- Email: [email protected]
Early Vitrectomy for Vitreous Hemorrhage Associated With Retinal Tears
Early Vitrectomy for Vitreous Hemorrhage Associated With Retinal Tears H. STEVIE TAN, MARCO MURA, AND HEICO M. BIJL ● PURPOSE: To evaluate outcome of early surgery in vitreous
hemorrhage, presumably associated with retinal tears. ● DESIGN: Retrospective, noncomparative interventional case series. ● METHODS: We included 40 consecutive cases in 39 patients treated with early vitrectomy for vitreous hemorrhage. Main outcome measures were incidence of tears identified preoperatively and intraoperatively, visual acuity, and complications. ● RESULTS: Vitrectomy took place after a mean delay of 2.7 days after presentation. A total of 69 tears were found in 40 eyes; 53.7% of tears were identified preoperatively by funduscopy or ultrasound (U/S) and the remaining 46.3% of tears were only identified during vitrectomy. Visual acuity improved significantly from 1/60 to 0.8. The sensitivity of U/S tear detection was 55.9%. A history of predisposing factors was not related to the risk of presence of unsupported tears. In 2 cases a retinal detachment developed between the U/S evaluation and vitrectomy. Postoperative complications were cataract (30%), macular pucker (2.5%), and retinal detachment (5.0%). ● CONCLUSIONS: Our findings illustrate the potential dangers of a conservative approach to vitreous hemorrhage and show that early vitrectomy has good outcome with acceptable complication rates. Prospective studies on optimal treatment of vitreous hemorrhage associated with tears are needed. (Am J Ophthalmol 2010;150: 529 –533. © 2010 by Elsevier Inc. All rights reserved.)
A
PATIENT PRESENTING WITH A FUNDUS-OBSCURING
vitreous hemorrhage, presumably caused by a retinal tear, poses a serious clinical dilemma. Many clinicians prefer a conservative approach. Positioning the head upright to allow blood to settle down in the eye, combined with immobilization and bilateral patching, enhances resolution of blood.1–3 These measures promote clearance of blood in the superior part of the eye, enabling visualization and treatment of superior retinal tears. The trade-off of a conservative approach is the risk of progression to a rhegmatogenous retinal detachment (RRD) before clearing of the vitreous hemorrhage allows full evaluation of the fundus and treatment of retinal tears.
Accepted for publication Apr 12, 2010. From the Department of Ophthalmology, Academic Medical Center, University of Amsterdam, The Netherlands. Inquiries to H. Stevie Tan, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands; e-mail: [email protected] 0002-9394/$36.00 doi:10.1016/j.ajo.2010.04.005
©
2010 BY
If the vitreous hemorrhage is too dense to allow funduscopic evaluation, ultrasound (U/S) investigation of the eye can identify retinal tears, pointing to a rhegmatogenous origin. It has been shown, however, that U/S detection rate of retinal tears is suboptimal, especially if tears are small,4 and this has also been our clinical impression. This makes assessment of risk of progression to RRD unreliable, undermining the conservative approach. For this reason, some have favored a more aggressive strategy.5 This paper describes our experiences and outcome of vitrectomy in an early stage of the disease.
METHODS MEDICAL RECORDS OF CONSECUTIVE PATIENTS WHO UN-
derwent early vitrectomy for a vitreous hemorrhage between February 1, 2007 and February 1, 2009 were reviewed. We included cases of vitrectomy for vitreous hemorrhage with presence or suspicion of retinal tears, that had a follow-up of at least 3 months. We excluded cases with other causes of vitreous hemorrhage like trauma and vascular disease and eyes in which a retinal detachment was seen on funduscopy or U/S. If a tear was visible at funduscopy, we attempted to laser the defect if we assessed the hemorrhage to be penetrable by laser. If it was not possible to get a satisfactory laser barrage, the tear was considered unsupported and the patient was advised to undergo vitrectomy. If no tear was found because the hemorrhage was too dense to allow proper funduscopic evaluation of the periphery of the retina, a U/S investigation was performed. If a tear was found at U/S investigation, we advised the patient to have surgery. If no tear was found, we discussed options with the patient and advised a vitrectomy if predisposing factors were identified. In certain cases without the presence of predisposing factors, still a mutual decision was made to opt for vitrectomy. After a decision was made to operate, the patient was scheduled for surgery within 7 days, but an attempt was always made to perform surgery as fast as possible. The time between decision and actual surgery was recorded as “delay.” All patients were operated at the Academic Medical Center, Amsterdam, a tertiary academic referral center. The operations were performed by 1 of 3 surgeons (H.S.T., M.M., or H.M.B.). Data were retrieved from an electronic patient file, containing structured operation notes and
ELSEVIER INC. ALL
RIGHTS RESERVED.
