Arteriolar abnormalities" M/123 44.8 42.9 43.8 Absent 32.6 30.7 31.6 ;?M/45 31.5 35.4 33.5 0-D/1 36.5 41.3 38.9

Fibrous proliferations on or ;?D/23 30.9 28.0 29.5 within 1 DD of the optic disc"

~DM 26.0 31.2 28.6 M 43.1 38.1 40.5 ;?MS 30.9 30.7 30.8

SD = standard deviation; DD = disc diameter. * None of the differences (early-conventional) is significantly different from zero at the 5% level. t Patients with both eyes entered occur in both early and conventional columns. t Duration of diabetes is defined as age at baseline minus age at diagnosis of diabetes. § Type I diabetes is defined as age at diagnosis of diabetes no older than 20 years and taking insulin at time of randomization. The mixed group is

defined as age at diagnosis of diabetes 21 to 39 years and taking insulin at time of randomization. Type II diabetes is defined as not taking insulin at time of randomization or age at diagnosis of diabetes no younger than 40 years. Mean ages at randomization for the type I, mixed, and type II groups were 30, 50, and 58 years, respectively.

II Severity categories for lesions graded in fields 3 to 7 are of the form (maximum severity/extent), where maximum severity can be absent, questionable (0), definite (D), moderate (M), or severe (S), and extent is the number of fields at that severity level. For example, M/23 means there are 2 or 3 fields from fields 3 to 7 with moderate severity (and none with higher severity). See Appendix for definitions of severity levels.

1309

OPHTHALMOLOGY • OCTOBER 1988 • VOLUME 95 • NUMBER 10

Table 2. Cumulative Frequency Distribution of Visual Acuity at 2, 3, and 4 Years by Treatment Group

2 Yrs 3 Yrs 4 Yrs

Early Conventional Early Conventional Early Conventional Visual Cumulative Cumulative Cumulative Cumulative Cumulative Cumulative Acuity % % P % % P % % P

10/10 10.98 6.79 0.26 17.76 6.16 0.004 16.55 5.80 0.008 10/15 32.32 20.99 0.03 36.84 17.81 0.0004 36.55 21.74 0.009 10/20 41.46 30.86 0.06 47.37 24.66 0.0001 44.14 28.26 0.008 10/30 48.17 41.98 0.31 53.29 36.30 0.005 49.66 45.65 0.58 10/40 53.05 48.15 0.44 54.61 41.78 0.04 53.10 49.28 0.60 10/50 54.27 51.23 0.66 57.24 48.63 0.17 55.86 52.90 0.70 10/70 57.93 53.09 0.44 60.53 52.74 0.22 60.00 56.52 0.64 10/100 60.98 57.41 0.59 66.45 58.22 0.18 67.59 63.04 0.50 10/160 64.02 66.05 0.79' 69.74 61 .64 0.18 68.97 63.77 0.42 10/200 67.68 67.28 0.99 70.39 62.33 0.18 70.34 64.49 0.36 5/200 69.51 71 .60 0.77 72.37 67.81 0.46 72.41 65.94 0.29 2/200 70.12 77.78 0.15 73.68 70.55 0.64 72.41 71.74 0.99

HM 78.66 83.95 0.28 75.66 80.14 0.43 76.55 78.99 0.73 LP 80.49 87.65 0.11 78.29 83.56 0.31 77.24 81.16 0.51 NLP 100.00 100.00 100.00 100.00 100.00 100.00

No. of visits completed

164 162 152 146 145 138

No. deceased

12 20 21 30 27 41

No. of visits missed

5 7 8 13 9 10

Total no. randomized

181 189 181

HM = hand motions; LP = light perception; NLP = no light perception.

failed to clear in 1 month and strongly encouraged after 4 months.

In the early vitrectomy group, scatter photocoagulation was carried out postoperatively if the appearance or pro­gression of iris and/or anterior chamber angle neovas­cularization suggested that neovascular glaucoma was imminent, or already present. Postoperative scatter pho­tocoagulation for active retinal new vessels, with or with­out vitreous hemorrhage, was permissible, but prior con­sultation with the Executive Committee was required on a case-by-case basis.

In the conventional management group, vitrectomy was carried out if retinal detachment involving the center of the macula developed and visual acuity had decreased by three or more lines since the randomization visit, or was 10/100 or worse. If ophthalmoscopic examination of the posterior pole was precluded by opacities of the media, ultrasound examination was carried out and vitrectomy was allowed if there was clear-cut evidence of extensive

1310

189 181 189

retinal detachment involving the center of the macula. Vitrectomy was also permitted for severe vitreous hem­orrhage that reduced visual acuity to 5/200 or worse and persisted for 6 months or more. Twenty-one months after beginning recruitment (when 154 eyes had been enrolled) an additional indication for vitrectomy in the conven­tional management group was added, allowing early vi­trectomy in eyes with a decrease in visual acuity from 10/ 50 or better at study entry to 10/100 or worse during follow-up when this decrease was due to traction on optic nerve, peripapillary retina or macula, or to fibrous pro­liferations in front of the macula, that could possibly be reversed by vitreous surgery. Photocoagulation was per­mitted during follow-up at the discretion of the DR VS ophthalmologist.

STATISTICAL METHODS

The principal outcome variables were visual acuity of 10/20 or better, less than 5/200, and no light perception.

