- Email: [email protected]
The Effect of Postoperative Face-Down Positioning and of Long- versus Short-Acting Gas in Macular Hole Surgery
The Effect of Postoperative Face-Down Positioning and of Long- versus Short-Acting Gas in Macular Hole Surgery Results of a Registry-Based Study Rohan W. Essex, MBBS, MBiostat,1 Zabrina S. Kingston, BBioMedSc, MBBS,1 Margarita Moreno-Betancur, PhD,2,3 Bruce Shadbolt, PhD,4 Alex P. Hunyor, MBBS,5 William G. Campbell, MBChB,6 Paul P. Connell, MBBCHBAO, MD,7 Ian L. McAllister, MBBS, DM,8 for the Australian and New Zealand Society of Retinal Specialists Macular Hole Study Group Purpose: To determine whether sulfur hexafluoride (SF6) gas is noninferior to longer-acting gases in macular hole surgery and whether withholding postoperative face-down positioning (FDP) is noninferior to FDP. Design: Registry-style, prospective, nonrandomized, observational cohort study. Participants: Patients with idiopathic macular holes undergoing primary surgery. Methods: Surgeons were invited to submit clinical details of all macular hole cases receiving surgery. Baseline demographic and clinical information were collected, as well as details of surgical intervention and postoperative posturing advice. Primary follow-up data were collected 3 months postoperatively. Main Outcome Measures: Macular hole closure at 3 months. A noninferiority approach was used, with a noninferiority margin set at 5% decreased frequency of success. Results: A total of 2456 eyes of 2367 patients were included in the study. Outcomes were available in 94.9% of cases (2330/2456). The rate of macular hole closure was 95.0% (2214/2330). Sulfur hexafluoride gas was found to be noninferior to longer-acting gases (95% confidence interval [CI] for adjusted effect on success, 1.76 to þ2.25), and noninferiority was demonstrated regardless of macular hole size. Although withholding FDP was found to be noninferior to FDP for the study population as a whole (95% CI for adjusted effect on success, 4.21 to þ0.64), the result was inconclusive in holes >400 mm in diameter (95% CI, 9.31 to þ1.04). Lack of internal limiting membrane (ILM) peel, increasing hole size, hole duration 9 months, increasing age, and 20-gauge surgery all were associated with lower odds of success. Vitreous attachment to the hole margin was not associated with outcome when corrected for hole size, and combined phacovitrectomy surgery was not observed to affect the odds of success in phakic eyes. Conclusions: Sulfur hexafluoride gas tamponade was noninferior to longer-acting gases in the surgical management of macular hole. Withholding FDP was noninferior to FDP in holes 400 mm in diameter. In holes >400 mm in diameter, noninferiority of withholding FDP could not be concluded. We would advise caution if posturing is withheld in this group on the basis of the results of this study and of others. Ophthalmology 2016;-:1e8 ª 2016 by the American Academy of Ophthalmology. Supplemental material is available at www.aaojournal.org.
Macular hole surgery was first described in 1991 by Kelly and Wendel1 and Wendel et al.2 Randomized controlled studies subsequently demonstrated the superiority of surgery over conservative management. These studies used perfluoropropane (C3F8) gas as a tamponade agent, rather than the originally described sulfur hexafluoride (SF6), and advised 2 weeks of face-down positioning (FDP).3e5 The surgical technique has been refined over the past 2 decades.6 Recent studies have demonstrated the
2016 by the American Academy of Ophthalmology Published by Elsevier Inc.
additional benefit of internal limiting membrane (ILM) peeling at the time of surgery, and hole closure rates in most recent series are greater than 90%.7e10 There remains broad variability in clinical practice and management of this condition, with no consensus regarding the best surgical approach, particularly regarding the choice of intravitreal tamponade and duration of FDP.11e13 Many investigators have reported good results without FDP.14e25 Various mechanisms of action of the gas tamponade are
http://dx.doi.org/10.1016/j.ophtha.2015.12.039 ISSN 0161-6420/16
1
Ophthalmology Volume -, Number -, Month 2016 postulated in macular hole surgery,26 with all assuming that bubble-fovea contact is relevant. Therefore, it would seem intuitive that a larger, longer-acting bubble combined with FDP should be beneficial because this would both facilitate and prolong this apposition. However, there is optical coherence tomography evidence that hole closure occurs very early in the postoperative period, often within the first 24 hours.27e29 As such, longer-acting gases and prolonged (or indeed any) FDP may be unnecessary. By using a noninferiority study design applied to a large prospective (nonrandomized) registry-based cohort, we aimed to observe whether withholding FDP was noninferior to FDP (of any duration) and whether SF6 gas was noninferior to longer-acting gas tamponades. The present article presents the anatomic outcomes.
