Purpose
To report the frequency of postoperative vitreous hemorrhage (VH) in eyes that underwent primary 23-gauge pars plana vitrectomy (PPV) for nonclearing VH resulting from proliferative diabetic retinopathy, as well as associated risk factors.
Design
Retrospective, consecutive, interventional case series.
Methods
setting : Institutional (Retina Service of Wills Eye Institute). study population : One hundred seventy-three eyes of 157 patients. intervention : Twenty-three–gauge PPV for nonclearing diabetic VH. main outcome measures : Percentage of eyes in which postoperative VH developed, categorized as early, delayed, or severe persistent, as well as percentage requiring reoperation.
Results
During a mean follow-up of 32 weeks, 56 (32%) of 173 eyes demonstrated postoperative VH, categorized as early (8 eyes; 5%), delayed (13 eyes; 8%), or severe persistent (35 eyes; 20%). Twenty-two (13%) of 173 eyes required reoperation: 4 (50%) of 8 in the early group, 8 (62%) of 13 in the delayed group, and 10 (29%) of 35 in the severe persistent group. Mean preoperative logarithm of the minimum angle of resolution visual acuity was 1.5 (Snellen equivalent, approximately 20/600); mean postoperative VA was 0.65 (Snellen equivalent, approximately 20/90), a gain of 0.85 ( P < .0001). Thirty-four (27%) of 127 eyes with complete scatter photocoagulation before undergoing PPV compared with 22 (48%) of 46 eyes with incomplete scatter photocoagulation before undergoing PPV demonstrated postoperative VH ( P = .002). Other factors associated with postoperative VH included younger age ( P = .022) and phakia ( P = .036).
Conclusions
Postoperative VH was not uncommon after initial 23-gauge PPV for diabetic VH and was associated with incomplete scatter photocoagulation, younger age, and phakia before PPV. However, only a minority of patients required reoperation.
Postoperative vitreous hemorrhage (VH) after diabetic vitrectomy has been reported to occur in 12% to 63% of cases. This complication, which impedes visual recovery, may appear within the first few weeks after surgery or occasionally months later. Depending on the severity of the VH, additional intervention may be necessary. Previous studies using 20-gauge pars plana vitrectomy (PPV) have suggested a variety of potential sources for the bleeding. These include incomplete removal of fibrovascular tissue, liberation of residual blood trapped in the peripheral vitreous skirt, fibrovascular proliferation from sclerotomy sites or the vitreous base, incomplete intraoperative hemostasis, and iatrogenic injury to retinal vessels.
The rate of postoperative VH after diabetic vitrectomy performed with small-incision PPV instrumentation has not been well established. Using 25-gauge PPV, Lee and Yu identified a similar incidence of postoperative VH (11.8%) compared with 20-gauge PPV in eyes with proliferative diabetic retinopathy (PDR). Park and associates compared 23-gauge and 20-gauge PPV in patients with PDR and found that despite a higher rate of transient hypotony with 23-gauge PPV, both methods had a similar single-surgery success rate. Our study defined better the rate of postoperative VH and reoperation after initial vitrectomy and examined associated risk factors after 23-gauge PPV in eyes with nonclearing VH resulting from PDR.
Methods
Data Collection
After prospective approval by the Institutional Review Board of Wills Eye Institute (protocol 10-999E), a retrospective chart review was performed on consecutive patients who had undergone 23-gauge PPV between January 2009 and December 2011 at Wills Eye Institute, Philadelphia, Pennsylvania. The authors confirm that they are in compliance with the Institutional Review Board requirements. Patients who met the inclusion criteria had undergone primary, 3-port 23-gauge PPV for nonclearing diabetic VH with a minimum postoperative follow-up of 6 weeks. All patients were required to have had some scatter photocoagulation at least 6 weeks before vitrectomy. Exclusion criteria included the presence of known or suspected diabetic tractional retinal detachment, cases with rubeosis iridis, or neovascular glaucoma requiring filtering or tube shunt surgery. The patients’ medical and operative records were reviewed and the following data were collected: age, gender, preoperative and postoperative pinhole Snellen visual acuity (VA), preoperative and postoperative intraocular pressure, lens status, type of endotamponade (air or gas) when used, and the extent of preoperative scatter photocoagulation. Incomplete scatter photocoagulation was defined as a lack of pre-existing photocoagulation in all 4 quadrants, or fewer than a total of 1000 laser spots. Additional collected data were systemic past medical history, including anticoagulation status and hemoglobin A1c levels, whether intravitreal bevacizumab (IVB) was administered, intraoperative and postoperative complications, and the intraoperative findings regarding the source of postoperative VH (obtained from the operative records), if reoperation was required.