DRVS • DRVS REPORT 3

The numbers of eyes in each treatment group reaching these endpoints at each yearly follow-up visit were eval­uated for significance using the chi-square test for 2 X 2 tables, corrected for continuity. The logistic regression models was used to generate P values for interactions of treatment group with baseline characteristics taken singly, and then with several in the model. Interactions with baseline factors of three or more ordered categories were evaluated with a single regression constant, testing for lin­ear trend. In addition, P values were generated with the logistic regression model to test for linear trend of each baseline factor with each visual acuity endpoint at the 2-, 3-, and 4-year visits for the early and conventional groups separately. For all analyses, only the significance proba­bility of the test is provided. Adjustment has not been made for mUltiple testing. Thus, the nominal significance levels are smaller than the real significance levels.

Figures are provided to depict results for up to 4 years of follow-up. More detailed tables are given only for 4 years. As these more detailed tables were reviewed, anal­ogous tables for 2 and 3 years were studied to protect against reporting unrepresentative results.

RESULTS

Table 1 presents baseline characteristics by treatment group. In no case was there a large or statistically signif­icant difference between groups. Statistics for character­istics of all eyes combined are presented in the last column.

Table 2 gives the visual acuity distributions of each group at the 2-, 3-, and 4-year follow-up visits. The early vitrectomy group had a more favorable outcome than did the group managed conventionally. The percentage of eyes with a visual acuity of 10/20 or better was higher in the early vitrectomy group at the 2-, 3-, and 4-year visits. The proportion with very poor visual outcome, defined by any criterion, was similar in the two groups. The treatment effect did not differ among clinics.

Figure 1 depicts the percentages of eyes with visual acu­ity of 10/20 or better ("good vision") at each visit. The early vitrectomy group had a higher percentage of eyes with good vision than the group managed conventionally at every follow-up visit after the 3-month visit. These dif­ferences were statistically significant (P < 0.05) at each follow-up visit except those at 6 and 24 months. Figures 2 and 3 are similar to Figure 1, but display percentages of eyes with visual acuity ofless than 5/200 ("poor vision") and no light perception, respectively. At no time was there a large or statistically significant difference between the two treatment groups in the frequency of either level of vision, although the percent with no light perception was higher in the early vitrectomy group at every visit.

Tables 3 and 4 present the percentages of eyes with good vision, poor vision, and no light perception at the 4-year visit for each treatment group, for subgroups of eyes defined by baseline clinical and fundus photographic factors. Columns are also provided for differences between

60r------------------------------------,

o 50

'" o AI 40

~ ~

'i 'E " 20 e &

10

Early

... • .... ___ _----------............. Conventional _ --- -- ... _---------

O~ __ _L __ ~ ____ L_ __ ~ __ ~ ____ L_ __ ~ __ ~

o 6 12 18 24 30 36 42 48

Time from Randomization in Months

Fig 1. Percent of eyes with visual acuity (VA) of 10/20 or better at each follow-up visit in the early vitrectomy (solid line) and conventional man­agement (dashed line) groups.

60~------------------------------------~

g 50

~ II)

v 40 -< > :€ ~

~ ~

D..

" ,l

Conventional

----------~~~~-~-:-:-~---------~ ---=~---­,;-- Early

" " "

0~ __ -7 __ ~~---L--~~--~---L----L---~ o 48

Time from Randomization in Months

Fig 2. Percent of eyes with visual acuity (VA) ofless than 5/200 at each follow-up visit in the early vitrectomy (solid line) and conventional man­agement (dashed line) groups.

60

50 D.. ...J Z

" 40 -< > j 30

'E " 20 0 G;

D..

10

0 0 6 12 18 24 30 36 42 48

Time Randomization in Months

Fig 3. Percent of eyes with visual acuity (V A) of no light perception (NLP) at each follow-up visit in the early vitrectomy (solid line) and conventional management (dashed line) groups.

treatment groups (E-C for good vision and C-E for poor vision and no light perception, so that a positive difference favors early vitrectomy in all parts of the tables). The tables can be used to answer two related questions. (1) Does

1311

OPHTHALMOLOGY • OCTOBER 1988 • VOLUME 95 • NUMBER 10

Table 3. Percent of Eyes with Specified Visual Acuity at the 4-year Visit for Early Vitrectomy and Conventional Management Groups, and Differences Between Groups, by Baseline Clinical Factors

No. of Eyes ~10/20 <5/200 NLP

% Difference % Difference % Difference E C E C E-C E C C-E E C C-E

Visual acuity ~10/20 50 49 66.0 49.0 17.0 16.0 24.5 8 .5 14.0 18.4 4.4 10/30-10/50 37 47 35.1 17.0 18.1 21.6 36.2 14.5 18.9 19.1 0 .2 <10/50 58 42 31 .0 16.7 14.4 41.4 42.9 1.5 32.8 19.0 -13.7 P 0.0004 0.0007 0.6231 0.0037 0.0652 0.4130 0.0217 0.9318 0.1360

Diabetes type Type I 90 86 58.9 .34.9 24.0 20.0 29.1 9.1 18.9 18.6 -0.3 Mixed 34 34 23.5 20.6 2.9 38.2 41.2 2.9 26.5 23.5 -2.9 Type II 21 18 14.3 11.1 3.2 42.9 44.4 1.6 33.3 11.1 -22.2 p 0.0000 0.0244 0.2440 0.0130 0.1247 0.5099 0.1276 0.7176 0.2133

History of past photocoagulation

Yes 92 91 52.2 35.2 17.0 20.7 23.1 2.4 15.2 17.6 2.4 No 53 47 30.2 14.9 15.3 39.6 55.3 15.7 35.8 21 .3 -14.6 P 0.0112 0.0150 0.7276 0.0152 0.0002 0.3628 0.0054 0.5994 0.1389

NLP = no light perception; E = early vitrectomy; C = conventional management.