529
TABLE 1. Early Vitrectomy for Vitreous Hemorrhage: Preoperative Characteristics (n ⫽ 40) Gender Female Male Mean age: yrs (range) Eye OD OS Predisposing factors None Flashes Previous tear Previous RRD Tear fellow eye RRD fellow eye Tear detection U/S no tears U/S tears Visual Total # identified tears Lens status Phakic Pseudophakic Mean delay: days (range)
TABLE 2. Early Vitrectomy for Vitreous Hemorrhage: Intraoperative Characteristics Anesthesia General Local Type of operation Vitrectomy Vitrectomy ⫹ phaco Vitrectomy platform 20-gauge 25-gauge Total # tears # unsupported tears # supported tears # new tears # new RRD Location of tears (cases) Superior Inferior Superior ⫹ inferior Tamponade No (BSS) Air SF6 C3F8 (0.5cc)
19 (47.5%) 21 (52.5%) 61.1 (37-86) 16 (40.0%) 24 (60.0%) 19 (47.5%) 5 (12.5%) 10 (25.0%) 3 (7.5%) 2 (5.0%) 1 (2.5%) 10 (25.0%) 17 (42.5%) 13 (32.5%) 38 (95.0%) 30 (75.0%) 10 (25.0%) 2.7 (0-11)
standardized fundus drawings with detailed localization of encountered and treated tears. All operations were performed with the Alcon Accurus 600 DS (Alcon Laboratories, Fort Worth, Texas, USA), BIOM wide-angle viewing system (Binocular Indirect Ophthalmo Microscope; Oculus Inc, Wetzlar, Germany). Operations were performed with either a 20-gauge or 25-gauge platform, depending on the surgeon’s preference. In 20-gauge cases infusion pressure was set at 20 mm Hg with cutting rate varying between 1000 and 2500 cuts/ minute. In 25-gauge procedures infusion pressure was set at 30 mm Hg, cutting rate varied between 1000 and 1500 cuts/minute, and light was provided by a xenon light source (Photon 2; Synergetics, O’Fallon, Missouri, USA). In combined procedures, we always started with phacoemulsification of the lens, implantation of an intraocular lens, and suturing of the cataract wound. During vitrectomy, a preexisting posterior vitreous detachment was encountered intraoperatively in all eyes. Core vitrectomy was performed, followed by vitrectomy of the periphery. The extent of peripheral vitrectomy depended on lens status of the eye, with less complete vitrectomy in phakic eyes. Extensive internal search was performed in all cases using visualization with the BIOM system and scleral indentation. All tears were treated with external cryo application. We also retreated the tears that were already supported by earlier laser barrage. If tears were detected, an air or gas tamponade was used. AMERICAN JOURNAL
36 (90.0%) 4 (10.0%) 16 (40.0%) 24 (60.0%) 69 62 (89.9%) 7 (10.1%) 32 (46.4%) 2 (2.9%) 17 (42.5%) 3 (7.5%) 13 (32.5%) 2 (5.0%) 22 (55.0%) 15 (37.5%) 1 (2.5%)
BSS⫽ balanced salt solution; C3F8 ⫽ octafluoropropane gas; RRD ⫽ rhegmatogenous retinal detachment; SF6 ⫽ sulfur hexafluoride gas.
OD ⫽ right eye; OS ⫽ left eye; RRD ⫽ rhegmatogenous retinal detachment; U/S ⫽ ultrasound; yrs ⫽ years.
530
5 (12.5%) 35 (87.5%)
Preoperative parameters retrieved were patient characteristics, visual acuity, lens status, predisposing factors, and number and location of tears detected preoperatively by means of funduscopy and by means of ultrasound. Intraoperative parameters were type of surgery, type of anesthesia, number and location of tears detected intraoperatively, intraoperative complications, and mode of tamponade. Postoperatively we collected visual acuity and postoperative complications. Total number of tears is defined as the number of tears found intraoperatively. We only considered tears with persistent vitreous traction. Operculated tears, round holes in lattice, and other degenerative tears were always treated during vitrectomy but not included in our analysis. Supported tears are defined as tears that were treated adequately by any means of retinopexy before operation. New tears are defined as tears that were encountered intraoperatively and had not been identified by means of funduscopy or U/S preoperatively. New retinal detachments are defined as presence of retinal detachment as encountered intraoperatively, which had not been present at the last preoperative funduscopic or ultrasonographic evaluation. Unsupported tears are defined as those tears that were identified during surgery but were not, or not adequately, treated. Statistical analysis was performed using SPSS software for Windows version 16.0 (SPSS Inc, Chicago, Illinois, USA). OF
OPHTHALMOLOGY
OCTOBER 2010
RESULTS FORTY EYES FROM 39 PATIENTS WERE INCLUDED. BASELINE
patient and ocular characteristics are summarized in Table 1. The mean follow-up was 19.1 months (range 6-34 months). Predisposing factors were present in 16 of 40 cases. Of the 24 cases without predisposing factors, only 5 mentioned sensation of flashes. Of 40 eyes, tears were visible by funduscopy in 13 cases but we were not able to adequately secure these tears with laser photocoagulation because of the density of the hemorrhage. In 17 eyes 1 or more defects were identified with U/S and in 10 eyes no tears could be found on U/S. Intraoperative details are depicted in Table 2. All 40 cases underwent vitrectomy for removal of blood and treatment of defects. Sixteen of 40 cases were operated with the conventional 20-gauge system and the remaining 24 cases with the sutureless 25-gauge platform. In all cases, a preexistent posterior vitreous detachment (PVD) was present over the posterior pole. Separation of the posterior vitreous membrane and retina in more peripheral areas varied but was not recorded systematically. Delay between decision to treat and actual treatment was on average 2.7 days. There were no differences in retinal tear incidence or complications between the 20-gauge and the 25-gauge cases. A total of 69 retinal tears with persistent vitreoretinal traction were found intraoperatively. Four eyes had 4 tears, 6 eyes had 3 tears, 10 eyes had 2 tears, and 15 eyes had 1 single tear. Sixty-two were unsupported tears, of which 32 were newly discovered tears (46.4%), the rest being tears that had been identified preoperatively but not lasered or not lasered adequately. Seven tears were supported tears, meaning that these tears had had adequate retinopexy in the past. In 1 case, U/S identified a tear but no tears could be found intraoperatively. Number of tears was evenly distributed over 20- and 25-gauge cases. Preoperative lens status is depicted in