Methods All surgeons who were members of the Australian and New Zealand Society of Retinal Specialists were invited to participate. The study was approved by the ethics committee of the Royal Australian and New Zealand College of Ophthalmologists, and local institutional review boards where relevant, and adhered to the tenets of the Declaration of Helsinki. Primary idiopathic macular holes as determined by the surgeon were included. Holes that had already undergone surgery were excluded. Traumatic holes, holes associated with pathologic myopia, eyes with refractive errors 6 diopters spherical equivalent, and “stage 1 macular holes” (i.e., no full-thickness retinal defect) also were excluded, as were those for which the surgeon documented the cause to be “other.” Surgeons prospectively completed a datasheet at the time of surgery, collecting key baseline demographic and clinical features, and surgical information including tamponade used and FDP advice (Table 1, available at www.aaojournal.org). Surgeons were asked to measure hole size as the minimum hole width at the narrowest hole point in the mid retina, as defined by the Vitreomacular Traction study group.30 The study commenced in November 2008 using a paper-based system and then transitioned to an online data-entry system in June 2012. The primary outcome was hole closure at 3 months with 1 surgery, chosen to avoid the confounding effects of cataract on visual outcome. Hole closure was the absence of a full-thickness retinal defect at the site of the original hole. Secondary outcomes were visual acuity and safety outcomes (retinal detachment and endophthalmitis). These secondary outcomes are not the subject of the present article and will be reported in a separate publication. If the hole failed to close after the primary procedure and subsequent surgery had been successfully performed, this was still regarded as a failure, and the acuity at the time of the failure was used as the outcome acuity. Retrospective data entry was allowed provided surgeons entered all the cases they had performed in a given period with no selective omission of cases (e.g., omission of unsuccessful cases). Outcomes from 1 week onward were accepted if no 3month outcome data were available, provided the tamponade had resorbed to the point that the surgeon could determine whether surgery was successful or not. The principal research questions were whether it is reasonable to withhold FDP and to use SF6 gas in place of longer-acting gases in the surgical management of macular hole. Sulfur hexafluoride as an intraocular tamponade agent was compared with longer-acting gases as a group (C2F6 and C3F8). Eyes receiving air or oil were excluded from the study. Surgeons were asked to record the
2
“duration of face-down posturing advised (days)” and advised to round up to the nearest day. No FDP (zero days) was compared with FDP for any length of time, using an intention-to-treat approach (i.e., advice given to the patient, whether or not it was followed). No attempt was made to record compliance with postoperative positioning advice. A noninferiority approach to the study was used, evaluating whether no FDP was noninferior to FDP and whether SF6 gas was noninferior to longer-acting gas. The noninferiority margin was set at 5% (i.e., if the effect of a surgical decision altered the surgical success rate by <5%, it was deemed to be noninferior). The effect of FDP was of particular interest to the investigators and used as the basis of a sample size calculation. After data for the first 400 patients were collected, the surgical success rate was noted to be 95%, and the ratio of postured to not-postured patients was 9:1. By using a noninferiority margin of 5%, a power of 90%, and type 1 error rate of 5%, the required sample size was n ¼ 1810 to demonstrate noninferiority. This is different than the sample size calculation performed by Yorston et al31 because our treatment groups were of unequal size and we used higher power. Data analysis was performed using Stata 13 (StataCorp LP, College Station, TX). In the primary analysis, the probability of success was modeled using logistic regression including all baseline covariates. Cases with missing data in any of the variables were excluded in the regression. From this model, the confounderadjusted differences in success rates (average marginal differences e referred to as “adjusted effect”) comparing no FDP with FDP and SF6 gas with longer-acting gases were estimated. The effect of FDP and gas choice was estimated globally and by hole size category. A separate analysis was performed on phakic eyes to explore the effect of combined phacovitrectomy surgery on surgical success. Several sensitivity analyses were performed to assess the robustness of the conclusions of the primary analysis. First, a reduced model without the nonsignificant predictors (P > 0.05) was considered. Second, multiple imputation was used as an alternative approach to handle missing data. Finally, patient and surgeon clustering effects, due to bilateral eyes and repeated cases from each site, respectively, were explored using generalized estimating equations methodology.
Results There were 2689 macular holes in the dataset. Of these, 87 were persistent, 5 were recurrent, 29 were traumatic, and 84 were “other” (including 47 with pathologic myopia). These were excluded from the dataset. Five cases received air as a postoperative tamponade, and 13 cases received silicone oil, and these were also excluded. An additional 6 cases were found to be duplicates and 4 had almost total lack of data, leaving a total of 2456 primary idiopathic holes from 2367 patients (89 bilateral), contributed by 50 surgeons and 3 institutions (i.e., 53 sites). The median number of cases per site was 29 (range, 1e173). Seven sites contributed >100 cases, and 9 sites contributed less than 5 cases. The baseline characteristics of the group are summarized in Table 2, and operative intervention is shown in Table 3 and Figure 1. Numbers of cases with missing data also are presented. Refraction was recorded in only 367 cases (mean þ0.46 diopters) and could not be included in the analysis. The majority of cases were submitted from private practices, receiving surgery from consultant surgeons rather than trainees. The median duration of macular hole at the time of surgery (estimated by the operating surgeon) was 3 months. Median hole size was 281 mm, and less than one quarter were >400 mm in diameter. Two thirds received SF6 gas tamponade. Vitreous was attached to the hole margin in
Essex et al
Macular Hole Surgery Outcomes
Table 2. Baseline Demographic and Clinical Features Missing/2456 Mean age
69.1 yrs (SD, 7.6; range, 37e95)
Gender n (%) Male Female Side Right Left Duration of hole (months) <9 mos 9 mos Median hole size (mm) 250 mm 251e400 mm >400 mm Vitreous attached to hole edge Lens status Phakic Pseudophakic Visual acuity (letters)
844 (34.5%) 1606 (65.6%) 1242 (50.8%) 1205 (49.2%) 3 median (range, 1e240) 2059 (87.4%) 298 (12.6%) 281 (IQR, 200e400) 1021 (44.5%) 741 (32.1%) 535 (23.4%) 1158/2326 (49.8%)
13
6
9 99
159
130 21
1909 (78.4%) 526 (21.6%) 48.1 (SD, 15.3)
46
IQR ¼ interquartile range; SD ¼ standard deviation.