Table 4. Percent of Eyes with Specified Visual Acuity at the 4-year Visit for Early Vitrectomy and Conventional Management Groups, and Differences Between Groups, by Baseline Fundus Photographic Factors

No. of ~10/20 <5/200 NLP Eyes

% Difference % Difference % Difference E C E C E-C E C C-E E C C-E

New vessels combined Least severe 35 34 42.9 41.2 1.7 22.9 17.6 -5.2 20.0 5.9 -14.1 Moderately severe 46 36 50.0 36.1 13.9 30.4 25.0 -5.4 23.9 13.9 -10.0 Severe 39 49 43.6 20.4 23.2 20.5 42.9 22.3 17.9 22.4 4.5 Very severe 25 19 36.0 10.5 25.5 40.0 57.9 17.9 32.0 42.1 10.1 P 0.5340 0.0058 0.0784 0.3585 0.0010 0.0708 0.4764 0.0019 0.0419

New vessels on or within 1 DO of the optic disc·

<0 51 45 47.1 40.0 7.1 27.5 17.8 -9.7 23.5 8.9 -14.6 DM-M 53 56 45.3 26.8 18.5 26.4 33.9 7.5 20.8 17.9 -2.9 >MS 41 37 39.0 16.2 22.8 29.3 54.1 24.8 24.4 32.4 8.0 P 0.4503 0.0185 0.1754 0.8590 0.0008 0.0197 0.9451 0.0090 0.0480

New vessels elsewhere· <M/1 31 29 41.9 51.7 -9.8 25.8 20.7 -5.1 22.6 13.8 -8.8 M/123 66 59 48.5 27.1 21.4 24.2 30.5 6.3 19.7 11.9 -7.8 >M45 48 50 39.6 16.0 23.6 33.3 46.0 12.7 27.1 30.0 2.9 p 0.7301 0.0014 0.0266 0.3996 0.0188 0.2861 0.5629 0.0424 0.2413

Fibrous proliferations on or within 1 DO of the optiC disc·

<OM 33 41 48.5 34.1 14.3 30.3 26.8 -3.5 27.3 17.1 -10.2 M 65 51 49.2 31.4 17.9 23.1 31.4 8.3 16.9 15.7 -1.2 >MS 47 46 34.0 19.6 14.5 31.9 43.5 11.6 27.7 23.9 -3.7 p 0.1624 0.1288 0.8884 0.7741 0.1006 0.3697 0.8268 0.4027 0.6539

(Table 4 continues)

1312

DRVS • DRVS REPORT 3

Table 4 (continued)

No. of ~ 10/20 <5/200 NLP Eyes

% Difference % Difference % Difference E C E C E-C E C C-E E C C-E

Fibrous proliferations elsewhere*

<M/1 36 42 41.7 35.7 6.0 30.6 33.3 2.8 27.8 19.0 -8.7 M/2345 52 36 50.0 27.8 22.2 23.1 25.0 1.9 19.2 16.7 -2.6 >S/1 57 60 40.4 23.3 17.0 29.8 40.0 10.2 22.8 20.0 -2.8 p 0.7876 0.1777 0.4299 0.9611 0.4141 0.6088 0.6548 0.8761 0.6718

Retinal elevation* Absent 46 35 54.3 34.3 20.1 23.9 31.4 7.5 19.6 17.1 -2.4 O-D/23 44 52 47.7 25.0 22.7 22.7 38.5 15.7 18.2 25.0 6.8 >D/45 55 51 32.7 27.5 5.3 34.5 31.4 -3.2 29.1 13.7 -15.4 P 0.0288 0.5419 0.3409 0.2199 0.9177 0.3493 0.2408 0.5749 0.2322

Preretinal hemorrhage* Absent 58 43 44.8 39.5 5.3 25.9 23.3 -2.6 19.0 16.3 -2.7 0-D/23 44 62 40.9 24.2 16.7 25.0 35.5 10.5 22.7 19.4 -3.4 >D/45 43 33 46.5 21.2 25.3 32.6 45.5 12.9 27.9 21.2 -6.7 p 0.9003 0.0690 0.1275 0.4820 0.0432 0.3005 0.2928 0.5791 0.8164

Vitreous hemorrhage * Absent 27 26 33.3 23.1 10.3 22.2 34.6 12.4 14.8 15.4 0.6 0-D/23 61 69 47.5 31.9 15.7 26.2 34.8 8.6 21.3 20.3 -1.0 >D/45 57 43 45.6 25.6 20.0 31.6 32.6 1.0 28.1 18.6 -9.5 p 0.3916 0.9576 0.6077 0.3461 0.8386 0.4170 0.1648 0.8032 0.4554

Hemorrhages and/or microaneurysms*

<M/1 28 30 57.1 30.0 27.1 21.4 20.0 -1.4 10.7 13.3 2.6 M/123 72 62 45.8 35.5 10.3 26.4 29.0 2.6 23.6 17.7 -5.9 >M/45 45 46 33.3 17.4 15.9 33.3 50.0 16.7 28.9 23.9 -5.0 p 0.0443 0.1569 0.7240 0.2542 0.0056 0.2600 0.0858 0.2393 0.7282

