- Email: [email protected]
Intravitreal Long-Acting Gas in the Prevention of Early Postoperative Vitreous Hemorrhage in Diabetic Vitrectomy
Intravitreal Long-Acting Gas in the Prevention of Early Postoperative Vitreous Hemorrhage in Diabetic Vitrectomy Chung-May Yang, MD,1,2 Po-Ting Yeh, MD,1,2,3 Chang-Hao Yang, MD, PhD1,2 Objective: To evaluate the hemostatic effects of intravitreal infusion of 10% C3F8 in patients undergoing diabetic vitrectomy on the occurrence of early postoperative recurrent vitreous hemorrhage. Design: Prospective, randomized, observational case series. Participants: Sixty-one eyes (59 patients) that underwent primary pars plana vitrectomy for complications of proliferative diabetic retinopathy from September 2004 to April 2005, with postoperative retinal reattachment ⱖ 3 months and follow-up ⬎ 6 months were enrolled. Methods: Sixty-one cases were randomly divided into either group 1 (intravitreal 10% C3F8 infusion at the end of surgery) or group 2 (no intravitreal gas). Ultrasound biomicroscopy (UBM) examination of the 3 sclerotomy sites was performed at ⱖ 2 months postoperatively. Demographic data, history, intraoperative findings, and management of recurrent vitreous hemorrhage were recorded. Main Outcome Measures: Initial time to vitreous clearing (ITVC), percentage of prolonged ITVC (⬎ 5 weeks), and early (ⱕ 4 weeks) versus late (⬎ 4 weeks) manifest postoperative recurrent vitreous hemorrhage in groups 1 and 2 were compared to determine the effects of 10% C3F8 on prevention of early recurrent vitreous hemorrhage. Multiple logistic regression analyses were performed to examine risk factors related to early recurrent vitreous hemorrhage. Results: Group 1 ITVC was 13.2⫾9.6 days, and group 2 ITVC was 11.3⫾11.1 days (P ⫽ 0.26). Prolonged ITVC (⬎ 5 weeks) in each group was 1/31 (3.2%) and 2/30 (6.7%; P ⫽ 0.53). Early manifest recurrent vitreous hemorrhage rates in groups 1 and 2 were 0/31 (0%) and 5/30 (16.7%), respectively (P ⫽ 0.02). Early manifest recurrent hemorrhage plus prolonged ITVC in the 2 groups were 1/31 (3.2%) and 7/30 (23.3%), respectively (P ⫽ 0.02). The incidences of elevated intraocular pressure, iris neovascularization, and significant cataract formation among the 2 groups were too low to detect statistical significance. No evidence of fibrovascular ingrowth was found by UBM examination in either group. Multiple logistic regression analyses in non– gas-infused cases showed that an increased extent of membrane peeling raised the possibility of significant early vitreous rebleeding. Conclusions: Intraocular tamponade with 10% C3F8 may be a useful adjunct to vitrectomy for proliferative diabetic retinopathy in the reduction of early postoperative recurrent vitreous hemorrhage. Ophthalmology 2007; 114:710 –715 © 2007 by the American Academy of Ophthalmology.
Recurrent vitreous hemorrhage after vitrectomy for diabetic retinopathy complications is a common occurrence with an incidence of 12% to 63%.1– 4 The hemorrhage may appear within the first few weeks after surgery or months later. This
Originally received: December 9, 2005. Accepted: July 11, 2006. Manuscript no. 2005-1212. 1 Department of Ophthalmology, National Taiwan University Hospital, Taipei, Taiwan. 2 National Taiwan University College of Medicine, Taipei, Taiwan. 3 En Chu Kong Hospital, Taipei, Taiwan. The authors have no financial interest in any materials or equipment used in the study. Correspondence to Chung-May Yang, MD, Department of Ophthalmology, National Taiwan University Hospital, College of Medicine, National Taiwan University. No. 7, Chung Shan South Road, Taipei, Taiwan. E-mail: [email protected].
710
© 2007 by the American Academy of Ophthalmology Published by Elsevier Inc.
complication may delay visual rehabilitation significantly, and sometimes requires additional procedures or surgery. The causes of bleeding are diverse. Although evidence suggests fibrovascular proliferation from the sclerotomy sites or from the vitreous base may be an important source of recurrent vitreous hemorrhage,5– 8 other sources of bleeding exist. These include residual blood in the peripheral vitreous skirt, iatrogenic intraoperative injury of retinal vessels, and incomplete removal of fibrovascular tissues.2– 4 These sites may be major origins of early postoperative vitreous hemorrhage as well. We have reported on the possible benefit of peripheral retinal cryotherapy and cryotherapy treatment of sclerotomy sites to inhibit fibrovascular proliferation along the inner surface of sclerotomy sites and to prevent delayed-onset recurrent vitreous hemorrhage.8 However, many patients still experience disturbing vitreous hemorrhage within the first postoperative month, which causes prolonged poor ISSN 0161-6420/07/$–see front matter doi:10.1016/j.ophtha.2006.07.047
Yang et al 䡠 Intravitreal Long-Acting Gas in Diabetic Vitrectomy vision or even necessitates further surgical procedures. Systemic administration of antifibrinolytic drugs9,10 and intravitreal infusion of short-acting gas11,12 have been used in an attempt to prevent early recurrent vitreous hemorrhage, but have not produced satisfactory results. We hypothesize that a long-acting gas bubble within the vitreous cavity may mechanically tamponade the fragile retinal vessels and concentrate the coagulation factors close to bleeding sites. This action would thus allow the gradual reestablishment of vascular integrity, and reduce the occurrence of early postoperative recurrent vitreous hemorrhage. To investigate this hypothesis, a clinical prospective study was undertaken to examine the effect of a long-acting gas infused into the vitreous cavity at the end of a diabetic vitrectomy to prevent early recurrent vitreous hemorrhage.
Patients and Methods From September 2004 to April 2005, consecutive patients undergoing primary pars plana vitrectomy for complications of proliferative diabetic retinopathy were recruited for the prospective study. Exclusion criteria were as follows: preoperative or postoperative anticoagulant therapy, blood diseases associated with abnormal coagulation, and silicone oil infusion or 15% C3F8 tamponade as a mandatory intraoperative procedure during retina reattachment. Informed consent was obtained from each patient before surgery. The clinical trial was approved by the review board and National Taiwan University Hospital research ethics committee (No. 9461700621). All cases were randomized into 2 groups near the end of surgery. Group 1 had 10% C3F8 infused into the vitreous cavity at the end of surgery whereas group 2 did not receive intravitreal gas. A single surgeon (CMY) performed all of the operations.
