Augmentation Laser for Proliferative Diabetic Retinopathy that Fails to Respond to Initial Panretinal Photocoagulation Bernard H. Doft, MD, Donna]. Metz, RN, BSN, Sheryl F. Kelsey, PhD Purpose: A study was performed to determine if diabetic subjects who fail to respond to initial panretinal photocoagulation with regression of retinopathy risk factors do better with supplemental panretinal photocoagulation. Methods: Thirty-five patients with 3 or more retinopathy risk factors who failed to respond to pan retinal photocoagulation with regression to less than 3 retinopathy risk factors by 3 weeks after initial pan retinal photocoagulation were prospectively randomized to augmentation laser panretinal photocoagulation (MORE) or to no additional treatment (NOMORE). Results: Six months after initial treatment, the MORE group (n = 16) had regressed a mean of -0.94 retinopathy risk factors (with 95% confidence interval [CI] -1.60 to -0.26), compared with -0.21 retinopathy risk factors (95% CI -0.69 to 0.27) in the NOMORE (n = 19) group (P = 0.055). However, by 1 year, there was no statistically significant difference in the amount of regression of retinopathy risk factors with a mean decrease of -1.12 (95% CI -2.0 to -0.24) versus -1.05 retinopathy risk factors (95% CI -1.80 to -0.28) in the 2 groups, respectively. Similarly, for visual acuity, there was no difference in outcome. For all study patients, the persistence of three or more retinopathy risk factors was associated with a poorer visual result than if there was regression to less than three retinopathy risk factors. Conclusion: . This study shows that although augmentation pan retinal photocoagulation achieved faster regression of retinopathy risk factors, by 1 year, there was no difference in either mean regression of retinopathy risk factors or visual acuity between eyes treated or not treated with augmentation panretinal photocoagulation. In addition, the study shows that the persistence of 3 or more retinopathy risk factors 1 year after treatment was associated with a poorer visual result. Because sample size limited the power of the study to find small differences between groups, and because in proliferative diabetic retinopathy small differences could be important clinically, the authors do not recommend changes in current clinical practice. Ophthalmology 1992;99: 1728-1735
Originally received: October 22, 1991. Revision accepted: July 13, 1992. From Retina-Vitreous Consultants, Pittsburgh. Presented at the American Academy of Ophthalmology Annual Meeting, Anaheim, October 1991. Reprint requests to Bernard H. Doft, MD, Retina-Vitreous Consultants, 3501 Forbes Ave, Pittsburgh, PA 15213.
1728
The Diabetic Retinopathy Study identified four retinopathy factors (vitreous or preretinal hemorrhage, new vessels, location of new vessels on the optic disc, and moderate or severe extent of new vessels) that increase the risk of developing severe visual loss. Eyes were found to be at high risk for such visual loss if three or more of these retinopathy risk factors were present. 1 The Diabetic Retinopathy Study also established that argon laser panretinal
Doft et al . Augmentation Laser for Proliferative Diabetic Retinopathy ablation decreases the risk of developing severe visual loss.
It is presumed that the beneficial effect of this treatment
is related to a reduction in the number of retinopathy risk factors that occur after panretinal photocoagulation. Doft and Blankenship2 demonstrated that the amount of regression of retinopathy risk factors found 3 weeks after panretinal photocoagulation is a good indicator of longer-term treatment effect. Two thirds of patients who did not regress from a high (3 or 4 risk factors) to a low risk category (2 or fewer risk factors) by 3 weeks after panretinal ablation never regressed to a low-risk category during the 6-month course of their study. The ability to make early identification of patients who fail to respond to initial laser treatment makes it possible to try to answer an important question about the management of proliferative diabetic retinopathy. Do diabetic patients who fail to respond to initial panretinal photocoagulation with regression of retinopathy risk factors do better with supplemental panretinal photocoagulation? In this article, we report the results of a prospective randomized study designed to answer that question.
Methods Patients with proliferative diabetic retinopathy were screened for the study if in the involved eye they were found to have all of the following: (1) three or more retinopathy risk factors; (2) new vessels on the disc; (3) no traction detachment of the macula; (4) ocular media clear enough to allow panretinal photocoagulation treatment; (5) intraocular pressure of 22 mmHg or less; and (6) visual acuity of 20/200 or better. Within 1 day of evaluation, patients with the above characteristics were treated with argon green laser panretinal photocoagulation in a single session using a Coherent Model 900 laser (Palo Alto, CA) and a Rodenstock panfunduscopic lens (Sola Optical, Petaluma, CA). Between 700 and 1000 bums were placed using 500-JLm spot size (at the laser control), O.l-second duration, and with a power setting sufficient to achieve a moderately white bum. Spots were placed one half bum width apart in a regular pattern as far to the periphery as possible with a Rodenstock lens, starting at points on an oval defined as 2 disc diameters above, below, and temporal to the center of the macula and 500 JLm nasal to the nasal half of the disc. Retrobulbar Xylocaine was used for anesthesia on most patients. Patients were seen in follow-up 3 weeks after initial treatment. Only patients whose eyes had media clear enough to allow supplemental laser treatment, and who still showed 3 or 4 retinopathy risk factors were considered eligible for entry into the study. Eligible patients who gave consent were randomly assigned to either no additional laser treatment (NOMORE group) or to immediate supplemental panretinal photocoagulation (MORE group). A block of 50 opaque envelopes had been prepared in advance, with 25 indicating MORE and 25 indicating NOMORE. These envelopes were prepared, sealed, thoroughly mixed, sequentially
numbered, and then used in numbered sequential order. Eyes assigned to the MORE group received a minimum of 500 additional bums of argon laser with the Rodenstock lens filling in between previous laser bums, with efforts made not to retreat over previous treatment spots. In addition, where there was room for additional treatment in the far periphery, the Goldmann three-mirror lens was used to treat this area (placing spots one half bum width apart and using 500-JLm spot size at the laser control) if the area could not be reached with the Rodenstock lens. Study follow-up data were collected 6 and 12 months after initial treatment. Visual acuity was measured using a Snellen chart by an experienced examiner of visual acuity who was masked to the patients' treatment status. Visual acuity was measured with the then-current glasses prescription and pinhole. On study visits, data capture forms were completed before seeing past ophthalmic records. Although not masked, when seeing patients on study visits 6 months after they had been randomized, the examiner (BHD) was not aware of the group to which the patient had been randomized. At any time, if a particular retinopathy risk factor could not be assessed because oflimited view due to vitreous hemorrhage, the risk factor was assumed to be present in its most severe extent. This occurred in 3 eyes at the I-year follow-up visit. Hypertension was defined as being present if the patient was being treated with systemic antihypertensive medication at the time of the initial examination. The onset of diabetes was dated from the time the patient began taking any medication for the disease. Insulin-dependent diabetes was defined as the patient using insulin on a regular basis at the time of the initial study examination. With the exception of the designation regarding panretinal photocoagulation treatment, the study protocol allowed for patients to receive any ophthalmologic care required, based on their ophthalmic status. Statistical analysis was carried out using the chi-square test for proportions and t tests for mean number of retinopathy risk factors. A one-tailed not two-tailed test was used because one would only want to recommend augmentation laser if it were better than observation. Thus, if augmentation laser and observation were shown to produce similar outcome overall, observation would be the treatment of choice since it is less invasive. It is recognized that augmentation laser could indeed result in worse outcome, but it is not important to distinguish the difference between augmentation laser worse and augmentation laser the same, since in either case one would choose not to perform augmentation laser. Ninety-five percent confidence intervals for differences over time in retinopathy risk factors are, however, reported as two-sided.