Table 1. Of the 30 phakic eyes, 23 appeared to have 1 or more unsupported tears (76.7%), while of the 10 pseudophakic eyes, 7 had unsupported tears (70.0%). This small difference was not statistically significant (P ⫽ .673, 2). Thus, preoperative lens status was not predictive of presence of unsupported tears. In 2 cases, a retinal detachment was encountered where no detachment was present at the last evaluation. We designated these 2 cases as new detachments. Both retinal detachments were macula-on and were repaired intraoperatively. Among the 40 cases, 1 or more tears were present in the superior half of the eye in 17 cases and in only 3 cases tears were confined to the inferior half. In 13 cases tears were found superiorly as well as inferiorly (Table 2). Thus, although there is a predilection for a superior location, in 40% of cases (16/40) tears occurred in the lower quadrants. Choice of tamponade was made on an individual basis. In 1 case, a small bubble of pure C3F8 was considered adequate. VOL. 150, NO. 4
EARLY VITRECTOMY
FOR
During follow-up, 9 of the total of 30 phakic eyes developed cataract. One case had to be reoperated for a macular pucker after 3 months. Two eyes developed an RRD, both occurring 4 months after surgery. Both cases were macula-on detachments and were repaired by repeat vitrectomy. In both cases the retina stayed attached after the second vitrectomy. Interestingly, we had not identified any tears in 1 of these cases during the primary surgery. In the second case, the causative tear was not documented during the first vitrectomy and should thus be considered as either missed or having developed after the first vitrectomy. Visual acuity was recorded preoperatively and postoperatively. There was improvement in visual acuity in 38 of 40 cases. The median preoperative VA was 1/60. The median postoperative VA was 0.8. This improvement is large and statistically significant (P ⬍ .001; Wilcoxon signed rank test). We assessed the sensitivity of our U/S in a selection of 27 cases in which too much blood was present for funduscopic evaluation and a full U/S investigation was performed. In these 27 eyes, a total of 19 tears were identified by U/S. During operation, 34 tears were identified in these 27 eyes, leading to a sensitivity of preoperative detection of tears by U/S of 55.9%. In 8 cases, surgery was still performed while no tears were detected by U/S. In 2 cases, there were predisposing factors (retinal tear in 1, RRD in the other) that justified a decision to treat. In the remaining 6 cases, however, the decision to operate was not based on the presence of predisposing factors. In the 2 cases with predisposing factors, no tears were found during operation. In the other 6 cases, tears were detected in 4 and no tears were found in 2. In 9 cases, no unsupported tears were found during surgery, so in these cases the operation could be considered unnecessary in hindsight. In 6 of these 9 cases, 1 or more predisposing factors were identified preoperatively. We assessed the relation between preoperative predisposing factors and presence of unsupported tears intraoperatively, needing treatment. Of 24 cases without predisposing factors, 21 cases appeared to have 1 or more unsupported tears (87.5%). In the 16 cases that did have a history of predisposing factors, only 9 had unsupported tears (56.3%). This unexpected inverse relation was actually statistically significant (P ⫽ .034, Fisher exact test). So a history of predisposing factors in our series was not related to the presence of an unsupported tear.
DISCUSSION HISTORICALLY, THE MANAGEMENT OF VITREOUS HEMOR-
rhage of unknown cause has been conservative.1 Upright positioning combined with immobilization and bilateral patching were advocated to allow faster clearance of blood in the upper vitreous cavity.2,3 The current preferred practice patterns of the American Academy of Ophthalmology6 states that in eyes in which a retinal tear is VITREOUS HEMORRHAGE
531
suspected, a conservative approach with repeat U/S should be followed, and a recent survey of management practices in the UK showed that most ophthalmologists prefer awaiting spontaneous resolution of the hemorrhage.7 Predisposing factors as rhegmatogenous incidents in either eye could help the clinician in their assessment of risk. A retinal detachment of the contralateral eye in the past history leads to a 75% incidence of retinal detachment, when confronted with a dense vitreous hemorrhage.8 In our series, however, eliciting a predisposing factor in the history did not influence the chance of finding an unsupported tear during vitrectomy. Earlier reports8 –10 have mentioned that retinal tears occur more frequently in the upper quadrants in the setting of a PVD. In our series, the predilection for the upper quadrants was corroborated, but still in as much as 40% of cases tears were found in the lower quadrants. The presumption that it suffices to evaluate only the upper quadrants for presence of tears is for this reason debatable. In cases with retina-obscuring vitreous hemorrhage, B-scan U/S is the best mode of evaluation of the integrity of the posterior segment. Sensitivity of tear detection was determined by several studies and varied between 92% in a small series and as low as 44%.4,9,10 One possible cause of this large variety is that the sensitivity in detection of large tears is high, but lower in detecting smaller tears. Another source of variety is the large dependence of accuracy on experience of the individual U/S operator. The findings in our current series suggest that the accuracy of tear detection is suboptimal, with a sensitivity of only 55.9%. Earlier studies closely followed cases of dense vitreous hemorrhage.8,11 During follow-up, 1 or more tears developed in 75% of cases. Even more seriously, 44% of cases developed an RRD, of which 28.6% were macula-off.8 Another 44.4% of eyes ultimately required vitrectomy for nonclearing vitreous hemorrhage. This agrees with a conservative, prospective follow-up,11 where 67% of eyes with dense vitreous hemorrhage appeared to have tears and even 39% had frank retinal detachment. One report investigated outcome of early vitrectomy in the context of dense vitreous hemorrhage with presumed retinal tears.