Venous caliber abnormalities *

Absent 49 40 55.1 40.0 15.1 22.4 17.5 -4.9 18.4 12.5 -5.9 0-D/1 66 61 42.4 29.5 12.9 24.2 36.1 11.8 19.7 16.4 -3.3 >D/23 30 37 30.0 13.5 16.5 43.3 48.6 5.3 36.7 29.7 -6.9 P 0.0286 0.0117 0.6653 0.0667 0.0047 0.4944 0.0877 0.0593 0.8035

Intraretinal microvascular abnormalities*

Absent 87 81 43.7 27.2 16.5 27.6 30.9 3.3 23.0 16.0 -6.9 0-D/1 33 32 45.5 28.1 17.3 36.4 37.5 1.1 30.3 21.9 -8.4 ~D/23 25 25 44.0 32.0 12.0 16.0 40.0 24.0 12.0 24.0 12.0 P 0.9365 0.6570 0.7828 0.4842 0.3476 0.2503 0.4573 0.3221 0.2204

Hard exudate* Absent 48 42 50.0 21.4 28.6 35.4 35.7 0.3 27.1 16.7 -10.4 0-D/45 54 45 48.1 28.9 19.3 20.4 35.6 15.2 20.4 26.7 6.3 >M/1 43 51 32.6 33.3 -0.8 27.9 31.4 3.5 20.9 13.7 -7.2 P 0.1014 0.2094 0.0422 0.3962 0.6505 0.7536 0.4746 0.6522 0.8688

Arteriolar abnormalities* Absent 47 38 53.2 36.8 16.3 27.7 18.4 -9.2 25.5 10.5 -15.0 0-D/1 54 59 42.6 20.3 22.3 24.1 40.7 16.6 16.7 23.7 7.1 ~D/23 44 41 36.4 31.7 4.7 31.8 39.0 7.2 27.3 19.5 -7.8 p 0.1068 0.6443 0.4844 0.6682 0.0604 0.2903 0.8644 0.3248 0.5242

NLP = no light perception; E = early vitrectomy; C = conventional management. * Severity categories for lesions graded in fields 3 to 7 are of the form (maximum severity/extent), where maximum severity can be absent, questionable

(a), definite (D), moderate (M), or severe (S), and extent is the number of fields at that severity level. For example, M/23 means there are two or three fields from fields 3 to 7 with moderate severity (and none with higher severity). See the appendix for definitions of severity levels.

1313

OPHTHALMOLOGY • OCTOBER 1988 • VOLUME 95 • NUMBER 10

visual outcome differ by any of these baseline factors? (2) Does the treatment effect vary in different subgroups? The P-values in the E and C columns relate to the first question, those in the difference columns (E-C or C-E) to the second.

As shown in Table 3, the chance of having good vision at the 4-year visit decreased as baseline vision decreased. There were corresponding increases in the chance for poor vision in both treatment groups and for no light perception in the early vitrectomy group. There was little evidence that the advantage of early vitrectomy varied by baseline visual acuity. Similarly, visual outcome was better for pa­tients with type I diabetes and for those with previous photocoagulation, but again there was little evidence that the advantage of early vitrectomy varied by either of these factors. Results were similar at 2 and 3 years.

Table 4 presents similar analyses for baseline fundus photographic factors, classified as described in the Ap­pendix. The first section in this table presents results for the four combined new vessel severity (NYC) subgroups and the next two sections deal separateiy with NYD and NYE. With increasing severity of new vessels by any of these measures, outcome worsened in the conventional management group for each endpoint. This relationship seemed to be almost completely neutralized by early vi­trectomy. As a result, the advantage of early vitrectomy tended to increase with increasing severity of new vessels. For NYC, the evidence for interaction between treatment and new vessel severity was borderline for each endpoint, whereas for NYD it was somewhat more impressive for visual acuity of less than 5/200 and for NYE for visual acuity of 10/20 or better. Results were similar at 2 and 3 years.

In the remainder of Table 4 only suggestive relation­ships were seen: for a lesser chance of good vision in the early vitrectomy group with increasing retinal elevation, for poorer visual outcome in the conventional manage­ment group with increasing preretinal hemorrhage, and for poorer visual outcome with increasing hemorrhages and/or microaneurysms or venous caliber abnormalities, with the most impressive results for the latter lesion, for which outcome was worse for all three endpoints for both treatment groups. There was little to suggest that the ad­vantage of early vitrectomy varied by severity of any of these lesions. The tendency for the advantage of early vitrectomy to decrease with increasing hard exudate (for the good vision endpoint only) was not supported by the 2- and 3-year results. Fundus lesions that were graded, but are not included in Table 3 because they were infre­quent, were soft exudates, macular edema, and other le­sions graded only in field 2.

