Iris N eovascularization in Proliferative Vitreoretinopathy Mark R. Comaratta, MD, Stanley Chang, MD, Janet Sparrow, PhD
Purpose: The purpose of this study is to report on the prevalence, incidence, and associated risk factors of iris neovascularization in nondiabetic patients undergoing vitrectomy for retinal detachment complicated by proliferative vitreoretinopathy (PVR). Methods: The authors conducted a retrospective review of 141 consecutive nondiabetic patients undergoing vitrectomy for recurrent retinal detachment resulting from PVR. Univariate and multivariate analyses were performed on all patients to determine which preoperative, intraoperative, and postoperative factors were associated with the development of postoperative iris neovascularization. Results: Twenty-seven of the 141 (19%) patients were noted with preoperative and/or postoperative iris neovascularization. Four of eight patients presenting with preoperative iris neovascularization had complete regression after successful reattachment of the retina. Results of analysis of the remaining 133 patients without iris neovascularization preoperatively showed residual retinal detachment as the most significant risk factor for postoperative iris neovascularization. In the absence of panretinal photocoagulation, none of the 27 patients developed neovascular glaucoma. Conclusions: The development of iris neovascularization preoperatively or postoperatively is not necessarily a predictor of a poor anatomic and/or visual result. Iris neovascularization in PVR rarely if ever progresses to neovascular glaucoma. Pan retinal photocoagulation is not indicated in these patients. Retinal reattachment is the most important factor in the prevention and/or resolution of postoperative iris neovascularization. The development of iris neovascularization in PVR appears to be a multifactorial process requiring multiple variables acting in concert. Ophthalmology 1992;99:898-905
Iris neovascularization is a well-known complication of proliferative diabetic retinopathy and venous occlusive disease. Retinal detachment has been noted by several authors to be complicated by iris neovascularization. 1•2 It also has been reported as a complication of scleral buckling surgery. 3- 5 The combination of lensectomy, vitrectomy, and rhegmatogenous retinal detachment has been shown to produce iris neovascularization in experimental animal models. 6 Originally received: June 27, 1991. Revision accepted: January 23, 1992. From the Department of Ophthalmology, The New York HospitalCornell University Medical Center, New York. Supported by the Vitreous Research Fund, Cornell University Medical College, New York, New York. Reprint requests to Stanley Chang, MD, Department of Ophthalmology, The New York Hospital, Starr Pavilion Room 817, 525 E 68th St, New York, NY 1002 I.
898
To our knowledge, no previous reports have examined the presence of iris neovascularization in nondiabetic patients undergoing surgery for retinal detachment complicated by proliferative vitreoretinopathy (PVR). This study reviewed retrospectively 141 nondiabetic patients undergoing vitrectomy for PVR. The prevalence and incidence of iris neovascularization and associated risk factors are reported.
Subjects and Methods We examined the records of 163 patients who underwent vitreous surgery performed by one surgeon (SC) for recurrent retinal detachment resulting from PVR between 1983 and 1989. Each patient's age, sex, and ocular history were recorded. Patients with the following conditions were excluded: (1) retinal vascular disease (i.e., diabetic retinopathy); (2) infectious or inflammatory posterior uveitis;
Comaratta et al · Iris Neovascularization in Proliferative Vitreoretinopathy or (3) intraocular tumor. Retinal detachments were classified as those associated with: ( 1) rhegmatogenous retinal detachment; (2) giant retinal tear; or (3) trauma (blunt or penetrating). The preoperative characteristics were recorded from the findings of the initial eye exam. They included: (1) visual acuity; (2) intraocular pressure (lOP); (3) iris neovascularization; (4) the presence of vitreous hemorrhage; (5) the grade of proliferative vitreoretinopathy (Retina Society Classification Committee); 7 and (6) the presence/absence of anterior proliferative vitreoretinopathy. Details of all initial and secondary surgical procedures performed on referral were recorded. These included: ( 1) the total amount of preoperative and postoperative laser (recorded as number of laser burns) and/ or cryotherapy; (2) the number of fluid-gas exchanges; (3) the use ofperfluorocarbon liquids; (4) the use of relaxing retinotomy or retinectomies; (5) the use of silicone oil; and (6) the use ofintravitreal gas tamponade. Followup data were recorded from the outpatient charts and from the referring ophthalmologist. All patients with less than 3 months offollow-up were excluded from the study. Data were recorded regarding: ( 1) the length of followup; (2) final visual acuity; (3) lOP; (4) final lens status (aphakic, phakic, anterior chamber or posterior chamber intraocular lens); (5) iris neovascularization, including the length of time from onset and/or regression; and (6) anatomic status of the retina (fully attached, partially detached posterior to the scleral buckle, detached anterior to the scleral buckle, total detachment). Patients were categorized as to the presence or absence of iris neovascularization both preoperatively and postoperatively. A univariate analysis was performed on all the patients who presented without iris neovascularization preoperatively to determine which preoperative, intraoperative, and postoperative factors were associated with the development of postoperative iris neovascularization. Factors scored as present or absent were initially tested for a statistically significant association with iris neovascularization by Fisher's exact test for two-by-two contingency tables. Factors that were graded (PVR or visual acuity), counted (number of scleral buckles), or measured on a continuous scale (age or lOP) were initially tested using a Student's two-sample t test. Factors even moderately significant univariately (P < 0.10) were then considered in a stepwise multiple logistic regression analysis. This was done to examine the association of a given variable with iris neovascularization after adjusting for all other variables. The strength of the association between postoperative iris neovascularization formation and a given patient variable was determined by using the relative odds ratio. This was calculated by the odds of developing postoperative iris neovascularization, given the high-risk category of a factor, divided by the odds given the lowrisk category of a factor.
Surgical Method Patients requiring a scleral buckle or revision of a previous scleral buckle underwent the placement of a circumfer-
ential3.2 X 7.5 mm grooved sponge with a #240 silicone encircling band, which was used to support the vitreous base for 360°. A standard three-port vitrectomy infusion system was used. A lensectomy was performed in those cases with extensive anterior proliferation at the vitreous base. All epiretinal membranes were removed as completely as possible, until the retina appeared mobile. Attention was then directed to the periphery and the removal of all anterior proliferation. Liquid perfluorocarbon was occasionally used at this point to stabilize the peripheral retina and flatten the posterior retina, mobilizing subretinal fluid and/or hemorrhage anteriorly. Extensive bimanual anterior dissection was then performed using the light pick and membrane forceps, to release all traction for 360°. It was occasionally necessary to perform relaxing retinotomies and retinectomies in areas where traction remained despite extensive dissection. A fluid-air exchange was then performed with drainage of all subretinal fluid with a soft-tipped flute needle through an existing retinal break or a retinotomy. Endophotocoagulation was then used to surround all identifiable retinal breaks. Two additional rows of photocoagulation were then placed on the anterior edge of the scleral buckle in areas not previously treated. Postoperatively, as the inferior meniscus of the gas bubble began to rise, new or previously unidentified retinal breaks caused reaccumulation of fluid in a dependent fashion. If the retina was mobile, a fluidgas exchange was performed with subsequent photocoagulation to seal all retinal breaks. If reproliferation occurred, resulting in significant tractional retinal elevation, repeat vitrectomy was then performed.