Their delay between presentation and vitrectomy was 6.3 days. New retinal tears were discovered intraoperatively in 75% of cases. Our findings are in agreement with these earlier reports. We were confronted with the identification of a high number of new tears (46.4%), tears that were encountered only during vitrectomy and missed by funduscopy and U/S preoperatively, and with the discovery of frank RRD in 2 cases at the start of vitrectomy, where no RRD was found in the last preoperative evaluation, meaning that the RRD had developed in the direct preoperative period. The presence of blood in an eye with a retinal detachment is known to increase the risk of developing proliferative vitreoretinopathy (PVR).12–15 In 1 study in which cases with dense vitreous hemorrhage were followed, the risk of PVR was 19%.8 In the small number of cases of retinal detachment in our current series, no PVR occurred. Our findings illustrate the inherent risk of a conservative strategy. With our current diagnostic techniques only part of retinal tears can be identified. In the context of an acute posterior vitreous detachment, the presence of an unsupported or untreated retinal tear carries an unacceptable risk of progression to RRD with increased risk of PVR, leading to a high risk of irreversible damage of visual function. This poses to the ophthalmologist a classic dilemma. Vitrectomy is a highly invasive procedure with well-known risks involved. In an elective setting, vitrectomy induces retinal tears with the potential to cause retinal detachments. As in any intraocular procedure, there is a risk of endophthalmitis and choroidal hemorrhage. In phakic cases, vitrectomy has a 100% risk of accelerating the development of cataract. Adopting a more aggressive strategy implies accepting that we impose these risks on individual cases that could have had good outcome with a conservative approach. Although the outcome in our series in terms of visual acuity and fast recovery of visual function was good and complications were acceptable, the small size and lack of any control or randomization limits us in drawing firm conclusions. There is great need for prospective controlled studies to guide the clinician in this important issue.
THE AUTHORS INDICATE NO FINANCIAL SUPPORT OR FINANCIAL CONFLICT OF INTEREST. INVOLVED IN DESIGN OF THE STUDY (H.S.T.); conduct of the study (H.S.T., M.M., H.M.B.); collection (H.S.T., M.M., H.M.B.), management (H.S.T.), analysis, and interpretation of the data (H.S.T., M.M., H.M.B.); and preparation, review, and approval of the manuscript (H.S.T., M.M., H.M.B.). The institutional review board at the University of Amsterdam declared that this type of retrospective study waived the need for IRB approval, in accordance with Dutch law on human clinical trials.
4. Rabinowitz R, Yagev R, Shoham A, Lifshitz T. Comparison between clinical and ultrasound findings in patients with vitreous hemorrhage. Eye 2004;18(3):253–256. 5. Dhingra N, Pearce I, Wong D. Early vitrectomy for fundusobscuring dense vitreous haemorrhage from presumptive retinal tears. Graefes Arch Clin Exp Ophthalmol 2007;245(2):301– 304. 6. Preferred Practice Pattern. Management of Posterior Vitreous Detachment, Retinal Breaks, and Lattice Degenera-
REFERENCES 1. Spraul CW, Grossniklaus HE. Vitreous hemorrhage. Surv Ophthalmol 1997;42(1):3–39. 2. Lincoff H, Kreissig I, Wolkstein M. Acute vitreous haemorrhage: a clinical report. Br J Ophthalmol 1976;60(6):454 – 458. 3. Wilkinson CP. What ever happened to bilateral patching? Retina 2005;25(4):393–394.
532
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2010
7.
8.
9.
10.
tion. San Francisco: American Academy of Ophthalmology; 2008. Vote BJ, Membrey WL, Casswell AG. Vitreous haemorrhage without obvious cause: national survey of management practices. Eye 2005;19(7):770 –777. Sarrafizadeh R, Hassan TS, Ruby AJ, et al. Incidence of retinal detachment and visual outcome in eyes presenting with posterior vitreous separation and dense fundusobscuring vitreous hemorrhage. Ophthalmology 2001;108(12): 2273–2278. DiBernardo C, Blodi B, Byrne SF. Echographic evaluation of retinal tears in patients with spontaneous vitreous hemorrhage. Arch Ophthalmol 1992;110(4):511–514. Nischal KK, James JN, McAllister J. The use of dynamic ultrasound B-scan to detect retinal tears in spontaneous vitreous haemorrhage. Eye 1995;9(4):502–506.
11. Lindgren G, Sjödell L, Lindblom B. A prospective study of dense spontaneous vitreous hemorrhage. Am J Ophthalmol 1995;119(4):458 – 465. 12. Yeung L, Yang KJ, Chen TL, et al. Association between severity of vitreous haemorrhage and visual outcome in primary rhegmatogenous retinal detachment. Acta Ophthalmol 2008;86(2):165–169. 13. Tseng W, Cortez RT, Ramirez G, Stinnett S, Jaffe GJ. Prevalence and risk factors for proliferative vitreoretinopathy in eyes with rhegmatogenous retinal detachment but no previous vitreoretinal surgery. Am J Ophthalmol 2004;137(6):1105–1115. 14. Asaria RH, Kon CH, Bunce C, et al. How to predict proliferative vitreoretinopathy: a prospective study. Ophthalmology 2001;108(7):1184 –1186. 15. Cowley M, Conway BP, Campochiaro PA, Kaiser D, Gaskin H. Clinical risk factors for proliferative vitreoretinopathy. Arch Ophthalmol 1989;107(8):1147–1151.
AJO History of Ophthalmology Series Paul Sullivan and Intraocular Sulfur Hexafluoride
T
he use of air to attempt to reattach the retina predates Jules Gonin’s discovery of the underlying etiology of rhegmatogenous retinal detachment. In the 1930’s, Rosengren frequently used air to improve the anatomic success rate of his surgical procedures. In the 1950’s and 1960’s many retinal surgeons used intraocular air but struggled with the main limitation of the device, that is, the relatively short duration of the gas within the vitreous cavity. Fortuitous events
VOL. 150, NO. 4
EARLY VITRECTOMY
FOR
allowed Dr. Paul Sullivan to introduce the first expansile gas, sulfur hexafluoride (SF6), to Dr. Edward W.D. Norton and his colleagues at the University of Miami in the late 1960’s. The use of SF6 helped revolutionize retinal detachment surgery and allowed for further advances in treating other posterior segment diseases. Submitted by Christopher F. Blodi from the Cogan Ophthalmic History Society.
