A O S 2001

34

Hemostatic effects of SF6 afterdiabetic vitrectomy for vitreoushemorrhageCh. N. Koutsandrea, M. N. Apostolopoulos, D. Z. Chatzoulis,E. A. Parikakis and G. P. Theodossiadis

University of Athens, Department of Ophthalmology, General Hospital of Athens,Cholargos, Athens, Greece

ABSTRACT.Purpose: To investigate the hemostatic effects of SF6 gas in preventing post-operative vitreous hemorrhage in diabetic vitrectomy.Methods: A prospective, randomized study of 33 diabetic eyes with vitreoushemorrhage, treated by vitrectomy. In 17 of our cases SF6 20% was injectedinto the eye at the end of the operation, while in 16 cases BSS remained in thevitreous cavity.Results: The incidence of vitreous hemorrhage recurrence was 17.6% for theSF6 group and 12.5% for the BSS group (statistically not significant). Pro-gression of lens opacities was observed in 23.5% of the SF6 group, and in 18.8%of the BSS group (statistically not significant, with a higher incidence in theSF6 group).Conclusions: SF6 gas did not show hemostatic effects in the cases studied. Fur-thermore, it may have contributed to cataract progression. Therefore we suggestthat the use of SF6 is not recommended as a treatment modality in preventingnew vitreous hemorrhage after diabetic vitrectomy.

Key words: diabetic retinopathy – vitrectomy – vitreous hemorrhage recurrence.

Acta Ophthalmol. Scand. 2001: 79: 34–38Copyright c Acta Ophthalmol Scand 2001. ISSN 1395-3907

One of the most serious compli-cations of vitrectomy for prolifer-

ative diabetic retinopathy is vitreoushemorrhage recurrence. A review of sev-eral studies indicates a wide range of per-centage for this complication (8–75%)(Aaberg 1979; Blankenship et al. 1979;Novak et al. 1984; Peyman et al. 1976;Rice & Michels 1980; Schachat et al.1983).

Advances in surgical technique and theuse of more sophisticated instrumen-tation, including endodiathermy and en-dolaser have reduced the incidence of thevitreous hemorrhage recurrences to 25%(Blumenkranz 1992).

The age of the patients, the duration ofthe operation and the poor preoperativevisual acuity are considered to be amongthe predisposing factors for vitreous

hemorrhage recurrences (Tolentino 1989).Various substances have been used to

prevent vitreous hemorrhage recurrencesafter diabetic vitrectomy. Thompson etal. (1987) used SF6 gas in some of theircases trying to prevent early postopera-tive bleeding. Joondeph and Blankenship(1989) have studied the hemostatic effectsof air injection into the vitreous cavity indiabetic eyes, treated by vitrectomy dueto vitreous hemorrhage or tractional reti-nal detachment.

Compared to former series, a greaternumber of more carefully selected dia-betic cases were studied for the purposesof this study, for a longer follow-upperiod. The eyes underwent vitrectomyfor vitreous hemorrhage. Eyes with trac-tional retinal detachment, rubeosis iridis,or iatrogenic tears were not included. The

aim of this prospective, controlled studywas to analyze the effects, especially thehemostatic ones, of SF6 gas injected inthe vitreous cavity.

Material and MethodsThirty-four diabetic eyes with vitreoushemorrhage, without retinal detachment,were operated on in the Department ofOphthalmology of the University ofAthens, between January 1994 and De-cember 1995. The eyes were mainly se-lected by biomicroscopic and indirectophthalmoscope examination and byultrasonography. All the patients werephakic with clear lenses.

The 34 cases were randomly divided intwo groups, using a sealed envelope sys-tem with red and blue cards (17 red and17 blue). It was predecided to tamponadethe vitreous cavity of the eyes of groupA (red cards) with SF6 at the end of theoperation, and to retain BSS-plus in thevitreous cavity of the eyes of group B(blue cards). One out of the 34 cases wasexcluded from the study because a retinalbreak was created intraoperatively. Thecurrent study finally included 33 eyes, 17in group A (SF6 group) and 16 in groupB (BSS group).