Selected Case Study A 49-year-old man was referred with a diagnosis of recurrent retinal detachment in the right eye. He had previously undergone a single scleral buckling procedure for a retinal detachment 6 months previously. On examination, the visual acuity was hand motions in the right eye (with a -4.25 sphere) and 20/15 in the left (with a -5.00 sphere). Intraocular pressure was 6 mmHg in the right eye and 15 mmHg in the left. Results of slit-lamp examination were unremarkable in both eyes. There was no iris neovascularization noted in either eye. Results of examination of the posterior segment of the right eye showed a vitreous hemorrhage and a total retinal detachment. Examination of the left fundus was normal. The patient underwent lensectomy and vitrectomy surgery in the right eye. After pars plana lensectomy, a dense vitreous hemorrhage was found along with the posterior hyaloid, which was contracted anteriorly with severe circumferential traction. The retinal detachment was a D3 PVR configuration. Extensive membranes were removed from the surface ofthe retina, using the Michel's pick and membrane forceps. Liquid perfluorocarbon was then infused over the optic disc, stabilizing the peripheral retina. Extensive anterior dissection was performed for 360° using the light pick and membrane forceps in an attempt to release all tractional elements. A retinotomy was performed from the 9-o'clock to 12-o'clock position, just posterior to the scleral buckle due to extensive anterior adhesion by membranes and hemorrhage to the peripheral retina. After a fluid-gas exchange, with complete removal of the perfluorocarbon liquid, the retina
899
Ophthalmology
Volume 99, Number 6, June 1992
was flat. Endophotocoagulation was placed for 360° on the scleral buckle. The eye was then flushed with a 20% mixture of C3F8 gas. Six weeks postoperatively, the retina was completely attached posterior to the scleral buckle. There was early anterior reproliferation and mild retinal elevation on the anterior crest of the buckle. A few days later, the patient complained of flashing lights and was re-examined. The very mobile retina was totally detached posterior to the scleral buckle. There was a retinal break noted on the crest of the buckle at the 12-o'clock position. A fluid-gas exchange was performed at that time, followed by argon green laser photocoagulation to surround the retinal break several days later. Three months postoperatively, his lOP was 22 mmHg and the patient was noted with early iris neovascularization at the pupillary margin. Two weeks later, frank rubeosis iridis developed. Results of gonioscopy showed intermittent areas of peripheral anterior synechiae for approximately 270°. The retina was completely attached posterior to the scleral buckle. There was significant reproliferation and retinal elevation anterior to the scleral buckle. Large areas of bare retinal pigment epithelium (RPE) were noted anterior to the scleral buckle from the 9-o'clock to !-o'clock position. There was no neovascularization on the retinal surface or optic disc. Fluorescein angiography at that time showed an irregular choroidal pattern with normal arteriole and venous transit times, and no areas of retinal capillary nonperfusion. The patient was re-examined 3 weeks later. His iris neovascularization had begun to regress and he had developed mild uveal ectropion. He eventually underwent therapeutic ultrasound due to slowly progressive angle closure. His lOP was chronically maintained at approximately 30 mmHg with timolol maleate. His retinal examination was unchanged. The patient's examination results have remained unchanged on subsequent visits.
Results Of the 163 patients undergoing vitrectomy for retinal detachment complicated by PVR, 22 were excluded because of inadequate follow-up. Mean follow-up time was 18
months (range, 3 to 84 months). Twenty-seven of theremaining 141 ( 19%) patients were noted with preoperative and/or postoperative iris neovascularization ranging from fine iris vessels to frank rubeosis iridis with uveal ectropion. Of the 27 patients noted with iris neovascularization, 8 presented to our office with iris neovascularization preoperatively. Nineteen developed iris neovascularization postoperatively. The results of a univariate analysis performed on the 133 patients who presented without iris neovascularization is shown in Table 1. Preoperative factors associated with postoperative iris neovascularization included vitreous hemorrhage (P = 0.051) and severe PVR, grade D2 or D3 (P = 0.049). Of the intraoperative factors, the creation of a retinotomy (P = 0.0026) was associated with postoperative iris neovascularization. The use of perfluorocarbon liquids in the management of anterior PVR was found to be a protective factor against developing postoperative iris neovascularization. Anterior (and/or total) retinal detachment was by far the single most significant postoperative factor associated with postoperative iris neovascularization. The risk was found to be similar in those patients with anterior detachment alone, compared with those with total detachment, so that the two categories were combined for further data analysis. Forty-six percent (17 of 37) of patients with anterior (and/ or total) detachment developed iris neovascularization compared with only 2% (2 of95) of patients without anterior (and/or total) detachment (P = 0.00000001 by Fisher's exact test). Other postoperative factors showing significant association with iris neovascularization by Student's t test included the performance of 3 or more fluid-gas exchanges (P = 0.037), postoperative visual acuity (P = 0.076), and final lOP (P = 0.10). When stepwise multiple logistic regression analysis was done on the entire group of patients, fluid-gas exchange remained statistically significant (P = 0.0 13) after adjusting for the presence of retinal detachment. No other factor remained significant in the model. Because only 2% of
Table 1. Univariate Analysis of Variables Associated with Postoperative Iris Neovascularization
Vitreous hemorrhage PVR Retinotomy No. of fluid-gas exchanges Perfluorocarbon liquid Retinal detachment (anterior and/or total)
Low Risk
No. with IN/ Total(%)
Absent C2-D1 Absent
12/108 (11.1%)* 5/64 (7.8%) 9/102 (8.8%)
1 or 2 Used Absent
High Risk
No. with IN/ Total(%)
PValue
Odds Ratio
Present D2-D3 Present
7/25 (28%)* 14/69 (20.3%) 10/31 (32.3%)
0.051 0.049 0.0026
3.11 3.00 4.92
6/77 (7.8%) 2/41 (4.9%)
23 Not used
11/53 (20.8%) 15/89 (16.9%)
0.037 0.091
3.10 3.95
2/95 (2.1%)
Present
17/37 (45.9%)
1.8 x 10-9
39.53
IN = iris neovascularization; PVR = proliferative vitreoretinopathy. Total = number of patients in that risk category. • Of 108 patients with vitreous hemorrhage absent preoperatively (low risk), 12 developed iris neovascularization postoperatively (11.1%). Seven of the 25 patients with vitreous hemorrhage present preoperatively (high risk) developed iris neovascularization postoperative (28%). This was statistically significant (P = 0.051).
900
Comaratta et al · Iris Neovascularization in Proliferative Vitreoretinopathy patients without anterior (and/or total) detachment developed postoperative iris neovascularization, further analysis was performed separately on those patients with anterior (and/or total) detachment. Fluid-gas exchange remained the most significant variable univariately (P = 0.013) followed by the absence ofperfluorocarbon liquid (P = 0.101 ), and use of retinotomy (P = 0.101) intraoperatively (Table 2). Multivariate logistic regression analysis of those patients with anterior (and/or total) detachment revealed that fluid-gas exchange, perfluorocarbon liquid, and retinotomy were only slightly more predictive of postoperative iris neovascularization (P = 0.0006 by the log likelihood ratio chi-squared statistic) than fluid-gas exchange alone (P = 0.013). The use of 3 or more fluid-gas exchanges was the second most predictive factor of postoperative iris neovascularization. Postoperative iris neovascularization developed in 71% of those patients who had both significant postoperative retinal detachment and 3 or more fluid-gas exchanges. Several factors we believed might be associated with postoperative iris neovascularization were not found to be statistically significant (Tables 3A and B). Preoperative factors such as anterior PVR, giant retinal tears, and trauma were not found to be associated risk factors. The average number of scleral buckles and vitrectomies, separately and combined, were only slightly greater in the patients with postoperative iris neovascularization. Revision of the scleral buckle did not correlate with the development of postoperative iris neovascularization. The amount of intraoperative and postoperative photocoagulation was not significantly greater in patients in whom postoperative iris neovascularization developed. Final lens status was not a risk factor. As illustrated in the case report, neovascularization of the retinal surface or optic disc was not present in any of the eyes that developed iris neovascularization. We divided patients with iris neovascularization into two groups for further data review (Table 4). Group 1 (preoperative iris neovascularization) consisted of the eight patients who presented with iris neovascularization on the initial visit to our office. Seven of these eight patients had previously undergone unsuccessful scleral buckling surgery. In three cases, this was followed by vitrectomy. Of the eight eyes,
four had complete regression of the iris neovascularization after successful reattachment of the retina. This occurred as quickly as 2 weeks postoperatively, and by 4 months in the other 3 eyes. The four remaining eyes had no regression of iris neovascularization after surgical intervention. All four of these eyes ultimately developed total retinal detachment with very poor visual acuity. Factors that were significantly associated with the development of iris neovascularization in the 133 patients of primary analysis (no preoperative iris neovascularization) also showed similar patterns when repeat analysis included these 8 patients. The only additional risk factor was preoperative iris neovascularization, which was a predisposing factor univariately, but not multivariately, when corrected for persistent retinal detachment postoperatively. Group 2 consisted of the 19 eyes that developed iris neovascularization postoperatively. Iris neovascularization was noted as early as 2 weeks and delayed as long as 7 months after surgery. The average onset was at approximately 3 months. We further subdivided this group of eyes by lOP. Group 2A included 13 patients with iris neovascularization and an lOP greater than 5 mmHg. Group 2B consisted of 6 patients noted with iris neovascularization and hypotony (lOP < 5 mmHg). A statistical comparison between the patients in these two groups was limited by the small sample size. We did not identify any statistically significant discriminating factors. However, there are a few trends worth noting. The presence of preoperative anterior PVR or intraoperative retinotomy did not appear to influence the development of hypotony. Retinal detachment was more prevalent in the group developing hypotony. Only 2 eyes in group 2A (n = 13) had total retinal detachment. Five of six eyes in group 2B were found to have a total retinal detachment. The visual acuity results paralleled the anatomic success rate of the two groups. Only 1 patient in group 2B had a visual acuity of better than light perception. Much better visual acuity was achieved by the majority of patients in group 2A and by those patients in group 1 who had successful reattachment. The final lOP was below 21 mmHg in 25 of 27 eyes with iris neovascularization. None of the eyes developed acute neovascular glaucoma. One eye developed slowly
Table 2. Univariate Analysis of Variables Associated with Postoperative Iris Neovascularization in Patients with Retinal Detachment Low Risk No. of fluid-gas exchanges Perfluorocarbon liquids Retinotomy
1 or 2 Used Absent
No. with IN/ Total(%)
5/20 (25%)* 1/8 (12.5%) 8/23 (34.8%)
High Risk
2':3 Not med Present
No. with IN/ Total(%)
PVa1ue
Odds Ratio
10/14 (71.4%)* 14/27 (51.9%) 9/14 (64.3%)
0.013 0.101 0.101
7.5 7.54 3.38
IN = iris neovascularization. Total= Number of patients in that risk category. • Of 20 patients having 1 or 2 fluid-gas exchanges (low risk), only 5 developed iris neovascularization postoperatively (25%). Ten of 14 patients having 3 or more fluid-gas exchanges (high risk) developed postoperative iris neovascularization (71.4%). This was statistically significant (P = 0.013).