Figure 4 depicts the percentages of eyes with good vision at each visit for each treatment group within the NYC subgroups. For eyes in the least severe group there was little difference between early vitrectomy and conventional management, but with increasing severity of new vessels early vitrectomy appeared to become increasingly advan­tageous. Figure 5 presents a similar plot for visual acuity

1314

~,-----------------------------,

o 50 OJ o ;:; 40 0(

>

i i 20

l 10

Least Severe New Vessels (NVC-l)

~.-----------------------------~

~ 50

~ 40 0(

> .<=

'i i 20

l 10

Moderately Severe New Vessels (NVC-2)

Early

'-------____ --------...... Conventional .".._-----........ _-----

~.-----------------------------~

o 50 OJ o A. 40

'" >

¥ i

10

Severe New Vessels (NVC-3)

Early

~.-----------------------------~

o 50 ~ ~ Al 40

.c 30,_ '"'" 'j

\ \

Very Severe New Vessels (NVC-4)

\ .......... - _____ ........ Conventional

....... _-------------------OL-__ L-__ L-~~~--~---L---L--~

~ ~ M ~ ~ g ~

Time 'rom Randomization in Months

Fig 4. Percent of eyes with visual acuity (V A) of 10/20 or better at each follow-up visit in the early vitrectomy (solid line) and conventional man­agement (dashed line) groups, by combined new vessel severity (NYC) subgroup (see Methods section and Appendix for definitions of subgroups). As new vessel severity increases, there is a trend for early vitrectomy to become more advantageous. P values for tests for this interaction at the 12-, 18-,24- 36-, and 48-month visits were 0.11,0.11, 0.05,0.07, and 0.20, respectively (adjusted for differences in proportions of eyes with visual acuity (VA) of 10/20 or better at baseline).

of less than 5/200. Early vitrectomy appeared disadvan­tageous in the least severe category, about equal to con­ventional management in the moderately severe category, and advantageous in the severe and very severe categories.

DRVS • DRVS REPORT 3

60r-----------------------------____ -.

~ 50 ., • 40 .. > ~

i i l

0

2 ., . .. > € • i ~ "-

8 !:! '" · ~ € • i u

l

g !:! '" • .. > £ 'j

c G U

~ "-

60

50

40

0

60

50

60

10

least Severe New Vessels ( NVC - 1)

....... -------__ Early

Conventional _-----

,", ---------------------------­, ,--

Moderately Severe New Vessels (NVC-2)

............... ",'

Severe New Vessels (NVC-3)

Early

------Convent6ona1

Very Severe New Vessels (NVC-4)

,~---___ ---------------c~;~~~~:~ , Early

12 18 24 30 36 42 48

Time from Randomization In Months

Fig S. Percent of eyes with visual acuity (Y A) of less than 5/200 at each follow-up visit in the early vitrectomy (solid line) and conventional man­agement (dashed line) groups, by combined new vessel severity (NYC) subgroup (see Methods section and Appendix for definitions of subgroups). As new vessel severity increases, there is a trend for early vitrectomy to become more advantageous. P-values for tests for this interaction atthe 12-, 18-,24-,36-, and 48-month visits were 0.01, 0.07, 0.09,0.02, and 0.07, respectively.

Figure 6 presents a similar plot for no light perception, with early vitrectomy appearing disadvantageous in the two less severe groups and about equal to conventional management in the two more severe groups.

Multiple logistic regression models were constructed

60

50

~ " 40 .. > E 30

• i 20

~ "- 10

0

60

50 "-i " 40 .. > ~ 30

• " 20 G

~ "- 10

0

60

50

~ II 40 .. > ~ 30 j

60

50 "--' Z , 40 .. > ~ 30 j

i 20

~ "- 10

o

least Severe New Vessels (NVC-1)

Early

Conventional

--------------------------------

Moderately Severe New Vessels (NVC-2)

... --------_ ... ,,,,'"

early

------------------.. ---- Conventionlll

Severe New Vessels (NVC-3)

Convent~

--------------Early

Very Severe New Vessels (NVC-4)

Conventional

,'" ",'

" ,'" early

,----------------,... ,... ",' ... '"

12 18 24 30 36 42 48

Time from Randomization in Months

Fig 6. Percent of eyes with visual acuity (Y A) of no light perception (NLP) at each follow-up visit in the early vitrectomy (solid line) and conventional management (dashed line) groups, by combined new vessel severity (NYC) subgroup (see Methods section and Appendix for defi­nitions of subgroups). As new vessel severity increases, there is a trend for early vitrectomy to become less disadvantageous. P-values for tests for this interaction at the 12-, 18-, 24-, 36-, and 48-month visits were 0.11 , 0.03, 0.09, 0.12, and 0.04, respectively.

with each of the three visual acuity endpoints (~tO/20, <5/200, and no light perception) as the dependent variable at the 18-, 24-, 36-, and 48-month visits and the various clinical and fundus photographic factors as the indepen­dent variables. In one set of models, both NVE and NVD

1315

OPHTHALMOLOGY • OCTOBER 1988 • VOLUME 95 • NUMBER 10

Table 5. Incidence of Untoward Events at any Time during 4 Years after Randomization by Treatment Group

Early Conventional (n = 181) (n = 189) P

Enucleation, phthisis, retrobulbar alcohol injection 9.4 7.4 0.62

Sympathetic unveitis, endophalmitis 0.6 0.0 0.98

Corneal edema, epithelial abnormality 7.7 4.8 0.33

Neovascular glaucoma 7.7 5.3 0.46 Any of above 18.8 11.6 0.08 Retinal detachment 21.0 37.0 0.001

100 ,.. l5 u

80 Very Severe (NVC-4) , ___ 1----------------

~ > ,---E

60 ~

E ., e

40 ., IL ., > j 20 ::J E ::J 0

,----,--­,--r , ,-

i--, ,-­,­r;:

18 24

Severe (NVC-3)

Moderatel Severe (NVC-2) ..