Fifteen out of the 33 patients were menand 18 women. The age of our patientsranged from 44 to 81 years (Mean age:59.8 years old).

Six of the 33 patients had diabetes typeI, while the remaining 27 patients haddiabetes type II, or belonged to the mixedgroup. Three of the patients with diabetestype I belonged to the group A and 3 tothe group B.

A O S 2001

35

The preoperative ocular examinationincluded measurement of the best cor-rected visual acuity, slit-lamp examina-tion of the anterior segment especially theiris, the angle of the anterior chamberand the lens, which was photographedusing color slit-lamp images and retroil-lumination images. The intraocularpressure was measured. Biomicroscopy,indirect ophthalmoscopy and ultrasono-graphy were also performed. The pre-operative visual acuity ranged from lightperception to 1/40.

All the eyes included in this studywere phakic with clear lenses accordingto the LOCS III classification. The pre-operative intraocular pressure waswithin normal limits in all our patients.Fifteen of 33 patients (7 from group Aand 8 from group B) had undergonepreoperative laser treatment (Table 1).The duration of the preoperative vit-reous hemorrhage in the eyes studiedranged from 2–8 months.

The patients underwent pars-plana vi-trectomy, which was performed undergeneral anesthesia or more often underlocal anesthesia. Vascularized preretinalmembranes were removed. Cauterizationof the new vascular formation was per-formed by endodiathermy. Special atten-tion was given to the complete removalof the cortical vitreous. Scatter laserphotocoagulation was applied in all ourpatients. None of our patients have beentreated with combined vitrectomy andcataract extraction.

At the end of the operation we per-formed fluid/air and then air/SF6 ex-change in the eyes of the group A. Theeyes of group B remained with BSS-plusin the vitreous cavity.

Postoperative follow-up examinationwas performed weekly during the firstmonth, and then monthly for the first sixpostoperative months after the operation.Further follow-up examinations wereperformed every six months.

The postoperative examination in-cluded measurement of the visual acuity,slit lamp examination, measurement ofthe intraocular pressure, biomicroscopy,indirect ophthalmoscopy and fluoresceinangiography after the first postoperativemonth. Additional laser treatment wasapplied when necessary.

We defined as recurrence of vitreoushemorrhage the situation in which visual-ization of retinal vessels and fundus de-tails was not possible and the fundus re-flex was red-orange or dark red. With thisdefinition, we have excluded small post-

operative vitreous hemorrhages, spon-taneously resolving.

We also defined as cataract formationthe progression of lens opacities of 1 ormore grades, according to the LOCS IIIsystem.

The follow-up period ranged from 5 to28 months; mean follow-up 18.3 months.

The statistical analysis of the incidenceof vitreous hemorrhage recurrences andcataract formation between the twogroups was made by the Fisher’s exacttest.

ResultsVitreous hemorrhage recurrence was ob-served in 3 out of the 17 cases (17.6%) ofgroup A (SF6 group), while recurrencewas noticed in 2 out of the 16 cases

Table 1. Preoperative clinical findings.

LaserPatient Visual acuity treatment Gas at the end Duration of VHNo Sex Age (preop) (preop) of the operation (preop) months

1 M 59 CF π π 42 M 62 CF - π 53 F 55 LP - π 34 M 73 CF - π 65 M 60 1/40 π π 66 M 51 LP - π 47 F 66 LP π π 78 F 64 LP - π 49 M 44 HM - π 2

10 F 57 1/40 - π 311 F 58 HM π π 512 M 62 HM π π 613 M 63 HM π π 714 F 58 CF - π 415 M 52 CF π π 316 F 72 LP π π 517 M 68 HM - π 418 F 57 HM π - 419 F 63 CF π - 620 M 65 HM - - 521 M 68 1/40 π - 322 F 58 CF - - 423 M 44 CF π - 224 M 59 HM π - 425 F 54 HM - - 426 F 50 HM - - 327 F 60 CF π - 528 F 68 LP - - 629 F 81 LP π - 830 M 60 HM π - 431 F 55 HM - - 332 M 60 LP - - 433 F 50 LP - - 3

MΩmale, FΩfemale, LPΩlight perception, HMΩhand movements, CFΩcounting fingers, NΩwithin normal limits, VHΩvitreous hemorrhage.