901
Ophthalmology
Volume 99, Number 6, June 1992
Table 3A. Variables Not Associated with Postoperative Iris Neovascularization Without Postoperative Iris Neovascularization Age (yrs) Initial lOP (rnmHg) Initial VA No. SB* No. VXt No. Totalt No. Laser§
With Postoperative Iris Neovascularization
Mean
No.
Standard Error
Mean
No.
Standard Error
48.63 10.73 5.84 1.62 1.74 2.84 353.04
102 94 100 114 114 114 68
2.1 0.73 0.12 0.08 0.08 0.12 24.47
55.39 11 5.77 1.82 2.05 3.16 384.71
18 17 13 19 19 19 14
4.06 2.04 0.47 0.2 0.21 0.29 45.17
lOP = intraocular pressure; VA = visual acuity; SB = scleral buckle; VX = vitrectomy. • Number of primary or revision scleral buckles.
t t
Number of primary of revision vitrectomies. Total number of operations.
§ Total number of laser burns.
progressive chronic angle closure glaucoma, which required therapeutic ultrasound. One eye developed pupillary block glaucoma with subsequent secondary chronic angle closure, requiring ciliary body endophotocoagulation. Four eyes in group 2A developed postoperative lOP rises (believed to be secondary to expansion of the intraocular gas tamponade), requiring temporary medical treatment with either oral carbonic anhydrase inhibitors and/or topical beta blockers. Only 4 eyes required chronic topical antiglaucoma medication, with a maintenance of a normal lOP (8 to 21 mmHg) in 3 eyes.
Discussion Traditionally, the presence of iris neovascularization in PVR had been believed to be associated with the development of a phthisical or prephthisical ocular state after failure to reattach the retina at the time of surgery. 8 Our Table 3B. Variables Not Associated with Postoperative Iris Neovascularization No. of Patients
Trauma GRT Division SB* Silicone Oil Anterior PVR
Without Postoperative Iris N eovascularization (n = 114)
With Postoperative Iris N eovascularization (n = 19)
21 (18.4%) 22 (19.2%) 33 (28.9%) 8(7%) 54 (47.3%)
3 (15.8%) 3 (15.8%) 7 (36.8%) 1 (5.2%) 13 (68.6%)
ORT = giant retinal tear; SB = scleral buckle; PVR = proliferative vitreoretinopathy. • Revision scleral buckle performed on referral.
902
results show a much higher prevalence and incidence of iris neovascularization than we had previously believed to occur. The development of iris neovascularization preoperatively or postoperatively was not necessarily a predictor of a poor anatomic and/ or visual result after surgery. Although the retinal reattachment rate and final visual acuity were worse in those patients in whom iris neovascularization developed, they were not as poor as one might have expected. One eye had a final visual acuity of20/50 with regression of the iris neovascularization. None of the eyes developed intractable neovascular glaucoma, and only 3 eyes required chronic topical glaucoma medication to maintain a normal lOP (8 to 21 mmHg). One additional patient required therapeutic ultrasound and medical therapy to maintain an lOP of 30 mmHg. This contrasts with a reported incidence of neovascular glaucoma in up to 75% of patients developing iris neovascularization from all causes. 2 The relatively normal lOPs noted in these patients with compromised filtering mechanisms may be due in part to posterior outflow secondary to chronic residual retinal detachment and/or exposed RPE. The lack of surface neovascularization of the retina and optic disc and the absence of capillary nonperfusion on fluorescein angiography suggest that retinal ischemia may not play a significant role in the development of iris neovascularization in these patients. This, in conjunction with the failure of any of these eyes to develop intractable neovascular glaucoma, leads us to conclude that panretinal photocoagulation is not indicated in these patients. Surprisingly, the iris neovascularization was noted to disappear clinically in four patients after successful reattachment of the retina without panretinal photocoagulation. The development of postoperative iris neovascularization in patients with nondiabetic PVR is strongly correlated with the presence of residual retinal detachment postoperatively. This is well illustrated by noting that preoperative iris neovascularization (group 1 patients) was a risk factor for postoperative iris neovascularization only in the presence of persistent retinal detachment postop-
\Q
8
2
CF 15
50 16
16
CF 16
200 NA
16
8
NA
+
4 1 150 318
3
6
17
+
+
38
+
20
+
+
+ NA
20
23
NA
+
40
4 1 NA 417
+
25
L E
+
+
B
A
R
E
3 408
4
1 3 4
+ 8
+ 8
28
23
A B L E
R
NA
+
27
+
+
26
+ 6
20
+
31
+
2 NA
30
+ 12
24
+ 12
40
+ 20
12
12
+
+ +
+
5 757
16
+
12
+
lE 208 R A B L 18 E
NA
+
NA
+
4 NA
NA
+
10
2 121
3
4 0 2 4
p
2
1 3
N
C3
1
C2
1 3
+
3 256
+
2 3 4
D3
17 NA HM HM
8
10
0
16
17
9
10
8
12
49
10
15
7
6
6
5
5
0
0
0
2
0
0
TD NLP
NA
::r +
Number of primary or revision vitrectomies.
H Final visual acuity: 50 = 20/50; 200 = 20/200; 400 = 20/400.
11 Time (in weeks) of initial onset of iris neovascularization. •• Time (in weeks) until regression of iris neovascularization. tt Final retinal status: A = complete retinal attachment; AD = detached anterior to the buckle; PD = detached posterior to the buckle; TD = total retinal detachment.
II Number of laser burns
t Total number of operations. § Number of fluid-gas exchanges.
t
!
~
~.
0
~
~. ~ < ......
'"I
(t'
g. ......
'"I
'"1::J
§
~.
...... N
()
se;
~
~
z
'"I ......
........
~
~
....
~
....~
~
~
0
(")
NA
NA
4
1
3
+
D2
3 NA
6 NA
5 64
4 73
+
D3
+
2 NA
10
3 18
HM
2 28
+
D2
NA NA
1 58
+
4 3 2 3 3 3 443 293 417 448 649 425
+
2 3 3
E
2 1 2
+
2 276
+
1 3 4
D2
D3
+
2 2 2
+
D2
+
+
p
0
3
2 2 3
+
D2
16 HM
LP
+
+
20
3 268
+
6
4
0
+
2
N p
3 3
3 3
3 1
1 2
+
D2
13 65
5
12 51
uu~~~~a~~~a~~~~~~~uuuu
+ +
+
+
1 230
2
4 2 860 799
+
2 3
4 2
+
C3
8 LP
11 75
Group 2B( Hypotony)
A A A A mmmmmmA mAmmmmm mmmmmmm m
8
22
16
+
+
+
2 2 217 509
+
3 3
1 2
2 4
1 4
+
2
1
2
4
+ I N
+
+
+
+
D3 D2
+
+
+
LP
+
10 49
6 LP
9 43
14 39 400 NA
8 71
9 NA
7 56
7
6 63
2
+
+
+
+
D3
D3
D3
D3
Dl
D3
D1
+
D3
D3
D2
C2
D1
40 NA
2 LP
28 LP
7 22 10 LP HM HM
9 9 CF HM
D3
13 10 CF HM
10 LP
5 13 NA HM
+
+
+
+
+
+
+
5 59
4 24
3 58
2 74
+
68
8 68
7 72
6 12
5 14
4 59
3 76
2
18
56
1
Group 2A (Postoperative Iris Neovascularization lOP > 5)
lOP = intraocular pressure; VA = visual acuity; CF = counting fingers; HM = hand motions; LP = light perception; NA = not available; VH = vitreous hemorrhage; PVR = proliferative vitreoretinopathy; SB = scleral buckle; VX = vitrectomy; FOX = fluid-gas exchange; PFC = perfluorocarbon liquid; NLP = no light perception. • Number of primary or revision scleral buckles.