Surgical Technique
The microincision vitrectomy surgical technique was similar for all eyes. Briefly, retrobulbar or peribulbar anesthesia was administered, followed by aseptic preparation of the periorbital skin and eyes using 5% povidone–iodine (Betadine; Purdue Fredrick Co, Norwalk, Connecticut, USA). All procedures were performed with the Accurus or the Constellation 23-gauge surgical vitrectomy systems (Alcon Laboratories Inc, Fort Worth, Texas, USA). The conjunctiva and Tenon capsule were displaced over the sclera to avoid direct communication between the conjunctival and scleral entry sites and to minimize postoperative fluid leakage. Insertion of cannulas was performed at a 30-degree angle using beveled trocars in the inferotemporal, superotemporal, and superonasal quadrants, 3.5 to 4.0 mm posterior to the limbus. A core vitrectomy was performed, and a posterior vitreous separation was created if not already present. Fill-in panretinal endolaser photocoagulation was performed in all cases. In the case of reoperation, careful examination to search for the source of VH was conducted during surgery. A fluid–air exchange or fluid–gas exchange using a nonexpansile concentration of sulfur hexafluoride (SF6) at 16% to 20% or perfluoropropane (C 3 F 8 ) at 12% to 14% was used in some cases, depending on the surgeon’s discretion ( Table 1 ). Before cannula removal near the end of each operation, a peripheral retinal examination was performed using scleral depression with a wide-field viewing system to examine the retina for breaks or other pathologic features. If the vitreous cavity was filled with air or gas at the end of surgery, wound competency was confirmed by checking for air bubbling through the sclerotomy sites. If the vitreous cavity was left filled with fluid, bleb formation typically was used as an indicator of wound leakage. If a wound leak was found, transconjunctival sutures were placed to close the sclerotomy.
| Demographic Variable | Postoperative VH | P Value | |
|---|---|---|---|
| No (n = 117 Eyes, 111 Subjects) | Yes (n = 56 Eyes, 46 Subjects) | ||
| Age (y) | |||
| Mean (range) | 61.7 (27 to 87) | 55.6 (33 to 81) | .022 |
| Gender, n (%) | |||
| Male | 54 (46) | 34 (61) | .075 |
| Female | 63(54) | 22 (39) | |
| Lens status, no. (%) | |||
| Pseudophakic | 65 (56) | 21 (38) | .036 |
| Phakic | 52 (44) | 35 (63) | |
| HbA1c (%) | |||
| Mean (range) | 7.6 (5.3 to 13.1) | 7.9 (5.5 to 13.1) | .26 |
| Preoperative bevacizumab, n (%) | 6 (5) | 6 (11) | .19 |
| Anticoagulants, n (%) | |||
| None | 44 (37) | 28 (50) | .36 |
| ASA | 47 (40) | 16 (29) | |
| ASA + clopidogrel | 20 (18) | 5 (9) | |
| Warfarin | 6 (5) | 7 (13) | |
| None | 44 (37) | 28 (50) | .13 |
| Any | 73 (63) | 28 (50) | |
| Tamponade, n (%) | |||
| Air | 73 (90) | 37 (97) | .19 |
| C 3 F 8 | 2 (2) | 0 | |
| SF 6 | 6 (7) | 1 (3) | |
| None (BSS) | 36 (31) | 18 (32) | .85 |
| Gas (air, C 3 F 8 , or SF 6 ) | 81 (69) | 38 (68) | |
Statistical Analysis
The main outcome measure was the percentage of eyes with postoperative VH. The VH was classified as early postoperative (VH fewer than 6 weeks from surgery but not present on postoperative day 1), delayed postoperative (VH 6 weeks or more from surgery), or severe persistent (nonclearing for more than 6 weeks and present since postoperative day 1). Note that the early and delayed cases did not have VH on postoperative day 1, whereas the persistent cases had VH present from the first day after surgery. Key secondary outcome measures included the need for reoperation, the average change in VA from baseline, intraoperative complications, and postoperative complications. Visual acuities as measured by pinhole on a Snellen acuity chart were converted to logarithm of minimal angle of resolution (logMAR) units for statistical analysis. Differences in rates of postoperative VH were evaluated using Poisson regression, adjusted for differing follow-up times. Patients with VH were compared with those without VH with respect to baseline factors using logistic regression or linear regression, as appropriate. Scatter photocoagulation groups (incomplete vs complete) were compared with respect to baseline characteristics and posttreatment outcomes using logistic regression or linear regression, as appropriate. When necessary, continuous outcomes were log-transformed to approximate better the assumption of normality. All P values were calculated using generalized estimating equation methods to account for correlation of multiple eyes from the same subject. P values were not adjusted for multiple comparisons. P < .05 was considered statistically significant.