Results As of the cutoff date for the analysis, 35 patients had completed the I-year follow-up examination. Entered into the study but not included in this report were 3 patients who were not yet due for final examination as of that date, 2
1729
Ophthalmology Volume 99, Number 11, November 1992 Table 1. Baseline Characteristics Item Number Number right eye Mean age (range) Male sex (number) Mean duration diabetes, yrs (range) Percent insulin requiring Percent hypertensive
More
All Patients
No More
35 20 40 (18-70) 17
19 12 40 (19-70) 9
16 8 42 (18-62) 8
17 (2-29) 94% 29%
15.5 (2-28) 95% 32%
18.8 (10-29) 94% 25%
(I from each study group) who died before the I-year examination, and I (from the NOMORE group) who refused to return for follow-up. Of the 35 patients followed for I year, 4 examinations (initial, 3 weeks, 6 months, and 1 year) were performed on all patients except for 2 (both from the MORE group) who missed their 6-month visit. For these two missed visits, the number of retinopathy risk factors were interpolated between an earlier and later visit. Sixteen eyes were assigned at random to supplemental laser (MORE), and 19 to no additional laser (NOMORE). Demographic information concerning the patients is included in Table 1. Information about selected specific patients referred to within individual sections below is found in Table 2.
Intraocular Pressure There was no difference in the mean intraocular pressure by treatment group either at study entry (NOMORE = 15.9 versus MORE = 17.4), 6 months (15.5 versus 15.8), or 1 year (15.2 versus 15.9). At 1 year, a single study patient in each group had an intraocular pressure of 6 mmHg (patients 10 1 and 108) (Table 2). All other patients in both groups had an intraocular pressure between 11 and 23 mmHg at 1 year.
Rubeosis At the initial, 3-week, and 6-month visits, 1 patient in the NOMORE group had rubeosis. At the I-year visit, 1 patient from each group had rubeosis (patients 101 and 108).
Table 2. Table of Selected Patients Study Visual lOP Number Group· Acuity (mmHg) Rubeosis Mediat
Retina
Additional Surgery
51 101
M M
CF HM
21 6
No Yes
1 2
Macular edema Posterior pole attached with some peripheral detachment
No PPY and SB, two revisions, last with silicone oil
159
M
HM
14
No
3
Tractional retinal detachment involves macula
PPY
M NM
HM NLP
17
No Yes
2
108
3
Retina attached Retina detached
164 167
NM NM
CF CF
20
16
No No
4
No PPY and SB one revision No PPY, PPL, lens implant placement
62
NM
20/100
15
No
32
6
1
Macular edema No retinal detachment (ultrasound)
Tractional retinal detachment extrafoveal
Comment By 2 yrs, total retinal detachment with LP acuity Retina attached after PPY, but retinal vessels sclerotic Prephthisical
Patient's surgery is 10 wks after the 1-yr follow-up and final visual acuity = 20/50
No
lOP = intraocular pressure; CF = counting fingers; HM = hand motions; PPY = pars plana vitrectomy; SB = scleral buckle; LP = light perception; NLP = no light perception; PPL = pars plana lensectomy. •M
=
t
more panretinal photocoagulation; NM
=
no more panretinal photocoagulation.
1 = clear; 2 = blood blocks less than one quarter of the retinal area; 3 = blood blocks one quarter or more of the retinal area; 4 = blood obscures all view of retina.
1730
Dolt et al
Augmentation Laser for Proliferative Diabetic Retinopathy
Retinopathy Risk Factors By definition (study entry criteria), 100% of eyes had 3 or more risk factors at the time of study entry, which was 3 weeks after the initial panretinal photocoagulation. Six months after treatment, 37.5% of MORE and 21% of NOMORE eyes had regressed from a high risk (3 or more risk factors) to a low risk category (less than 3 risk factors). The percent of MORE eyes that had regressed to a lowrisk category increased by only an additional 6.5% to 44% at 1 year, but the percent of NOMORE eyes that had regressed to less than 3 risk factors changed from 21 % at 6 months to 58% at 1 year. For all patients, regardless of treatment group, 51.4% of eyes had regressed to less than 3 risk factors by 1 year. The percentage of eyes with 0 risk factors at 1 year was 25% and 26% in the MORE and NOMORE groups, respectively. There was no advantage in regression of risk factors to the MORE group over the NOMORE group at the final study visit at 1 year. These data are summarized in Table 3. The mean number of risk factors decreased in both groups with time from treatment. The mean number of risk factors was 3.50 versus 3.21 at the initial visit, 2.56 versus 3.00 at 6 months, and 2.38 versus 2.16 at the 1year visit in the MORE versus NOM ORE groups, as shown in Table 3. Eyes in the MORE group achieved most of their regression of mean retinopathy risk factors within 6 months of treatment (-0.94 of a total of -1.12), with
only little additional regression (-0.18 risk factors) occurring during the second 6 months of follow up. In contrast, eyes in the NOM ORE group showed a mean decrease of just 0.21 risk factors in the first 6 months, with substantial additional regression in the second 6-month period (an additional -0.84 risk factors to a total of -1.05 at 1 year). The difference in amount of regression for 6 months between groups was significant at P = 0.055. However, for 1 year, there was no statistical difference in the mean risk factors regression between groups. The NOMORE group was slower in achieving regression, but its end result (-1.05 mean risk factors) at 1 year was essentially the same amount of risk factors regression achieved by the MORE group (-1.12 risk factors). Ninetyfive percent confidence intervals for these differences are shown in Table 3. Reflecting this sample size, the confidence intervals are relatively wide and overlap between the MORE and NOMORE treatment groups. The study data show the standard deviation of the change in number of risk factors from initial to I year was 1.3. Power analysis showed the study with 35 patients had an 80% power to detect a difference of 1.1 risk factors between groups at I year. To have 80% power to detect a difference of 0.5 risk factors, 168 eyes would have been required. The number of retinopathy risk factors was looked at for eyes with and without evidence of vitreous hemorrhage at I year. The mean number of risk factors in study eyes with no evidence of vitreous hemorrhage at 1 year was 1.38 (n = 21), with a mean of 1.33 (n = 12) for the NO-
Table 3. Retinopathy Risk Factors by Treatment and Time of Observation
Mean Number RRF Initial 6 Months 1 Year
NOMORE (n = 19)
MORE (n = 16)
3.21 3.00 2.16
3.50 2.56 2.38
Mean Change in Number RRF (95% Confidence Interval) Initial to 6 months· 6 months to 1 year Initial to 1 year
-0.21 (-0.69, 0.27) -0.84 (-1 .60, -0.04) -1.05 (-1.80, -0.28)
-0.94 (-1.60, -0.26) -0.18 (-0.72, 0.34) -1.12 (-2.00, -0.24)
Percent of Patients with 2 or Fewer RRF (95% Confidence Interval) Initial 6 Months 1 Year
0 21.1 (3, 39) 57.9 (36, 80)
0 37.5 (14, 61) 43.8 (19, 68)
5.3 26.3
14.3 25.0
Percent Patients with No RRF 6 Months 1 Year
RRF = retinopathy risk factor . • p = 0.055 for one-sided t test between MORE and NOMORE treatment groups. Two patients in the MORE group missed the 6-month follow-up visit. Data for these two patients at this time point were interpolated between the prior and subsequent visit.