Operative Technique During primary pars plana vitrectomy for diabetic retinopathy complications, anterior–posterior traction, fibrovascular tissues, and old and fresh blood were removed as completely and safely as possible. Blood clots adherent to the peripheral vitreous skirt were removed to the best possible extent. Hemostasis was obtained by raising the infusion bottle, mechanical compression using a softtipped cannula, endodiathermy, or a combination of these techniques. Additional blood clots were removed carefully, except from the bleeding sites where they were trimmed to small islands. Panretinal photocoagulation in non–laser-treated eyes or supplementary laser in previously laser-treated eyes extending beyond the level of the equator was performed. The assignment of gas injection was randomly made at this point, which was performed after further peripheral retinal cryotherapy (10 –12 spots in 1 row). Finally, after wound closure, cryotherapy of the sclerotomy sites (2 spots, each 6 seconds, for 3 sclerotomy sites) was performed. After surgery, patients in Group 1 were kept in a prone position overnight, and maintained a head-down position during waking hours. The patients were then allowed to lie on either side during sleep for 3 weeks thereafter. No specific head position was required for group 2 patients. Ophthalmologic examinations were performed in the first 4 days after surgery, then weekly for 4 weeks, biweekly for 1 month, and then monthly for at least 6 months. Ultrasound biomicroscopy (UBM) of the 3 sclerotomy sites was performed 2 to 3 months after surgery in each case to detect any evidence of fibrovascular ingrowth (FVIG). The UBM images for FVIG have been described previously.8
The preoperative, intraoperative, and postoperative data were collected for each patient. These demographics and clinical findings included age, gender, study eye, types and duration of diabetes mellitus, systemic diseases such as hypertension, renal insufficiency (24-hour creatinine clearance estimated by the Cockcroft–Gault equation), intraoperative diagnosis, extent of membrane peeling required, significant intraoperative bleeding, duration of the surgery, combined lens extraction, and the use of scleral buckle. Data regarding the time, duration, frequency and treatment of recurrent vitreous hemorrhage, UBM findings, and the duration of postoperative follow-up were also compiled. Using a modified version of the grading system by Eliott, the extent of fibrovascular proliferation requiring a membrane peel was graded as follows: grade 0, none or focal adhesion ⱕ 3 sites; grade 1, broad adhesion ⱖ 1 site(s) or vitreous-retinal adhesion at the disc, macula, and arcade; or grade 2, vitreous-retinal attachment extending to the periphery or no posterior vitreous detachment.13 Significant intraoperative bleeding was defined as fresh blood clot formation covering half or more of the posterior pole during tissue dissection. Results of ophthalmologic examinations, including best-corrected visual acuity, intraocular pressure, lens status, and intravitreal gas amount, were recorded. We defined initial time to vitreous clearing (ITVC) as the interval between the time when the fundus first lost its detail from dispersion of vitreous blood, if the condition developed within 3 days after surgery, and the time when visualization of retinal vessels regained below the gas bubble (for group 1 cases) or in any part of the fundus (for group 2 cases). Initial time to vitreous clearing of ⱖ5 weeks was considered prolonged. We defined manifest recurrent vitreous hemorrhage as recurrent hemorrhage that obscured the retinal vessels (grade 2 or above in the Diabetic Retinopathy Vitrectomy Study) for ⬎ 10 days after ITVC or after the third postoperative day. Both early (ⱕ4 weeks) and late (⬎4 weeks) manifest recurrent vitreous hemorrhage were recorded. The severity of vitreous hemorrhage was classified according to the scale defined in the Diabetic Retinopathy Vitrectomy Study,14 and was reconfirmed by another ophthalmologist for every patient. The ITVC, the rate and treatment of recurrent vitreous hemorrhage, and the change of bestcorrected visual acuity were compared between groups 1 and 2. Visual acuity was graded into 3 levels: low (ⱕ1 m counting fingers), moderate (⬎1 m counting fingers, but ⬍20/200), and good (ⱖ20/200).10
Statistical Analysis To examine the differences between groups 1 and 2, including patients’ demographic data, surgical findings, additional surgical procedures, surgical outcomes, and frequency of manifest recurrent vitreous hemorrhage plus prolonged ITVC, we performed statistical analysis with chi-square test (if the sample size was ⬍ 5, Fisher exact test was performed). Age, duration of diabetes, surgical time, follow-up time, and ITVC were presented as mean values ⫾ standard deviation. The Wilcoxon rank sum test was performed to make comparisons between groups 1 and 2. To further verify the effect of gas infusion and to examine other possible factors affecting early recurrent vitreous hemorrhage, we performed multivariate logistic regression analysis to determine the significance of the following factors: age (ⱕ 45 years, 45– 60 years, or ⬎ 60 years), gender, duration of diabetes (⬍ 10 years or ⱖ 10 years), hypertension, renal insufficiency, extent of membrane peeling, duration of surgery (⬍ 1 hour, 1–2 hours, or ⱖ 2 hours), and intravitreal gas infusion. Logistic regression analysis was also performed on group 2 data to reveal possible risk factors in eyes without intravitreal gas. All of the statistical analyses were performed using STATA 8.2 software (StataCorp LP, College Station, TX). A P-value ⬍ 0.05 was considered statistically significant.
711
Ophthalmology
Volume 114, Number 4, April 2007
Results
Table 2. Surgical Procedures and Clinical Findings
In our study, group 1 consisted of 31 cases (30 patients) and group 2 of 30 cases (29 patients). After pars plana vitrectomy with or without gas infusion, all study patients had attached retina at the end of a minimum of a 6-month follow-up. Demographic data are summarized in Table 1. Surgical findings and additional surgical procedures are shown in Table 2. Patients’ characteristics were similar between groups 1 and 2; no statistically significant differences were noted between the groups. All cases with systemic hypertension in the 2 groups were under medical control. After gas infusion, ⬎ 75% of the vitreous cavity remained filled with gas 1 day after surgery in all cases. The gas volume reduced to about 30% of the vitreous cavity at 2 weeks after surgery, and was reabsorbed completely in 6 weeks. One case in group 1 had no clearly visible ocular fundus until the postoperative week 7. In group 2, vessel detail was still obstructed by a significant amount of vitreous blood after the week 5 in 2 cases. In both patients, the fundus became clear by the postoperative week 8. The surgical outcome and average ITVC are shown in Table 3. The difference of ITVC between groups 1 and 2 was not statistically significant (13.2⫾9.6 days vs. 11.3⫾11.1 days; P ⫽ 0.26). Early manifest recurrent vitreous hemorrhage did not occur in any of the cases in group 1, but appeared in 5 cases in group 2. The difference between the 2 groups was statistically significant (P ⫽ 0.02). The total number of cases with prolonged ITVC in combination with cases with early manifest recurrent hemorrhage was Table 1. Baseline Features of Pars Plana Vitrectomy Patients with and without 10% C3F8 Intravitreal Gas Infusion Characteristics
Group 1*
Group 2†
Number of eyes Right eye Left eye Age (yrs) (mean ⫾ SD) Gender Male Female Type of DM (n) 1 (IDDM) 2 (NIDDM) Duration of DM (n) ⱕ10 yrs ⬎10 yrs Systemic hypertension (n) Renal insufficiency (n) Lens status (n) Phakic eye Pseudophakic eye Diagnosis (n) Localized TRD Active FVP VH only Duration of follow-up (mean ⫾ SD) (mos)
31 15 16 54.6⫾14.2
30 14 16 57.8⫾12.5
15 16
11 19
4 27
3 27
8 23 18 17
4 26 16 10
21 10
22 8
4 20 7 7.6⫾0.9
3 19 8 7.6⫾0.8
P Value 0.89 0.40 0.36 0.72 0.22 0.71 0.09 0.63 1.00
0.85
DM ⫽ diabetes mellitus; FVP ⫽ fibrovascular proliferation; IDDM ⫽ insulin-dependent diabetes mellitus; NIDDM ⫽ non–insulin-dependent diabetes mellitus; SD ⫽ standard deviation; TRD ⫽ traction retinal detachment; VH ⫽ vitreous hemorrhage. Statistical analysis of age, DM duration, and follow-up between groups 1 and 2 were performed using Wilcoxon rank sum test. Others analyses were performed using chi-square test. If n⬍5, Fisher exact test was performed. P⬍0.05 was considered statistically significant. *Patients who received intravitreal 10% C3F8 gas. † Patients who did not receive gas.
712
Surgical Procedures
Group 1* (n ⴝ 31)
Group 2† (n ⴝ 30) P Value
Extent of membrane peeling [n (%)] Grade 0 5 (16.1) 9 (30.0) Grade 1 19 (61.3) 14 (46.7) Grade 2 7 (22.6) 7 (23.3) Scleral buckle [n (%)] 7 (22.6) 6 (20.0) Lens extraction [n (%)] 1 (3.2) 2 (6.7) Intraoperative bleeding [n (%)] 8 (25.8) 6 (20.0) Duration of surgery [(mean ⫾ SD) 104.8⫾34.6 92.2⫾23.7 (min)]
0.39
0.81 0.53 0.59 0.59
SD ⫽ standard deviation. Statistical comparison between groups were performed using chi-square test (if n⬍5, Fisher exact test was performed). Duration of the surgery was compared by Wilcoxon rank sum test. P⬍0.05 was considered statistically significant. Grading of extent of membrane peeling required was defined as follows: grade 0 ⫽ none or focal adhesion ⱕ 3 sites; grade 1 ⫽ broad adhesion ⱖ 1 sites or vitreous-retinal adhesion at disc, macula, or arcade; grade 2 ⫽ vitreous-retinal attachment extending to the peripheral or no posterior vitreous detachment. Intraoperative bleeding was defined as fresh blood clot formation covering half or more of the posterior pole during tissue dissection. *Patients received intravitreal 10% C3F8 gas. † Patients received no gas.
also statistically different between the 2 groups (P ⫽ 0.02). Rebleeding occurred in postoperative week 2 in 3 cases and week 4 in 2 cases. Two patients were observed with eventually clearing of the media within 2 months. Two patients received vitreous lavage; the other underwent lavage combined with additional cryotherapy. Eight cases in group 1 had a transient increase of intraocular pressure. One case developed acute angle-closure glaucoma after complete reabsorption of gas, which required laser peripheral iridectomy. Five cases in group 2 had a transient increase of intraocular pressure. All cases in this study had negative UBM findings for FVIG. Late manifest recurrent vitreous hemorrhage was noted in 2 patients in group 1 and 3 cases in group 2. In group 1, 1 case occurred in postoperative week 15, and was observed without surgery. The other group 1 case of late manifest recurrent vitreous hemorrhage occurred in postoperative week 9. This patient, who also had rubeosis iridis, received vitreous lavage combined with additional cryotherapy. In group 2, 1 case of late manifest recurrent vitreous hemorrhage occurred in week 10, another in week 15, and the third case in week 22. Iris neovascularization was noted in the patient with hemorrhage in week 10. The patient received vitreous lavage combined with additional cryotherapy. The other 2 group 2 cases were observed without further surgery. The preoperative and postoperative best-corrected visual acuity levels for the 2 study groups are shown in Table 4. There were no statistical significant differences between these 2 groups. The results of logistic regression analysis (either univariate or multivariate analysis) showed that an increased extent of peeling increased vitreous hemorrhage, whereas intravitreal gas infusion decreases the possibility of early postoperative recurrent vitreous hemorrhage (Table 5). Logistic regression analysis in group 2 (n ⫽ 30) to explore possible risk factors for early recurrent vitreous hemorrhage showed that in univariate analysis, age (ⱕ 45 years vs. ⬎ 60 years) and extent of peeling were significant factors associated with early recurrent hemorrhage. However, only the extent of peeling showed statistical significance in multivariate logistic regression analysis (Table 6).