49.8%, and the breakdown of attachment by hole size is shown in Table 4. The baseline characteristics of cases with no FDP are compared with those who were advised any duration of FDP in Table 5 (available at www.aaojournal.org). The groups were well balanced with respect to all baseline features except vitreous attachment to hole margin, which was slightly less common in the FDP group (47% vs. 56%, P < 0.001). Patients who were
Table 3. Operative Intervention Missing/2456 ILM peel Yes No Tamponade used SF6 C2F6 C3F8 Postoperative FDP (any duration) Yes No Combined cataract and vitrectomy surgery Grade of surgeon Consultant Trainee Gauge of instruments used 20G 23G 25G
2392 (97.7%) 57 (2.3%)
7
1653 (67.7%) 87 (3.6%) 702 (28.8%)
14
1806 (74.2%) 628 (25.8%) 250 (13.3% of 1880 phakic eyes)
22
2282 (94.9%) 124 (5.2%)
50
79 (3.3%) 1353 (55.6%) 1001 (41.1%)
23
29
Figure 1. Duration of postoperative face-down positioning (FDP) advised (days, rounded up to nearest day).
advised not to posture face-down postoperatively were more likely to receive SF6 gas. They were also more likely to have small-gauge surgery, be operated on by a trainee, and to have the ILM peeled. The statistical analysis for the primary outcomes (i.e., the adjusted effects of no FDP and of SF6 gas use) corrects for any imbalances in baseline covariates.
Primary Outcome Outcome was available in 94.9% of cases (2330/2456), and the rate of primary macular hole closure among these was 95.0% (2214/ 2330). There was complete baseline and outcome data in 82.2% of cases (2019/2456), and this group was used for the primary analysis. The 95% confidence interval (CI) for the adjusted effect of no FDP was a 4.21% worse success rate to a 0.64% better success rate compared with FDP. This does not cross the predetermined 5% noninferiority margin, demonstrating that withholding FDP was noninferior to FDP for the study population as a whole. When the effect of posturing was estimated by hole size category (<250 mm, 251e400 mm, and >400 mm), the 95% CI straddled the noninferiority margin in holes >400 mm (Fig 2). Noninferiority in larger holes cannot be concluded. We did not observe FDP to be superior to no FDP at any hole size (all CIs also straddled zero effect). Sulfur hexafluoride gas use was noninferior to longer-acting gas tamponades (95% CI for adjusted effect on success, 1.76
Table 4. Preoperative Optical Coherence Tomography Grading of Macular Holes Hole Size N (%*)
C3F8 ¼ perfluoropropane; FDP ¼ face-down positioning; ILM ¼ internal limiting membrane; SF6 ¼ sulfur hexafluoride.
Vitreous attachment to hole margin
Not attached Attached
250 mm
251e400 mm
>400 mm
420 (18.8%)
383 (17.2%)
336 (15.1%)
588 (26.4%)
337 (15.1%)
167 (7.5%)
*Percent of number with both hole size and vitreous attachment recorded in dataset (¼2231).
3
Ophthalmology Volume -, Number -, Month 2016
Figure 2. The adjusted effects of withholding face-down positioning (FDP) and of sulfur hexafluoride (SF6) gas use on the probability of hole closure. The adjusted effect for the entire study population and the effect in the 3 hole size subgroups for each of the 2 interventions are shown. *Noninferiority of withholding FDP cannot be concluded in the subgroup of holes >400 mm. CI ¼ confidence interval.
to þ2.25). The noninferiority of SF6 gas was observed at all macular hole sizes (Fig 2). The effects of the other covariates on the odds of success are presented in Table 6. Lack of ILM peel, increasing hole size, hole duration 9 months, increasing age, and 20-gauge surgery all were associated with lower odds of success. The number of eyes in which the ILM was not peeled (n ¼ 57, 2.3%) and with 20-gauge surgery (n ¼ 79, 3.3%) was small. Vitreous attachment to the hole
margin was not associated with outcome when corrected for hole size. Removing nonsignificant predictors from the model had no effect on these conclusions, nor did accounting for intrasurgeon correlations (which were found to be negligible: þ0.011) or intrapatient correlation (also low: 0.077). Analyses using multiple imputation as an alternative method for managing missing data also did not affect the conclusions.