(12.5%) of group B (BSS group). Thenew hemorrhage appeared between the5th and 20th postoperative day in allcases. One out of the 6 patients with dia-betes type I, belonging to the group A,developed vitreous hemorrhage recur-rence. The eyes which developed vitreoushemorrhage recurrence were all success-fully treated, after a 2–3-month waitingperiod, in the operating theater, with‘‘lavage’’, endodiathermy and additionalendolaser. All these cases had no othervitreous hemorrhage recurrence duringthe follow-up period.

No other serious complications wererecorded in our patients during the fol-low-up period (Tables 2, 3).

Three months to one year postopera-tivelly cataract formation was noticed in7 out of the 17 patients of group A(41.5% of the cases) and 4 out of the 16patients of the group B (24.9% of the

A O S 2001

36

Table 2. Results of Group A (with Gas).

Recurrence of Intraocular Rubeosis VisualPatient vitreous Cataract pressure iridis acuity Follow-upNo Gas hemorrhage formation (post) (post) (post) (months)

1 π - - - - 2/10 232 π - π - - 1/10 273 π - - ∫ - 3/10 284 π π - π π L.P. 95 π - - - - 5/10 266 π - π ∫ - 2/10 287 π π - - - 2/10 178 π - - ∫ - 4/10 279 π π - - - 5/10 26

10 π - - - - 6/10 2811 π - - ∫ - 2/10 2612 π - - - - 3/10 1513 π - - - - 1/10 2614 π - - ∫ - 1/20 1215 π - π - - 2/10 1716 π - - - - 2/10 2117 π - π ∫ - 4/10 19

(∫)Ωtransient increase of the intraocular pressure, (π)Ωpermanent increase of the intraocularpressure, L.P.Ωlight perception

Table 3. Results of Group B (without Gas).

Vitreous Intraocular Rubeosis VisualPatient hemorrhage Cataract pressure iridis acuity Follow-upNo Gas recurrence formation (post) (post) (post) (months)

18 - - - - - 4/10 919 - - - ∫ - 3/10 1620 - - π - - 1/10 721 - - - - - 6/10 2222 - - - - - 2/10 2623 - - - - - 2/10 2624 - - - - - 3/10 525 - π - - - 3/10 1526 - - - ∫ - 2/10 927 - π - - - 5/10 2728 - - - - - 2/10 1029 - - π - - 1/40 730 - - - - - 4/10 1231 - - - - - 3/10 732 - - π - - 1/20 1433 - - - - - 3/10 19

(∫)Ωtransient increase of the intraocular pressure (π)Ωpermanent increase of the intraocularpressure.

Table 4. Comparison of the incidence of vitreous hemorrhage recurrences between the SF6 group(group A) and the BSS group (group B).

Vitreous hemorrhage SF6 group BSS grouprecurrences (group A) (group B) Total

Yes 3 2 5No 14 14 28

Total 17 16 33

pΩ0.53 (Fisher’s exact test).

cases). One case of group A (3%) de-veloped rubeosis iridis, due to excessiveischemia of the retina.

The intraocular pressure presented atransient ( up to 20 days) and moderateincrease (22–25 mm Hg Appl) in 6 eyesof group A and in 2 eyes of group B.These cases were treated with a beta-blocker. One eye of group A developeda permanent increase of the intraocularpressure (the patient with the rubeosis ir-idis). No particular problem from thecornea was observed.

Postoperative visual acuity rangedfrom light perception to 6/10 for group Aand from 1/40 to 6/10 for group B, and itwas equal or better than 2/10 in 81.2% ofthe cases of both groups (Fig. 1).

Regarding the vitreous hemorrhagerecurrences the statistical analysisshowed no significant difference betweenthe groups A and B. In the Fisher’s ex-act test the p value was pΩ0.53 (Table4). The cataract formation revealed nostatistically significant differencebetween the two groups as well. In theFisher’s exact test the p value was pΩ0.27 (Table 5).