1731
Ophthalmology Volume 99, Number 11, November 1992 MORE and 1.44 (n = 9) for the MORE group. In contrast, the mean number of risk factors for eyes that showed any amount of vitreous hemorrhage at 1 year was 3.57 (n = 14), with a separate mean of 3.57 for the MORE and NOMORE groups (n = 7 in both). Visual acuity distributions were assessed based on persistent risk factors present 1 year after treatment. For patients with 3 or more persistent risk factors at 1 year regardless of whether they were in the MORE or NOMORE group, the visual acuity distribution was worse than for patients who had less than 3 persistent risk factors. Eightynine percent (16 of 18) of patients with less than 3 risk factors at 1 year had visual acuity of 20/200 or better, compared with 59% (10 of 17) of those with more than 3 risk factors. This difference was statistically significant at P = 0.04, based on chi-square analysis. For visual acuity of 20/100 or better, the percentages were 78% and 53%, respectively, and for visual acuity of 20/50 or better, the percentages were 61 % and 41 %, respectively. These differences were not statistically significant at P = 0.05 (Fig 1). Poor results can be defined in a variety of ways. If one defines an eye with a poor result at 1 year as an eye with any of the following: traction detachment more than 4 disc areas in size, or vitreous hemorrhage blocking one quarter or more of the retinal area, or visual acuity worse than 20/400, then there were 7 eyes (3 in the NOMORE group and 4 in the MORE group) with a poor result. A poor result was present in 6 of 17 eyes (35%) with 3 or more risk factors at 1 year, but in only 1 of 18 eyes (6%) with less than 3 risk factors at that time. This difference is significant at P = 0.016. The persistence of 3 or more risk factors 1 year after treatment is therefore associated with a poor result as herein defined.
Visual Acuity Before initial treatment, all patients had visual acuity of 20/200 or better. At the 6-month follow-up visit, visual acuity of 20/50 or better was present in 57% of eyes in the MORE group and in 63% of eyes in the NOMORE group; visual acuity of 20/200 or better was found in 71 % VISUAL ACUITY DISTRIBUTION BY PERSISTENCE OF RETINOPATHY RISK FACTORS (RRF) AT 1 YR
120.------------------------------, = 3RRF -+-
I-
z
W
=20/25 >=20150 >=201100 >-201200 >"'101200 >=5/200
>""LP
>=-NLP
ACUITY
Figure 2. Cumulative visual acuity distribution at 1 year. The cumulative percent of eyes with various acuity levels is shown far the MORE and NOMORE groups.
and 68% of the 2 groups, respectively. At the I-year followup visit, 50% of the MORE group and 53% of the NOMORE group had visual acuity of 20/50 or better, with 75% of the MORE group and 74% of the NOMORE group having visual acuity of 20/200 or better. At the I-year visit, there were 4 eyes in the MORE group and 3 in the NOMORE group with visual acuity worse than 20/400, as shown in Table 2. Figure 2 shows cumulative visual acuity distribution at 1 year. There is no evidence of a harmful or beneficial effect on visual acuity of either MORE or NOMORE treatment.
Vitreous Hemorrhage Vitreous hemorrhage was graded during each examination as being absent, as obscuring the visualization ofless than one quarter of the retina, or as obscuring one quarter or more of the retina. At the I-year follow-up visit, there were 7 eyes in each group with vitreous hemorrhage. In the MORE group, 6 of the 7 eyes had obscuration ofless than one quarter of the retinal area, with one of these having a retinal detachment (patient 101). A single patient (patient 159) had one quarter or more of the retinal area obscured by vitreous hemorrhage. In the NOMORE group, 5 of the 7 eyes with vitreous hemorrhage had obscuration ofless than one quarter of the retinal area, and none of these had retinal detachment. Two eyes had vitreous hemorrhage obscuring one quarter or more of the retina (Table 2).
60
Detachment of the Retina
40
At the I-year visit, there were four patients (Table 2) with retinal detachment, two in each group.
20
~=20125
>=20/50 >=20/100 >=20/200 >=101200 >=51200
>=LP
>=NLP
ACUITY
Figure 1. Cumulative visual acuity distribution at 1 year by low (less than 3) versus high (3 or more) retinopathy risk factors category.
1732
Additional Surgical Procedures Two eyes in the MORE group and 2 in the NOMORE group required additional surgery during the course of
Dolt et al . Augmentation Laser for Proliferative Diabetic Retinopathy the study, or within 10 weeks of the last study visit. Details are presented in Table 2.
Discussion The Diabetic Retinopathy Study identified 4 retinopathy risk factors that are associated with severe visual loss, and found that the presence of 3 or more of these factors carried a risk of 26% to 37% for severe visual loss at 2 years, whereas the presence ofless than 3 retinopathy risk factors carried a risk of 8.5% or less. J Eyes with 3 or more risk factors were therefore defined as eyes with high-risk characteristics. The Diabetic Retinopathy Study recommended laser photocoagulation for eyes with high-risk characteristics, since results demonstrated that panretinal photocoagulation decreases the risk of severe visual loss by approximately 50% compared with untreated patients. 3 It is presumed, but not known, that the beneficial effect of panretinal photocoagulation is related to reduction of number of retinopathy risk factors that occur after panretinal photocoagulation treatment. Doft and B1ankenship 2 assessed for rapidity of regression of retinopathy risk factors after panretinal photocoagulation. They showed that risk factors regress rapidly after photocoagulation, and that the 3-week response was a good indicator oflonger-term risk factor regression. All of their patients had three or more retinopathy risk factors before treatment. By 3 weeks after treatment, 72% had regressed to less than 3 risk factors, and 28% continued to have high-risk characteristics. Of patients who did not respond to initial treatment by regression to less than 3 risk factors by 3 weeks, 64% continued to have high-risk characteristics (3 or more risk factors) for the full duration of the study during which no additional panretinal photocoagulation treatment was placed. The Diabetic Retinopathy Studl showed that the risk of severe visual loss falls with increasing amount of initial treatment (as measured by treatment density in the posttreatment photograph of standard field 6). However that study did not look specifically at the subgroup of patients who had persistent high-risk characteristics after initial treatment, or what would happen if additional treatment was placed to those eyes. Roge1l 5 described a protocol for incremental laser treatment iffollow-up after initial treatment did not result in substantial regression of neovascularization. However, there were no controls to determine what would happen if additional treatment was not placed. Vine 6 reported on eyes that failed to respond with regression to less than 3 risk factors by 8 weeks after an initial panretinal photocoagulation of 3000 calculated burns. Calculated burns were determined by multiplying the number of burns placed with a pan funduscopic lens by 2.7 to convert to Goldmann burn equivalents, based on the work of Barr.7 Twenty-three nonresponder eyes were treated with an additional 1000 to 2000 Goldmann burns. The final result was assessed 8 weeks after the last laser treatment session. Twelve of the 23 eyes responded (less than 3 risk factors) and 11 did not (maintained 3 or
more risk factors). There was little adverse effect on acuity due to additional laser treatment since acuity was within I line of pretreatment acuity in 10 of the 12 responder eyes. Of the 11 failure eyes, 6 had visual acuity of 20/80 or better, but 5 had visual acuity of counting fingers or worse, and 3 had neovascular glaucoma. While Vine's study showed that about one half of initial nonresponder eyes respond to augmentation panretinal photocoagulation, his study was retrospective and did not include a control group. Thus, it is impossible to know if a similar group of patients who had not received additional laser would have done worse, better, or the same as the patients treated with augmentation. The current study uses a prospective randomized design to determine the effect of augmentation panretinal photocoagulation in patients who fail to respond to less than 3 risk factors 3 weeks after initial panretinal photocoagulation. The results show that MORE treatment results in more rapid regression of retinopathy risk factors. Six months after treatment, the MORE group had regressed a mean of -0.94 risk factors, compared with -0.21 risk factors in the NOMORE group. However, by I year, there was no statistically significant difference in the amount of regression of risk factors by treatment group, with a mean decrease of -1.12 versus -1.05 risk factors in the 2 groups, respectively. While the NOMORE group is slower in achieving regression, by 1 year the same amount of regression of retinopathy risk factors occurs in both groups. As far as visual acuity is concerned, there was no difference between groups, as shown in Figure 2. The study deals with another issue. While the Diabetic Retinopathy Study established that the presence of retinopathy risk factors be/ore treatment was associated with risk of severe visual loss, the Diabetic Retinopathy Study did not deal with the question of whether persistence of risk factors a./iertreatment had already been placed is also a risk factor for visual loss. To examine that question, in this study visual acuity distributions were assessed based on the number of persistent risk factors present I year after treatment. For all patients with 3 or more risk factors at 1 year, regardless of whether they were in the MORE or NOMORE group, the visual acuity distribution was worse than for patients who had less than 3 risk factors, as shown in Figure 1. Therefore, results of the study show that the persistence of retinopathy risk factors after treatment is associated with a poorer visual acuity result. Similar information is shown by assessing for eyes with poor result as defined in this study. A poor result was present in 35% (6 of 17) of eyes with 3 or more risk factors at 1 year, but in only 6% (I of 18) of eyes with less than 3 risk factors. While there was a relationship between persistence of 3 or more risk factors and poor result, these findings do not necessarily allow a conclusion that augmentation panretinal photocoagulation will lead to a lesser chance of a poor result. There are certain strengths and weaknesses to this study. It is prospective and uses random assignment of patients to a treatment group. However, the number of patients in the study is too small to allow the statistical power necessary to detect small differences between treat-