Iris neovascularization postoperatively Onset (wks)1! Regression•• Retinatt Final YAH Final lOP
VXt Totalt Revision SB FOX§ Laserll PFC Retinotomy Maximum postoperative lOP
Patient no. Age (yrs) Preoperative Iris Neovascularization Initial lOP Initial VA VH PVR Anterior PVR SB*
Group 1 (Preoperative Iris. Neovascularization)
Table 4. Patients with Iris Neovascularization
Ophthalmology
Volume 99, Number 6, June 1992
eratively. In addition, only 2% (n = 95) of patients without anterior and/or total detachment developed postoperative iris neovascularization. It has been shown in animal models that lensectomy, vitrectomy, and creation of a rhegmatogenous retinal detachment can induce iris neovascularization.6 The pathophysiologic mechanism inducing iris neovascularization in the presence of retinal detachment is not clear. Experimental studies have implicated RPE cells as modulators of ocular neovascularization via the release of both inhibitors and stimulators of endothelial cell proliferation (Connor and Glaser, unpublished data). The breakdown of the blood-ocular barrier allows passage of serum factors into the vitreous cavity. These factors may then interact with dispersed RPE cells to promote neovascularization. Hernandez and Blumenkranz (unpublished data) observed that, in rabbits, lensectomy, vitrectomy, and retinectomy (> 90% removal) resulted in florid rubeosis and suggested that vasoactive substances from the RPE and ciliary body may play an important role in iris neovascularization. The eyes with postoperative iris neovascularization displayed a tendency toward a larger amount of RPE exposure and blood-ocular barrier breakdown via an increased presence of preoperative vitreous hemorrhage, more severe PVR, increased incidence ofretinectomy, retinopexy, and residual anterior retinal traction and detachment. The use of perfluorocarbon liquid intraoperatively appeared to lower the risk of postoperative iris neovascularization. Perfluorocarbon liquids allow for improved epiretinal membrane dissection and release of traction at the vitreous base. This results in a lower rate of relaxing retinectomies and retinotomies, residual or recurrent anterior proliferation, and retinal detachment postoperatively-all noted risk factors for postoperative iris neovascularization. Our finding that anterior detachment has a significant association with the development of iris neovascularization is consistent with the findings of Elner et al 8 in their report on the pathology of anterior PVR. They noted a 57% incidence offibrovascular tissue originating from the iris in these eyes. In this study, 52% of the eyes with postoperative retinal traction and detachment limited to the anterior retina developed iris neovascularization. Ten of 19 eyes (52%) with postoperative iris neovascularization had retinal detachment limited to the anterior portion of the scleral buckle. Postoperative anterior proliferation and detachment may be related to postoperative iris neovascularization only by the increased amount of RPE exposure and blood-ocular barrier breakdown it produces. This may be enhanced by the increased use ofretinectomy and retinopexy in those patients. An alternative mechanism may be that recurrent anterior proliferation produces traction on the ciliary body and subsequent ischemia, resulting in the formation of iris neovascularization. An additional factor that may play a role in the development of iris neovascularization, but which was not revealed by our statistical analysis, is anterior segment ischemia. A broad, high encircling element, commonly used to support the vitreous base, is a well-recognized cause of impaired circulation to the anterior segment. 5 Cohen et al 3 recently reported a case ofrubeosis iridis and
904
peripheral retinal neovascularization after a circular buckling operation. The broad base of the buckle has the capacity to impinge on the vortex veins, which experimentally has produced iris neovascularization in the primate eye. 9 A posteriorly placed encircling element may compromise the vortex vein system, producing anterior segment ischemia in some of those eyes developing postoperative iris neovascularization. Extensive ablation of the retina anterior to the high, broad encircling buckle probably contributes to the ischemic state. Photocoagulation, both intraoperatively and postoperatively, was not found to be statistically greater in the iris neovascularization group. This may be misleading because of an inability to document the amount of cryopexy or photocoagulation performed before referral to our office. Retinopexy also may contribute to the formation of iris neovascularization by the liberation of RPE cells and breakdown of the blood-ocular barrier. Acutely elevated lOP, after expansile fluid-gas exchange, may promote neovascularization by compromising the anterior vascular system. The performance of three or more fluid-gas exchanges was the second most significant risk factor for developing postoperative iris neovascularization. An alternative explanation for the significance of fluid-gas exchange is its disruption of the blood-ocular barrier and promotion of reproliferation, especially anteriorly. It is our conclusion that the pathophysiologic mechanism of iris neovascularization in PVR is multifactorial in nature. Analysis of the two groups of eyes with iris neovascularization previously described serves to illustrate this point (Table 4). The resolution of preoperative iris neovascularization (group 1) after successful and complete reattachment of the retina emphasizes the importance of limiting RPE exposure and dispersion and of restoring the integrity of the blood-ocular barrier. Failure of regression of iris neovascularization in the remaining four eyes can be linked primarily to failure to reattach the retina. We divided group 2 eyes into two subgroups, 2A and 2B, on the basis ofiOP. We believe the causative mechanisms of the iris neovascularization may be somewhat different. In group 2A (normal lOP), the prevalence ofretinotomy and residual detachment was suspected of promoting iris neovascularization but was not related to the development of hypotony. The high rate of anterior traction and detachment in this group suggests anterior proliferation may play a role. Multiple factors discussed previously may have contributed to a relative state of anterior segment ischemia in these patients. We suggest that the severe hypotony in all 6 eyes in group 2B represents a more severe form of anterior segment ischemia resulting in ciliary body dysfunction and the development of iris neovascularization. The alterative mechanism for the hypotony and iris neovascularization in this group is the failure to reattach the retina at the time of surgery. None of the patients in whom postoperative iris neovascularization developed had all of the characteristics noted previously, but all patients had at least one characteristic and many had two or more. Conversely, many patients in whom iris neovascularization did not develop postoperatively had one or more factors (including retinal
Comaratta et al · Iris Neovascularization in Proliferative Vitreoretinopathy detachment) associated with postoperative iris neovascularization. This suggests a multifactorial process requiring multiple variables acting in concert to induce the development of iris neovascularization.
References l. Schulze RR. Rubeosis iridis. Am J Ophthalmol 1967;63:
487-95. 2. Anderson DM, Morin JD, Hunter WS. Rubeosis iridis. Can J Ophthalmol 1971 ;6: 183-8. 3. Cohen S, Kremer I, Yassur Y, Ben-Sira I. Peripheral retinal neovascularization and rubeosis iridis after bilateral circular buckling operation. Ann Ophthalmol 1988;20:153-6.
4. Gartner S, Henkind P. Neovascularization of the iris (rubeosis iridis). Surv Ophthalmol1978;22:291-312. 5. Ansem RPJ, Bastiaensen LAK. Glaucoma following retinal detachment operations. Doc Ophthalmol 1987;67: 19-24. 6. Stefansson E, Landers MB III, Wolbarsht ML, Klintworth GK. Neovascularization of the iris: an experimental model in cats. Invest Ophthalmol Vis Sci 1984;25:361-4. 7. The Retina Society Terminology Committee. The classification of retinal detachment with proliferative vitreoretinopathy. Ophthalmology 1983;90: 121-5. 8. Elner SG, Elner VM, Diaz-Rohena R, et al. Anterior proliferative vitreoretinopathy: clinicopathologic, light microscopic, and ultrastructural findings. Ophthalmology 1988;95: 1349-57. 9. Hayreh SS, Baines JAB. Occlusion of the vortex veins. An experimental study. Br J Ophthalmol 1973;57:217-38.
