Heavy silicone oil versus standard silicone oil in as vitreous tamponade in inferior PVR (HSO Study): interim analysis

Abstract

Purpose: The Heavy Silicone Oil versus Standard Silicone Oil Study (HSO study) is designed to answer the question whether a heavier-than-water tamponade improves the prognosis of eyes with proliferative vitreoretinopathy (PVR) of the lower retina. Methods: The HSO Study is a multicentre, randomized, prospective, controlled clinical trial stratified by surgeon comparing two endotamponades within a two-arm parallel-group design. Patients with inferiorly and posteriorly located PVR grade C-A6 were randomized to either HSO or standard silicone oil as a tamponading agent. The main end-point criteria are complete retinal attachment at 12 months and change in visual acuity (VA) 12 months postoperatively compared to the preoperative VA. Results: Forty-six patients treated with HSO were compared to 47 patients treated with standard silicone oil. There was no difference among the groups regarding baseline data. Three patients in the HSO and five patients in the standard silicone oil group fulfilled intraoperative exclusion criteria. There was no significant difference between both groups regarding anatomical success. Neither noninferiority nor superiority was shown with regard to final acuity. Conclusions: The HSO Study is the first randomized prospective clinical trial to compare heavy and standard silicone oil in patients with PVR of the lower retina. The intermediate results failed to demonstrate superiority of a heavy tamponade. The Heavy Silicone Oil Study (HSO study) was initiated to investigate whether a specific tamponade with HSO is able to reduce the rate of proliferative vitreoretinopathy (PVR) in aphakic or pseudophakic patients with inferior retinal tear or inferior PVR. Heavy silicone oil is a 25–75% mixture of perfluorohexyloctane and silicone oil, i.e. a mixture of 5000 cSt silicone oil (specific gravity of 0.98) and of 3.4 cSt F6H8 (specific gravity 1.33) resulting in a solution with a viscosity of 1347 cst and a gravity of 1.06. There are several reports on the use of heavier-than-water tamponades (Stappler et al. 2008; Wickham et al. 2010) that demonstrate the theoretical ability of heavier-than-water tamponades to reduce the rate of tractional late retinal redetachments through displacement of the PVR-stimulating environment from the inferior retina. The HSO study, however, is the first randomized trial to investigate the efficacy of heavier-than-water tamponades (Joussen et al. 2007). The HSO study aimed to answer the question whether there is a difference in the anatomical success rate that is ‘retinal attachment’ between an endotamponade lighter than water (standard silicone oil) versus an endotamponade heavier than water (HSO). Furthermore, the functional success that is ‘visual acuity’ (VA) was used to assess whether one of the tamponades offers advantages compared to the other. The HSO study was conducted as a prospective, multicentre, randomized, controlled, clinical trial comparing two different surgical procedures based on the application of different endotamponades in parallel treatment groups. The ethics committee of RWTH Aachen (study no. 124/03, EUDRA CT 2005-004154-27) and all local ethics committees agreed to the study protocol. Most details of the study design, e.g. surgical procedures and the follow-up, are described in HSO study report No.1 (Joussen et al. 2007). The alterations to the study protocol are described here. Inclusion and exclusion criteria are shown in Table 1. Every patient gave a written informed consent before participation in the study and agreed to be examined at four scheduled follow-up visits: within the first week after surgery; before and 8 weeks after removal surgery; and 1 year after surgery. Additional examinations were performed in case of reoperations or at any unscheduled visit (for example, because of concomitant therapy or adverse events). The disease under consideration can sufficiently be treated only by surgical intervention. Pars plana vitrectomy (PPV) was performed as described previously (Joussen et al. 2007): encircling band according to the surgeon’s preference, PPV via conventional three-port approach, removal of the flap of the retinal tear, if present, usage of perfuordecalin fluid (PFCL) to unfold the retina, retinopexy by cryopexy or laser coagulation, and relaxing retinotomies, if necessary. After verification of the eligibility criteria (including informed consent), randomization (opening of the prepared sealed envelopes) took place immediately prior to the filling of the tamponade that is after all vitreoretinal preparations were completed and the retina was flattened to the choroid by perfluorocarbons or air. PFCL – standard silicone exchange or PFCL-air-silicone exchange was performed according to the individual surgeon’s preferences. The tamponades were supposed to be in the eye for a minimum of 2 months and a maximum of 6 months. Retinotomies or peeling of tractional membranes during removal surgery was considered retinal affecting procedure. Reoperations and other medical interventions can be performed at any time depending on the surgeon’s decision and on the patient’s