Yang et al 䡠 Intravitreal Long-Acting Gas in Diabetic Vitrectomy Table 3. Surgical Outcome of Pars Plana Vitrectomy Patients with and without Intravitreal 10% C3F8 Gas Infusion Surgical Outcome
Group 1* (n ⴝ 31)
Group 2† (n ⴝ 30)
P Value
Intraocular pressure ⬎ 25 mmHg (ⱖ 1 week) [n (%)] ITVC (days) [mean ⫾ SD] (median) Early (ⱕ 4 weeks) manifest recurrent vitreous hemorrhage [n (%)] Prolonged ITVC (⬎ 5 wks) [n (%)] Early manifest recurrent vitreous hemorrhage ⫹ prolonged ITVC [n (%)] Late (⬎ 4 weeks) manifest recurrent vitreous hemorrhage [n (%)] Iris neovascularization [n (%)] Significant cataract formation [n (%)]
8 (25.8) 13.2 ⫾ 9.6 (14) 0 (0) 1 (3.2) 1 (3.2) 2 (6.5) 1 (3.2) 1 (3.2)
5 (16.7) 11.3 ⫾ 11.1 (7) 5 (16.7) 2 (6.7) 7 (23.3) 3 (10.0) 1 (3.3) 2 (6.7)
0.38 0.26 0.02 0.53 0.02 0.61 0.98 0.53
ITVC ⫽ initial time to vitreous clearing; SD ⫽ standard deviation. P⬍0.05 was considered to be statistically significant. Statistical comparison between groups were performed using chi-square test (if n⬍5, the Fisher exact test was performed) and Wilcoxon rank sum test (for ITVC). *Study patients who received 10% C3F8 gas infusion. † Study patients without gas infusion.
Discussion Early recurrent vitreous hemorrhage after successful operation for diabetic retinopathy can occur in as high as 63% of cases.3 To prevent recurrent vitreous hemorrhage, -aminocaproic acid and tranexamic acid have been used systemically9; thrombin,15,16 sodium hyaluronate,17,18 and silicone oil19,20 have been used intravitreally. Mechanical tamponade of the bleeding sites by air or nonexpansile SF6–air mixture also have been tried for hemorrhage prevention.11,12 In the present study, 10% C3F8 was used intravitreally to obtain better hemostasis. The frequency of manifest early recurrent hemorrhage in the gas-treated group was significantly lower than in the non– gas-treated group. The combined number of cases suffering manifest early recurrent hemorrhage plus prolonged ITVC (⬎ 5 weeks) was also statistically lower in the gas-treated group. The significance persisted in logistic regression analysis for risk factors among all study cases. These results suggested that the incidence of significant early recurrent vitreous hemorrhage was reduced in patients receiving the intravitreal long-acting gas 10% C3F8. Early recurrent hemorrhage may develop soon after surgery and mix with persistent hemorrhage from lysis of residual clots. Hemorrhage may also occur shortly after the vitreous cavity becomes clear. Clinical examinations to differentiate hemorrhage from lysed clots and very early recurrent hemorrhage may be difficult.21 In this study, we did not attempt to distinguish the former from the latter during the first 3 postoperative days. We defined ITVC to
represent either the time to clearing of persistent hemorrhage or the time required for reabsorption of blood from both persistent and very early rebleeding. We compared the ITVC between the 2 groups based on the reasoning that excessive and long ITVC implied significant recurrent vitreous hemorrhage on top of persistent hemorrhage. The criterion for selection of cases with possible significant very early recurrent hemorrhage was ITVC persisting ⬎ 5 weeks. Our clinical experience suggests that with careful manipulation during surgery to minimize retained blood clots, pure persistent hemorrhage seldom lasts ⬎ 1 month. Admittedly, with our study design, early minor recurrent bleeding may escape detection. This limitation in study design is acceptable; our major aim was to investigate clinically meaningful recurrent hemorrhage and its prevention. The fact that we excluded rebleeding lasting ⬍ 10 days as manifest recurrent hemorrhage also relates to this study goal. The overall incidence of manifest recurrent vitreous hemorrhage over at least a 6-month follow-up period in this study was 16.4% (10/61). In the group without intravitreal gas, the incidence was 26.7% (8/30), and in the gas-filled group, manifest recurrent vitreous hemorrhage was a much lower 6.6% (2/31). Our results compare favorably with those in previous reports.1– 4 Aside from our stringent criteria in case selection and in the definition of recurrent hemorrhage, modern surgical techniques to more effectively reduce retinal traction and new vessel formation may also contribute to the low frequency of significant recurrent vitreous hemorrhage. In this study, the UBM examination did not detect evidence of FVIG in any of the cases. One
Table 4. Visual Acuity (VA) Levels at Preoperative and Final Visit in Pars Plana Vitrectomy Patients with and without Intravitreal 10% C3F8 Infusion Preoperative VA
Postoperative VA
VA Level
Low
Moderate
Good
P Value
Low
Moderate
Good
P Value
Group 1* [n (%)] Group 2† [n (%)]
21 (67.7) 22 (73.3)
9 (29.0) 8 (26.7)
1 (3.2) 0 (0)
0.59
8 (25.8) 3 (10.0)
11 (35.5) 13 (43.3)
12 (38.7) 14 (46.7)
0.28
Statistical comparison between groups was performed using the Fisher exact test. P⬍0.05 was considered to be statistically significant. *Study patients who received 10% C3F8 gas infusion. † Study patients without gas infusion.
713
Ophthalmology
Volume 114, Number 4, April 2007
Table 5. Logistic Regression Analyses of Risk Factors Associated with Early Postoperative Recurrent Vitreous Hemorrhage in Study Patients Analysis (n ⴝ 61) Univariate logistic regression Extent of peeling Intravitreal gas infusion Multivariate logistic regression Age (per year) Gender (male vs. female) DM duration (⬍ 10 years vs. ⱖ 10 years) Extent of peeling Renal insufficiency Hypertension Duration of surgery (per hour) Intravitreal gas infusion Multivariate logistic regression Extent of peeling Intravitreal gas infusion
Odds Ratio (95% CI)
P Value
6.74 (1.57–28.87) 0.11 (0.01–0.95)
0.01 0.045
1.01 (0.90–1.15) 3.24 (0.37–28.34) 1.09 (0.07–17.42)
0.82 0.29 0.95
10.64 (1.48–77.50) 0.44 (0.04–4.81) 4.87 (0.32–73.38) 2.73 (0.13–56.87) 0.03 (0.00–0.75)
0.02 0.50 0.25 0.52 0.03
8.81 (1.76–44.2) 0.07 (0.01–0.70)
0.008 0.024
CI ⫽ confidential interval. P⬍0.05 was considered to be statistically significant.