905
Purpose: The purpose of this study is to report on the prevalence, incidence, and associated risk factors of iris neovascularization in nondiabetic patients undergoing vitrectomy for retinal detachment complicated by proliferative vitreoretinopathy (PVR). Methods: The authors conducted a retrospective review of 141 consecutive nondiabetic patients undergoing vitrectomy for recurrent retinal detachment resulting from PVR. Univariate and multivariate analyses were performed on all patients to determine which preoperative, intraoperative, and postoperative factors were associated with the development of postoperative iris neovascularization. Results: Twenty-seven of the 141 (19%) patients were noted with preoperative and/or postoperative iris neovascularization. Four of eight patients presenting with preoperative iris neovascularization had complete regression after successful reattachment of the retina. Results of analysis of the remaining 133 patients without iris neovascularization preoperatively showed residual retinal detachment as the most significant risk factor for postoperative iris neovascularization. In the absence of panretinal photocoagulation, none of the 27 patients developed neovascular glaucoma. Conclusions: The development of iris neovascularization preoperatively or postoperatively is not necessarily a predictor of a poor anatomic and/or visual result. Iris neovascularization in PVR rarely if ever progresses to neovascular glaucoma. Pan retinal photocoagulation is not indicated in these patients. Retinal reattachment is the most important factor in the prevention and/or resolution of postoperative iris neovascularization. The development of iris neovascularization in PVR appears to be a multifactorial process requiring multiple variables acting in concert. Ophthalmology 1992;99:898-905
Iris neovascularization is a well-known complication of proliferative diabetic retinopathy and venous occlusive disease. Retinal detachment has been noted by several authors to be complicated by iris neovascularization. 1•2 It also has been reported as a complication of scleral buckling surgery. 3- 5 The combination of lensectomy, vitrectomy, and rhegmatogenous retinal detachment has been shown to produce iris neovascularization in experimental animal models. 6 Originally received: June 27, 1991. Revision accepted: January 23, 1992. From the Department of Ophthalmology, The New York HospitalCornell University Medical Center, New York. Supported by the Vitreous Research Fund, Cornell University Medical College, New York, New York. Reprint requests to Stanley Chang, MD, Department of Ophthalmology, The New York Hospital, Starr Pavilion Room 817, 525 E 68th St, New York, NY 1002 I.
898
To our knowledge, no previous reports have examined the presence of iris neovascularization in nondiabetic patients undergoing surgery for retinal detachment complicated by proliferative vitreoretinopathy (PVR). This study reviewed retrospectively 141 nondiabetic patients undergoing vitrectomy for PVR. The prevalence and incidence of iris neovascularization and associated risk factors are reported.
Subjects and Methods We examined the records of 163 patients who underwent vitreous surgery performed by one surgeon (SC) for recurrent retinal detachment resulting from PVR between 1983 and 1989. Each patient's age, sex, and ocular history were recorded. Patients with the following conditions were excluded: (1) retinal vascular disease (i.e., diabetic retinopathy); (2) infectious or inflammatory posterior uveitis;
Comaratta et al · Iris Neovascularization in Proliferative Vitreoretinopathy or (3) intraocular tumor. Retinal detachments were classified as those associated with: ( 1) rhegmatogenous retinal detachment; (2) giant retinal tear; or (3) trauma (blunt or penetrating). The preoperative characteristics were recorded from the findings of the initial eye exam. They included: (1) visual acuity; (2) intraocular pressure (lOP); (3) iris neovascularization; (4) the presence of vitreous hemorrhage; (5) the grade of proliferative vitreoretinopathy (Retina Society Classification Committee); 7 and (6) the presence/absence of anterior proliferative vitreoretinopathy. Details of all initial and secondary surgical procedures performed on referral were recorded. These included: ( 1) the total amount of preoperative and postoperative laser (recorded as number of laser burns) and/ or cryotherapy; (2) the number of fluid-gas exchanges; (3) the use ofperfluorocarbon liquids; (4) the use of relaxing retinotomy or retinectomies; (5) the use of silicone oil; and (6) the use ofintravitreal gas tamponade. Followup data were recorded from the outpatient charts and from the referring ophthalmologist. All patients with less than 3 months offollow-up were excluded from the study. Data were recorded regarding: ( 1) the length of followup; (2) final visual acuity; (3) lOP; (4) final lens status (aphakic, phakic, anterior chamber or posterior chamber intraocular lens); (5) iris neovascularization, including the length of time from onset and/or regression; and (6) anatomic status of the retina (fully attached, partially detached posterior to the scleral buckle, detached anterior to the scleral buckle, total detachment). Patients were categorized as to the presence or absence of iris neovascularization both preoperatively and postoperatively. A univariate analysis was performed on all the patients who presented without iris neovascularization preoperatively to determine which preoperative, intraoperative, and postoperative factors were associated with the development of postoperative iris neovascularization. Factors scored as present or absent were initially tested for a statistically significant association with iris neovascularization by Fisher's exact test for two-by-two contingency tables. Factors that were graded (PVR or visual acuity), counted (number of scleral buckles), or measured on a continuous scale (age or lOP) were initially tested using a Student's two-sample t test. Factors even moderately significant univariately (P < 0.10) were then considered in a stepwise multiple logistic regression analysis. This was done to examine the association of a given variable with iris neovascularization after adjusting for all other variables. The strength of the association between postoperative iris neovascularization formation and a given patient variable was determined by using the relative odds ratio. This was calculated by the odds of developing postoperative iris neovascularization, given the high-risk category of a factor, divided by the odds given the lowrisk category of a factor.
Surgical Method Patients requiring a scleral buckle or revision of a previous scleral buckle underwent the placement of a circumfer-
ential3.2 X 7.5 mm grooved sponge with a #240 silicone encircling band, which was used to support the vitreous base for 360°. A standard three-port vitrectomy infusion system was used. A lensectomy was performed in those cases with extensive anterior proliferation at the vitreous base. All epiretinal membranes were removed as completely as possible, until the retina appeared mobile. Attention was then directed to the periphery and the removal of all anterior proliferation. Liquid perfluorocarbon was occasionally used at this point to stabilize the peripheral retina and flatten the posterior retina, mobilizing subretinal fluid and/or hemorrhage anteriorly. Extensive bimanual anterior dissection was then performed using the light pick and membrane forceps, to release all traction for 360°. It was occasionally necessary to perform relaxing retinotomies and retinectomies in areas where traction remained despite extensive dissection. A fluid-air exchange was then performed with drainage of all subretinal fluid with a soft-tipped flute needle through an existing retinal break or a retinotomy. Endophotocoagulation was then used to surround all identifiable retinal breaks. Two additional rows of photocoagulation were then placed on the anterior edge of the scleral buckle in areas not previously treated. Postoperatively, as the inferior meniscus of the gas bubble began to rise, new or previously unidentified retinal breaks caused reaccumulation of fluid in a dependent fashion. If the retina was mobile, a fluidgas exchange was performed with subsequent photocoagulation to seal all retinal breaks. If reproliferation occurred, resulting in significant tractional retinal elevation, repeat vitrectomy was then performed.