Combined Phacovitrectomy Surgery Table 6. Effect of Baseline and Other Surgical Features on the Odds of Surgical Success
No ILM peel Increasing age by 10 yrs Female gender Left eye Increase hole size by 100 mm Vitreous attachment to hole margin Hole duration >9 mos Pseudophakic lens status Trainee surgeon Small-gauge instruments used (23G or 25G vs. 20G)
Odds Ratio (of Success)
95% CI
P Value
0.22 0.69 1.06 1.27 0.75 1.12
0.092e0.54 0.51e0.94 0.67e1.67 0.83e1.93 0.69e0.82 0.72e1.75
0.001 0.018 0.808 0.274 <0.001 0.604
0.53 1.24 0.68 2.53
0.32e0.87 0.72e2.14 0.26e1.82 1.08e5.84
0.012 0.437 0.445 0.031
Boldface indicates P < 0.05. CI ¼ confidence interval; ILM ¼ internal limiting membrane.
4
A separate analysis was performed on phakic eyes to investigate the effect of combined phacovitrectomy surgery on closure rate. The numbers undergoing combined surgery were relatively small (n ¼ 250/1880 phakic eyes). Combined surgery was not observed to have any effect on the odds of hole closure overall (odds ratio, 0.95; 95% CI, 0.48e1.86; P ¼ 0.875). There was no interaction observed between combined surgery and gas choice (P ¼ 0.20) or FDP (P ¼ 0.12), although we had low power to estimate these interactions with confidence. A further analysis was performed on the entire study group using “lens status at the end of the procedure” as a covariate. Again, pseudophakia and combined surgery were not associated with anatomic outcome, and the significance of the other covariates was not affected (results not shown).
Discussion The present study has demonstrated the noninferiority of SF6 gas in macular hole surgery (when compared with longer-acting gases) and that withholding FDP
Essex et al
Macular Hole Surgery Outcomes
postoperatively is noninferior to FDP in holes 400 mm in diameter. Noninferiority of no FDP could not be concluded in holes >400 mm in diameter. A noninferiority approach is the appropriate way to investigate a treatment if “.it is expected to cause fewer side effects, or lead to improved quality of life.” when compared with the standard treatment.32 This is the case for withholding FDP (compared with FDP) and for SF6 gas (compared with longer-acting gas). Only 1 previous study33 has used a noninferiority design to investigate whether withholding posturing is noninferior to a more intrusive FDP regimen. However, the investigators chose a noninferiority margin of 15%, which in our opinion was far too generous. At the commencement of the present study, the study steering committee determined that a 5% noninferiority margin was most appropriate, and Yorston et al31 have subsequently independently published sample size calculations using the same 5% noninferiority margin. This margin balances the desire for hole closure with the onerous nature of FDP, which itself has potential adverse outcomes (reviewed by Gupta26). Attempts to study the effect of withholding FDP have been made difficult by the generally high success rates of surgery, the fact that most studies have been small and lacked sufficient power, and the confounding effects of hole size and concomitant cataract surgery.31,33e41 The majority of published series state that no differences could be observed between the posture and nonposture groups then state or imply that there is no difference. The lack of a noninferiority approach in many of these studies invalidates these conclusions. Attempted (overlapping) meta-analyses have been challenging because of differing study designs and lack of randomization, and have yielded conflicting results regarding the benefits of FDP.42e44 The present study failed to demonstrate noninferiority of withholding FDP in holes >400 mm in diameter. Two small randomized studies have concluded that FDP is associated with better hole closure rates in holes >400 mm in diameter.37,38 On the basis of the results of these 2 studies37,38 and this study, the authors would advise caution in withholding postoperative FDP for eyes with holes >400 mm. Although others have suggested that FDP may have a differential effect depending on whether or not combined phacovitrectomy surgery is performed, the present study failed to identify any interaction between combined surgery and the effect of FDP on outcome.31,37 In addition, the present study failed to demonstrate a beneficial effect of combined surgery itself or pseudophakic lens status on hole closure. Less attention has been given in the literature to the choice of gas tamponade. Most studies are not randomized, all are small, and several confound the question by using different surgical techniques in the treatment groups (e.g., cataract surgery or differing posturing advice).13,38,45e51 Most studies fail to demonstrate any difference between SF6 and longer-acting gas, although they are again flawed by the lack of a noninferiority design. Tognetto et al7 found a higher success rate (uncorrected) with SF6 gas in their large retrospective study. The present study presents strong evidence supporting the noninferiority of SF6 gas.