DiscussionVitrectomy is the appropriate treatmentfor proliferative diabetic retinopathy withpersistent vitreous hemorrhage. The timeof surgical treatment has been discussedby the Diabetic Retinopathy VitrectomyStudy Group (1985; 1988; 1988; 1989;1990). The incidence of postoperative vit-reous hemorrhage recurrence in thesecases varies (Aaberg 1979; Blankenship etal. 1979; Novak et al. 1984; Peyman et al.1976; Rice ( Michels 1980; Schachat etal.1983; Blumenkranz 1992).

Recurrence of vitreous hemorrhagecan interfere with the fundus examina-tion and the completion of panretinalphotocoagulation. It may compromisethe final visual acuity, and it can also con-tribute to cell proliferation (Campochia-ro et al. 1984).

Several substances, such as thrombin(Thompson et al. 1986), aminocaproidacid (De Bustros et al.1985), and sodiumhyaluronate, have been used in order tominimise the incidence of vitreoushemorrhage recurrences after diabetic vi-trectomy. The sodium hyaluronate has amechanical action, trapping the hemor-rhage (Folk et al. 1986). As reported byCharles (1987) silicone oil decreases

A O S 2001

37

Fig. 1. Scattergram showing the preoperative visual acuity and the visual acuity at last visit.

Table 5. Comparison of the incidence of cataract formation between the SF6 group (group A)and the BSS group (group B).

SF6 group BSS groupCataract formation (group B) (group A) Total

Yes 7 4 11No 10 12 22

Total 17 16 33

pΩ0.27 (Fisher’s exact test).

hemorrhage, through its ‘‘recompart-mentalization’’ of the eye, by preventingdiffusion of the clotting substances awayfrom the bleeding sites.

Injection of air into the vitreous cavityis believed to act mechanically as a tem-porary tamponade of the bleeding sitesof the retina, obtaining hemostasis bynaturally occurring clotting mechanisms.Joondeph and Blankenship (1989) in aclinical and experimental study con-cluded that the use of air in the vitreouscavity after diabetic vitrectomy did notreduce the number of vitreous hemor-rhage recurrences. They also reportedcataract formation as an additional prob-lem of the air injection. The experimental

model used in this investigation was notconsidered to be completely analogous tothe human diabetic eye during and aftervitrectomy. The clinical portion of theirstudy included diabetic eyes with vitreoushemorrhage, as well as eyes with severeproliferative diabetic retinopathy withtractional retinal detachment involvingthe macula. The authors refer to an het-erogeneous material, with more severeretinal pathology. Such cases usually de-mand lengthy procedures, with multipleand more traumatic manipulations.These could be the reasons of the in-creased incidence of vitreous hemorrhagerecurrence, as well as the increased inci-dence of cataract formation. The trac-

tional retinal detachment involving themacula in some cases might have led to apoorer final visual acuity.

Thompson et al. (1987) mention theuse of SF6 gas in some of their cases inorder to tamponade suspected retinalbreaks or trying to prevent perioperativevitreous hemorrhage. The authors do notreport any data for the effect of the gason the incidence of vitreous hemorrhagerecurrence.

In the current study diabetic eyes wereselected and operated on for vitreoushemorrhage exclusively. Cases with trac-tional retinal detachment, rubeosis iridisor cases in which iatrogenic breaks wereintraoperatively created, were not in-cluded. We had no additional variablesinfluencing the results of our study.Therefore we were able to evaluate the ef-fect of SF6, versus BSS more accurately.

Our cases were randomly assigned anddivided in two equal groups. The dur-ation of the operating procedure wasrather short, because of the limited path-ology of the cases. Previous panretinallaser photocoagulation had been per-formed in a similar number of cases inboth groups.

The overall incidence of vitreoushemorrhage recurrence in our study was15.1% (5/33 eyes), an even lower percen-tage than the recently reported one (Blu-menkranz 1992).

This could be due to the strict criteriathat we used in selecting our patients, ex-cluding cases with extended retinal path-ology. An additional reason could be thedefinition of vitreous hemorrhage recur-rence as a diffuse, rather dense hemor-rhage, according to which small hemor-rhages were excluded.