Selected Case Study A 49-year-old man was referred with a diagnosis of recurrent retinal detachment in the right eye. He had previously undergone a single scleral buckling procedure for a retinal detachment 6 months previously. On examination, the visual acuity was hand motions in the right eye (with a -4.25 sphere) and 20/15 in the left (with a -5.00 sphere). Intraocular pressure was 6 mmHg in the right eye and 15 mmHg in the left. Results of slit-lamp examination were unremarkable in both eyes. There was no iris neovascularization noted in either eye. Results of examination of the posterior segment of the right eye showed a vitreous hemorrhage and a total retinal detachment. Examination of the left fundus was normal. The patient underwent lensectomy and vitrectomy surgery in the right eye. After pars plana lensectomy, a dense vitreous hemorrhage was found along with the posterior hyaloid, which was contracted anteriorly with severe circumferential traction. The retinal detachment was a D3 PVR configuration. Extensive membranes were removed from the surface ofthe retina, using the Michel's pick and membrane forceps. Liquid perfluorocarbon was then infused over the optic disc, stabilizing the peripheral retina. Extensive anterior dissection was performed for 360° using the light pick and membrane forceps in an attempt to release all tractional elements. A retinotomy was performed from the 9-o'clock to 12-o'clock position, just posterior to the scleral buckle due to extensive anterior adhesion by membranes and hemorrhage to the peripheral retina. After a fluid-gas exchange, with complete removal of the perfluorocarbon liquid, the retina
899
Ophthalmology
Volume 99, Number 6, June 1992
was flat. Endophotocoagulation was placed for 360° on the scleral buckle. The eye was then flushed with a 20% mixture of C3F8 gas. Six weeks postoperatively, the retina was completely attached posterior to the scleral buckle. There was early anterior reproliferation and mild retinal elevation on the anterior crest of the buckle. A few days later, the patient complained of flashing lights and was re-examined. The very mobile retina was totally detached posterior to the scleral buckle. There was a retinal break noted on the crest of the buckle at the 12-o'clock position. A fluid-gas exchange was performed at that time, followed by argon green laser photocoagulation to surround the retinal break several days later. Three months postoperatively, his lOP was 22 mmHg and the patient was noted with early iris neovascularization at the pupillary margin. Two weeks later, frank rubeosis iridis developed. Results of gonioscopy showed intermittent areas of peripheral anterior synechiae for approximately 270°. The retina was completely attached posterior to the scleral buckle. There was significant reproliferation and retinal elevation anterior to the scleral buckle. Large areas of bare retinal pigment epithelium (RPE) were noted anterior to the scleral buckle from the 9-o'clock to !-o'clock position. There was no neovascularization on the retinal surface or optic disc. Fluorescein angiography at that time showed an irregular choroidal pattern with normal arteriole and venous transit times, and no areas of retinal capillary nonperfusion. The patient was re-examined 3 weeks later. His iris neovascularization had begun to regress and he had developed mild uveal ectropion. He eventually underwent therapeutic ultrasound due to slowly progressive angle closure. His lOP was chronically maintained at approximately 30 mmHg with timolol maleate. His retinal examination was unchanged. The patient's examination results have remained unchanged on subsequent visits.
Results Of the 163 patients undergoing vitrectomy for retinal detachment complicated by PVR, 22 were excluded because of inadequate follow-up. Mean follow-up time was 18
months (range, 3 to 84 months). Twenty-seven of theremaining 141 ( 19%) patients were noted with preoperative and/or postoperative iris neovascularization ranging from fine iris vessels to frank rubeosis iridis with uveal ectropion. Of the 27 patients noted with iris neovascularization, 8 presented to our office with iris neovascularization preoperatively. Nineteen developed iris neovascularization postoperatively. The results of a univariate analysis performed on the 133 patients who presented without iris neovascularization is shown in Table 1. Preoperative factors associated with postoperative iris neovascularization included vitreous hemorrhage (P = 0.051) and severe PVR, grade D2 or D3 (P = 0.049). Of the intraoperative factors, the creation of a retinotomy (P = 0.0026) was associated with postoperative iris neovascularization. The use of perfluorocarbon liquids in the management of anterior PVR was found to be a protective factor against developing postoperative iris neovascularization. Anterior (and/or total) retinal detachment was by far the single most significant postoperative factor associated with postoperative iris neovascularization. The risk was found to be similar in those patients with anterior detachment alone, compared with those with total detachment, so that the two categories were combined for further data analysis. Forty-six percent (17 of 37) of patients with anterior (and/ or total) detachment developed iris neovascularization compared with only 2% (2 of95) of patients without anterior (and/or total) detachment (P = 0.00000001 by Fisher's exact test). Other postoperative factors showing significant association with iris neovascularization by Student's t test included the performance of 3 or more fluid-gas exchanges (P = 0.037), postoperative visual acuity (P = 0.076), and final lOP (P = 0.10). When stepwise multiple logistic regression analysis was done on the entire group of patients, fluid-gas exchange remained statistically significant (P = 0.0 13) after adjusting for the presence of retinal detachment. No other factor remained significant in the model. Because only 2% of
Table 1. Univariate Analysis of Variables Associated with Postoperative Iris Neovascularization
Vitreous hemorrhage PVR Retinotomy No. of fluid-gas exchanges Perfluorocarbon liquid Retinal detachment (anterior and/or total)
Low Risk
No. with IN/ Total(%)
Absent C2-D1 Absent
12/108 (11.1%)* 5/64 (7.8%) 9/102 (8.8%)
1 or 2 Used Absent
High Risk
No. with IN/ Total(%)
PValue
Odds Ratio
Present D2-D3 Present
7/25 (28%)* 14/69 (20.3%) 10/31 (32.3%)
0.051 0.049 0.0026
3.11 3.00 4.92
6/77 (7.8%) 2/41 (4.9%)
23 Not used
11/53 (20.8%) 15/89 (16.9%)
0.037 0.091
3.10 3.95
2/95 (2.1%)
Present
17/37 (45.9%)
1.8 x 10-9
39.53
IN = iris neovascularization; PVR = proliferative vitreoretinopathy. Total = number of patients in that risk category. • Of 108 patients with vitreous hemorrhage absent preoperatively (low risk), 12 developed iris neovascularization postoperatively (11.1%). Seven of the 25 patients with vitreous hemorrhage present preoperatively (high risk) developed iris neovascularization postoperative (28%). This was statistically significant (P = 0.051).
900
Comaratta et al · Iris Neovascularization in Proliferative Vitreoretinopathy patients without anterior (and/or total) detachment developed postoperative iris neovascularization, further analysis was performed separately on those patients with anterior (and/or total) detachment. Fluid-gas exchange remained the most significant variable univariately (P = 0.013) followed by the absence ofperfluorocarbon liquid (P = 0.101 ), and use of retinotomy (P = 0.101) intraoperatively (Table 2). Multivariate logistic regression analysis of those patients with anterior (and/or total) detachment revealed that fluid-gas exchange, perfluorocarbon liquid, and retinotomy were only slightly more predictive of postoperative iris neovascularization (P = 0.0006 by the log likelihood ratio chi-squared statistic) than fluid-gas exchange alone (P = 0.013). The use of 3 or more fluid-gas exchanges was the second most predictive factor of postoperative iris neovascularization. Postoperative iris neovascularization developed in 71% of those patients who had both significant postoperative retinal detachment and 3 or more fluid-gas exchanges. Several factors we believed might be associated with postoperative iris neovascularization were not found to be statistically significant (Tables 3A and B). Preoperative factors such as anterior PVR, giant retinal tears, and trauma were not found to be associated risk factors. The average number of scleral buckles and vitrectomies, separately and combined, were only slightly greater in the patients with postoperative iris neovascularization. Revision of the scleral buckle did not correlate with the development of postoperative iris neovascularization. The amount of intraoperative and postoperative photocoagulation was not significantly greater in patients in whom postoperative iris neovascularization developed. Final lens status was not a risk factor. As illustrated in the case report, neovascularization of the retinal surface or optic disc was not present in any of the eyes that developed iris neovascularization. We divided patients with iris neovascularization into two groups for further data review (Table 4). Group 1 (preoperative iris neovascularization) consisted of the eight patients who presented with iris neovascularization on the initial visit to our office. Seven of these eight patients had previously undergone unsuccessful scleral buckling surgery. In three cases, this was followed by vitrectomy. Of the eight eyes,
four had complete regression of the iris neovascularization after successful reattachment of the retina. This occurred as quickly as 2 weeks postoperatively, and by 4 months in the other 3 eyes. The four remaining eyes had no regression of iris neovascularization after surgical intervention. All four of these eyes ultimately developed total retinal detachment with very poor visual acuity. Factors that were significantly associated with the development of iris neovascularization in the 133 patients of primary analysis (no preoperative iris neovascularization) also showed similar patterns when repeat analysis included these 8 patients. The only additional risk factor was preoperative iris neovascularization, which was a predisposing factor univariately, but not multivariately, when corrected for persistent retinal detachment postoperatively. Group 2 consisted of the 19 eyes that developed iris neovascularization postoperatively. Iris neovascularization was noted as early as 2 weeks and delayed as long as 7 months after surgery. The average onset was at approximately 3 months. We further subdivided this group of eyes by lOP. Group 2A included 13 patients with iris neovascularization and an lOP greater than 5 mmHg. Group 2B consisted of 6 patients noted with iris neovascularization and hypotony (lOP < 5 mmHg). A statistical comparison between the patients in these two groups was limited by the small sample size. We did not identify any statistically significant discriminating factors. However, there are a few trends worth noting. The presence of preoperative anterior PVR or intraoperative retinotomy did not appear to influence the development of hypotony. Retinal detachment was more prevalent in the group developing hypotony. Only 2 eyes in group 2A (n = 13) had total retinal detachment. Five of six eyes in group 2B were found to have a total retinal detachment. The visual acuity results paralleled the anatomic success rate of the two groups. Only 1 patient in group 2B had a visual acuity of better than light perception. Much better visual acuity was achieved by the majority of patients in group 2A and by those patients in group 1 who had successful reattachment. The final lOP was below 21 mmHg in 25 of 27 eyes with iris neovascularization. None of the eyes developed acute neovascular glaucoma. One eye developed slowly
Table 2. Univariate Analysis of Variables Associated with Postoperative Iris Neovascularization in Patients with Retinal Detachment Low Risk No. of fluid-gas exchanges Perfluorocarbon liquids Retinotomy
1 or 2 Used Absent
No. with IN/ Total(%)
5/20 (25%)* 1/8 (12.5%) 8/23 (34.8%)
High Risk
2':3 Not med Present
No. with IN/ Total(%)
PVa1ue
Odds Ratio
10/14 (71.4%)* 14/27 (51.9%) 9/14 (64.3%)
0.013 0.101 0.101
7.5 7.54 3.38
IN = iris neovascularization. Total= Number of patients in that risk category. • Of 20 patients having 1 or 2 fluid-gas exchanges (low risk), only 5 developed iris neovascularization postoperatively (25%). Ten of 14 patients having 3 or more fluid-gas exchanges (high risk) developed postoperative iris neovascularization (71.4%). This was statistically significant (P = 0.013).