Least Severe (NVC-1)

30 36 42 48

Time from Randomization in Months

Fig 7. Cumulative percent of eyes receiving vitrectomy among survivors in the conventional management group, by combined new vessel severity (NVC) subgroup (see Methods section and Appendix for definitions of subgroups).

were included in the group of possible independent vari­ables and in another set NYC was used in place of them. The statistical significance of the interaction between new vessel severity and treatment with respect to visual out­come remained borderline. To explore this interaction further, univariate logistic regression analyses like those summarized in Tables 3 and 4 were carried out for NYC, NVD, and NYE using three additional visual acuity end­points, 10/30 or better, 10/40 or better, and 10/50 or bet­ter. The strength of the interaction was slightly increased, but remained borderline.

Table 5 gives the incidence of untoward events at any time during the 4 years after randomization, by treatment group. For the most serious complications, and those that might be expected to occur together and to be caused by vitrectomy, the incidence of each event was somewhat greater in the early vitrectomy group, and the incidence of anyone or more of these events was about 50% greater in the early group (P = 0.08). Retinal detachment was about 75% more frequent in the conventional manage­ment group (P = 0.001).

1316

Table 6. Course Followed by Eyes in the Conventional Management Group

Course No. of Eyes

Vitrectomy carried out For retinal detachment involving center of macula For nonclearing vitreous hemorrhage

No vitrectomy Visual acuity never decreased by ~4 times initial

visual angle Visual acuity decreased by ~4 times initial visual

angle Cause of decrease was vitreous hemorrhage,

which cleared, with recovery of visual acuity to <2 times initial visual angle at last visit

Visual acuity did not recover to <2 times initial visual angle at last visit

Retinal detachment with poor prognosis, and/ or poor general health, poor outcome in fellow eye after early vitrectomy, or patient died before planned surgery

Vitreous hemorrhage noted at last follow-up visit, outcome unknown

Cause of poor vision not amenable to vitrectomy

Edema, degeneration, or dragging of the macula and/or marked retinal ischemia

Cataract Neovascular glaucoma (occurred 3 months

after severe vitreous hemorrhage while waiting for clearing)

Total randomized

10

4 1

6

3

15

19

24

91 72 19

98 55

43

189

Figure 7 presents life-table estimates of cumulative in­cidence of vitrectomy in the conventional management group in each ofthe NYC subgroups. After 1 year offol­low-up, the rate was 63% in the very severe group, 36% in the severe group, and about 17% in the two less-severe groups. After 4 years, more than 80% of conventionally treated eyes in the very severe group had had vitrectomy, compared with 40 to 50% in the other three severity groups.

Table 6 summarizes the course followed by all eyes in the conventional management group. Yitrectomy was carried out in a total of 91 eyes, in 72 of them because of retinal detachment involving the center of the macula and in 19 because of nonclearing vitreous hemorrhage. Yitrectomy was not done in 98 eyes, 55 of which never had worsening of visual acuity by as much as a quadru­pling of the initial visual angle (at the last recorded follow­up visit only 3 of these 55 eyes had a visual acuity decrease of as much as a doubling of the initial visual angle). Of the remaining 43 eyes in which vitrectomy was not done, it was unnecessary in 19 because there was spontaneous recovery from vitreous hemorrhage. In the remaining 24 eyes, vitrectomy was not done despite a substantial de­crease in visual acuity from which recovery had not oc-

DRVS • DRVS REPORT 3

curred at the patient's last follow-up visit, and the reasons for this are listed in Table 6.

DISCUSSION

This trial was designed to assess the risks and benefits of early vitrectomy in eyes with advanced, active, PDR and useful vision. We wanted to determine whether the visual prognosis might be improved by removal of severe fibrovascular proliferations before their contraction led to distortion or detachment of the macula. But we also were concerned that vitrectomy in eyes with active retinopathy and extensive new vessels might often be attended by se­rious complications leading to poor visual outcome.6,7 Our findings indicate that in eyes with advanced, active, PDR early vitrectomy can improve the prognosis for good vision without increasing the risk of poor vision (Table 2; Figs 1-3).

SEVERITY OF NEW VESSELS

The interaction between new vessel severity and treat­ment, although of borderline statistical significance, is consistent with the rationale originally proposed for early vitrectomy and important in applying these results to clinical practice. Eyes in the least severe NVC subgroup, of which one half were barely eligible by photographic criteria and one half were eligible only by clinical exam­ination, did not appear to benefit from early vitrectomy. Within this subgroup, eyes managed conventionally had a similar chance of good vision and a smaller risk of poor vision or no light perception than eyes undergoing early vitrectomy (Figs 4-6; NVC-l panels). Furthermore, vi­trectomy was necessary in only 40% of conventionally managed eyes in this subgroup during the 4-year follow­up period (Fig 7).

For eyes in the moderately severe NYC subgroup, early vitrectomy appeared to increase the chance of good vision ( 10/20 or better) without increasing the risk of poor vision, but was associated with a small increase in the proportion with no light perception (Figs 4-6; NVC-2 panels). As in the least severe subgroup, only 40% of conventionally managed eyes underwent vitrectomy.

In the severe and very severe subgroups, early vitrec­tomy provided a substantial increase in the chance of good vision and a corresponding decrease in the risk of poor vision at 3 and 4 years, with little change in the risk of no light perception (Figs 4-6; NVC-3 and NVC-4 panels). By the end of the first year of follow-up, the rate of vi­trectomy in eyes managed conventionally was 35% in the severe subgroup and 65% in the very severe subgroup, and at 4 years these rates were 50 and 85%, respectively (Fig 7).