The incidence of vitreous hemorrhagerecurrence (17.6% for group A and 12.5%for group B ) did not show any statisti-cally significant difference. Consideringalso the fact that the vitreous hemor-rhage recurrences usually appeared soonafter the operation, even when SF6 wasstill present, there appears to be no ad-vantage in using it.

For the purposes of this study factorsrelated to the incidence of vitreoushemorrhage recurrence such as youngage, or poor preoperative visual acuity(Tolentino et al. 1989) were not included.The duration of the operating procedurewas not correlated to the incidence of vit-reous hemorrhage recurrences, since itwas similar in all cases.

Blankenship (1986) reported an inci-dence of rebleeding of 12% for the aphak-

A O S 2001

38

ic eyes and 18% for the phakic eyes. No-vak and co-workers (1984) had found thehigher incidence of vitreous hemorrhagein phakic eyes compared to aphakic eyesto be statistically significant. In our seriesall the treated eyes were phakic with clearlenses,according to the LOCS III classi-fication.

The cases which presented vitreoushemorrhage recurrence were treated inthe operating theater with ‘‘lavage’’(Blankenship 1986) , endodiathermy andadditional laser treatment. The pro-gression of lens opacities had an inci-dence of 41.5% for group A and 24.9%for group B. Although the analysis re-vealed no statistically significant differ-ence, the percentage of the cataract for-mation in the gas group was higher. Thereported percentage of cataract forma-tion (Thompson et al. 1986, 1987) is alsohigher compared to that of our study.This could be attributed to the shorterduration of the operation in our cases,with less manipulations, due to thelimited pathology.

Our findings presented a higher inci-dence (81.2%) of good final visual acuity(equal or more than 2/10), compared tothe results of other investigators, since allour cases had the macula attached. Thefinal visual acuity was similar betweengroups A and B (equal or more than 2/10 in 81.2 % of the cases), regardless theuse of gas. This result does not supportthe conclusion of Thompson et al. (1986;1987) that the use of gas has a bad prog-nosis for the final visual acuity. On theother hand, it agrees with the findings ofJoondeph and Blankenship (1989), whoreported that the use of gas does not in-terfere with the final visual outcome.

Only one case of group A developedrubeosis iridis, due to excessive ischemia,and ended up with a permanent increaseof the intraocular pressure. The rest ofour cases had only a transient and mod-erate increase of the intraocular pressure,in a higher rate (35.2%) for group A,compared to group B. Local treatmentwith a b-blocker for a short period oftime, was enough for all our cases, exceptfor the one which developed neovascularglaucoma.

The results of the current study indi-cate that there are no hemostatic effectsof the SF6 gas, when it is injected into thevitreous cavity at the end of a diabeticvitrectomy. It was also noticed that the

presence of gas in the vitreous cavityseemed to cause a trend toward cataractprogression.

We conclude from the findings of ourprospective, controlled study of carefullyselected diabetic patients, that there ap-pears to be no advantage in using SF6 gasin diabetic vitrectomy, in order to preventa postoperative recurrence of vitreoushemorrhage.

ReferencesAaberg TM (1979): Clinical results in vitrecto-

my for diabetic traction retinal detachment.Am J Ophthalmol 88: 246–53.

Blankenship G, Cortez R & Machemer R(1979): The lens and pars plana vitrectomyfor diabetic retinopathy complications. ArchOphthalmol 97: 1263–1267.

Blankenship GW (1986): Management of vit-reous cavity hemorrhage following parsplana vitrectomy for diabetic retinopathy.Ophthalmology 93: 39–44.

Blumenkranz M.S (1992): New Developmentin Diabetic Vitrectomy. The VitreoretinalFrontier Dallas, Texas 352.

Campochiaro PA, Jerdan JA & Glasser BM(1984): Serum contains chemoattractants forhuman retinal pigment epithelial cells. ArchOphthalmol 102: 1830–1833.

Charles S (1987): Vitreous Microsurgery, 2nded. Baltimore: Williams & Wilkins 79–122.

de Bustros S, Glaser BM, Michels RG & AuerC (1985): Effect of epsilon-aminocaproicacid on postvitrectomy hemorrhage. ArchOphthalmol 103: 219–221.