VITREOUS HEMORRHAGE
533
hemorrhage, presumably associated with retinal tears. ● DESIGN: Retrospective, noncomparative interventional case series. ● METHODS: We included 40 consecutive cases in 39 patients treated with early vitrectomy for vitreous hemorrhage. Main outcome measures were incidence of tears identified preoperatively and intraoperatively, visual acuity, and complications. ● RESULTS: Vitrectomy took place after a mean delay of 2.7 days after presentation. A total of 69 tears were found in 40 eyes; 53.7% of tears were identified preoperatively by funduscopy or ultrasound (U/S) and the remaining 46.3% of tears were only identified during vitrectomy. Visual acuity improved significantly from 1/60 to 0.8. The sensitivity of U/S tear detection was 55.9%. A history of predisposing factors was not related to the risk of presence of unsupported tears. In 2 cases a retinal detachment developed between the U/S evaluation and vitrectomy. Postoperative complications were cataract (30%), macular pucker (2.5%), and retinal detachment (5.0%). ● CONCLUSIONS: Our findings illustrate the potential dangers of a conservative approach to vitreous hemorrhage and show that early vitrectomy has good outcome with acceptable complication rates. Prospective studies on optimal treatment of vitreous hemorrhage associated with tears are needed. (Am J Ophthalmol 2010;150: 529 –533. © 2010 by Elsevier Inc. All rights reserved.)
A
PATIENT PRESENTING WITH A FUNDUS-OBSCURING
vitreous hemorrhage, presumably caused by a retinal tear, poses a serious clinical dilemma. Many clinicians prefer a conservative approach. Positioning the head upright to allow blood to settle down in the eye, combined with immobilization and bilateral patching, enhances resolution of blood.1–3 These measures promote clearance of blood in the superior part of the eye, enabling visualization and treatment of superior retinal tears. The trade-off of a conservative approach is the risk of progression to a rhegmatogenous retinal detachment (RRD) before clearing of the vitreous hemorrhage allows full evaluation of the fundus and treatment of retinal tears.
Accepted for publication Apr 12, 2010. From the Department of Ophthalmology, Academic Medical Center, University of Amsterdam, The Netherlands. Inquiries to H. Stevie Tan, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands; e-mail: [email protected] 0002-9394/$36.00 doi:10.1016/j.ajo.2010.04.005
©
2010 BY
If the vitreous hemorrhage is too dense to allow funduscopic evaluation, ultrasound (U/S) investigation of the eye can identify retinal tears, pointing to a rhegmatogenous origin. It has been shown, however, that U/S detection rate of retinal tears is suboptimal, especially if tears are small,4 and this has also been our clinical impression. This makes assessment of risk of progression to RRD unreliable, undermining the conservative approach. For this reason, some have favored a more aggressive strategy.5 This paper describes our experiences and outcome of vitrectomy in an early stage of the disease.
METHODS MEDICAL RECORDS OF CONSECUTIVE PATIENTS WHO UN-
derwent early vitrectomy for a vitreous hemorrhage between February 1, 2007 and February 1, 2009 were reviewed. We included cases of vitrectomy for vitreous hemorrhage with presence or suspicion of retinal tears, that had a follow-up of at least 3 months. We excluded cases with other causes of vitreous hemorrhage like trauma and vascular disease and eyes in which a retinal detachment was seen on funduscopy or U/S. If a tear was visible at funduscopy, we attempted to laser the defect if we assessed the hemorrhage to be penetrable by laser. If it was not possible to get a satisfactory laser barrage, the tear was considered unsupported and the patient was advised to undergo vitrectomy. If no tear was found because the hemorrhage was too dense to allow proper funduscopic evaluation of the periphery of the retina, a U/S investigation was performed. If a tear was found at U/S investigation, we advised the patient to have surgery. If no tear was found, we discussed options with the patient and advised a vitrectomy if predisposing factors were identified. In certain cases without the presence of predisposing factors, still a mutual decision was made to opt for vitrectomy. After a decision was made to operate, the patient was scheduled for surgery within 7 days, but an attempt was always made to perform surgery as fast as possible. The time between decision and actual surgery was recorded as “delay.” All patients were operated at the Academic Medical Center, Amsterdam, a tertiary academic referral center. The operations were performed by 1 of 3 surgeons (H.S.T., M.M., or H.M.B.). Data were retrieved from an electronic patient file, containing structured operation notes and
ELSEVIER INC. ALL
RIGHTS RESERVED.