case in each group had rubeosis. This complication was also associated with recurrent vitreous hemorrhage, which suggests ocular or retinal ischemia may be an ongoing factor in late recurrent vitreous hemorrhage despite adequate preventive measures. Our results differ from 2 previous studies evaluating hemostatic effects of intravitreal gas after diabetic vitrectomy.11,12 The difference may be due to the following reasons. The integrity of blood vessels from surgical injuries may take ⬎ 2 to 3 weeks to reestablish. In our study, some recurrent hemorrhage in the control group did not occur until 3 to 4 weeks after surgery; another study reported that early recurrent vitreous hemorrhage could occur after 20 days postoperatively.12 Short-acting intravitreal gas thus may not provide a long-lasting tamponade in such cases. In previous studies, either air or 20% SF6 were used intravitreally,11,12 with the tamponade to the posterior retina lasting no more than 10 days. In contrast, 10% C3F8 was used in this study. With proper head position, the gas provided tamponade to the posterior retina for at least 3 weeks. The posterior retina is the area that may have sustained the most surgical manipulations and may contain vessels susceptible for rebleeding. Additionally, to accurately assess the hemostatic effect of a long-acting gas, other causes of recurrent hemorrhage, such as blood entrapped in the vitreous skirt, residual retinal traction or FVIG at the sclerotomy sites, should be counteracted as thoroughly as possible. Otherwise, the possibility of recurrent vitreous hemorrhage not related to fragile vessels may interfere with a correct assessment of the hemostatic effect of the intravitreal gas. In this study, upon meticulous removal of blood in the vitreous base, release of all possible tractions, and cryotherapy of the anterior retina and 3 sclerotomy sites to prevent FVIG, we believe early recurrent hemorrhage from causes other than breakdown of retinal vessels have been significantly reduced. Finally, it should be noted that very early postoperative vitreous hemorrhage may in fact be blood dispersed
714
from lysed clots or vitreous base, and may not represent actual recurrent bleeding. However, if there was no significant recurrent bleeding, the vitreous should become clear within a relatively short period. In this study, ITVC and manifest recurrent hemorrhage were defined and used as parameters of comparison between 2 groups. With this study design, although minor recurrent bleeding may be ignored, vitreous opacity from persistent hemorrhage was also excluded. Cases with notable and clinically important recurrent fresh hemorrhage may be identified, and the clinical usefulness of a long-acting gas may be better assessed. The exact mechanism of intravitreal long-acting gas to reduce the rate of recurrent vitreous hemorrhage is unclear. The mechanical tamponade effect on the fragile vessels by the gas may be one factor11,12; the biochemical effect of antifibrinolytic substances compartmentalized close to the bleeding sites similar to situations in silicone oil-filled eyes may be another.19,20 Concentrated procoagulants in the limited fluid space unoccupied by the gas may also help maintain the integrity of retinal vessels during the early postoperative period. Complications associated with intravitreal long-acting gas are high intraocular pressure,11 cataract,11,21,22 intravitreal fibrin formation, and retinal breaks.23–25 The 10% C3F8 we used was nonexpansile. Therefore, this gas was less likely to cause a rise in intraocular pressure or induce breaks through stretching of the peripheral vitreous. It was reabsorbed more slowly than nonexpansile 20% SF6, but faster than nonexpansive 15% C3F8. The slow reabsorption allowed the gas mixture to provide a longer tamponade than 20% SF6 and avoids a long delay of visual recovery and high incidence of complicated cataract. Excessive fibrin formation and retinal traction are potential complications secondary to concentration of proliferative factors by gas.20 Careful examination and timely steroid treatment (intravitreal or subtenon route) may reduce inflammation and possible intraocular fibrin.26 In this study, we did not identify significantly more cataract or high intraocular pressure in Table 6. Logistic Regression Analyses of Risk Factors in Association with Early Postoperative Recurrent Vitreous Hemorrhage in Group 2 Study Patients Analysis (n ⴝ 30) Univariate logistic regression Age (yrs) ⱕ 45 vs. ⬎ 60 ⬎ 45 but ⱕ 60 vs. ⬎ 60 Extent of peeling Multivariate logistic regression Age (yrs) Gender (male vs. female) DM duration (⬍ 10 yrs vs. ⱖ 10 yrs) Extent of peeling Renal insufficiency Hypertension Duration of surgery (hr)
Odds Ratio (95% CI)
P Value
0.91 (0.84–1.00) 12.00 (1.25–115.36) 0.67 (0.05–8.55) 17.41 (2.22–136.35)
0.046 0.03 0.76 ⬍0.01
0.94 (0.78–1.13) 3.96 (0.19–69.02) 18.79 (0.33–1075.18)
0.54 0.40 0.16
64.23 (1.14–3634.95) 2.03 (0.11–38.28) 0.12 (0.00–5.83) 1.55 (0.02–133.61)
0.04 0.64 0.28 0.85
CI ⫽ confidential interval; DM ⫽ diabetes mellitus. P⬍0.05 was considered to be statistically significant.
Yang et al 䡠 Intravitreal Long-Acting Gas in Diabetic Vitrectomy the gas-infused group as compared with the control. Additionally, best-corrected visual acuity was compatible between groups 1 and 2. However, the small size of each group and low incidence render the power of the statistics not strong enough. Longer follow-up and larger sample size may show the adverse effect of the addition of intravitreal gas in accelerating the progression of cataract. Because vitrectomy itself may induce cataract formation, and many patients require lens extraction at later times after surgery even without gas infusion, the disadvantage of intravitreal gas in this regard may not be important. In this study, logistic regression analysis was performed on the group without gas infusion in an attempt to determine possible risk factors for recurrent hemorrhage. With this small sample (n ⫽ 30), 2 factors—young age (ⱕ 45 years) and extensive membrane peeling—were identified with a higher risk of significant early recurrent hemorrhage by univariate analysis. Extensive peeling became the only significant factor after multivariate analysis. Although this study was not designed primarily to examine risk factors for early vitreous hemorrhage and a larger sample size may be needed to confirm the results, these findings may provide some guidelines in the determination of who would benefit most from long-acting intravitreal gas infusion. In summary, our small study suggests that intraocular gas tamponade with 10% C3F8 may be a useful adjunct to vitrectomy for proliferative diabetic retinopathy in the reduction of early postoperative vitreous hemorrhage. The study also found that cases undergoing extensive membrane peeling may be at higher risk. Larger sample size with more inclusive recruiting criteria is necessary to definitely determine all possible factors associated with recurrent vitreous hemorrhage.
References 1. Yang CM. Surgical treatment for diabetic retinopathy: 5-year experience. J Formos Med Assoc 1998;97:477– 84. 2. Schachat AP, Oyakawa RT, Michels RG, et al. Complications of vitreous surgery for diabetic retinopathy. II. Postoperative complications. Ophthalmology 1983;90:522–30. 3. Novak MA, Rice TA, Michels RG, et al. Vitreous hemorrhage after vitrectomy for diabetic retinopathy. Ophthalmology 1984; 91:1485–9. 4. West JF, Gregor ZJ. Fibrovascular ingrowth and recurrent haemorrhage following diabetic vitrectomy. Br J Ophthalmol 2000;84:822–5. 5. Bhende M, Agraharam SG, Gopal L, et al. Ultrasound biomicroscopy of sclerotomy sites after pars plana vitrectomy for diabetic vitreous hemorrhage. Ophthalmology 2000;107: 1729 –36. 6. Hershberger VS, Augsburger JJ, Hutchins RK, et al. Fibrovascular ingrowth at sclerotomy sites in vitrectomized diabetic eyes with recurrent vitreous hemorrhage. Ultrasound biomicroscopy findings. Ophthalmology 2004;111:1215–21. 7. Kreiger AE. Wound complications in pars plana vitrectomy. Retina 1993;13:335– 44. 8. Yeh PT, Yang CM, Yang CH, et al. Cryotherapy of the anterior retina and sclerotomy sites in diabetic vitrectomy to prevent recurrent vitreous hemorrhage. An ultrasound biomicroscopy study. Ophthalmology 2005;112:2095–102.
9. de Bustros S, Glaser BM, Michels RG, et al. Effect of epsilonaminocaproic acid on postvitrectomy hemorrhage. Arch Ophthalmol 1985;103:219 –21. 10. Ramezani AR, Ahmadieh H, Ghaseminejad AK, et al. Effect of tranexamic acid on early postvitrectomy diabetic haemorrhage: a randomised clinical trial. Br J Ophthalmol 2005;89:1041– 4. 11. Joondeph BC, Blankenship GW. Hemostatic effects of air versus fluid in diabetic vitrectomy. Ophthalmology 1989;96: 1701– 6, discussion 1706 –7. 12. Koutsandrea CN, Apostolopoulos MN, Chatzoulis DZ, et al. Hemostatic effects of SF6 after diabetic vitrectomy for vitreous hemorrhage. Acta Ophthalmol Scand 2001;79:34 – 8. 13. Eliott D, Lee MS, Abrams GS. Proliferative diabetic retinopathy: principles and techniques of surgical treatment. In: Ryan SJ, Hinton DR, Schachat AP, Wilkinson P, eds. Retina. Vol. 3. Surgical Retina. 4th ed. Philadelphia: Elsevier Mosby; 2006:2413–5. 14. Two-year course of visual acuity in severe proliferative diabetic retinopathy with conventional management. Diabetic Retinopathy Vitrectomy Study (DRVS) report #1. Ophthalmology 1985;92:492–502. 15. Verdoorn C, Hendrikse F. Intraocular human thrombin infusion in diabetic vitrectomies. Ophthalmic Surg 1989;20: 278 –9. 16. Thompson JT, Glaser BM, Michels RG, et al. The use of intravitreal thrombin to control hemorrhage during vitrectomy. Ophthalmology 1986;93:279 – 82. 17. Packer AJ, McCuen BW, 2nd II, Hutton WL, et al. Procoagulant effects of intraocular sodium hyaluronate (Healon) after phakic diabetic vitrectomy: a prospective, randomized study. Ophthalmology 1989;96:1491– 4. 18. Folk JC, Packer AJ, Weingeist TA, et al. Sodium hyaluronate (Healon) in closed vitrectomy. Ophthalmic Surg 1986;17: 299 –306. 19. Bodanowitz S, Kir N, Hesse L. Silicone oil for recurrent vitreous hemorrhage in previously vitrectomized diabetic eyes. Ophthalmologica 1997;211:219 –22. 20. Charles S, Katz A, Wood B. Vitreous Microsurgery. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2002:104 –25. 21. Thompson JT, de Bustros S, Michels RG, et al. Results and prognostic factors in vitrectomy for diabetic tractionrhegmatogenous retinal detachment. Arch Ophthalmol 1987; 105:503–7. 22. Thompson JT, de Bustros S, Michels RG, et al. Results and prognostic factors in vitrectomy for diabetic traction retinal detachment of the macula. Arch Ophthalmol 1987;105:497– 502. 23. Hilton GF, Tornambe PE, Retinal Detachment Study Group. Pneumatic retinopexy: an analysis of intraoperative and postoperative complications. The Retinal Detachment Study Group. Retina 1991;11:285–94. 24. McAllister IL, Meyers SM, Zegarra H, et al. Comparison of pneumatic retinopexy with alternative surgical techniques. Ophthalmology 1988;95:877– 83. 25. Tornambe PE, Hilton GF, Retinal Detachment Study Group. Pneumatic retinopexy: a multicenter randomized controlled clinical trial comparing pneumatic retinopexy with scleral buckling. The Retinal Detachment Study Group. Ophthalmology 1989;96:772– 83. 26. Chalam KV, Malkani S, Shah VA. Intravitreal dexamethasone effectively reduces postoperative inflammation after vitreoretinal surgery. Ophthalmic Surg Lasers Imaging 2003;34: 188 –92.