TYPE OF DIABETES

We have previously reported more favorable results af­ter 2 years of follow-up for eyes with recent severe diabetic

Fig AI. Standard Photograph JOC of the Modified Airlie House Clas­sification defining the lower limit of severe new vessels on or within I DD of the optic disc,3

Fig A2. Standard Photograph 7 of the Modified Airlie House Oassification defining the lower limit of severe new vessels elsewhere (> 1 DD from the optic disc). 3

vitreous hemorrhage that had early vitrectomy, as com­pared with deferral of vitrectomy for 1 year.2 In patients with type I diabetes there was a clear-cut advantage for early vitrectomy. No such advantage was seen in the type II diabetes or mixed groups, but evidence that this ad­vantage differed by diabetes type was of borderline statis­tical significance. We speculated that this difference might be explained in part by the more severe new vessels and fibrous proliferations found in the type I diabetes patients in that study, as assessed during early vitrectomy. We had no way to estimate severity of new vessels and fibrous proliferations in the conventional management group in

1317

OPHTHALMOLOGY • OCTOBER 1988 • VOLUME 95 • NUMBER 10

Fig A3. Standard Photograph lOA of the Modified Airlie House Clas­sification defining the lower limit of moderate new vessels on or within 1 DD of the optic disc.3

Fig A4. Standard Photograph 11 of the Modified Airlie House Classifi­cation defining the lower limit of severe fibrous proliferations elsewhere (> 1 DD from the optic disc).3

Table A1. Minimum Severity Required for Grades D, M, and S for Specified Lesions

Lesion D

New vessels on or within 1 DD of the optic Definitely present disc

New vessels elsewhere Definitely present Fibrous proliferations on or within 1 DD of the Definitely present

optic disc Fibrous proliferations elsewhere Definitely present Retinal elevation 1 disc area Preretinal hemorrhage Definitely present

Vitreous hemorrhage 1 disc area Hemorrhages and/or microaneurysms Definitely present Venous caliber abnormalities Definitely present Intraretinal microvascular abnormalities Definitely present Hard exudate Definitely present Arteriolar abnormalities Definitely present

D = definite; M = moderate; S = severe; DD = disc diameter.

the Group H Study, because the fundi of these eyes at baseline were obscured by vitreous hemorrhage. In the current study, severe PDR was an eligibility requirement, and thus it is not surprising that we found no substantial difference in retinopathy severity by diabetes type. The interaction found in this study between severity of new vessels and treatment group is consistent with our inter­pretation of the findings in Group H.

PHOTOCOAGULATION

In both treatment groups, outcome was better in eyes with previous photocoagulation, as has been reported by others,8 but there was no indication that the advantage of early vitrectomy was influenced by this factor. Two

1318

Grade

M S

Standard Photograph 10A Standard Photograph 1 DC (Fig A3) (Fig A1)

1/2 disc area Standard Photograph 7 (Fig A2) Standard Photograph 1 DB 2 disc areas

1/2 disc area Standard Photograph 11 (Fig A4) Standard Photograph 12 - (4-step scale only) Standard Photograph 9 or 1/2 of field

13 1/2 of field obscured All of field obscured Standard Photograph 1 Standard Photograph 2A Standard Photograph 6A Standard Photograph 6B Standard Photograph 8A Standard Photograph 8B Standard Photograph 3 Standard Photograph 4 Standard Photograph 7 - (4-step scale only)

thirds of the eyes in our study had had photocoagulation before enrollment and by 3 years offollow-up 85% of eyes in each treatment group had had such treatment. Our study was carried out at a time when argon laser endo­photocoagulation was not widely available. For this rea­son, and to avoid the difficulty in interpreting study results that might have occurred, if eyes assigned to early vitrec­tomy received more photocoagulation than those assigned to conventional management, scatter photocoagulation during vitrectomy was not a part of the treatment protocol.

CONCLUSION

Our results indicate that early vitrectomy should be considered in eyes with useful vision and advanced, active,

DRVS • DRVS REPORT 3

Table A2. Grading Scale for Lesions Graded in Multiple Photographic Fields

Grade Maximum Severity/No. of Fields *

A Q/1 Q/23 Q/45 0/1 0/23 0/45 M/1 M/23 M/45 S/1 S/23 S/45

A/1-5t or AQ/1 Q/1 or AQ/2-5 Q/2 or 3 Q/4 or 5 or QO/1 0/1 or QO/2-5 0/2 or 3 0/4 or 5 or OM/1 M/1 or OM/2-5 M/2 or 3 M/4 or 5 or MS/1 S/1 or MS/2-5 S/2 or 3 S/4 or 5

* Severity categories for lesions graded in fields 3 to 7 are of the form (maximum severity/extent), where maximum severity can be absent (A), questionable (Q), definite (0), moderate (M), or severe (S), and extent is the number of fields at that severity level. For example, M/23 means there are 2 or 3 fields from fields 3 to 7 with moderate severity (and none with higher severity). See Appendix for definitions of severity levels.

t If the maximum grade is A but this appears in less than five fields, the remaining fields are missing or ungradable.

PDR with extensive new vessels. We do not advocate vi­trectomy as an alternative to photocoagulation, but as an adjunct when photocoagulation has not been followed by substantial regression of new vessels, or is precluded by vitreous hemorrhage. The more severe the new vessels and the poorer the outlook for improvement with addi­tional photocoagulation, the more seriously early vitrec­tomy should be considered. yitrectomy is, of course, a major procedure never to be undertaken without careful consideration of all aspects of the patient's situation. The companion article to this one presents several DRYS cases illustrating the types of eyes in which we believe it is rea­sonable to consider early vitrectomy.