consent. Admissible and inadmissible concomitant therapies and surgical procedures are depicted in detail in the design paper (Joussen et al. 2007). The study aimed to determine the success rate of vitrectomy with heavy silicone compared to vitrectomy with standard silicone oil in eyes with PVR and retinal tears confined to the lower quadrants of the eye (PVR retinal detachment of grade C-A6 (10 over 6 to 2) P-12 according to Machemer (1968). The difference between the surgical procedures refers to the application of the different, above-mentioned endotamponades. The outcome measures were anatomical success rate that is ‘retinal attachment’ and ‘visual acuity’. Anatomical success was defined as complete retinal reattachment at the 12-month visit without any retina affecting preceding intervention, and VA means change in VA 12 months postoperatively when compared to the preoperative value (Joussen et al. 2007). The secondary end-point criteria combined the evaluation of the complete retinal attachment before endotamponade removal, a quality of life analysis, and the evaluation of the number of retina affecting reoperations within 1 year of follow-up. The current report describes the interim analysis of the HSO trial. The primary end-point ‘12-month attachment’ as well as the secondary end-points ‘attachment (before removal)’ and the ‘number of reoperations’ was assessed by masked examiners within the end-point committee meetings based on photographic documentation or surrogates. From the initiation of the study, the recruitment rate remained below expectance. Several surgeons indicated that the use of silicone oil 1000 cSt as the standard oil for the comparison to the heavy silicone would hold them off to recruit. Thus, with a vote from the ethics committee of the University of Cologne, the use of either silicone 1000 cSt or silicone 5000 cSt was up to the surgeon′s decision. Nevertheless, despite initiation of several international study centres, the recruitment rate did not improve, and the aim to recruit 700 patients within 4 years seemed to be impossible. Further because of the low recruitment rate, the study design was altered to a two-step adaptive design (Bauer & Köhne 1994). This substantial amendment was approved by the ethics committee of the University of Cologne on 6.4.2008. To prove our initial study hypothesis, we chose a significance level of 0.0233 in the first phase. However, according to the procedure, if the p-value is 0.5 or higher, our study has to be stopped with the decision of no treatment effect. The patient flow is displayed in Fig 1. A total of 94 patients (46 in the HSO group, 48 in the standard silicone oil group) were recruited to the study. All patients received the allocated treatment during the initial surgery. One control patient having no data except the randomization sheet was excluded from the analysis. Localization of the redetachment after tamponade with either HSO (A) or standard silicone oil (B) (clock hours of retached retina). After tamponade with heavy sillicone oil a redetachment is predominantly found in the upper quadrants. Between December 2003 and February 2008, 23 surgeons in 15 centres out of seven countries recruited 93 patients. The median number of patients per surgeon was 2 (range 1–22). The median number of patients was 1 (range 0–10) in the standard silicone oil group and 1 (range 0–12) in the HSO group. The treatment allocation was fairly balanced. We observed the differences in the number of patients per treatment was one for 15 surgeons, two for two surgeons and three for only one surgeon. Of the 93 patients included in the study, PVR was present in 86 patients (92.5%), giant retinal tears were present in four patients (4.3%), and both PVR and giant tears were found in three patients (3.2%). None of the patients met any of the exclusion criteria that could be assessed preoperatively. Intraoperatively, eight patients presented with at least one exclusion criterion. In seven patients, the retinotomy/holes/tears were found in the upper quadrant within the 10 over 12 to 2 hr position. Superior anterior PVR grade C A6 in this area was present in 3 eyes. In 85 patients, no exclusion criterion was present. Three patients in the HSO group and 5 patients in the standard silicone group fulfilled intraoperative exclusion criteria. Treatment cross-over was only possible in patients with at least one reoperation. In the HSO group, 13 patients (32%) changed to another tamponade (gas), and 13 to standard silicone oil (32%). Heavy silicone oil was maintained as a tamponade in 14 patients. In contrast, a tamponade with standard silicone oil remained unchanged in 27 (67%) patients. In eight patients standard silicone oil was changed to gas, and in five to HSO. There was no difference with respect to the total number of surgeries among both groups. At baseline, there was no difference between the two groups: HSO (n = 46) and standard silicone oil (n = 47) with respect to age, gender, or general health (see Table 1); neither did they differ with respect to symptoms as in both groups deterioration of VA and visual field defects were the main symptoms. Previous treatment was carried out in more than 80% of the patients in both groups and included