715
Recurrent vitreous hemorrhage after vitrectomy for diabetic retinopathy complications is a common occurrence with an incidence of 12% to 63%.1– 4 The hemorrhage may appear within the first few weeks after surgery or months later. This
Originally received: December 9, 2005. Accepted: July 11, 2006. Manuscript no. 2005-1212. 1 Department of Ophthalmology, National Taiwan University Hospital, Taipei, Taiwan. 2 National Taiwan University College of Medicine, Taipei, Taiwan. 3 En Chu Kong Hospital, Taipei, Taiwan. The authors have no financial interest in any materials or equipment used in the study. Correspondence to Chung-May Yang, MD, Department of Ophthalmology, National Taiwan University Hospital, College of Medicine, National Taiwan University. No. 7, Chung Shan South Road, Taipei, Taiwan. E-mail: [email protected].
710
© 2007 by the American Academy of Ophthalmology Published by Elsevier Inc.
complication may delay visual rehabilitation significantly, and sometimes requires additional procedures or surgery. The causes of bleeding are diverse. Although evidence suggests fibrovascular proliferation from the sclerotomy sites or from the vitreous base may be an important source of recurrent vitreous hemorrhage,5– 8 other sources of bleeding exist. These include residual blood in the peripheral vitreous skirt, iatrogenic intraoperative injury of retinal vessels, and incomplete removal of fibrovascular tissues.2– 4 These sites may be major origins of early postoperative vitreous hemorrhage as well. We have reported on the possible benefit of peripheral retinal cryotherapy and cryotherapy treatment of sclerotomy sites to inhibit fibrovascular proliferation along the inner surface of sclerotomy sites and to prevent delayed-onset recurrent vitreous hemorrhage.8 However, many patients still experience disturbing vitreous hemorrhage within the first postoperative month, which causes prolonged poor ISSN 0161-6420/07/$–see front matter doi:10.1016/j.ophtha.2006.07.047
Yang et al 䡠 Intravitreal Long-Acting Gas in Diabetic Vitrectomy vision or even necessitates further surgical procedures. Systemic administration of antifibrinolytic drugs9,10 and intravitreal infusion of short-acting gas11,12 have been used in an attempt to prevent early recurrent vitreous hemorrhage, but have not produced satisfactory results. We hypothesize that a long-acting gas bubble within the vitreous cavity may mechanically tamponade the fragile retinal vessels and concentrate the coagulation factors close to bleeding sites. This action would thus allow the gradual reestablishment of vascular integrity, and reduce the occurrence of early postoperative recurrent vitreous hemorrhage. To investigate this hypothesis, a clinical prospective study was undertaken to examine the effect of a long-acting gas infused into the vitreous cavity at the end of a diabetic vitrectomy to prevent early recurrent vitreous hemorrhage.
Patients and Methods From September 2004 to April 2005, consecutive patients undergoing primary pars plana vitrectomy for complications of proliferative diabetic retinopathy were recruited for the prospective study. Exclusion criteria were as follows: preoperative or postoperative anticoagulant therapy, blood diseases associated with abnormal coagulation, and silicone oil infusion or 15% C3F8 tamponade as a mandatory intraoperative procedure during retina reattachment. Informed consent was obtained from each patient before surgery. The clinical trial was approved by the review board and National Taiwan University Hospital research ethics committee (No. 9461700621). All cases were randomized into 2 groups near the end of surgery. Group 1 had 10% C3F8 infused into the vitreous cavity at the end of surgery whereas group 2 did not receive intravitreal gas. A single surgeon (CMY) performed all of the operations.
Operative Technique During primary pars plana vitrectomy for diabetic retinopathy complications, anterior–posterior traction, fibrovascular tissues, and old and fresh blood were removed as completely and safely as possible. Blood clots adherent to the peripheral vitreous skirt were removed to the best possible extent. Hemostasis was obtained by raising the infusion bottle, mechanical compression using a softtipped cannula, endodiathermy, or a combination of these techniques. Additional blood clots were removed carefully, except from the bleeding sites where they were trimmed to small islands. Panretinal photocoagulation in non–laser-treated eyes or supplementary laser in previously laser-treated eyes extending beyond the level of the equator was performed. The assignment of gas injection was randomly made at this point, which was performed after further peripheral retinal cryotherapy (10 –12 spots in 1 row). Finally, after wound closure, cryotherapy of the sclerotomy sites (2 spots, each 6 seconds, for 3 sclerotomy sites) was performed. After surgery, patients in Group 1 were kept in a prone position overnight, and maintained a head-down position during waking hours. The patients were then allowed to lie on either side during sleep for 3 weeks thereafter. No specific head position was required for group 2 patients. Ophthalmologic examinations were performed in the first 4 days after surgery, then weekly for 4 weeks, biweekly for 1 month, and then monthly for at least 6 months. Ultrasound biomicroscopy (UBM) of the 3 sclerotomy sites was performed 2 to 3 months after surgery in each case to detect any evidence of fibrovascular ingrowth (FVIG). The UBM images for FVIG have been described previously.8
The preoperative, intraoperative, and postoperative data were collected for each patient. These demographics and clinical findings included age, gender, study eye, types and duration of diabetes mellitus, systemic diseases such as hypertension, renal insufficiency (24-hour creatinine clearance estimated by the Cockcroft–Gault equation), intraoperative diagnosis, extent of membrane peeling required, significant intraoperative bleeding, duration of the surgery, combined lens extraction, and the use of scleral buckle. Data regarding the time, duration, frequency and treatment of recurrent vitreous hemorrhage, UBM findings, and the duration of postoperative follow-up were also compiled. Using a modified version of the grading system by Eliott, the extent of fibrovascular proliferation requiring a membrane peel was graded as follows: grade 0, none or focal adhesion ⱕ 3 sites; grade 1, broad adhesion ⱖ 1 site(s) or vitreous-retinal adhesion at the disc, macula, and arcade; or grade 2, vitreous-retinal attachment extending to the periphery or no posterior vitreous detachment.13 Significant intraoperative bleeding was defined as fresh blood clot formation covering half or more of the posterior pole during tissue dissection. Results of ophthalmologic examinations, including best-corrected visual acuity, intraocular pressure, lens status, and intravitreal gas amount, were recorded. We defined initial time to vitreous clearing (ITVC) as the interval between the time when the fundus first lost its detail from dispersion of vitreous blood, if the condition developed within 3 days after surgery, and the time when visualization of retinal vessels regained below the gas bubble (for group 1 cases) or in any part of the fundus (for group 2 cases). Initial time to vitreous clearing of ⱖ5 weeks was considered prolonged. We defined manifest recurrent vitreous hemorrhage as recurrent hemorrhage that obscured the retinal vessels (grade 2 or above in the Diabetic Retinopathy Vitrectomy Study) for ⬎ 10 days after ITVC or after the third postoperative day. Both early (ⱕ4 weeks) and late (⬎4 weeks) manifest recurrent vitreous hemorrhage were recorded. The severity of vitreous hemorrhage was classified according to the scale defined in the Diabetic Retinopathy Vitrectomy Study,14 and was reconfirmed by another ophthalmologist for every patient. The ITVC, the rate and treatment of recurrent vitreous hemorrhage, and the change of bestcorrected visual acuity were compared between groups 1 and 2. Visual acuity was graded into 3 levels: low (ⱕ1 m counting fingers), moderate (⬎1 m counting fingers, but ⬍20/200), and good (ⱖ20/200).10
Statistical Analysis To examine the differences between groups 1 and 2, including patients’ demographic data, surgical findings, additional surgical procedures, surgical outcomes, and frequency of manifest recurrent vitreous hemorrhage plus prolonged ITVC, we performed statistical analysis with chi-square test (if the sample size was ⬍ 5, Fisher exact test was performed). Age, duration of diabetes, surgical time, follow-up time, and ITVC were presented as mean values ⫾ standard deviation. The Wilcoxon rank sum test was performed to make comparisons between groups 1 and 2. To further verify the effect of gas infusion and to examine other possible factors affecting early recurrent vitreous hemorrhage, we performed multivariate logistic regression analysis to determine the significance of the following factors: age (ⱕ 45 years, 45– 60 years, or ⬎ 60 years), gender, duration of diabetes (⬍ 10 years or ⱖ 10 years), hypertension, renal insufficiency, extent of membrane peeling, duration of surgery (⬍ 1 hour, 1–2 hours, or ⱖ 2 hours), and intravitreal gas infusion. Logistic regression analysis was also performed on group 2 data to reveal possible risk factors in eyes without intravitreal gas. All of the statistical analyses were performed using STATA 8.2 software (StataCorp LP, College Station, TX). A P-value ⬍ 0.05 was considered statistically significant.