1733
Ophthalmology
Volume 99, Number 11, November 1992
ment groups. For example, to have a standard 80% power to detect a difference of 0.5 risk factors between groups, 168 eyes would have been required. Visual acuity, although measured by a masked examiner, used only pinhole over previous refraction and was not obtained by refracting the patient at the time visual acuity was measured. A Bailey-Lovie acuity chart was not available at study initiation, so Snellen acuities were used. Snellen acuities may be less useful, especially when visual acuities are worse than 20/ tOO. In this study, visual acuity was measured the same way in all patients, and by a masked, experienced observer who used previous refraction and pinhole. If there were errors in acuity measurement, they should have occurred equally in either treatment group. This study did not address certain important issues. There is a possibility that augmentation laser treatment could result in adverse side effects. This could occur if treatment were applied to previously treated areas since treatment of thinned retina may increase the risk of damage to the nerve fiber layer, and possibly increase the amount of field loss. 8 The current study made no attempt to assess visual fields on these patients. In conclusion, in this article, we report the results of a prospective randomized clinical trial designed to determine if patients who fail to regress from high-risk characteristics with argon laser panretinal photocoagulation will benefit from augmentation laser. The study showed that although augmentation panretinal photocoagulation achieved faster regression of risk factors, by 1 year, there was no difference in either mean regression of risk factors or visual acuity between eyes treated with and without augmentation panretinal photocoagulation. In addition, results of the study showed that the persistence of risk factors after treatment is associated with a poorer visual acuity result. The power of this study to find small differences between treatment groups was limited by sample size. In
proliferative diabetic retinopathy, even a small difference in outcome could be very meaningful clinically. Because of the limitations imposed by sample size, the authors do not recommend any change in management of diabetic patients who do not regress their retinopathy risk factors. A common practice is to treat such patients with further laser therapy. There is no reason to abandon such a practice based on this report. However, a larger clinical trial with sample size adequate to detect small differences in outcome between treatment groups should be performed.
References 1. The Diabetic Retinopathy Study Research Group. Four risk factors for severe visual loss in diabetic retinopathy. The third report from the Diabetic Retinopathy Study. Arch Ophthalmol 1979;97:654-5. 2. Doft BR, Blankenship G. Retinopathy risk factor regression after laser pametinal photocoagulation for proliferative diabetic retinopathy. Ophthalmology 1984;91:1453-7. 3. The Diabetic Retinopathy Study Research Group. Photocoagulation treatment of proliferative diabetic retinopathy: the second report of the Diabetic Retinopathy Study findings. Ophthalmology 1978;85:82-106. 4. Kaufman SC, Ferris FL III, Seigel DG, et al. Factors associated with visual outcome after photocoagulation for diabetic retinopathy. Diabetic Retinopathy Study Report #13. Invest Ophthalmol Vis Sci 1989;30:23-8. 5. Rogell GD. Incremental pametinal photocoagulation. Results in treating proliferative diabetic retinopathy. Retina 1983;3:308-11. 6. Vine AK. The efficacy of additional argon laser photocoagulation for persistent, severe proliferative diabetic retinopathy. Ophthalmology 1985;92:1532-7. 7. Barr Cc. Estimation of the maximum number of argon laser burns possible in pametinal photocoagulation. Am J Ophthalmol 1984;97:697-703. 8. Frank RN. Visual fields and electroretinography following extensive photocoagulation. Arch Ophthalmol 1975;93:591-8.
Discussion
by
Andrew K. Vine, MD The management of eyes with persistent proliferative diabetic retinopathy after pametinal photocoagulation is a recurrent problem faced by all ophthalmologists who treat this disease. Possible options include additional peripheral laser therapy, early vitrectomy,l or no additional intervention. The Diabetic Retinopathy Study report number 13 2 showed that the risk of significant visual loss after pametinal photocoagulation decreased with increasing amount of initial treatment as measured by "treatment density" in the post-treatment photograph of standard field number 6. This strong correlation supported the common clinical practice of adding additional photocoagulation to eyes where the initial treatment did not cause sufficient reduction of the retinal and disc neovascularization. From the W. K. Kellogg Eye Center, University of Michigan, Ann Arbor.
1734
In this randomized, controlled prospective clinical trial, Doft and Metz assess the efficacy of additional peripheral scatter photocoagulation in diabetic eyes that continued to show 3 or more retinopathy risk factors 3 weeks after initial pametinal photocoagulation treatment. At the 6-month assessment, 4 of the 19 eyes in the NOMORE group and 6 of the 14 eyes in the MORE group showed less than 3 retinopathy risk factors. At the I-year assessment, 11 of the 19 eyes in the NOMORE group and 7 of the 16 eyes in the MORE group showed less than 3 retinopathy risk factors. The number of eyes that showed less than 3 retinopathy risk factors in the MORE group versus the NOMORE group is not statistically significant at either 6 months or 1 year. This study has a number of problems. The additional photocoagulation treatment that was given to eyes in the MORE group was not standardized. Eyes received a "minimum of 500 additional burns," but eyes in this category may have received
Doft et al . Augmentation Laser for Proliferative Diabetic Retinopathy a large range of additional laser burns. More importantly, the number of eyes involved in this study is much too small to make any conclusions whether additional scatter photocoagulation is beneficial in eyes with three or more persistent retinopathy risk factors after photocoagulation. Because of the limited sample size, the authors do not advocate abandoning the present clinical practice of additional scatter photocoagulation in eyes that fail to respond to initial panretinal photocoagulation. The Diabetic Retinopathy Study report number 33 identified four retinopathy risk factors before laser therapy associated with severe visual loss: the presence of new vessels, vessels on or near the disc, the severity of new vessels, and preretinal or vitreous hemorrhage. The significance of persistent retinopathy risk factors after laser therapy is unclear. Among the six risk factors for severe visual loss after photocoagulation therapy identified in Diabetic Retinopathy Study report number 13,2 neovascularization on or near the disc is the most significant risk factor. Thus, persistence or recurrence of some retinopathy risk factors appears to increase the risk of severe visual loss. The study of Doft et al and previous reports4 confirm that the persistence of three or more retinopathy risk factors after laser therapy is associated with a poor visual outcome and a greater risk of a "poor result." Eyes that fail to show an adequate response to initial peripheral panretinal photocoagulation remain relatively uncharacterized. It is possible that the extent of retinal capillary closure,5.6 as determined by panfunduscopic angiography, is much more extensive in these nonresponder eyes. The presence of an attached or detached posterior vitreous face also may playa role
in these unresponsive eyes. Additional investigation is clearly needed to try to characterize these nonresponder eyes and to determine the most effective management protocol. References
I. The Diabetic Retinopathy Vitrectomy Study Research Group. Early vitrectomy for severe proliferative diabetic retinopathy in eyes with useful vision. Results of a randomized trial-Diabetic Retinopathy Vitrectomy Study Report 3. Ophthalmology 1988;95: 1307-20. 2. Kaufman SC, Ferris FL III, Seigel 00, et al. Factors associated with visual outcome after photocoagulation for diabetic retinopathy. Diabetic Retinopathy Study Report 13. Invest Ophthalmol Vis Sci 1989;30:23-8. 3. The Diabetic Retinopathy Study Research Group. Four risk factors for severe visual loss in diabetic retinopathy. The third report from the Diabetic Retinopathy Study. Arch Ophthalmol 1979;97:654-5. 4. Vine AK. The efficacy of additional argon laser photocoagulation for persistent, severe proliferative diabetic retinopathy. Ophthalmology 1985;92:1532-7. 5. Shimizu K, Kobayashi Y, Muraoka K. Mid peripheral fundus involvement in diabetic retinopathy. Ophthalmology 1981;88:601-12. 6. Niki T, Muraoka K, Shimizu K. Distribution of capillary non perfusion in early stage diabetic retinopathy. Ophthalmology 1984;91:1431-9.