Although we did not find SF6 gas to be superior to longer-acting gases, it is logical to prefer the shorteracting SF6 gas in light of our results. We found that lack of ILM peel and increasing hole size were associated with a lower odds of success. The importance of ILM peel has been well demonstrated by other investigators,7e10 as has the effect of hole size.5,52,53 Longer hole duration was observed to be associated with a lower probability of success by Wendel et al2 in their 1993 article, and Tognetto et al7 observed longer duration of symptoms to be associated with a lower odds of success. Our results support these earlier findings. The effect of hole duration was not consistently recorded in the UK National Ophthalmology Database.13 However, good success rates are still reported for chronic holes.54 Increasing age was reported by Tognetto et al7 to be associated with a lower odds of success, and our results are consistent with this finding. Smaller studies have failed to observe an effect with age,5 but these lacked the power to detect a small effect. It is unclear whether this association is present at the extremes of the age range, particularly in younger patients in whom pathogenesis may differ from the more typical idiopathic macular holes. It is important to emphasize that increasing patient age and duration of hole are not contraindications for surgery: Surgeons and patients simply should be aware that the probability of a successful outcome is moderately reduced. We observed patients receiving 20-gauge surgery to have a lower odds of success; however, the numbers of patients receiving 20-gauge surgery were small (3.3%), and this result may be anomalous or confounded by other unmeasured factors (e.g., socioeconomic). We are not aware of other published studies documenting better macular hole outcomes with small-gauge surgery. The international vitreomacular traction study group has recommended that vitreous attachment to hole margin be included in the optical coherence tomographyebased grading system of macular hole.30 This is particularly relevant to pharmacologic vitreolysis with ocriplasmin, but we failed to observe any association between vitreomacular attachment and hole closure postvitrectomy.55 In this study, 41.5% of eyes in the dataset had vitreomacular attachment and hole size less than 400 mm and would have been eligible for the Microplasmin for Intravitreous InjectiondTraction Release without Surgical Treatment (MIVI-TRUST) studies. Our observed (unadjusted) surgical success rate was 97.0% in this subgroup (data not shown). Study Limitations The present study is not randomized. It is possible that there is selection bias in the dataset and that patients with a higher risk of failure preferentially received longer-acting gases or were allocated to FDP. If this were the case, it could mask a beneficial effect of these interventions. However, the dataset includes many baseline covariates that are all included in the logistic regression. Logistic regression corrects for any imbalance providing there are no unmeasured determinants of outcome that are also imbalanced between the study groups. This cannot be known and would be addressed only
5
Ophthalmology Volume -, Number -, Month 2016 by a randomized study. It was reassuring to observe that the observed baseline covariates including hole size were reasonably well balanced in the FDP and no FDP groups. The initial sample size calculation for this study was based on an observed ratio between the nonposture and posture groups of 1:9. During the course of the study, this evolved to closer to 1:3. This, combined with a push encouraging surgeons to catch up with their missing data and cases, resulted in us collecting a dataset more powerful than was necessary to address the primary study questions. This did not present any ethical issues because no patients were inconvenienced in any way by the study. However, power calculations were made for the study group as a whole and did not consider subgroup analyses. The raw (unadjusted) success rates in holes 250 mm, 251 to 400 mm, and >400 mm were 97.8%, 95.6%, and 88.8%, respectively. The adjusted estimate of the effect of FDP in the large hole group was a 4% improvement in success rate, but with a CI that extended from a 9% benefit to 1% worse. By using this as the basis for a sample size calculation (using a type 1 error rate of 5% and a power of 80%), a study designed to demonstrate this 4% difference would require 2070 participants with holes >400 mm in diameter. Such a study is unlikely ever to occur, and for now we are left with the present study and others on which to base our clinical decisions. Ongoing data collection using this and other registries, such as the IRIS registry and the UK National Ophthalmology Database, may better inform us in time. It is important these registries collect posturing information, gas use, and hole size in their datasets in addition to other baseline covariates known to be relevant to outcome. Ongoing data collection is vital to better inform our decision making. We have demonstrated the noninferiority of SF6 gas tamponade in the surgical management of macular hole. We also have demonstrated the noninferiority of withholding postoperative FDP in holes 400 mm in diameter. We cannot offer definitive advice regarding the efficacy of FDP in holes >400 mm, but would advise caution if posturing is withheld in this group on the basis of the results of this study and others. Measurement of the hole size using optical coherence tomography is an important component of the preoperative assessment of patients with macular hole.
References 1. Kelly NE, Wendel RT. Vitreous surgery for idiopathic macular holes. Results of a pilot study. Arch Ophthalmol 1991;109:654–9. 2. Wendel RT, Patel AC, Kelly NE, et al. Vitreous surgery for macular holes. Ophthalmology 1993;100:1671–6. 3. Kim JW, Freeman WR, Azen SP, et al. Prospective randomized trial of vitrectomy or observation for stage 2 macular holes. Vitrectomy for Macular Hole Study Group. Am J Ophthalmol 1996;121:605–14. 4. Freeman WR, Azen SP, Kim JW, et al. Vitrectomy for the treatment of full-thickness stage 3 or 4 macular holes. Results of a multicentered randomized clinical trial. The Vitrectomy for Treatment of Macular Hole Study Group. Arch Ophthalmol 1997;115:11–21.