Results
Vitreous Hemorrhage Frequency
A total of 173 eyes from 157 patients met the study inclusion criteria. Comparisons of the postoperative VH groups with respect to subject demographics and preoperative characteristics are listed in Table 1 . During a mean overall follow-up of 32 weeks (range, 6 to 97 weeks), 56 (32%) of the 173 eyes had postoperative VH characterized as early postoperative VH (8 eyes; 5%), delayed postoperative VH (13 eyes; 8%), or severe persistent VH (35 eyes; 20%). Twenty-two (13%) of 173 eyes required reoperation: 4 (50%) of the 8 with early postoperative VH, 8 (62%) of the 13 with delayed postoperative VH, and 10 (29%) of the 35 with severe persistent VH. The mean time from initial surgery to repeat PPV was 15 weeks (range, 0 to 53 weeks). The remaining 34 patients had sufficient spontaneous resolution of VH and did not require repeat PPV. Postoperative VH was observed in only 1 of the eyes that underwent a second vitrectomy.
Visual Acuity
Mean preoperative logMAR VA was 1.5 (Snellen equivalent, approximately 20/600) and mean postoperative logMAR VA was 0.65 (Snellen equivalent, approximately 20/90) at the last follow-up examination, a gain in logMAR VA of 0.85 ( P < .0001). The mean follow-up time was 33 weeks after initial PPV for those patients who did not require repeat vitrectomy and 25 weeks after the subsequent PPV in those patients who did require repeat vitrectomy. Of the eyes that had postoperative VH requiring reoperation, the logMAR VA before the second vitrectomy was 1.47 (Snellen equivalent, approximately 20/600) and that after vitrectomy was 0.8 (Snellen equivalent, approximately 20/125), with a gain of 0.67 ( P < .0001). The mean preoperative VA of patients who had incomplete scatter photocoagulation was worse (1.65 logMAR; Snellen equivalent, approximately 20/900) than those who had complete scatter photocoagulation (1.45 logMAR; Snellen equivalent, approximately 20/560) before PPV ( P = .021).
Risk Factors for Vitreous Hemorrhage
Of the 127 eyes that had complete scatter photocoagulation before PPV, 34 (27%) demonstrated postoperative VH, compared with 22 (48%) of the 46 eyes that had incomplete scatter photocoagulation before PPV ( P = .002; Table 2 ). Eyes with incomplete scatter photocoagulation were more likely to have postoperative VH requiring repeat PPV (17%) compared with eyes that had complete scatter photocoagulation (11%; P = .038). The rate ratio for developing postoperative VH requiring repeat PPV for incomplete versus complete scatter photocoagulation was 2.26 (95% confidence interval, 1.05 to 4.88).
| Demographic Variables | Scatter Photocoagulation (n = 173 Eyes, 157 Subjects) | P Value | |
|---|---|---|---|
| Incomplete (n = 46) | Complete (n = 127) | ||
| Age (y) | |||
| Mean (range) | 61.0 (38 to 85) | 59.5 (27 to 87) | .42 |
| Gender, n (%) | |||
| Male | 26 (57) | 62 (49) | .40 |
| Female | 20 (43) | 65 (51) | |
| IOP (mm Hg) | |||
| Preoperative, median (range) | 14.0 (9 to 28) | 16.0 (7 to 36) | .75 |
| Postoperative, mean (range) | 15.76 (10 to 23) | 15.93 (7 to 42) | .80 |
| Visual acuity (logMAR) | |||
| Preoperative, mean (range) | 1.65 (0.4 to 2.5) | 1.45 (0.2 to 2.5) | .021 |
| Change, a mean (range) | 0.99 (−0.9 to 2.1) | 0.81 (−1.6 to 1.8) | .85 |
| Postoperative vitreous hemorrhage, n (%) | 22 (48) | 34 (27) | .002 |
| Postoperative vitreous hemorrhage rate ratio, incomplete vs complete (95% CI) | 2.56 (1.64 to 4.00) | ||
| Postoperative vitreous hemorrhage requiring repeat PPV, n (%) | 8 (17) | 14 (11) | .038 |
| Postoperative vitreous hemorrhage requiring repeat PPV rate ratio, incomplete vs complete (95% CI) | 2.26 (1.05 to 4.88) | ||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