1735
Originally received: October 22, 1991. Revision accepted: July 13, 1992. From Retina-Vitreous Consultants, Pittsburgh. Presented at the American Academy of Ophthalmology Annual Meeting, Anaheim, October 1991. Reprint requests to Bernard H. Doft, MD, Retina-Vitreous Consultants, 3501 Forbes Ave, Pittsburgh, PA 15213.
1728
The Diabetic Retinopathy Study identified four retinopathy factors (vitreous or preretinal hemorrhage, new vessels, location of new vessels on the optic disc, and moderate or severe extent of new vessels) that increase the risk of developing severe visual loss. Eyes were found to be at high risk for such visual loss if three or more of these retinopathy risk factors were present. 1 The Diabetic Retinopathy Study also established that argon laser panretinal
Doft et al . Augmentation Laser for Proliferative Diabetic Retinopathy ablation decreases the risk of developing severe visual loss.
It is presumed that the beneficial effect of this treatment
is related to a reduction in the number of retinopathy risk factors that occur after panretinal photocoagulation. Doft and Blankenship2 demonstrated that the amount of regression of retinopathy risk factors found 3 weeks after panretinal photocoagulation is a good indicator of longer-term treatment effect. Two thirds of patients who did not regress from a high (3 or 4 risk factors) to a low risk category (2 or fewer risk factors) by 3 weeks after panretinal ablation never regressed to a low-risk category during the 6-month course of their study. The ability to make early identification of patients who fail to respond to initial laser treatment makes it possible to try to answer an important question about the management of proliferative diabetic retinopathy. Do diabetic patients who fail to respond to initial panretinal photocoagulation with regression of retinopathy risk factors do better with supplemental panretinal photocoagulation? In this article, we report the results of a prospective randomized study designed to answer that question.
Methods Patients with proliferative diabetic retinopathy were screened for the study if in the involved eye they were found to have all of the following: (1) three or more retinopathy risk factors; (2) new vessels on the disc; (3) no traction detachment of the macula; (4) ocular media clear enough to allow panretinal photocoagulation treatment; (5) intraocular pressure of 22 mmHg or less; and (6) visual acuity of 20/200 or better. Within 1 day of evaluation, patients with the above characteristics were treated with argon green laser panretinal photocoagulation in a single session using a Coherent Model 900 laser (Palo Alto, CA) and a Rodenstock panfunduscopic lens (Sola Optical, Petaluma, CA). Between 700 and 1000 bums were placed using 500-JLm spot size (at the laser control), O.l-second duration, and with a power setting sufficient to achieve a moderately white bum. Spots were placed one half bum width apart in a regular pattern as far to the periphery as possible with a Rodenstock lens, starting at points on an oval defined as 2 disc diameters above, below, and temporal to the center of the macula and 500 JLm nasal to the nasal half of the disc. Retrobulbar Xylocaine was used for anesthesia on most patients. Patients were seen in follow-up 3 weeks after initial treatment. Only patients whose eyes had media clear enough to allow supplemental laser treatment, and who still showed 3 or 4 retinopathy risk factors were considered eligible for entry into the study. Eligible patients who gave consent were randomly assigned to either no additional laser treatment (NOMORE group) or to immediate supplemental panretinal photocoagulation (MORE group). A block of 50 opaque envelopes had been prepared in advance, with 25 indicating MORE and 25 indicating NOMORE. These envelopes were prepared, sealed, thoroughly mixed, sequentially
numbered, and then used in numbered sequential order. Eyes assigned to the MORE group received a minimum of 500 additional bums of argon laser with the Rodenstock lens filling in between previous laser bums, with efforts made not to retreat over previous treatment spots. In addition, where there was room for additional treatment in the far periphery, the Goldmann three-mirror lens was used to treat this area (placing spots one half bum width apart and using 500-JLm spot size at the laser control) if the area could not be reached with the Rodenstock lens. Study follow-up data were collected 6 and 12 months after initial treatment. Visual acuity was measured using a Snellen chart by an experienced examiner of visual acuity who was masked to the patients' treatment status. Visual acuity was measured with the then-current glasses prescription and pinhole. On study visits, data capture forms were completed before seeing past ophthalmic records. Although not masked, when seeing patients on study visits 6 months after they had been randomized, the examiner (BHD) was not aware of the group to which the patient had been randomized. At any time, if a particular retinopathy risk factor could not be assessed because oflimited view due to vitreous hemorrhage, the risk factor was assumed to be present in its most severe extent. This occurred in 3 eyes at the I-year follow-up visit. Hypertension was defined as being present if the patient was being treated with systemic antihypertensive medication at the time of the initial examination. The onset of diabetes was dated from the time the patient began taking any medication for the disease. Insulin-dependent diabetes was defined as the patient using insulin on a regular basis at the time of the initial study examination. With the exception of the designation regarding panretinal photocoagulation treatment, the study protocol allowed for patients to receive any ophthalmologic care required, based on their ophthalmic status. Statistical analysis was carried out using the chi-square test for proportions and t tests for mean number of retinopathy risk factors. A one-tailed not two-tailed test was used because one would only want to recommend augmentation laser if it were better than observation. Thus, if augmentation laser and observation were shown to produce similar outcome overall, observation would be the treatment of choice since it is less invasive. It is recognized that augmentation laser could indeed result in worse outcome, but it is not important to distinguish the difference between augmentation laser worse and augmentation laser the same, since in either case one would choose not to perform augmentation laser. Ninety-five percent confidence intervals for differences over time in retinopathy risk factors are, however, reported as two-sided.
Results As of the cutoff date for the analysis, 35 patients had completed the I-year follow-up examination. Entered into the study but not included in this report were 3 patients who were not yet due for final examination as of that date, 2
1729
Ophthalmology Volume 99, Number 11, November 1992 Table 1. Baseline Characteristics Item Number Number right eye Mean age (range) Male sex (number) Mean duration diabetes, yrs (range) Percent insulin requiring Percent hypertensive
More
All Patients
No More
35 20 40 (18-70) 17
19 12 40 (19-70) 9
16 8 42 (18-62) 8
17 (2-29) 94% 29%
15.5 (2-28) 95% 32%
18.8 (10-29) 94% 25%
(I from each study group) who died before the I-year examination, and I (from the NOMORE group) who refused to return for follow-up. Of the 35 patients followed for I year, 4 examinations (initial, 3 weeks, 6 months, and 1 year) were performed on all patients except for 2 (both from the MORE group) who missed their 6-month visit. For these two missed visits, the number of retinopathy risk factors were interpolated between an earlier and later visit. Sixteen eyes were assigned at random to supplemental laser (MORE), and 19 to no additional laser (NOMORE). Demographic information concerning the patients is included in Table 1. Information about selected specific patients referred to within individual sections below is found in Table 2.