photocoagulation (55%), cryotherapy (20), cataract surgery (55), buckling surgery (55) or vitrectomy (65). Twenty-two patients in each group had received endotamponade with standard silicone oil before, and one patient in the standard silicone group presented with HSO tamponade. In more than 95% of the patients in each group, there was sufficient visual access to the retina. There was no difference between the groups regarding lens status. The mean VA at baseline was better in the standard silicone oil group (1.59 ± 0.83 versus 1.82 ± 0.90 logMar). There was no difference between the two groups regarding retinal detachment, star folds anterior or posterior to the equator, and the presence of PVR. In 19 patients (40.4%) of the standard silicone oil group, the macula was attached compared to 13 (28.3%) in the HSO group. Baseline data are presented in Table 2. None of the randomized patients was excluded by the end-point committee. As patients who did not satisfy the major inclusion criteria were in all cases already randomized (performed preoperatively), all of them are therefore part of the full analysis set. One patient (‘8701001’) was excluded due to a lack of pre- and postsurgical assessment and data (only randomization sheet present). Three patients in the HSO group and five patients in the standard silicone group fulfilled intraoperative exclusion criteria. Concerning the evaluation of the anatomical success, in 20 cases the 12-month visit was substituted by the delayed 12-month visit. In cases where a 12-month visit was not available, the anatomical success was assessed as treatment failures. This is applied to 18 cases in the HSO group and 16 in the standard group. For the analysis of VA, in cases of no valid 12-month visit, the last available observation was used (last observation carried forward) for 18 patients in the HSO group and 15 patients in the standard group. Data entered into the analysis were collected at a medium time of 364 days (range 1–648 days) after the initial surgery and were derived from 1 week postoperative visit in 8%, pre-removal visit in 3%, postremoval visit in 13% and unscheduled visit in 7%. Only four of the 23 surgeons recruited 10 or more patients. According to the protocol, surgeons with <10 patients were grouped together. In the heavy silicone group, 13 of 46 patients (28.3%) showed complete retinal reattachment at the 12-month visit without any retina affecting preceding intervention. In the standard silicone group, we observed 19 of 47 patients (40.4%) with anatomical success. There was no statistically significant difference between the groups ( = 1.1972, df = 1, p = 0.2739). We next consider the main end-point criterion change in VA 12 months postoperatively when compared to the preoperative value. The adjusted mean logMAR VA from the SAS® Type II Sums of Squares analysis was 1.24 (SE 0.116) in the standard silicone oil group and 1.27 (SE 0.117) in the HSO group. This results in a confidence interval of [−0.2856; 0.3392] which is below the noninferiority margin of 0.2. Thus, noninferiority of heavy silicon oil compared to standard silicon oil could not be shown. The secondary end-point criterion‘number of retina affecting reoperations’ within the 1 year follow-up was evaluated in patients with at least one removal operation only. This led to a reduced number of 81 patients valid for analysis. In the HSO group, 21 of the 40 patients have had a reoperation compared to 14 of 41 patients in the standard silicone group. However, the difference in the number of retina affecting reoperations does not differ significantly between both study groups (van Elteren Test p = 0.1446). Redetachments occurred after primary silicone oil removal in six of 46 patients with HSO tamponade (13.04%) compared to eight of 47 (17.02%) in the standard silicon group. Redetachments under silicone oil were observed in 17 of 46 patients in the HSO group (36.96%) and in 13 of 47 patients (27.66%) in the standard silicone oil group (Table 3). Regarding the location of the redetachments under silicone oil, the attachment in each quadrant was assessed on the basis of fundus drawings and registered clock-hour wise (Fig. 1A,B). In the HSO group, redetachments were seen predominantly in the inferior quadrants between 3 over 6–9 o′clock. This is in contrast to the HSO group, where redetachments were more equally distributed to the upper and lower retina with a dominance of the upper quadrants. Four patients died during the follow-up period with no relation to the surgical procedure. Of 71 adverse events, 26 were retinal detachment, one persistent retinal detachment, five macular pucker, four clinically significant cataract, two capsular fibrosis and 22 glaucoma. No enucleation of an eye was reported. Our results showed no significant difference between HSO and standard silicone oil regarding anatomical success. Neither noninferiority nor superiority was shown with regard to final acuity. Thus, HSO is no better than ‘normal’ silicone oil. One could discuss whether a full recruitment of all patients would have had an effect on the result. The success rate in the standard silicone oil group was slightly higher (40%) compared to the HSO group (28%) in this interim analysis. A recalculation