6
5. Ezra E, Gregor ZJ; Morfields Macular Hole Study Group Report No 1. Surgery for idiopathic full-thickness macular hole: two-year results of a randomized clinical trial comparing natural history, vitrectomy, and vitrectomy plus autologous serum: Morfields Macular Hole Study Group Report no. 1. Arch Ophthalmol 2004;122:224–36. 6. Tornambe PE. The evolution of macular hole surgery twenty years after its original description. Am J Ophthalmol 2009;147:954–6. 7. Tognetto D, Grandin R, Sanguinetti G, et al. Internal limiting membrane removal during macular hole surgery: results of a multicenter retrospective study. Ophthalmology 2006;113: 1401–10. 8. Christensen UC, Kroyer K, Sander B, et al. Value of internal limiting membrane peeling in surgery for idiopathic macular hole stage 2 and 3: a randomised clinical trial. Br J Ophthalmol 2009;93:1005–15. 9. Brooks HL Jr. Macular hole surgery with and without internal limiting membrane peeling. Ophthalmology 2000;107: 1939–49. 10. Lois N, Burr J, Norrie J, et al. Internal limiting membrane peeling versus no peeling for idiopathic full-thickness macular hole: a pragmatic randomized controlled trial. Invest Ophthalmol Vis Sci 2011;52:1586–92. 11. Khan RS, Khan KN, Costen MT. Surgical management of macular holes: a national survey of current UK practice. Br J Ophthalmol 2009;93:553–4. 12. Stone T. ASRS 2015 Preferences and Trends Membership Survey. 2015. Chicago, IL: American Society of Retina Specialists; 2015. 13. Jackson TL, Donachie PH, Sparrow JM, Johnston RL. United Kingdom National Ophthalmology Database study of vitreoretinal surgery: report 2, macular hole. Ophthalmology 2013;120:629–34. 14. Iezzi R, Kapoor KG. No face-down positioning and broad internal limiting membrane peeling in the surgical repair of idiopathic macular holes. Ophthalmology 2013;120:1998–2003. 15. Yagi F, Takagi S, Tomita G. Combined idiopathic macular hole vitrectomy with phacoemulsification without face-down positioning. J Ophthalmol 2012;2012:571748. 16. Yagi F, Sato Y, Takagi ST, Tomita G. Idiopathic macular hole vitrectomy without postoperative face-down positioning performed by different surgeons. Asia Pac J Ophthalmol (Phila) 2013;2:20–2. 17. Nadal J, Delas B, Pinero A. Vitrectomy without face-down posturing for idiopathic macular holes. Retina 2012;32: 918–21. 18. Heath G, Rahman R. Combined 23-gauge, sutureless transconjunctival vitrectomy with phacoemulsification without face down posturing for the repair of idiopathic macular holes. Eye (Lond) 2010;24:214–20. quiz 21. 19. Yagi F, Sato Y, Takagi S, Tomita G. Idiopathic macular hole vitrectomy without postoperative face-down positioning. Jpn J Ophthalmol 2009;53:215–8. 20. Rubinstein A, Ang A, Patel CK. Vitrectomy without postoperative posturing for idiopathic macular holes. Clin Experiment Ophthalmol 2007;35:458–61. 21. Merkur AB, Tuli R. Macular hole repair with limited nonsupine positioning. Retina 2007;27:365–9. 22. Simcock PR, Scalia S. Phacovitrectomy without prone posture for full thickness macular holes. Br J Ophthalmol 2001;85: 1316–9. 23. Tornambe PE, Poliner LS, Grote K. Macular hole surgery without face-down positioning. A pilot study. Retina 1997;17: 179–85.
Essex et al
Macular Hole Surgery Outcomes
24. Madgula IM, Costen M. Functional outcome and patient preferences following combined phaco-vitrectomy for macular hole without prone posturing. Eye (Lond) 2008;22:1050–3. 25. Carvounis PE, Kopel AC, Kuhl DP, et al. 25-gauge vitrectomy using sulfur hexafluoride and no prone positioning for repair of macular holes. Retina 2008;28:1188–92. 26. Gupta D. Face-down posturing after macular hole surgery: a review. Retina 2009;29:430–43. 27. Eckardt C, Eckert T, Eckardt U, et al. Macular hole surgery with air tamponade and optical coherence tomographybased duration of face-down positioning. Retina 2008;28: 1087–96. 28. Kikushima W, Imai A, Toriyama Y, et al. Dynamics of macular hole closure in gas-filled eyes within 24 h of surgery observed with swept source optical coherence tomography. Ophthalmic Res 2015;53:48–54. 29. Muqit MM, Akram I, Turner GS, Stanga PE. Fourier-domain optical coherence tomography imaging of gas tamponade following macular hole surgery. Ophthalmic Surg Lasers Imaging 2010;41 Online:e1–6. 30. Duker JS, Kaiser PK, Binder S, et al. The International Vitreomacular Traction Study Group classification of vitreomacular adhesion, traction, and macular hole. Ophthalmology 2013;120:2611–9. 31. Yorston D, Siddiqui MA, Awan MA, et al. Pilot randomised controlled trial of face-down posturing following phacovitrectomy for macular hole. Eye (Lond) 2012;26:267–71. 32. Schumi J, Wittes JT. Through the looking glass: understanding non-inferiority. Trials 2011;12:106. 33. Tadayoni R, Vicaut E, Devin F, et al. A randomized controlled trial of alleviated positioning after small macular hole surgery. Ophthalmology 2011;118:150–5. 34. Feist RM Jr, Pomerleau DL, Feist R, et al. Nonsupine positioning is preferred by patients over face-down positioning and provides an equivalent closure rate in 25- and 23-gauge macular hole surgery. Retin Cases Brief Rep 2014;8:205–8. 35. Alberti M, la Cour M. Face-down positioning versus nonsupine positioning in macular hole surgery. Br J Ophthalmol 2015;99:236–9. 36. Forsaa VA, Raeder S, Hashemi LT, Krohn J. Short-term postoperative non-supine positioning versus strict face-down positioning in macular hole surgery. Acta Ophthalmol 2013;91:547–51. 37. Lange CA, Membrey L, Ahmad N, et al. Pilot randomised controlled trial of face-down positioning following macular hole surgery. Eye (Lond) 2012;26:272–7. 38. Guillaubey A, Malvitte L, Lafontaine PO, et al. Comparison of face-down and seated position after idiopathic macular hole surgery: a randomized clinical trial. Am J Ophthalmol 2008;146:128–34. 39. Tranos PG, Peter NM, Nath R, et al. Macular hole surgery without prone positioning. Eye (Lond) 2007;21:802–6.