Intraocular Pressure There was no difference in the mean intraocular pressure by treatment group either at study entry (NOMORE = 15.9 versus MORE = 17.4), 6 months (15.5 versus 15.8), or 1 year (15.2 versus 15.9). At 1 year, a single study patient in each group had an intraocular pressure of 6 mmHg (patients 10 1 and 108) (Table 2). All other patients in both groups had an intraocular pressure between 11 and 23 mmHg at 1 year.
Rubeosis At the initial, 3-week, and 6-month visits, 1 patient in the NOMORE group had rubeosis. At the I-year visit, 1 patient from each group had rubeosis (patients 101 and 108).
Table 2. Table of Selected Patients Study Visual lOP Number Group· Acuity (mmHg) Rubeosis Mediat
Retina
Additional Surgery
51 101
M M
CF HM
21 6
No Yes
1 2
Macular edema Posterior pole attached with some peripheral detachment
No PPY and SB, two revisions, last with silicone oil
159
M
HM
14
No
3
Tractional retinal detachment involves macula
PPY
M NM
HM NLP
17
No Yes
2
108
3
Retina attached Retina detached
164 167
NM NM
CF CF
20
16
No No
4
No PPY and SB one revision No PPY, PPL, lens implant placement
62
NM
20/100
15
No
32
6
1
Macular edema No retinal detachment (ultrasound)
Tractional retinal detachment extrafoveal
Comment By 2 yrs, total retinal detachment with LP acuity Retina attached after PPY, but retinal vessels sclerotic Prephthisical
Patient's surgery is 10 wks after the 1-yr follow-up and final visual acuity = 20/50
No
lOP = intraocular pressure; CF = counting fingers; HM = hand motions; PPY = pars plana vitrectomy; SB = scleral buckle; LP = light perception; NLP = no light perception; PPL = pars plana lensectomy. •M
=
t
more panretinal photocoagulation; NM
=
no more panretinal photocoagulation.
1 = clear; 2 = blood blocks less than one quarter of the retinal area; 3 = blood blocks one quarter or more of the retinal area; 4 = blood obscures all view of retina.
1730
Dolt et al
Augmentation Laser for Proliferative Diabetic Retinopathy
Retinopathy Risk Factors By definition (study entry criteria), 100% of eyes had 3 or more risk factors at the time of study entry, which was 3 weeks after the initial panretinal photocoagulation. Six months after treatment, 37.5% of MORE and 21% of NOMORE eyes had regressed from a high risk (3 or more risk factors) to a low risk category (less than 3 risk factors). The percent of MORE eyes that had regressed to a lowrisk category increased by only an additional 6.5% to 44% at 1 year, but the percent of NOMORE eyes that had regressed to less than 3 risk factors changed from 21 % at 6 months to 58% at 1 year. For all patients, regardless of treatment group, 51.4% of eyes had regressed to less than 3 risk factors by 1 year. The percentage of eyes with 0 risk factors at 1 year was 25% and 26% in the MORE and NOMORE groups, respectively. There was no advantage in regression of risk factors to the MORE group over the NOMORE group at the final study visit at 1 year. These data are summarized in Table 3. The mean number of risk factors decreased in both groups with time from treatment. The mean number of risk factors was 3.50 versus 3.21 at the initial visit, 2.56 versus 3.00 at 6 months, and 2.38 versus 2.16 at the 1year visit in the MORE versus NOM ORE groups, as shown in Table 3. Eyes in the MORE group achieved most of their regression of mean retinopathy risk factors within 6 months of treatment (-0.94 of a total of -1.12), with
only little additional regression (-0.18 risk factors) occurring during the second 6 months of follow up. In contrast, eyes in the NOM ORE group showed a mean decrease of just 0.21 risk factors in the first 6 months, with substantial additional regression in the second 6-month period (an additional -0.84 risk factors to a total of -1.05 at 1 year). The difference in amount of regression for 6 months between groups was significant at P = 0.055. However, for 1 year, there was no statistical difference in the mean risk factors regression between groups. The NOMORE group was slower in achieving regression, but its end result (-1.05 mean risk factors) at 1 year was essentially the same amount of risk factors regression achieved by the MORE group (-1.12 risk factors). Ninetyfive percent confidence intervals for these differences are shown in Table 3. Reflecting this sample size, the confidence intervals are relatively wide and overlap between the MORE and NOMORE treatment groups. The study data show the standard deviation of the change in number of risk factors from initial to I year was 1.3. Power analysis showed the study with 35 patients had an 80% power to detect a difference of 1.1 risk factors between groups at I year. To have 80% power to detect a difference of 0.5 risk factors, 168 eyes would have been required. The number of retinopathy risk factors was looked at for eyes with and without evidence of vitreous hemorrhage at I year. The mean number of risk factors in study eyes with no evidence of vitreous hemorrhage at 1 year was 1.38 (n = 21), with a mean of 1.33 (n = 12) for the NO-
Table 3. Retinopathy Risk Factors by Treatment and Time of Observation
Mean Number RRF Initial 6 Months 1 Year
NOMORE (n = 19)
MORE (n = 16)
3.21 3.00 2.16
3.50 2.56 2.38
Mean Change in Number RRF (95% Confidence Interval) Initial to 6 months· 6 months to 1 year Initial to 1 year
-0.21 (-0.69, 0.27) -0.84 (-1 .60, -0.04) -1.05 (-1.80, -0.28)
-0.94 (-1.60, -0.26) -0.18 (-0.72, 0.34) -1.12 (-2.00, -0.24)
Percent of Patients with 2 or Fewer RRF (95% Confidence Interval) Initial 6 Months 1 Year
0 21.1 (3, 39) 57.9 (36, 80)
0 37.5 (14, 61) 43.8 (19, 68)
5.3 26.3
14.3 25.0
Percent Patients with No RRF 6 Months 1 Year
RRF = retinopathy risk factor . • p = 0.055 for one-sided t test between MORE and NOMORE treatment groups. Two patients in the MORE group missed the 6-month follow-up visit. Data for these two patients at this time point were interpolated between the prior and subsequent visit.
1731
Ophthalmology Volume 99, Number 11, November 1992 MORE and 1.44 (n = 9) for the MORE group. In contrast, the mean number of risk factors for eyes that showed any amount of vitreous hemorrhage at 1 year was 3.57 (n = 14), with a separate mean of 3.57 for the MORE and NOMORE groups (n = 7 in both). Visual acuity distributions were assessed based on persistent risk factors present 1 year after treatment. For patients with 3 or more persistent risk factors at 1 year regardless of whether they were in the MORE or NOMORE group, the visual acuity distribution was worse than for patients who had less than 3 persistent risk factors. Eightynine percent (16 of 18) of patients with less than 3 risk factors at 1 year had visual acuity of 20/200 or better, compared with 59% (10 of 17) of those with more than 3 risk factors. This difference was statistically significant at P = 0.04, based on chi-square analysis. For visual acuity of 20/100 or better, the percentages were 78% and 53%, respectively, and for visual acuity of 20/50 or better, the percentages were 61 % and 41 %, respectively. These differences were not statistically significant at P = 0.05 (Fig 1). Poor results can be defined in a variety of ways. If one defines an eye with a poor result at 1 year as an eye with any of the following: traction detachment more than 4 disc areas in size, or vitreous hemorrhage blocking one quarter or more of the retinal area, or visual acuity worse than 20/400, then there were 7 eyes (3 in the NOMORE group and 4 in the MORE group) with a poor result. A poor result was present in 6 of 17 eyes (35%) with 3 or more risk factors at 1 year, but in only 1 of 18 eyes (6%) with less than 3 risk factors at that time. This difference is significant at P = 0.016. The persistence of 3 or more risk factors 1 year after treatment is therefore associated with a poor result as herein defined.