901
Ophthalmology
Volume 99, Number 6, June 1992
Table 3A. Variables Not Associated with Postoperative Iris Neovascularization Without Postoperative Iris Neovascularization Age (yrs) Initial lOP (rnmHg) Initial VA No. SB* No. VXt No. Totalt No. Laser§
With Postoperative Iris Neovascularization
Mean
No.
Standard Error
Mean
No.
Standard Error
48.63 10.73 5.84 1.62 1.74 2.84 353.04
102 94 100 114 114 114 68
2.1 0.73 0.12 0.08 0.08 0.12 24.47
55.39 11 5.77 1.82 2.05 3.16 384.71
18 17 13 19 19 19 14
4.06 2.04 0.47 0.2 0.21 0.29 45.17
lOP = intraocular pressure; VA = visual acuity; SB = scleral buckle; VX = vitrectomy. • Number of primary or revision scleral buckles.
t t
Number of primary of revision vitrectomies. Total number of operations.
§ Total number of laser burns.
progressive chronic angle closure glaucoma, which required therapeutic ultrasound. One eye developed pupillary block glaucoma with subsequent secondary chronic angle closure, requiring ciliary body endophotocoagulation. Four eyes in group 2A developed postoperative lOP rises (believed to be secondary to expansion of the intraocular gas tamponade), requiring temporary medical treatment with either oral carbonic anhydrase inhibitors and/or topical beta blockers. Only 4 eyes required chronic topical antiglaucoma medication, with a maintenance of a normal lOP (8 to 21 mmHg) in 3 eyes.
Discussion Traditionally, the presence of iris neovascularization in PVR had been believed to be associated with the development of a phthisical or prephthisical ocular state after failure to reattach the retina at the time of surgery. 8 Our Table 3B. Variables Not Associated with Postoperative Iris Neovascularization No. of Patients
Trauma GRT Division SB* Silicone Oil Anterior PVR
Without Postoperative Iris N eovascularization (n = 114)
With Postoperative Iris N eovascularization (n = 19)
21 (18.4%) 22 (19.2%) 33 (28.9%) 8(7%) 54 (47.3%)
3 (15.8%) 3 (15.8%) 7 (36.8%) 1 (5.2%) 13 (68.6%)
ORT = giant retinal tear; SB = scleral buckle; PVR = proliferative vitreoretinopathy. • Revision scleral buckle performed on referral.
902
results show a much higher prevalence and incidence of iris neovascularization than we had previously believed to occur. The development of iris neovascularization preoperatively or postoperatively was not necessarily a predictor of a poor anatomic and/ or visual result after surgery. Although the retinal reattachment rate and final visual acuity were worse in those patients in whom iris neovascularization developed, they were not as poor as one might have expected. One eye had a final visual acuity of20/50 with regression of the iris neovascularization. None of the eyes developed intractable neovascular glaucoma, and only 3 eyes required chronic topical glaucoma medication to maintain a normal lOP (8 to 21 mmHg). One additional patient required therapeutic ultrasound and medical therapy to maintain an lOP of 30 mmHg. This contrasts with a reported incidence of neovascular glaucoma in up to 75% of patients developing iris neovascularization from all causes. 2 The relatively normal lOPs noted in these patients with compromised filtering mechanisms may be due in part to posterior outflow secondary to chronic residual retinal detachment and/or exposed RPE. The lack of surface neovascularization of the retina and optic disc and the absence of capillary nonperfusion on fluorescein angiography suggest that retinal ischemia may not play a significant role in the development of iris neovascularization in these patients. This, in conjunction with the failure of any of these eyes to develop intractable neovascular glaucoma, leads us to conclude that panretinal photocoagulation is not indicated in these patients. Surprisingly, the iris neovascularization was noted to disappear clinically in four patients after successful reattachment of the retina without panretinal photocoagulation. The development of postoperative iris neovascularization in patients with nondiabetic PVR is strongly correlated with the presence of residual retinal detachment postoperatively. This is well illustrated by noting that preoperative iris neovascularization (group 1 patients) was a risk factor for postoperative iris neovascularization only in the presence of persistent retinal detachment postop-
\Q
8
2
CF 15
50 16
16
CF 16
200 NA
16
8
NA
+
4 1 150 318
3
6
17
+
+
38
+
20
+
+
+ NA
20
23
NA
+
40
4 1 NA 417
+
25
L E
+
+
B
A
R
E
3 408
4
1 3 4
+ 8
+ 8
28
23
A B L E
R
NA
+
27
+
+
26
+ 6
20
+
31
+
2 NA
30
+ 12
24
+ 12
40
+ 20
12
12
+
+ +
+
5 757
16
+
12
+
lE 208 R A B L 18 E
NA
+
NA
+
4 NA
NA
+
10
2 121
3
4 0 2 4
p
2
1 3
N
C3
1
C2
1 3
+
3 256
+
2 3 4
D3
17 NA HM HM
8
10
0
16
17
9
10
8
12
49
10
15
7
6
6
5
5
0
0
0
2
0
0
TD NLP
NA
::r +
Number of primary or revision vitrectomies.
H Final visual acuity: 50 = 20/50; 200 = 20/200; 400 = 20/400.
11 Time (in weeks) of initial onset of iris neovascularization. •• Time (in weeks) until regression of iris neovascularization. tt Final retinal status: A = complete retinal attachment; AD = detached anterior to the buckle; PD = detached posterior to the buckle; TD = total retinal detachment.
II Number of laser burns
t Total number of operations. § Number of fluid-gas exchanges.
t
!
~
~.
0
~
~. ~ < ......
'"I
(t'
g. ......
'"I
'"1::J
§
~.
...... N
()
se;
~
~
z
'"I ......
........
~
~
....
~
....~
~
~
0
(")
NA
NA
4
1
3
+
D2
3 NA
6 NA
5 64
4 73
+
D3
+
2 NA
10
3 18
HM
2 28
+
D2
NA NA
1 58
+
4 3 2 3 3 3 443 293 417 448 649 425
+
2 3 3
E
2 1 2
+
2 276
+
1 3 4
D2
D3
+
2 2 2
+
D2
+
+
p
0
3
2 2 3
+
D2
16 HM
LP
+
+
20
3 268
+
6
4
0
+
2
N p
3 3
3 3
3 1
1 2
+
D2
13 65
5
12 51
uu~~~~a~~~a~~~~~~~uuuu
+ +
+
+
1 230
2
4 2 860 799
+
2 3
4 2
+
C3
8 LP
11 75
Group 2B( Hypotony)
A A A A mmmmmmA mAmmmmm mmmmmmm m
8
22
16
+
+
+
2 2 217 509
+
3 3
1 2
2 4
1 4
+
2
1
2
4
+ I N
+
+
+
+
D3 D2
+
+
+
LP
+
10 49
6 LP
9 43
14 39 400 NA
8 71
9 NA
7 56
7
6 63
2
+
+
+
+
D3
D3
D3
D3
Dl
D3
D1
+
D3
D3
D2
C2
D1
40 NA
2 LP
28 LP
7 22 10 LP HM HM
9 9 CF HM
D3
13 10 CF HM
10 LP
5 13 NA HM
+
+
+
+
+
+
+
5 59
4 24
3 58
2 74
+
68
8 68
7 72
6 12
5 14
4 59
3 76
2
18
56
1
Group 2A (Postoperative Iris Neovascularization lOP > 5)
lOP = intraocular pressure; VA = visual acuity; CF = counting fingers; HM = hand motions; LP = light perception; NA = not available; VH = vitreous hemorrhage; PVR = proliferative vitreoretinopathy; SB = scleral buckle; VX = vitrectomy; FOX = fluid-gas exchange; PFC = perfluorocarbon liquid; NLP = no light perception. • Number of primary or revision scleral buckles.