40. Szurman P, Di Tizio FM, Lafaut B, et al. Significance of postoperative face-down positioning after surgery for idiopathic macular holes: consecutive case-control study. Klin Monbl Augenheilkd 2000;217:351–5. 41. Jaycock PD, Bunce C, Xing W, et al. Outcomes of macular hole surgery: implications for surgical management and clinical governance. Eye (Lond) 2005;19:879–84. 42. Tatham A, Banerjee S. Face-down posturing after macular hole surgery: a meta-analysis. Br J Ophthalmol 2010;94: 626–31. 43. Solebo AL, Lange CA, Bunce C, Bainbridge JW. Face-down positioning or posturing after macular hole surgery. Cochrane Database Syst Rev 2011;CD008228. 44. Hu Z, Xie P, Ding Y, et al. Face-down or no face-down posturing following macular hole surgery: a meta-analysis. Acta Ophthalmol 2015 Sep 19. doi: 10.1111/aos.12844. [Epub ahead of print] Review. 45. Briand S, Chalifoux E, Tourville E, et al. Prospective randomized trial: outcomes of SF(6) versus C(3)F(8) in macular hole surgery. Can J Ophthalmol 2015;50:95–100. 46. Mulhern MG, Cullinane A, Cleary PE. Visual and anatomical success with short-term macular tamponade and autologous platelet concentrate. Graefes Arch Clin Exp Ophthalmol 2000;238:577–83. 47. Gribomont AC. Surgery for idiopathic macular hole in stage III/IV: is long-acting gas tamponade indispensable? Bull Soc Belge Ophtalmol 2001;101–5. 48. Kumar A, Gogia V, Kumar P, et al. Evaluation of predictors for anatomical success in macular hole surgery in Indian population. Indian J Ophthalmol 2014;62:1141–5. 49. Rahman R, Madgula I, Khan K. Outcomes of sulfur hexafluoride (SF6) versus perfluoroethane (C2F6) gas tamponade for non-posturing macular-hole surgery. Br J Ophthalmol 2012;96:185–8. 50. Xirou T, Theodossiadis PG, Apostolopoulos M, et al. Macular hole surgery with short-acting gas and short-duration facedown positioning. Clin Ophthalmol 2012;6:1107–12. 51. Kim SS, Smiddy WE, Feuer WJ, Shi W. Outcomes of sulfur hexafluoride (SF6) versus perfluoropropane (C3F8) gas tamponade for macular hole surgery. Retina 2008;28:1408–15. 52. Ip MS, Baker BJ, Duker JS, et al. Anatomical outcomes of surgery for idiopathic macular hole as determined by optical coherence tomography. Arch Ophthalmol 2002;120:29–35. 53. Ullrich S, Haritoglou C, Gass C, et al. Macular hole size as a prognostic factor in macular hole surgery. Br J Ophthalmol 2002;86:390–3. 54. Shukla SY, Afshar AR, Kiernan DF, Hariprasad SM. Outcomes of chronic macular hole surgical repair. Indian J Ophthalmol 2014;62:795–8. 55. Stalmans P, Benz MS, Gandorfer A, et al. Enzymatic vitreolysis with ocriplasmin for vitreomacular traction and macular holes. N Engl J Med 2012;367:606–15.
Footnotes and Financial Disclosures Originally received: October 22, 2015. Final revision: December 20, 2015. Accepted: December 29, 2015. Available online: ---.
3
Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Australia.
4
Manuscript no. 2015-1844.
1
Centre for Advances in Epidemiology and IT, Canberra Hospital, Canberra, Australia.
Academic Unit of Ophthalmology, Australian National University, Canberra, Australia.
5
Sydney Eye Hospital, Sydney, Australia.
6
Royal Victorian Eye and Ear Hospital, Melbourne, Australia.
2
7
Mater Misericordiae University Hospital, Dublin, Ireland.
Clinical Epidemiology and Biostatistics Unit, Murdoch Childrens Research Institute, Melbourne, Australia.
8
Lions Eye Institute, Centre for Ophthalmology and Visual Science, University of Western Australia, Perth, Australia.