Visual Acuity Before initial treatment, all patients had visual acuity of 20/200 or better. At the 6-month follow-up visit, visual acuity of 20/50 or better was present in 57% of eyes in the MORE group and in 63% of eyes in the NOMORE group; visual acuity of 20/200 or better was found in 71 % VISUAL ACUITY DISTRIBUTION BY PERSISTENCE OF RETINOPATHY RISK FACTORS (RRF) AT 1 YR
120.------------------------------, = 3RRF -+-
I-
z
W
=20/25 >=20150 >=201100 >-201200 >"'101200 >=5/200
>""LP
>=-NLP
ACUITY
Figure 2. Cumulative visual acuity distribution at 1 year. The cumulative percent of eyes with various acuity levels is shown far the MORE and NOMORE groups.
and 68% of the 2 groups, respectively. At the I-year followup visit, 50% of the MORE group and 53% of the NOMORE group had visual acuity of 20/50 or better, with 75% of the MORE group and 74% of the NOMORE group having visual acuity of 20/200 or better. At the I-year visit, there were 4 eyes in the MORE group and 3 in the NOMORE group with visual acuity worse than 20/400, as shown in Table 2. Figure 2 shows cumulative visual acuity distribution at 1 year. There is no evidence of a harmful or beneficial effect on visual acuity of either MORE or NOMORE treatment.
Vitreous Hemorrhage Vitreous hemorrhage was graded during each examination as being absent, as obscuring the visualization ofless than one quarter of the retina, or as obscuring one quarter or more of the retina. At the I-year follow-up visit, there were 7 eyes in each group with vitreous hemorrhage. In the MORE group, 6 of the 7 eyes had obscuration ofless than one quarter of the retinal area, with one of these having a retinal detachment (patient 101). A single patient (patient 159) had one quarter or more of the retinal area obscured by vitreous hemorrhage. In the NOMORE group, 5 of the 7 eyes with vitreous hemorrhage had obscuration ofless than one quarter of the retinal area, and none of these had retinal detachment. Two eyes had vitreous hemorrhage obscuring one quarter or more of the retina (Table 2).
60
Detachment of the Retina
40
At the I-year visit, there were four patients (Table 2) with retinal detachment, two in each group.
20
~=20125
>=20/50 >=20/100 >=20/200 >=101200 >=51200
>=LP
>=NLP
ACUITY
Figure 1. Cumulative visual acuity distribution at 1 year by low (less than 3) versus high (3 or more) retinopathy risk factors category.
1732
Additional Surgical Procedures Two eyes in the MORE group and 2 in the NOMORE group required additional surgery during the course of
Dolt et al . Augmentation Laser for Proliferative Diabetic Retinopathy the study, or within 10 weeks of the last study visit. Details are presented in Table 2.
Discussion The Diabetic Retinopathy Study identified 4 retinopathy risk factors that are associated with severe visual loss, and found that the presence of 3 or more of these factors carried a risk of 26% to 37% for severe visual loss at 2 years, whereas the presence ofless than 3 retinopathy risk factors carried a risk of 8.5% or less. J Eyes with 3 or more risk factors were therefore defined as eyes with high-risk characteristics. The Diabetic Retinopathy Study recommended laser photocoagulation for eyes with high-risk characteristics, since results demonstrated that panretinal photocoagulation decreases the risk of severe visual loss by approximately 50% compared with untreated patients. 3 It is presumed, but not known, that the beneficial effect of panretinal photocoagulation is related to reduction of number of retinopathy risk factors that occur after panretinal photocoagulation treatment. Doft and B1ankenship 2 assessed for rapidity of regression of retinopathy risk factors after panretinal photocoagulation. They showed that risk factors regress rapidly after photocoagulation, and that the 3-week response was a good indicator oflonger-term risk factor regression. All of their patients had three or more retinopathy risk factors before treatment. By 3 weeks after treatment, 72% had regressed to less than 3 risk factors, and 28% continued to have high-risk characteristics. Of patients who did not respond to initial treatment by regression to less than 3 risk factors by 3 weeks, 64% continued to have high-risk characteristics (3 or more risk factors) for the full duration of the study during which no additional panretinal photocoagulation treatment was placed. The Diabetic Retinopathy Studl showed that the risk of severe visual loss falls with increasing amount of initial treatment (as measured by treatment density in the posttreatment photograph of standard field 6). However that study did not look specifically at the subgroup of patients who had persistent high-risk characteristics after initial treatment, or what would happen if additional treatment was placed to those eyes. Roge1l 5 described a protocol for incremental laser treatment iffollow-up after initial treatment did not result in substantial regression of neovascularization. However, there were no controls to determine what would happen if additional treatment was not placed. Vine 6 reported on eyes that failed to respond with regression to less than 3 risk factors by 8 weeks after an initial panretinal photocoagulation of 3000 calculated burns. Calculated burns were determined by multiplying the number of burns placed with a pan funduscopic lens by 2.7 to convert to Goldmann burn equivalents, based on the work of Barr.7 Twenty-three nonresponder eyes were treated with an additional 1000 to 2000 Goldmann burns. The final result was assessed 8 weeks after the last laser treatment session. Twelve of the 23 eyes responded (less than 3 risk factors) and 11 did not (maintained 3 or
more risk factors). There was little adverse effect on acuity due to additional laser treatment since acuity was within I line of pretreatment acuity in 10 of the 12 responder eyes. Of the 11 failure eyes, 6 had visual acuity of 20/80 or better, but 5 had visual acuity of counting fingers or worse, and 3 had neovascular glaucoma. While Vine's study showed that about one half of initial nonresponder eyes respond to augmentation panretinal photocoagulation, his study was retrospective and did not include a control group. Thus, it is impossible to know if a similar group of patients who had not received additional laser would have done worse, better, or the same as the patients treated with augmentation. The current study uses a prospective randomized design to determine the effect of augmentation panretinal photocoagulation in patients who fail to respond to less than 3 risk factors 3 weeks after initial panretinal photocoagulation. The results show that MORE treatment results in more rapid regression of retinopathy risk factors. Six months after treatment, the MORE group had regressed a mean of -0.94 risk factors, compared with -0.21 risk factors in the NOMORE group. However, by I year, there was no statistically significant difference in the amount of regression of risk factors by treatment group, with a mean decrease of -1.12 versus -1.05 risk factors in the 2 groups, respectively. While the NOMORE group is slower in achieving regression, by 1 year the same amount of regression of retinopathy risk factors occurs in both groups. As far as visual acuity is concerned, there was no difference between groups, as shown in Figure 2. The study deals with another issue. While the Diabetic Retinopathy Study established that the presence of retinopathy risk factors be/ore treatment was associated with risk of severe visual loss, the Diabetic Retinopathy Study did not deal with the question of whether persistence of risk factors a./iertreatment had already been placed is also a risk factor for visual loss. To examine that question, in this study visual acuity distributions were assessed based on the number of persistent risk factors present I year after treatment. For all patients with 3 or more risk factors at 1 year, regardless of whether they were in the MORE or NOMORE group, the visual acuity distribution was worse than for patients who had less than 3 risk factors, as shown in Figure 1. Therefore, results of the study show that the persistence of retinopathy risk factors after treatment is associated with a poorer visual acuity result. Similar information is shown by assessing for eyes with poor result as defined in this study. A poor result was present in 35% (6 of 17) of eyes with 3 or more risk factors at 1 year, but in only 6% (I of 18) of eyes with less than 3 risk factors. While there was a relationship between persistence of 3 or more risk factors and poor result, these findings do not necessarily allow a conclusion that augmentation panretinal photocoagulation will lead to a lesser chance of a poor result. There are certain strengths and weaknesses to this study. It is prospective and uses random assignment of patients to a treatment group. However, the number of patients in the study is too small to allow the statistical power necessary to detect small differences between treat-
1733
Ophthalmology
Volume 99, Number 11, November 1992
ment groups. For example, to have a standard 80% power to detect a difference of 0.5 risk factors between groups, 168 eyes would have been required. Visual acuity, although measured by a masked examiner, used only pinhole over previous refraction and was not obtained by refracting the patient at the time visual acuity was measured. A Bailey-Lovie acuity chart was not available at study initiation, so Snellen acuities were used. Snellen acuities may be less useful, especially when visual acuities are worse than 20/ tOO. In this study, visual acuity was measured the same way in all patients, and by a masked, experienced observer who used previous refraction and pinhole. If there were errors in acuity measurement, they should have occurred equally in either treatment group. This study did not address certain important issues. There is a possibility that augmentation laser treatment could result in adverse side effects. This could occur if treatment were applied to previously treated areas since treatment of thinned retina may increase the risk of damage to the nerve fiber layer, and possibly increase the amount of field loss. 8 The current study made no attempt to assess visual fields on these patients. In conclusion, in this article, we report the results of a prospective randomized clinical trial designed to determine if patients who fail to regress from high-risk characteristics with argon laser panretinal photocoagulation will benefit from augmentation laser. The study showed that although augmentation panretinal photocoagulation achieved faster regression of risk factors, by 1 year, there was no difference in either mean regression of risk factors or visual acuity between eyes treated with and without augmentation panretinal photocoagulation. In addition, results of the study showed that the persistence of risk factors after treatment is associated with a poorer visual acuity result. The power of this study to find small differences between treatment groups was limited by sample size. In
proliferative diabetic retinopathy, even a small difference in outcome could be very meaningful clinically. Because of the limitations imposed by sample size, the authors do not recommend any change in management of diabetic patients who do not regress their retinopathy risk factors. A common practice is to treat such patients with further laser therapy. There is no reason to abandon such a practice based on this report. However, a larger clinical trial with sample size adequate to detect small differences in outcome between treatment groups should be performed.