711
Ophthalmology
Volume 114, Number 4, April 2007
Results
Table 2. Surgical Procedures and Clinical Findings
In our study, group 1 consisted of 31 cases (30 patients) and group 2 of 30 cases (29 patients). After pars plana vitrectomy with or without gas infusion, all study patients had attached retina at the end of a minimum of a 6-month follow-up. Demographic data are summarized in Table 1. Surgical findings and additional surgical procedures are shown in Table 2. Patients’ characteristics were similar between groups 1 and 2; no statistically significant differences were noted between the groups. All cases with systemic hypertension in the 2 groups were under medical control. After gas infusion, ⬎ 75% of the vitreous cavity remained filled with gas 1 day after surgery in all cases. The gas volume reduced to about 30% of the vitreous cavity at 2 weeks after surgery, and was reabsorbed completely in 6 weeks. One case in group 1 had no clearly visible ocular fundus until the postoperative week 7. In group 2, vessel detail was still obstructed by a significant amount of vitreous blood after the week 5 in 2 cases. In both patients, the fundus became clear by the postoperative week 8. The surgical outcome and average ITVC are shown in Table 3. The difference of ITVC between groups 1 and 2 was not statistically significant (13.2⫾9.6 days vs. 11.3⫾11.1 days; P ⫽ 0.26). Early manifest recurrent vitreous hemorrhage did not occur in any of the cases in group 1, but appeared in 5 cases in group 2. The difference between the 2 groups was statistically significant (P ⫽ 0.02). The total number of cases with prolonged ITVC in combination with cases with early manifest recurrent hemorrhage was Table 1. Baseline Features of Pars Plana Vitrectomy Patients with and without 10% C3F8 Intravitreal Gas Infusion Characteristics
Group 1*
Group 2†
Number of eyes Right eye Left eye Age (yrs) (mean ⫾ SD) Gender Male Female Type of DM (n) 1 (IDDM) 2 (NIDDM) Duration of DM (n) ⱕ10 yrs ⬎10 yrs Systemic hypertension (n) Renal insufficiency (n) Lens status (n) Phakic eye Pseudophakic eye Diagnosis (n) Localized TRD Active FVP VH only Duration of follow-up (mean ⫾ SD) (mos)
31 15 16 54.6⫾14.2
30 14 16 57.8⫾12.5
15 16
11 19
4 27
3 27
8 23 18 17
4 26 16 10
21 10
22 8
4 20 7 7.6⫾0.9
3 19 8 7.6⫾0.8
P Value 0.89 0.40 0.36 0.72 0.22 0.71 0.09 0.63 1.00
0.85
DM ⫽ diabetes mellitus; FVP ⫽ fibrovascular proliferation; IDDM ⫽ insulin-dependent diabetes mellitus; NIDDM ⫽ non–insulin-dependent diabetes mellitus; SD ⫽ standard deviation; TRD ⫽ traction retinal detachment; VH ⫽ vitreous hemorrhage. Statistical analysis of age, DM duration, and follow-up between groups 1 and 2 were performed using Wilcoxon rank sum test. Others analyses were performed using chi-square test. If n⬍5, Fisher exact test was performed. P⬍0.05 was considered statistically significant. *Patients who received intravitreal 10% C3F8 gas. † Patients who did not receive gas.
712
Surgical Procedures
Group 1* (n ⴝ 31)
Group 2† (n ⴝ 30) P Value
Extent of membrane peeling [n (%)] Grade 0 5 (16.1) 9 (30.0) Grade 1 19 (61.3) 14 (46.7) Grade 2 7 (22.6) 7 (23.3) Scleral buckle [n (%)] 7 (22.6) 6 (20.0) Lens extraction [n (%)] 1 (3.2) 2 (6.7) Intraoperative bleeding [n (%)] 8 (25.8) 6 (20.0) Duration of surgery [(mean ⫾ SD) 104.8⫾34.6 92.2⫾23.7 (min)]
0.39
0.81 0.53 0.59 0.59
SD ⫽ standard deviation. Statistical comparison between groups were performed using chi-square test (if n⬍5, Fisher exact test was performed). Duration of the surgery was compared by Wilcoxon rank sum test. P⬍0.05 was considered statistically significant. Grading of extent of membrane peeling required was defined as follows: grade 0 ⫽ none or focal adhesion ⱕ 3 sites; grade 1 ⫽ broad adhesion ⱖ 1 sites or vitreous-retinal adhesion at disc, macula, or arcade; grade 2 ⫽ vitreous-retinal attachment extending to the peripheral or no posterior vitreous detachment. Intraoperative bleeding was defined as fresh blood clot formation covering half or more of the posterior pole during tissue dissection. *Patients received intravitreal 10% C3F8 gas. † Patients received no gas.
also statistically different between the 2 groups (P ⫽ 0.02). Rebleeding occurred in postoperative week 2 in 3 cases and week 4 in 2 cases. Two patients were observed with eventually clearing of the media within 2 months. Two patients received vitreous lavage; the other underwent lavage combined with additional cryotherapy. Eight cases in group 1 had a transient increase of intraocular pressure. One case developed acute angle-closure glaucoma after complete reabsorption of gas, which required laser peripheral iridectomy. Five cases in group 2 had a transient increase of intraocular pressure. All cases in this study had negative UBM findings for FVIG. Late manifest recurrent vitreous hemorrhage was noted in 2 patients in group 1 and 3 cases in group 2. In group 1, 1 case occurred in postoperative week 15, and was observed without surgery. The other group 1 case of late manifest recurrent vitreous hemorrhage occurred in postoperative week 9. This patient, who also had rubeosis iridis, received vitreous lavage combined with additional cryotherapy. In group 2, 1 case of late manifest recurrent vitreous hemorrhage occurred in week 10, another in week 15, and the third case in week 22. Iris neovascularization was noted in the patient with hemorrhage in week 10. The patient received vitreous lavage combined with additional cryotherapy. The other 2 group 2 cases were observed without further surgery. The preoperative and postoperative best-corrected visual acuity levels for the 2 study groups are shown in Table 4. There were no statistical significant differences between these 2 groups. The results of logistic regression analysis (either univariate or multivariate analysis) showed that an increased extent of peeling increased vitreous hemorrhage, whereas intravitreal gas infusion decreases the possibility of early postoperative recurrent vitreous hemorrhage (Table 5). Logistic regression analysis in group 2 (n ⫽ 30) to explore possible risk factors for early recurrent vitreous hemorrhage showed that in univariate analysis, age (ⱕ 45 years vs. ⬎ 60 years) and extent of peeling were significant factors associated with early recurrent hemorrhage. However, only the extent of peeling showed statistical significance in multivariate logistic regression analysis (Table 6).