529
TABLE 1. Early Vitrectomy for Vitreous Hemorrhage: Preoperative Characteristics (n ⫽ 40) Gender Female Male Mean age: yrs (range) Eye OD OS Predisposing factors None Flashes Previous tear Previous RRD Tear fellow eye RRD fellow eye Tear detection U/S no tears U/S tears Visual Total # identified tears Lens status Phakic Pseudophakic Mean delay: days (range)
TABLE 2. Early Vitrectomy for Vitreous Hemorrhage: Intraoperative Characteristics Anesthesia General Local Type of operation Vitrectomy Vitrectomy ⫹ phaco Vitrectomy platform 20-gauge 25-gauge Total # tears # unsupported tears # supported tears # new tears # new RRD Location of tears (cases) Superior Inferior Superior ⫹ inferior Tamponade No (BSS) Air SF6 C3F8 (0.5cc)
19 (47.5%) 21 (52.5%) 61.1 (37-86) 16 (40.0%) 24 (60.0%) 19 (47.5%) 5 (12.5%) 10 (25.0%) 3 (7.5%) 2 (5.0%) 1 (2.5%) 10 (25.0%) 17 (42.5%) 13 (32.5%) 38 (95.0%) 30 (75.0%) 10 (25.0%) 2.7 (0-11)
standardized fundus drawings with detailed localization of encountered and treated tears. All operations were performed with the Alcon Accurus 600 DS (Alcon Laboratories, Fort Worth, Texas, USA), BIOM wide-angle viewing system (Binocular Indirect Ophthalmo Microscope; Oculus Inc, Wetzlar, Germany). Operations were performed with either a 20-gauge or 25-gauge platform, depending on the surgeon’s preference. In 20-gauge cases infusion pressure was set at 20 mm Hg with cutting rate varying between 1000 and 2500 cuts/ minute. In 25-gauge procedures infusion pressure was set at 30 mm Hg, cutting rate varied between 1000 and 1500 cuts/minute, and light was provided by a xenon light source (Photon 2; Synergetics, O’Fallon, Missouri, USA). In combined procedures, we always started with phacoemulsification of the lens, implantation of an intraocular lens, and suturing of the cataract wound. During vitrectomy, a preexisting posterior vitreous detachment was encountered intraoperatively in all eyes. Core vitrectomy was performed, followed by vitrectomy of the periphery. The extent of peripheral vitrectomy depended on lens status of the eye, with less complete vitrectomy in phakic eyes. Extensive internal search was performed in all cases using visualization with the BIOM system and scleral indentation. All tears were treated with external cryo application. We also retreated the tears that were already supported by earlier laser barrage. If tears were detected, an air or gas tamponade was used. AMERICAN JOURNAL
36 (90.0%) 4 (10.0%) 16 (40.0%) 24 (60.0%) 69 62 (89.9%) 7 (10.1%) 32 (46.4%) 2 (2.9%) 17 (42.5%) 3 (7.5%) 13 (32.5%) 2 (5.0%) 22 (55.0%) 15 (37.5%) 1 (2.5%)
BSS⫽ balanced salt solution; C3F8 ⫽ octafluoropropane gas; RRD ⫽ rhegmatogenous retinal detachment; SF6 ⫽ sulfur hexafluoride gas.
OD ⫽ right eye; OS ⫽ left eye; RRD ⫽ rhegmatogenous retinal detachment; U/S ⫽ ultrasound; yrs ⫽ years.
530
5 (12.5%) 35 (87.5%)
Preoperative parameters retrieved were patient characteristics, visual acuity, lens status, predisposing factors, and number and location of tears detected preoperatively by means of funduscopy and by means of ultrasound. Intraoperative parameters were type of surgery, type of anesthesia, number and location of tears detected intraoperatively, intraoperative complications, and mode of tamponade. Postoperatively we collected visual acuity and postoperative complications. Total number of tears is defined as the number of tears found intraoperatively. We only considered tears with persistent vitreous traction. Operculated tears, round holes in lattice, and other degenerative tears were always treated during vitrectomy but not included in our analysis. Supported tears are defined as tears that were treated adequately by any means of retinopexy before operation. New tears are defined as tears that were encountered intraoperatively and had not been identified by means of funduscopy or U/S preoperatively. New retinal detachments are defined as presence of retinal detachment as encountered intraoperatively, which had not been present at the last preoperative funduscopic or ultrasonographic evaluation. Unsupported tears are defined as those tears that were identified during surgery but were not, or not adequately, treated. Statistical analysis was performed using SPSS software for Windows version 16.0 (SPSS Inc, Chicago, Illinois, USA). OF
OPHTHALMOLOGY
OCTOBER 2010
RESULTS FORTY EYES FROM 39 PATIENTS WERE INCLUDED. BASELINE
patient and ocular characteristics are summarized in Table 1. The mean follow-up was 19.1 months (range 6-34 months). Predisposing factors were present in 16 of 40 cases. Of the 24 cases without predisposing factors, only 5 mentioned sensation of flashes. Of 40 eyes, tears were visible by funduscopy in 13 cases but we were not able to adequately secure these tears with laser photocoagulation because of the density of the hemorrhage. In 17 eyes 1 or more defects were identified with U/S and in 10 eyes no tears could be found on U/S. Intraoperative details are depicted in Table 2. All 40 cases underwent vitrectomy for removal of blood and treatment of defects. Sixteen of 40 cases were operated with the conventional 20-gauge system and the remaining 24 cases with the sutureless 25-gauge platform. In all cases, a preexistent posterior vitreous detachment (PVD) was present over the posterior pole. Separation of the posterior vitreous membrane and retina in more peripheral areas varied but was not recorded systematically. Delay between decision to treat and actual treatment was on average 2.7 days. There were no differences in retinal tear incidence or complications between the 20-gauge and the 25-gauge cases. A total of 69 retinal tears with persistent vitreoretinal traction were found intraoperatively. Four eyes had 4 tears, 6 eyes had 3 tears, 10 eyes had 2 tears, and 15 eyes had 1 single tear. Sixty-two were unsupported tears, of which 32 were newly discovered tears (46.4%), the rest being tears that had been identified preoperatively but not lasered or not lasered adequately. Seven tears were supported tears, meaning that these tears had had adequate retinopexy in the past. In 1 case, U/S identified a tear but no tears could be found intraoperatively. Number of tears was evenly distributed over 20- and 25-gauge cases. Preoperative lens status is depicted in