A list of DRVS investigators appears at the end of DRVS Report I (Ophthalmology 1985; 92:492-502) and DRVS Report 2 (Arch Ophthalmol1985; 103:1644-52).

APPENDIX

ELIGIBILITY GRADING

The minimum required to meet the definition of severe new vessels was new vessels on or within I disc diameter of the optic disc (NVD) equaling or exceeding those in Figure AI, or new vessels elsewhere (NVE) equaling or exceeding those in Figure A2 in at least one photographic field, or total new vessels (NVD plus NVE, summed over all photographic fields with allowance for overlap between them) estimated to equal or exceed 4 disc areas.

The minimum required to meet the definition of moderate new vessels was NVE equaling or exceeding 0.5 disc area in at least two of fields 3 to 8, or NVE of this extent in one field plus NVD equaling or exceeding those in Figure A3, or total new vessels estimated to equal or exceed 2 disc areas.

NVE

so

S 0/23

0/45 Least Severe

MIl

M/23

NVD OMorM

Moderately

Severe

"Ml45 I L ____ s_e_ve_re ____ ..1

MS or S

Severe

Fig AS. Definitions of the four combined new vessel severity subgroups based on fundus photograph gradings.

The minimum required to meet the definition of severe fibrous proliferations was fibrous proliferations on or within I disc di­ameter of the optic disc (FPD) equaling or exceeding 2 disc areas, or fibrous proliferations elsewhere (FPE) equaling or ex­ceeding those in Figure A4 in at least one field, or total fibrous proliferations estimated to equal or exceed 4 disc areas.

The requirement for red vitreous hemorrhage was met by the definite presence of any amount of red preretinal or vitreous hemorrhage in any photographic field.

DETAILED BASELINE GRADING

Baseline stereoscopic fundus photographs were graded inde­pendently by each of two graders using the Modified Airlie House Classification.3 For most of the lesions listed in Tables I and 4, the scale used by each grader to grade each photographic field consisted of five steps: (A) absent; (Q) questionable (grader 50-90% sure the lesion is present); (D) definitely present; (M) mod­erate (~ a specified standard photograph or written definition); and (S) severe (~ a specified standard photograph or written definition). The definition of each step assumes that the definition of a higher step is not met. For two lesions, the definitions for grade Q differed from those used for the other lesions: for retinal elevation the minimum requirement was any definite elevation of the retina, including avulsion of a retinal vessel; for vitreous hemorrhage the minimum requirement was definitely visible hemorrhage. Table Al lists the minimum severity required for steps D, M, and S on the scale for each lesion included in Tables I and 4.

The gradings of each grader were entered into a computer file and compared for each lesion in each photographic field in which it was graded. Disagreements between graders of two steps or more were resolved by a senior grader, who reviewed the pho­tographs without knowledge of the original grades. Disagree­ments of one step were assigned to an intermediate step, ex­panding the five-step scale shown above to nine steps: A, AQ,

1319

OPHTHALMOLOGY • OCTOBER 1988 • VOLUME 95 • NUMBER 10

Q, QD, D, DM, M, MS, S. This scale is used in Tables 1 and 4 for NVD and FPD, since they are graded in only one field.

For lesions graded in fields 3 to 7, the maximum grade in any field and the number of fields in which this grade was chosen were used to produce the scale shown in Table A2. This scale was used for the remaining lesions in Tables I and 4, all of which were graded in fields 3 to 7, and in optional field 8, if it was taken to document eligibility. Some of these lesions were also graded in fields 1 and/or 2. If the maximum grade for a lesion occurred only in one or more fields other than fields 3 to 7, this maximum in one field was recorded as the grade for the eye; otherwise fields other than 3 to 7 were ignored. Figure AS depicts the combined new vessel scale.

REFERENCES

1. DRVS Research Group. Two·year course of visual acuity in severe proliferative diabetic retinopathy with conventional management. Ophthalmology 1985; 92:492-502.

1320

2. The Diabetic Retinopathy Vitrectomy Study Research Group. Early vitrectomy for severe vitreous hemorrhage in diabetic retinopathy. Two-year results of a randomized trial. Diabetic Retinopathy Vitrectomy Study Report 2. Arch Ophthalmol1985; 103:1644-52.

3. The Diabetic Retinopathy Study Research Group. Report 7. A modi· fication of the Airlie House classification of diabetic retinopathy. Invest Ophthalmol Vis Sci 1981; 21 :210-26.

4. The Diabetic Retinopathy Study Research Group. Report 6. Design, methods, and baseline results . Invest Ophthalmol Vis Sci 1981; 21 : 149-209.

5. Breslow NE, Day NE. Statistical Methods in Cancer Research: Volume 1-The Analysis of Case-Control Studies (IARC Scientific Publications No. 32). Lyon: International Agency for Research on Cancer, 1980.

6. Blankenship GW, Machemer R. Prophylactic vitrectomy in proliferative diabetic retinopathy. Mod Probl OphthalmoI1977; 18:236-41.

7. Kupfer C. A new patient group in the Diabetic Retinopathy Vitrectomy Study [Editorial]. Arch Ophthalmol 1981; 99:65.

8. de Bustros S, Thompson JT, Michels RG, Rice TA. Vitrectomy for progressive proliferative diabetic retinopathy. Arch Ophthalmol1987; 105:196-9.