The Diabetic Retinopathy Vitrectomy StudyGroup (1989): Two-year course of visualacuity in severe proliferative diabetic retino-pathy with conventional management. Oph-thalmology 92: 492–501.

Folk JC, Wengeist TA, Packer AJ & HowcroftMJ (1986): Sodium hyaluronate (Healon) inclosed vitrectomy. Ophthalmic Surg 17: 299–306.

Joondeph BC & Blankenship GW (1989):Hemostatic Effects of Air Versus Fluid inDiabetic Vitrectomy. Ophthalmology 96:1701–1707.

Novak MA, Rice TA, Michels RG & Auer C(1984): Vitreous hemorrhage after vitrecto-my for diabetic retinopathy. Ophthalmology91: 1485–1489.

Peyman GA, Raichand M, Huamonte FU,Nagpal KC, Goldberg MF & Sanders DR(1976): Vitrectomy in 125 eyes with diabeticvitreous hemorrhage. Br J Ophthalmol 60:752–755.

Rice TA & Michels RG (1980): Long-term an-atomic and functional results of vitrectomyfor diabetic retinopathy. Am J Ophthalmol90: 297–303.

Schachat AP, Oyakawa RT, Michels RG & RiceTA (1983): Complications of vitreoussurgery for diabetic retinopathy. II. Post-operative complications. Ophthalmology 90:522–530.

The Diabetic Retinopathy Vitrectomy StudyResearch Group (1985): Early Vitrectomyfor Severe Vitreous Hemorrhage in DiabeticRetinopathy. Two Years Results of a Ran-domized Trial. Diabetic Retinopathy Vitrec-tomy Study Report 2. Arch Ophthalmol103: 1644–1652.

The Diabetic Retinopathy Vitrectomy StudyResearch Group (1988): Early Vitrectomyfor Severe Proliferative diabetic Retinopathyin Eyes with Useful Vision: Results of a ran-domized trial – Diabetic Retinopathy Vitrec-tomy Study Report 3. Ophthalmology 95:1307–1320.

The Diabetic Retinopathy Vitrectomy StudyResearch Group (1988): Early Vitrectomyfor Severe Proliferative Diabetic Retinopa-thy in Eyes With Useful Vision. Clinical Ap-plication of Results of a Randomized Trial-Diabetic Retinopathy Vitrectomy Study Re-port 4. Ophthalmology 95: 1321–1334.

The Diabetic Retinopathy Vitrectomy StudyResearch Group (1990): Early Vitrectomyfor Severe Vitreous Hemorrhage in DiabeticRetinopathy: 4 year results of a randomizedtrial – Diabetic Retinopathy Study Report 5.Arch Ophthalmol 108: 958–964.

Thompson JT, Auer CL, de Bustros S, MichelsRG, Rice TA & Glaser BM (1986): Prognos-tic indicators of success and failure in vitrec-tomy for diabetic retinopathy. Ophthal-mology 93: 290–295.

Thompson JT, Glaser BM, Michels RG & deBustros S (1986): The use of intravitrealthrombin to control hemorrhage during vi-trectomy. Ophthalmology 93: 279.

Thompson JT, de Bustros S, Michels RG &Rice TA (1987): Results and prognostic fac-tors in vitrectomy for diabetic vitreoushemorrhage. Arch Ophthalmol 105: 191–5.

Thompson JT, de Bustros S, Michels RG &Rice TA (1987): Results and prognostic fac-tors in vitrectomy for diabetic traction reti-nal detachment of the macula. ArchOphthalmol 105: 497–502.

Tolentino FI, Cajita VN, Gancayco T & SkalesS (1989): Vitreous hemorrhage after closedvitrectomy for proliferative diabetic retino-pathy. Ophthalmology 96: 1495–1500.

Received on August 10th, 1999.Accepted on May 4th, 2000.

Corresponding author:

Chryssanthi N. Koutsandrea, MD31 Andritsaenis StreetAthens, 11142, Greece

Tel: 30 1 6812328Fax: 30 1 2231508