Yang et al 䡠 Intravitreal Long-Acting Gas in Diabetic Vitrectomy Table 3. Surgical Outcome of Pars Plana Vitrectomy Patients with and without Intravitreal 10% C3F8 Gas Infusion Surgical Outcome
Group 1* (n ⴝ 31)
Group 2† (n ⴝ 30)
P Value
Intraocular pressure ⬎ 25 mmHg (ⱖ 1 week) [n (%)] ITVC (days) [mean ⫾ SD] (median) Early (ⱕ 4 weeks) manifest recurrent vitreous hemorrhage [n (%)] Prolonged ITVC (⬎ 5 wks) [n (%)] Early manifest recurrent vitreous hemorrhage ⫹ prolonged ITVC [n (%)] Late (⬎ 4 weeks) manifest recurrent vitreous hemorrhage [n (%)] Iris neovascularization [n (%)] Significant cataract formation [n (%)]
8 (25.8) 13.2 ⫾ 9.6 (14) 0 (0) 1 (3.2) 1 (3.2) 2 (6.5) 1 (3.2) 1 (3.2)
5 (16.7) 11.3 ⫾ 11.1 (7) 5 (16.7) 2 (6.7) 7 (23.3) 3 (10.0) 1 (3.3) 2 (6.7)
0.38 0.26 0.02 0.53 0.02 0.61 0.98 0.53
ITVC ⫽ initial time to vitreous clearing; SD ⫽ standard deviation. P⬍0.05 was considered to be statistically significant. Statistical comparison between groups were performed using chi-square test (if n⬍5, the Fisher exact test was performed) and Wilcoxon rank sum test (for ITVC). *Study patients who received 10% C3F8 gas infusion. † Study patients without gas infusion.
Discussion Early recurrent vitreous hemorrhage after successful operation for diabetic retinopathy can occur in as high as 63% of cases.3 To prevent recurrent vitreous hemorrhage, -aminocaproic acid and tranexamic acid have been used systemically9; thrombin,15,16 sodium hyaluronate,17,18 and silicone oil19,20 have been used intravitreally. Mechanical tamponade of the bleeding sites by air or nonexpansile SF6–air mixture also have been tried for hemorrhage prevention.11,12 In the present study, 10% C3F8 was used intravitreally to obtain better hemostasis. The frequency of manifest early recurrent hemorrhage in the gas-treated group was significantly lower than in the non– gas-treated group. The combined number of cases suffering manifest early recurrent hemorrhage plus prolonged ITVC (⬎ 5 weeks) was also statistically lower in the gas-treated group. The significance persisted in logistic regression analysis for risk factors among all study cases. These results suggested that the incidence of significant early recurrent vitreous hemorrhage was reduced in patients receiving the intravitreal long-acting gas 10% C3F8. Early recurrent hemorrhage may develop soon after surgery and mix with persistent hemorrhage from lysis of residual clots. Hemorrhage may also occur shortly after the vitreous cavity becomes clear. Clinical examinations to differentiate hemorrhage from lysed clots and very early recurrent hemorrhage may be difficult.21 In this study, we did not attempt to distinguish the former from the latter during the first 3 postoperative days. We defined ITVC to
represent either the time to clearing of persistent hemorrhage or the time required for reabsorption of blood from both persistent and very early rebleeding. We compared the ITVC between the 2 groups based on the reasoning that excessive and long ITVC implied significant recurrent vitreous hemorrhage on top of persistent hemorrhage. The criterion for selection of cases with possible significant very early recurrent hemorrhage was ITVC persisting ⬎ 5 weeks. Our clinical experience suggests that with careful manipulation during surgery to minimize retained blood clots, pure persistent hemorrhage seldom lasts ⬎ 1 month. Admittedly, with our study design, early minor recurrent bleeding may escape detection. This limitation in study design is acceptable; our major aim was to investigate clinically meaningful recurrent hemorrhage and its prevention. The fact that we excluded rebleeding lasting ⬍ 10 days as manifest recurrent hemorrhage also relates to this study goal. The overall incidence of manifest recurrent vitreous hemorrhage over at least a 6-month follow-up period in this study was 16.4% (10/61). In the group without intravitreal gas, the incidence was 26.7% (8/30), and in the gas-filled group, manifest recurrent vitreous hemorrhage was a much lower 6.6% (2/31). Our results compare favorably with those in previous reports.1– 4 Aside from our stringent criteria in case selection and in the definition of recurrent hemorrhage, modern surgical techniques to more effectively reduce retinal traction and new vessel formation may also contribute to the low frequency of significant recurrent vitreous hemorrhage. In this study, the UBM examination did not detect evidence of FVIG in any of the cases. One
Table 4. Visual Acuity (VA) Levels at Preoperative and Final Visit in Pars Plana Vitrectomy Patients with and without Intravitreal 10% C3F8 Infusion Preoperative VA
Postoperative VA
VA Level
Low
Moderate
Good
P Value
Low
Moderate
Good
P Value
Group 1* [n (%)] Group 2† [n (%)]
21 (67.7) 22 (73.3)
9 (29.0) 8 (26.7)
1 (3.2) 0 (0)
0.59
8 (25.8) 3 (10.0)
11 (35.5) 13 (43.3)
12 (38.7) 14 (46.7)
0.28
Statistical comparison between groups was performed using the Fisher exact test. P⬍0.05 was considered to be statistically significant. *Study patients who received 10% C3F8 gas infusion. † Study patients without gas infusion.
713
Ophthalmology
Volume 114, Number 4, April 2007
Table 5. Logistic Regression Analyses of Risk Factors Associated with Early Postoperative Recurrent Vitreous Hemorrhage in Study Patients Analysis (n ⴝ 61) Univariate logistic regression Extent of peeling Intravitreal gas infusion Multivariate logistic regression Age (per year) Gender (male vs. female) DM duration (⬍ 10 years vs. ⱖ 10 years) Extent of peeling Renal insufficiency Hypertension Duration of surgery (per hour) Intravitreal gas infusion Multivariate logistic regression Extent of peeling Intravitreal gas infusion
Odds Ratio (95% CI)
P Value
6.74 (1.57–28.87) 0.11 (0.01–0.95)
0.01 0.045
1.01 (0.90–1.15) 3.24 (0.37–28.34) 1.09 (0.07–17.42)
0.82 0.29 0.95
10.64 (1.48–77.50) 0.44 (0.04–4.81) 4.87 (0.32–73.38) 2.73 (0.13–56.87) 0.03 (0.00–0.75)
0.02 0.50 0.25 0.52 0.03
8.81 (1.76–44.2) 0.07 (0.01–0.70)
0.008 0.024
CI ⫽ confidential interval. P⬍0.05 was considered to be statistically significant.
case in each group had rubeosis. This complication was also associated with recurrent vitreous hemorrhage, which suggests ocular or retinal ischemia may be an ongoing factor in late recurrent vitreous hemorrhage despite adequate preventive measures. Our results differ from 2 previous studies evaluating hemostatic effects of intravitreal gas after diabetic vitrectomy.11,12 The difference may be due to the following reasons. The integrity of blood vessels from surgical injuries may take ⬎ 2 to 3 weeks to reestablish. In our study, some recurrent hemorrhage in the control group did not occur until 3 to 4 weeks after surgery; another study reported that early recurrent vitreous hemorrhage could occur after 20 days postoperatively.12 Short-acting intravitreal gas thus may not provide a long-lasting tamponade in such cases. In previous studies, either air or 20% SF6 were used intravitreally,11,12 with the tamponade to the posterior retina lasting no more than 10 days. In contrast, 10% C3F8 was used in this study. With proper head position, the gas provided tamponade to the posterior retina for at least 3 weeks. The posterior retina is the area that may have sustained the most surgical manipulations and may contain vessels susceptible for rebleeding. Additionally, to accurately assess the hemostatic effect of a long-acting gas, other causes of recurrent hemorrhage, such as blood entrapped in the vitreous skirt, residual retinal traction or FVIG at the sclerotomy sites, should be counteracted as thoroughly as possible. Otherwise, the possibility of recurrent vitreous hemorrhage not related to fragile vessels may interfere with a correct assessment of the hemostatic effect of the intravitreal gas. In this study, upon meticulous removal of blood in the vitreous base, release of all possible tractions, and cryotherapy of the anterior retina and 3 sclerotomy sites to prevent FVIG, we believe early recurrent hemorrhage from causes other than breakdown of retinal vessels have been significantly reduced. Finally, it should be noted that very early postoperative vitreous hemorrhage may in fact be blood dispersed
714
from lysed clots or vitreous base, and may not represent actual recurrent bleeding. However, if there was no significant recurrent bleeding, the vitreous should become clear within a relatively short period. In this study, ITVC and manifest recurrent hemorrhage were defined and used as parameters of comparison between 2 groups. With this study design, although minor recurrent bleeding may be ignored, vitreous opacity from persistent hemorrhage was also excluded. Cases with notable and clinically important recurrent fresh hemorrhage may be identified, and the clinical usefulness of a long-acting gas may be better assessed. The exact mechanism of intravitreal long-acting gas to reduce the rate of recurrent vitreous hemorrhage is unclear. The mechanical tamponade effect on the fragile vessels by the gas may be one factor11,12; the biochemical effect of antifibrinolytic substances compartmentalized close to the bleeding sites similar to situations in silicone oil-filled eyes may be another.19,20 Concentrated procoagulants in the limited fluid space unoccupied by the gas may also help maintain the integrity of retinal vessels during the early postoperative period. Complications associated with intravitreal long-acting gas are high intraocular pressure,11 cataract,11,21,22 intravitreal fibrin formation, and retinal breaks.23–25 The 10% C3F8 we used was nonexpansile. Therefore, this gas was less likely to cause a rise in intraocular pressure or induce breaks through stretching of the peripheral vitreous. It was reabsorbed more slowly than nonexpansile 20% SF6, but faster than nonexpansive 15% C3F8. The slow reabsorption allowed the gas mixture to provide a longer tamponade than 20% SF6 and avoids a long delay of visual recovery and high incidence of complicated cataract. Excessive fibrin formation and retinal traction are potential complications secondary to concentration of proliferative factors by gas.20 Careful examination and timely steroid treatment (intravitreal or subtenon route) may reduce inflammation and possible intraocular fibrin.26 In this study, we did not identify significantly more cataract or high intraocular pressure in Table 6. Logistic Regression Analyses of Risk Factors in Association with Early Postoperative Recurrent Vitreous Hemorrhage in Group 2 Study Patients Analysis (n ⴝ 30) Univariate logistic regression Age (yrs) ⱕ 45 vs. ⬎ 60 ⬎ 45 but ⱕ 60 vs. ⬎ 60 Extent of peeling Multivariate logistic regression Age (yrs) Gender (male vs. female) DM duration (⬍ 10 yrs vs. ⱖ 10 yrs) Extent of peeling Renal insufficiency Hypertension Duration of surgery (hr)
Odds Ratio (95% CI)
P Value
0.91 (0.84–1.00) 12.00 (1.25–115.36) 0.67 (0.05–8.55) 17.41 (2.22–136.35)
0.046 0.03 0.76 ⬍0.01
0.94 (0.78–1.13) 3.96 (0.19–69.02) 18.79 (0.33–1075.18)
0.54 0.40 0.16
64.23 (1.14–3634.95) 2.03 (0.11–38.28) 0.12 (0.00–5.83) 1.55 (0.02–133.61)
0.04 0.64 0.28 0.85
CI ⫽ confidential interval; DM ⫽ diabetes mellitus. P⬍0.05 was considered to be statistically significant.