Table 1. Of the 30 phakic eyes, 23 appeared to have 1 or more unsupported tears (76.7%), while of the 10 pseudophakic eyes, 7 had unsupported tears (70.0%). This small difference was not statistically significant (P ⫽ .673, 2). Thus, preoperative lens status was not predictive of presence of unsupported tears. In 2 cases, a retinal detachment was encountered where no detachment was present at the last evaluation. We designated these 2 cases as new detachments. Both retinal detachments were macula-on and were repaired intraoperatively. Among the 40 cases, 1 or more tears were present in the superior half of the eye in 17 cases and in only 3 cases tears were confined to the inferior half. In 13 cases tears were found superiorly as well as inferiorly (Table 2). Thus, although there is a predilection for a superior location, in 40% of cases (16/40) tears occurred in the lower quadrants. Choice of tamponade was made on an individual basis. In 1 case, a small bubble of pure C3F8 was considered adequate. VOL. 150, NO. 4
EARLY VITRECTOMY
FOR
During follow-up, 9 of the total of 30 phakic eyes developed cataract. One case had to be reoperated for a macular pucker after 3 months. Two eyes developed an RRD, both occurring 4 months after surgery. Both cases were macula-on detachments and were repaired by repeat vitrectomy. In both cases the retina stayed attached after the second vitrectomy. Interestingly, we had not identified any tears in 1 of these cases during the primary surgery. In the second case, the causative tear was not documented during the first vitrectomy and should thus be considered as either missed or having developed after the first vitrectomy. Visual acuity was recorded preoperatively and postoperatively. There was improvement in visual acuity in 38 of 40 cases. The median preoperative VA was 1/60. The median postoperative VA was 0.8. This improvement is large and statistically significant (P ⬍ .001; Wilcoxon signed rank test). We assessed the sensitivity of our U/S in a selection of 27 cases in which too much blood was present for funduscopic evaluation and a full U/S investigation was performed. In these 27 eyes, a total of 19 tears were identified by U/S. During operation, 34 tears were identified in these 27 eyes, leading to a sensitivity of preoperative detection of tears by U/S of 55.9%. In 8 cases, surgery was still performed while no tears were detected by U/S. In 2 cases, there were predisposing factors (retinal tear in 1, RRD in the other) that justified a decision to treat. In the remaining 6 cases, however, the decision to operate was not based on the presence of predisposing factors. In the 2 cases with predisposing factors, no tears were found during operation. In the other 6 cases, tears were detected in 4 and no tears were found in 2. In 9 cases, no unsupported tears were found during surgery, so in these cases the operation could be considered unnecessary in hindsight. In 6 of these 9 cases, 1 or more predisposing factors were identified preoperatively. We assessed the relation between preoperative predisposing factors and presence of unsupported tears intraoperatively, needing treatment. Of 24 cases without predisposing factors, 21 cases appeared to have 1 or more unsupported tears (87.5%). In the 16 cases that did have a history of predisposing factors, only 9 had unsupported tears (56.3%). This unexpected inverse relation was actually statistically significant (P ⫽ .034, Fisher exact test). So a history of predisposing factors in our series was not related to the presence of an unsupported tear.
DISCUSSION HISTORICALLY, THE MANAGEMENT OF VITREOUS HEMOR-
rhage of unknown cause has been conservative.1 Upright positioning combined with immobilization and bilateral patching were advocated to allow faster clearance of blood in the upper vitreous cavity.2,3 The current preferred practice patterns of the American Academy of Ophthalmology6 states that in eyes in which a retinal tear is VITREOUS HEMORRHAGE
531
suspected, a conservative approach with repeat U/S should be followed, and a recent survey of management practices in the UK showed that most ophthalmologists prefer awaiting spontaneous resolution of the hemorrhage.7 Predisposing factors as rhegmatogenous incidents in either eye could help the clinician in their assessment of risk. A retinal detachment of the contralateral eye in the past history leads to a 75% incidence of retinal detachment, when confronted with a dense vitreous hemorrhage.8 In our series, however, eliciting a predisposing factor in the history did not influence the chance of finding an unsupported tear during vitrectomy. Earlier reports8 –10 have mentioned that retinal tears occur more frequently in the upper quadrants in the setting of a PVD. In our series, the predilection for the upper quadrants was corroborated, but still in as much as 40% of cases tears were found in the lower quadrants. The presumption that it suffices to evaluate only the upper quadrants for presence of tears is for this reason debatable. In cases with retina-obscuring vitreous hemorrhage, B-scan U/S is the best mode of evaluation of the integrity of the posterior segment. Sensitivity of tear detection was determined by several studies and varied between 92% in a small series and as low as 44%.4,9,10 One possible cause of this large variety is that the sensitivity in detection of large tears is high, but lower in detecting smaller tears. Another source of variety is the large dependence of accuracy on experience of the individual U/S operator. The findings in our current series suggest that the accuracy of tear detection is suboptimal, with a sensitivity of only 55.9%. Earlier studies closely followed cases of dense vitreous hemorrhage.8,11 During follow-up, 1 or more tears developed in 75% of cases. Even more seriously, 44% of cases developed an RRD, of which 28.6% were macula-off.8 Another 44.4% of eyes ultimately required vitrectomy for nonclearing vitreous hemorrhage. This agrees with a conservative, prospective follow-up,11 where 67% of eyes with dense vitreous hemorrhage appeared to have tears and even 39% had frank retinal detachment. One report investigated outcome of early vitrectomy in the context of dense vitreous hemorrhage with presumed retinal tears.