References 1. The Diabetic Retinopathy Study Research Group. Four risk factors for severe visual loss in diabetic retinopathy. The third report from the Diabetic Retinopathy Study. Arch Ophthalmol 1979;97:654-5. 2. Doft BR, Blankenship G. Retinopathy risk factor regression after laser pametinal photocoagulation for proliferative diabetic retinopathy. Ophthalmology 1984;91:1453-7. 3. The Diabetic Retinopathy Study Research Group. Photocoagulation treatment of proliferative diabetic retinopathy: the second report of the Diabetic Retinopathy Study findings. Ophthalmology 1978;85:82-106. 4. Kaufman SC, Ferris FL III, Seigel DG, et al. Factors associated with visual outcome after photocoagulation for diabetic retinopathy. Diabetic Retinopathy Study Report #13. Invest Ophthalmol Vis Sci 1989;30:23-8. 5. Rogell GD. Incremental pametinal photocoagulation. Results in treating proliferative diabetic retinopathy. Retina 1983;3:308-11. 6. Vine AK. The efficacy of additional argon laser photocoagulation for persistent, severe proliferative diabetic retinopathy. Ophthalmology 1985;92:1532-7. 7. Barr Cc. Estimation of the maximum number of argon laser burns possible in pametinal photocoagulation. Am J Ophthalmol 1984;97:697-703. 8. Frank RN. Visual fields and electroretinography following extensive photocoagulation. Arch Ophthalmol 1975;93:591-8.
Discussion
by
Andrew K. Vine, MD The management of eyes with persistent proliferative diabetic retinopathy after pametinal photocoagulation is a recurrent problem faced by all ophthalmologists who treat this disease. Possible options include additional peripheral laser therapy, early vitrectomy,l or no additional intervention. The Diabetic Retinopathy Study report number 13 2 showed that the risk of significant visual loss after pametinal photocoagulation decreased with increasing amount of initial treatment as measured by "treatment density" in the post-treatment photograph of standard field number 6. This strong correlation supported the common clinical practice of adding additional photocoagulation to eyes where the initial treatment did not cause sufficient reduction of the retinal and disc neovascularization. From the W. K. Kellogg Eye Center, University of Michigan, Ann Arbor.
1734
In this randomized, controlled prospective clinical trial, Doft and Metz assess the efficacy of additional peripheral scatter photocoagulation in diabetic eyes that continued to show 3 or more retinopathy risk factors 3 weeks after initial pametinal photocoagulation treatment. At the 6-month assessment, 4 of the 19 eyes in the NOMORE group and 6 of the 14 eyes in the MORE group showed less than 3 retinopathy risk factors. At the I-year assessment, 11 of the 19 eyes in the NOMORE group and 7 of the 16 eyes in the MORE group showed less than 3 retinopathy risk factors. The number of eyes that showed less than 3 retinopathy risk factors in the MORE group versus the NOMORE group is not statistically significant at either 6 months or 1 year. This study has a number of problems. The additional photocoagulation treatment that was given to eyes in the MORE group was not standardized. Eyes received a "minimum of 500 additional burns," but eyes in this category may have received
Doft et al . Augmentation Laser for Proliferative Diabetic Retinopathy a large range of additional laser burns. More importantly, the number of eyes involved in this study is much too small to make any conclusions whether additional scatter photocoagulation is beneficial in eyes with three or more persistent retinopathy risk factors after photocoagulation. Because of the limited sample size, the authors do not advocate abandoning the present clinical practice of additional scatter photocoagulation in eyes that fail to respond to initial panretinal photocoagulation. The Diabetic Retinopathy Study report number 33 identified four retinopathy risk factors before laser therapy associated with severe visual loss: the presence of new vessels, vessels on or near the disc, the severity of new vessels, and preretinal or vitreous hemorrhage. The significance of persistent retinopathy risk factors after laser therapy is unclear. Among the six risk factors for severe visual loss after photocoagulation therapy identified in Diabetic Retinopathy Study report number 13,2 neovascularization on or near the disc is the most significant risk factor. Thus, persistence or recurrence of some retinopathy risk factors appears to increase the risk of severe visual loss. The study of Doft et al and previous reports4 confirm that the persistence of three or more retinopathy risk factors after laser therapy is associated with a poor visual outcome and a greater risk of a "poor result." Eyes that fail to show an adequate response to initial peripheral panretinal photocoagulation remain relatively uncharacterized. It is possible that the extent of retinal capillary closure,5.6 as determined by panfunduscopic angiography, is much more extensive in these nonresponder eyes. The presence of an attached or detached posterior vitreous face also may playa role
in these unresponsive eyes. Additional investigation is clearly needed to try to characterize these nonresponder eyes and to determine the most effective management protocol. References
I. The Diabetic Retinopathy Vitrectomy Study Research Group. Early vitrectomy for severe proliferative diabetic retinopathy in eyes with useful vision. Results of a randomized trial-Diabetic Retinopathy Vitrectomy Study Report 3. Ophthalmology 1988;95: 1307-20. 2. Kaufman SC, Ferris FL III, Seigel 00, et al. Factors associated with visual outcome after photocoagulation for diabetic retinopathy. Diabetic Retinopathy Study Report 13. Invest Ophthalmol Vis Sci 1989;30:23-8. 3. The Diabetic Retinopathy Study Research Group. Four risk factors for severe visual loss in diabetic retinopathy. The third report from the Diabetic Retinopathy Study. Arch Ophthalmol 1979;97:654-5. 4. Vine AK. The efficacy of additional argon laser photocoagulation for persistent, severe proliferative diabetic retinopathy. Ophthalmology 1985;92:1532-7. 5. Shimizu K, Kobayashi Y, Muraoka K. Mid peripheral fundus involvement in diabetic retinopathy. Ophthalmology 1981;88:601-12. 6. Niki T, Muraoka K, Shimizu K. Distribution of capillary non perfusion in early stage diabetic retinopathy. Ophthalmology 1984;91:1431-9.
1735