Yang et al 䡠 Intravitreal Long-Acting Gas in Diabetic Vitrectomy the gas-infused group as compared with the control. Additionally, best-corrected visual acuity was compatible between groups 1 and 2. However, the small size of each group and low incidence render the power of the statistics not strong enough. Longer follow-up and larger sample size may show the adverse effect of the addition of intravitreal gas in accelerating the progression of cataract. Because vitrectomy itself may induce cataract formation, and many patients require lens extraction at later times after surgery even without gas infusion, the disadvantage of intravitreal gas in this regard may not be important. In this study, logistic regression analysis was performed on the group without gas infusion in an attempt to determine possible risk factors for recurrent hemorrhage. With this small sample (n ⫽ 30), 2 factors—young age (ⱕ 45 years) and extensive membrane peeling—were identified with a higher risk of significant early recurrent hemorrhage by univariate analysis. Extensive peeling became the only significant factor after multivariate analysis. Although this study was not designed primarily to examine risk factors for early vitreous hemorrhage and a larger sample size may be needed to confirm the results, these findings may provide some guidelines in the determination of who would benefit most from long-acting intravitreal gas infusion. In summary, our small study suggests that intraocular gas tamponade with 10% C3F8 may be a useful adjunct to vitrectomy for proliferative diabetic retinopathy in the reduction of early postoperative vitreous hemorrhage. The study also found that cases undergoing extensive membrane peeling may be at higher risk. Larger sample size with more inclusive recruiting criteria is necessary to definitely determine all possible factors associated with recurrent vitreous hemorrhage.
References 1. Yang CM. Surgical treatment for diabetic retinopathy: 5-year experience. J Formos Med Assoc 1998;97:477– 84. 2. Schachat AP, Oyakawa RT, Michels RG, et al. Complications of vitreous surgery for diabetic retinopathy. II. Postoperative complications. Ophthalmology 1983;90:522–30. 3. Novak MA, Rice TA, Michels RG, et al. Vitreous hemorrhage after vitrectomy for diabetic retinopathy. Ophthalmology 1984; 91:1485–9. 4. West JF, Gregor ZJ. Fibrovascular ingrowth and recurrent haemorrhage following diabetic vitrectomy. Br J Ophthalmol 2000;84:822–5. 5. Bhende M, Agraharam SG, Gopal L, et al. Ultrasound biomicroscopy of sclerotomy sites after pars plana vitrectomy for diabetic vitreous hemorrhage. Ophthalmology 2000;107: 1729 –36. 6. Hershberger VS, Augsburger JJ, Hutchins RK, et al. Fibrovascular ingrowth at sclerotomy sites in vitrectomized diabetic eyes with recurrent vitreous hemorrhage. Ultrasound biomicroscopy findings. Ophthalmology 2004;111:1215–21. 7. Kreiger AE. Wound complications in pars plana vitrectomy. Retina 1993;13:335– 44. 8. Yeh PT, Yang CM, Yang CH, et al. Cryotherapy of the anterior retina and sclerotomy sites in diabetic vitrectomy to prevent recurrent vitreous hemorrhage. An ultrasound biomicroscopy study. Ophthalmology 2005;112:2095–102.
9. de Bustros S, Glaser BM, Michels RG, et al. Effect of epsilonaminocaproic acid on postvitrectomy hemorrhage. Arch Ophthalmol 1985;103:219 –21. 10. Ramezani AR, Ahmadieh H, Ghaseminejad AK, et al. Effect of tranexamic acid on early postvitrectomy diabetic haemorrhage: a randomised clinical trial. Br J Ophthalmol 2005;89:1041– 4. 11. Joondeph BC, Blankenship GW. Hemostatic effects of air versus fluid in diabetic vitrectomy. Ophthalmology 1989;96: 1701– 6, discussion 1706 –7. 12. Koutsandrea CN, Apostolopoulos MN, Chatzoulis DZ, et al. Hemostatic effects of SF6 after diabetic vitrectomy for vitreous hemorrhage. Acta Ophthalmol Scand 2001;79:34 – 8. 13. Eliott D, Lee MS, Abrams GS. Proliferative diabetic retinopathy: principles and techniques of surgical treatment. In: Ryan SJ, Hinton DR, Schachat AP, Wilkinson P, eds. Retina. Vol. 3. Surgical Retina. 4th ed. Philadelphia: Elsevier Mosby; 2006:2413–5. 14. Two-year course of visual acuity in severe proliferative diabetic retinopathy with conventional management. Diabetic Retinopathy Vitrectomy Study (DRVS) report #1. Ophthalmology 1985;92:492–502. 15. Verdoorn C, Hendrikse F. Intraocular human thrombin infusion in diabetic vitrectomies. Ophthalmic Surg 1989;20: 278 –9. 16. Thompson JT, Glaser BM, Michels RG, et al. The use of intravitreal thrombin to control hemorrhage during vitrectomy. Ophthalmology 1986;93:279 – 82. 17. Packer AJ, McCuen BW, 2nd II, Hutton WL, et al. Procoagulant effects of intraocular sodium hyaluronate (Healon) after phakic diabetic vitrectomy: a prospective, randomized study. Ophthalmology 1989;96:1491– 4. 18. Folk JC, Packer AJ, Weingeist TA, et al. Sodium hyaluronate (Healon) in closed vitrectomy. Ophthalmic Surg 1986;17: 299 –306. 19. Bodanowitz S, Kir N, Hesse L. Silicone oil for recurrent vitreous hemorrhage in previously vitrectomized diabetic eyes. Ophthalmologica 1997;211:219 –22. 20. Charles S, Katz A, Wood B. Vitreous Microsurgery. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2002:104 –25. 21. Thompson JT, de Bustros S, Michels RG, et al. Results and prognostic factors in vitrectomy for diabetic tractionrhegmatogenous retinal detachment. Arch Ophthalmol 1987; 105:503–7. 22. Thompson JT, de Bustros S, Michels RG, et al. Results and prognostic factors in vitrectomy for diabetic traction retinal detachment of the macula. Arch Ophthalmol 1987;105:497– 502. 23. Hilton GF, Tornambe PE, Retinal Detachment Study Group. Pneumatic retinopexy: an analysis of intraoperative and postoperative complications. The Retinal Detachment Study Group. Retina 1991;11:285–94. 24. McAllister IL, Meyers SM, Zegarra H, et al. Comparison of pneumatic retinopexy with alternative surgical techniques. Ophthalmology 1988;95:877– 83. 25. Tornambe PE, Hilton GF, Retinal Detachment Study Group. Pneumatic retinopexy: a multicenter randomized controlled clinical trial comparing pneumatic retinopexy with scleral buckling. The Retinal Detachment Study Group. Ophthalmology 1989;96:772– 83. 26. Chalam KV, Malkani S, Shah VA. Intravitreal dexamethasone effectively reduces postoperative inflammation after vitreoretinal surgery. Ophthalmic Surg Lasers Imaging 2003;34: 188 –92.
715