Their delay between presentation and vitrectomy was 6.3 days. New retinal tears were discovered intraoperatively in 75% of cases. Our findings are in agreement with these earlier reports. We were confronted with the identification of a high number of new tears (46.4%), tears that were encountered only during vitrectomy and missed by funduscopy and U/S preoperatively, and with the discovery of frank RRD in 2 cases at the start of vitrectomy, where no RRD was found in the last preoperative evaluation, meaning that the RRD had developed in the direct preoperative period. The presence of blood in an eye with a retinal detachment is known to increase the risk of developing proliferative vitreoretinopathy (PVR).12–15 In 1 study in which cases with dense vitreous hemorrhage were followed, the risk of PVR was 19%.8 In the small number of cases of retinal detachment in our current series, no PVR occurred. Our findings illustrate the inherent risk of a conservative strategy. With our current diagnostic techniques only part of retinal tears can be identified. In the context of an acute posterior vitreous detachment, the presence of an unsupported or untreated retinal tear carries an unacceptable risk of progression to RRD with increased risk of PVR, leading to a high risk of irreversible damage of visual function. This poses to the ophthalmologist a classic dilemma. Vitrectomy is a highly invasive procedure with well-known risks involved. In an elective setting, vitrectomy induces retinal tears with the potential to cause retinal detachments. As in any intraocular procedure, there is a risk of endophthalmitis and choroidal hemorrhage. In phakic cases, vitrectomy has a 100% risk of accelerating the development of cataract. Adopting a more aggressive strategy implies accepting that we impose these risks on individual cases that could have had good outcome with a conservative approach. Although the outcome in our series in terms of visual acuity and fast recovery of visual function was good and complications were acceptable, the small size and lack of any control or randomization limits us in drawing firm conclusions. There is great need for prospective controlled studies to guide the clinician in this important issue.
THE AUTHORS INDICATE NO FINANCIAL SUPPORT OR FINANCIAL CONFLICT OF INTEREST. INVOLVED IN DESIGN OF THE STUDY (H.S.T.); conduct of the study (H.S.T., M.M., H.M.B.); collection (H.S.T., M.M., H.M.B.), management (H.S.T.), analysis, and interpretation of the data (H.S.T., M.M., H.M.B.); and preparation, review, and approval of the manuscript (H.S.T., M.M., H.M.B.). The institutional review board at the University of Amsterdam declared that this type of retrospective study waived the need for IRB approval, in accordance with Dutch law on human clinical trials.
4. Rabinowitz R, Yagev R, Shoham A, Lifshitz T. Comparison between clinical and ultrasound findings in patients with vitreous hemorrhage. Eye 2004;18(3):253–256. 5. Dhingra N, Pearce I, Wong D. Early vitrectomy for fundusobscuring dense vitreous haemorrhage from presumptive retinal tears. Graefes Arch Clin Exp Ophthalmol 2007;245(2):301– 304. 6. Preferred Practice Pattern. Management of Posterior Vitreous Detachment, Retinal Breaks, and Lattice Degenera-
REFERENCES 1. Spraul CW, Grossniklaus HE. Vitreous hemorrhage. Surv Ophthalmol 1997;42(1):3–39. 2. Lincoff H, Kreissig I, Wolkstein M. Acute vitreous haemorrhage: a clinical report. Br J Ophthalmol 1976;60(6):454 – 458. 3. Wilkinson CP. What ever happened to bilateral patching? Retina 2005;25(4):393–394.
532
AMERICAN JOURNAL
OF
OPHTHALMOLOGY
OCTOBER 2010
7.
8.
9.
10.
tion. San Francisco: American Academy of Ophthalmology; 2008. Vote BJ, Membrey WL, Casswell AG. Vitreous haemorrhage without obvious cause: national survey of management practices. Eye 2005;19(7):770 –777. Sarrafizadeh R, Hassan TS, Ruby AJ, et al. Incidence of retinal detachment and visual outcome in eyes presenting with posterior vitreous separation and dense fundusobscuring vitreous hemorrhage. Ophthalmology 2001;108(12): 2273–2278. DiBernardo C, Blodi B, Byrne SF. Echographic evaluation of retinal tears in patients with spontaneous vitreous hemorrhage. Arch Ophthalmol 1992;110(4):511–514. Nischal KK, James JN, McAllister J. The use of dynamic ultrasound B-scan to detect retinal tears in spontaneous vitreous haemorrhage. Eye 1995;9(4):502–506.
11. Lindgren G, Sjödell L, Lindblom B. A prospective study of dense spontaneous vitreous hemorrhage. Am J Ophthalmol 1995;119(4):458 – 465. 12. Yeung L, Yang KJ, Chen TL, et al. Association between severity of vitreous haemorrhage and visual outcome in primary rhegmatogenous retinal detachment. Acta Ophthalmol 2008;86(2):165–169. 13. Tseng W, Cortez RT, Ramirez G, Stinnett S, Jaffe GJ. Prevalence and risk factors for proliferative vitreoretinopathy in eyes with rhegmatogenous retinal detachment but no previous vitreoretinal surgery. Am J Ophthalmol 2004;137(6):1105–1115. 14. Asaria RH, Kon CH, Bunce C, et al. How to predict proliferative vitreoretinopathy: a prospective study. Ophthalmology 2001;108(7):1184 –1186. 15. Cowley M, Conway BP, Campochiaro PA, Kaiser D, Gaskin H. Clinical risk factors for proliferative vitreoretinopathy. Arch Ophthalmol 1989;107(8):1147–1151.
AJO History of Ophthalmology Series Paul Sullivan and Intraocular Sulfur Hexafluoride
T
he use of air to attempt to reattach the retina predates Jules Gonin’s discovery of the underlying etiology of rhegmatogenous retinal detachment. In the 1930’s, Rosengren frequently used air to improve the anatomic success rate of his surgical procedures. In the 1950’s and 1960’s many retinal surgeons used intraocular air but struggled with the main limitation of the device, that is, the relatively short duration of the gas within the vitreous cavity. Fortuitous events
VOL. 150, NO. 4
EARLY VITRECTOMY
FOR
allowed Dr. Paul Sullivan to introduce the first expansile gas, sulfur hexafluoride (SF6), to Dr. Edward W.D. Norton and his colleagues at the University of Miami in the late 1960’s. The use of SF6 helped revolutionize retinal detachment surgery and allowed for further advances in treating other posterior segment diseases. Submitted by Christopher F. Blodi from the Cogan Ophthalmic History Society.
VITREOUS HEMORRHAGE
533