Iris neovascularization postoperatively Onset (wks)1! Regression•• Retinatt Final YAH Final lOP
VXt Totalt Revision SB FOX§ Laserll PFC Retinotomy Maximum postoperative lOP
Patient no. Age (yrs) Preoperative Iris Neovascularization Initial lOP Initial VA VH PVR Anterior PVR SB*
Group 1 (Preoperative Iris. Neovascularization)
Table 4. Patients with Iris Neovascularization
Ophthalmology
Volume 99, Number 6, June 1992
eratively. In addition, only 2% (n = 95) of patients without anterior and/or total detachment developed postoperative iris neovascularization. It has been shown in animal models that lensectomy, vitrectomy, and creation of a rhegmatogenous retinal detachment can induce iris neovascularization.6 The pathophysiologic mechanism inducing iris neovascularization in the presence of retinal detachment is not clear. Experimental studies have implicated RPE cells as modulators of ocular neovascularization via the release of both inhibitors and stimulators of endothelial cell proliferation (Connor and Glaser, unpublished data). The breakdown of the blood-ocular barrier allows passage of serum factors into the vitreous cavity. These factors may then interact with dispersed RPE cells to promote neovascularization. Hernandez and Blumenkranz (unpublished data) observed that, in rabbits, lensectomy, vitrectomy, and retinectomy (> 90% removal) resulted in florid rubeosis and suggested that vasoactive substances from the RPE and ciliary body may play an important role in iris neovascularization. The eyes with postoperative iris neovascularization displayed a tendency toward a larger amount of RPE exposure and blood-ocular barrier breakdown via an increased presence of preoperative vitreous hemorrhage, more severe PVR, increased incidence ofretinectomy, retinopexy, and residual anterior retinal traction and detachment. The use of perfluorocarbon liquid intraoperatively appeared to lower the risk of postoperative iris neovascularization. Perfluorocarbon liquids allow for improved epiretinal membrane dissection and release of traction at the vitreous base. This results in a lower rate of relaxing retinectomies and retinotomies, residual or recurrent anterior proliferation, and retinal detachment postoperatively-all noted risk factors for postoperative iris neovascularization. Our finding that anterior detachment has a significant association with the development of iris neovascularization is consistent with the findings of Elner et al 8 in their report on the pathology of anterior PVR. They noted a 57% incidence offibrovascular tissue originating from the iris in these eyes. In this study, 52% of the eyes with postoperative retinal traction and detachment limited to the anterior retina developed iris neovascularization. Ten of 19 eyes (52%) with postoperative iris neovascularization had retinal detachment limited to the anterior portion of the scleral buckle. Postoperative anterior proliferation and detachment may be related to postoperative iris neovascularization only by the increased amount of RPE exposure and blood-ocular barrier breakdown it produces. This may be enhanced by the increased use ofretinectomy and retinopexy in those patients. An alternative mechanism may be that recurrent anterior proliferation produces traction on the ciliary body and subsequent ischemia, resulting in the formation of iris neovascularization. An additional factor that may play a role in the development of iris neovascularization, but which was not revealed by our statistical analysis, is anterior segment ischemia. A broad, high encircling element, commonly used to support the vitreous base, is a well-recognized cause of impaired circulation to the anterior segment. 5 Cohen et al 3 recently reported a case ofrubeosis iridis and
904
peripheral retinal neovascularization after a circular buckling operation. The broad base of the buckle has the capacity to impinge on the vortex veins, which experimentally has produced iris neovascularization in the primate eye. 9 A posteriorly placed encircling element may compromise the vortex vein system, producing anterior segment ischemia in some of those eyes developing postoperative iris neovascularization. Extensive ablation of the retina anterior to the high, broad encircling buckle probably contributes to the ischemic state. Photocoagulation, both intraoperatively and postoperatively, was not found to be statistically greater in the iris neovascularization group. This may be misleading because of an inability to document the amount of cryopexy or photocoagulation performed before referral to our office. Retinopexy also may contribute to the formation of iris neovascularization by the liberation of RPE cells and breakdown of the blood-ocular barrier. Acutely elevated lOP, after expansile fluid-gas exchange, may promote neovascularization by compromising the anterior vascular system. The performance of three or more fluid-gas exchanges was the second most significant risk factor for developing postoperative iris neovascularization. An alternative explanation for the significance of fluid-gas exchange is its disruption of the blood-ocular barrier and promotion of reproliferation, especially anteriorly. It is our conclusion that the pathophysiologic mechanism of iris neovascularization in PVR is multifactorial in nature. Analysis of the two groups of eyes with iris neovascularization previously described serves to illustrate this point (Table 4). The resolution of preoperative iris neovascularization (group 1) after successful and complete reattachment of the retina emphasizes the importance of limiting RPE exposure and dispersion and of restoring the integrity of the blood-ocular barrier. Failure of regression of iris neovascularization in the remaining four eyes can be linked primarily to failure to reattach the retina. We divided group 2 eyes into two subgroups, 2A and 2B, on the basis ofiOP. We believe the causative mechanisms of the iris neovascularization may be somewhat different. In group 2A (normal lOP), the prevalence ofretinotomy and residual detachment was suspected of promoting iris neovascularization but was not related to the development of hypotony. The high rate of anterior traction and detachment in this group suggests anterior proliferation may play a role. Multiple factors discussed previously may have contributed to a relative state of anterior segment ischemia in these patients. We suggest that the severe hypotony in all 6 eyes in group 2B represents a more severe form of anterior segment ischemia resulting in ciliary body dysfunction and the development of iris neovascularization. The alterative mechanism for the hypotony and iris neovascularization in this group is the failure to reattach the retina at the time of surgery. None of the patients in whom postoperative iris neovascularization developed had all of the characteristics noted previously, but all patients had at least one characteristic and many had two or more. Conversely, many patients in whom iris neovascularization did not develop postoperatively had one or more factors (including retinal
Comaratta et al · Iris Neovascularization in Proliferative Vitreoretinopathy detachment) associated with postoperative iris neovascularization. This suggests a multifactorial process requiring multiple variables acting in concert to induce the development of iris neovascularization.
References l. Schulze RR. Rubeosis iridis. Am J Ophthalmol 1967;63:
487-95. 2. Anderson DM, Morin JD, Hunter WS. Rubeosis iridis. Can J Ophthalmol 1971 ;6: 183-8. 3. Cohen S, Kremer I, Yassur Y, Ben-Sira I. Peripheral retinal neovascularization and rubeosis iridis after bilateral circular buckling operation. Ann Ophthalmol 1988;20:153-6.
4. Gartner S, Henkind P. Neovascularization of the iris (rubeosis iridis). Surv Ophthalmol1978;22:291-312. 5. Ansem RPJ, Bastiaensen LAK. Glaucoma following retinal detachment operations. Doc Ophthalmol 1987;67: 19-24. 6. Stefansson E, Landers MB III, Wolbarsht ML, Klintworth GK. Neovascularization of the iris: an experimental model in cats. Invest Ophthalmol Vis Sci 1984;25:361-4. 7. The Retina Society Terminology Committee. The classification of retinal detachment with proliferative vitreoretinopathy. Ophthalmology 1983;90: 121-5. 8. Elner SG, Elner VM, Diaz-Rohena R, et al. Anterior proliferative vitreoretinopathy: clinicopathologic, light microscopic, and ultrastructural findings. Ophthalmology 1988;95: 1349-57. 9. Hayreh SS, Baines JAB. Occlusion of the vortex veins. An experimental study. Br J Ophthalmol 1973;57:217-38.
905