of the number needed to treat for the second phase of the study resulted in 294 patients per treatment group [α = 0.0317 (results in a two-step design) and a power of 80%], allowing to prove superiority of the standard silicone oil compared to HSO. However, this was not the intention of the study. Further, it is questionable how long the recruitment of 600 more patients would take as recruitment of the initial 93 patients required 4 years. We thus do not consider the completion of the second part of the study. Nevertheless, our results supported the interpretation that HSO (Densiron-68®) as a tamponading agent in patients with complicated inferior detachments is improved upon by standard silicon oil with respect to retinal detachment. This is in agreement with other nonrandomized trials demonstrating a clinical efficacy of densiron. In our study, there was no difference in the rate of retinal reattachment after 12 months. In experimental and clinical investigations, significant inflammatory responses have not been reported using HSO (Mackiewicz et al. 2007a,b). In our study, almost no side-effects were reported, such as emulsification with consecutive glaucoma, corneal endothelial decompensation, subconjunctival cysts with fibrous scars and cataract promotion. There was no retinal toxicity reported, and there was no difference in the changes in VA over 12 months in patients treated with HSO and standard oil. Similarly, the number of reoperations was equally distributed among the groups, while the number of retina affecting reoperations was slightly higher in the HSO group, however, without statistical significance. The hypothesis investigated in the HSO study was that physically separating the PVR-stimulating environment from the effector cell – fibroblast or retinal pigment epithelial cells – inhibits or mitigates fibroplasia. Displacement of the PVR-stimulating environment seems to be possible over the inferior retina by vitreous tamponades heavier than water. Proliferative vitreoretinopathy is a complication of rhegmatogenous retinal detachment and of trauma involving the posterior segment of the eye (intraocular foreign bodies, penetration, perforation, rupture, contusion). Such membranes appear predominantly in the inferior circumference of the retina. Although vitreoretinal techniques have been improved over the past years, the rate of PVR did not decrease considerably (Joeres et al. 2006). Interestingly, there seems to be a shift in the PVR milieu when using HSO (Heimann et al. 2008, Joussen & Wong 2008; Engelmann & Becker 2009; Joussen et al. 2009). As expected, in the group treated with HSO, redetachments occurred in the lower part of the retina. While the number of redetachments did not differ, there was a more equal distribution of redetachments in all clock-hours in the Densiron-68® group. While there were still patients with inferior redetachments, most of the redetachments were seen above the meridian. This is in agreement with the previous data in a trauma model (Joussen et al. 2009). A heavier-than-water vitreous tamponade can shift PVR within the eye, but cannot prevent PVR altogether. This is the first time to show that PVR can be modulated by physical means. Nevertheless, the heavy tamponade approach to PVR is not sufficient to improve the prognosis. The lack of emulsification here may reflect the early removal of the oil, i.e. within 2–3 months. It also suggests that emulsification reported as prominent feature of low-viscosity oils in smaller series is more likely triggered by individual factors within the patient (blood ocular barrier) than by viscosity of oil. The patients included in this trial were with PVR and not with simple primary inferior rhegmatogenous detachments. Further studies are needed to assess HSO as a primary tamponade. In conclusion, the HSO study is the first randomized multicentre study investigating the efficacy of a heavier-than-water tamponade in the setting of PVR of the inferior retina. While there were no severe side-effects, superiority above standard oil could not be shown. The data suggest that for patients with PVR an individual decision for a heavy tamponade should be made. It could be advantageous to clear the problems in the lower periphery in a primary surgery with HSO and then in a second step address the problems in the upper part with gas or normal silicone oil. Future studies are also needed to detail the role of additional buckling surgery in the treatment of patients with inferior PVR. This study is funded by the Deutsche Forschungsgemeinschaft (DFG Ki 743/2-1, and DFG Hi 541/1-1). Bernd Kirchhof (Köln), Antonia M. Joussen (Berlin), Ralf-Dieter Hilgers (Aachen). Christina Lente (Aachen). Bernd Ralf-Dieter Antonia M. Ralf-Dieter Christina University of University of University of University of University of University of of for University of Cologne, University of University of University of University of University of University of University of University of Antonia M. University Bernd of for University of Cologne, University of University of The University of The University of The University of University of University of University of University of The RWTH University of University of (Berlin), (Berlin), (Köln), (Köln),