7
Ophthalmology Volume -, Number -, Month 2016 Members of the Australian and New Zealand Society of Retinal Specialists who contributed cases to the study are as follows: Dr. Penelope Allen, Dr. John Ambler, Dr. Robert Bourke, Dr. Michael Branley, Dr. Robert Buttery, Dr. William Campbell, Dr. Andrew Chang, Dr. Devinder Chauhan, Dr. Fred Chen, Dr. Simon Chen, Dr. Ben Clark, Dr. Mark Donaldson, Dr. John Downie, Dr. Rohan Essex, Dr. Kay Evans, Dr. David Fabinyi, Dr. Ben Fleming, Dr. Adrian Fung, Dr. J.S. Gilhotra, Dr. Mark Gorbatov, Dr. Erwin Groenveld, Dr. Stephen Guest, Dr. Peter Hadden, Dr. Anthony B. Hall, Dr. Wilson Heriot, Dr. I-Van Ho, Dr. Alex Hunyor, Dr. Timothy Isaacs, Dr. Andrew Jones, Dr. Tony Kwan, Dr. Hyong Kwon Kang, Dr. Stewart Lake, Dr. Lawrence Lee, Dr. Alan Luckie, Dr. Ian McAllister, Dr. Mark McCombe, Dr. David McKay, Dr. Mike O’Rourke, Dr. Joseph Park, Dr. Russell Phillips, Dr. Ian Reddie, Dr. Ed Roufail, Dr. Niladri Saha, Dr. Devaraj Subramaniam, Dr. Gina Tsanaktsidis, Dr. Kevin Vandeleur, Dr. Vilacorta-Sandez, Dr. Sarah Welch, Dr. H.C. Wong, Dr. Dimitri Yellachich, Royal Victorian Eye and Ear Hospital (various surgical fellows), Sydney Eye Hospital (various surgical fellows), Royal Perth Hospital (various surgical fellows). Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Supported in part by a grant from the Ophthalmic Research Institute of Australia, 2012. The funding organization had no role in the design or conduct of this research.
8
Preliminary results of this project were presented at: the annual meeting of the American Society of Retinal Specialists, August 9e13, 2014, San Diego, California. Author Contributions: Conception and design: Essex, Hunyor, Campbell, Connell, McAllister Data collection: Essex, Kingston, Moreno-Betancur, Shadbolt, Hunyor, Campbell, Connell, McAllister Analysis and interpretation: Essex, Moreno-Betancur, Shadbolt, Hunyor, Campbell, Connell, McAllister Obtained funding: Essex, Hunyor, Campbell, Connell, McAllister Overall responsibility: Essex, Moreno-Betancur, Shadbolt, Hunyor, Campbell, Connell, McAllister Abbreviations and Acronyms: CI ¼ confidence interval; C3F8 ¼ perfluoropropane; FDP ¼ face-down positioning; ILM ¼ internal limiting membrane; SF6 ¼ sulfur hexafluoride. Correspondence: Rohan W. Essex, MBBS, MBiostat, Canberra Hospital Department of Ophthalmology, Yamba Dr., Garran ACT 2605, Australia. E-mail: rohan. [email protected].
Macular hole surgery was first described in 1991 by Kelly and Wendel1 and Wendel et al.2 Randomized controlled studies subsequently demonstrated the superiority of surgery over conservative management. These studies used perfluoropropane (C3F8) gas as a tamponade agent, rather than the originally described sulfur hexafluoride (SF6), and advised 2 weeks of face-down positioning (FDP).3e5 The surgical technique has been refined over the past 2 decades.6 Recent studies have demonstrated the
2016 by the American Academy of Ophthalmology Published by Elsevier Inc.
additional benefit of internal limiting membrane (ILM) peeling at the time of surgery, and hole closure rates in most recent series are greater than 90%.7e10 There remains broad variability in clinical practice and management of this condition, with no consensus regarding the best surgical approach, particularly regarding the choice of intravitreal tamponade and duration of FDP.11e13 Many investigators have reported good results without FDP.14e25 Various mechanisms of action of the gas tamponade are
http://dx.doi.org/10.1016/j.ophtha.2015.12.039 ISSN 0161-6420/16
1
Ophthalmology Volume -, Number -, Month 2016 postulated in macular hole surgery,26 with all assuming that bubble-fovea contact is relevant. Therefore, it would seem intuitive that a larger, longer-acting bubble combined with FDP should be beneficial because this would both facilitate and prolong this apposition. However, there is optical coherence tomography evidence that hole closure occurs very early in the postoperative period, often within the first 24 hours.27e29 As such, longer-acting gases and prolonged (or indeed any) FDP may be unnecessary. By using a noninferiority study design applied to a large prospective (nonrandomized) registry-based cohort, we aimed to observe whether withholding FDP was noninferior to FDP (of any duration) and whether SF6 gas was noninferior to longer-acting gas tamponades. The present article presents the anatomic outcomes.
Methods All surgeons who were members of the Australian and New Zealand Society of Retinal Specialists were invited to participate. The study was approved by the ethics committee of the Royal Australian and New Zealand College of Ophthalmologists, and local institutional review boards where relevant, and adhered to the tenets of the Declaration of Helsinki. Primary idiopathic macular holes as determined by the surgeon were included. Holes that had already undergone surgery were excluded. Traumatic holes, holes associated with pathologic myopia, eyes with refractive errors 6 diopters spherical equivalent, and “stage 1 macular holes” (i.e., no full-thickness retinal defect) also were excluded, as were those for which the surgeon documented the cause to be “other.” Surgeons prospectively completed a datasheet at the time of surgery, collecting key baseline demographic and clinical features, and surgical information including tamponade used and FDP advice (Table 1, available at www.aaojournal.org). Surgeons were asked to measure hole size as the minimum hole width at the narrowest hole point in the mid retina, as defined by the Vitreomacular Traction study group.30 The study commenced in November 2008 using a paper-based system and then transitioned to an online data-entry system in June 2012. The primary outcome was hole closure at 3 months with 1 surgery, chosen to avoid the confounding effects of cataract on visual outcome. Hole closure was the absence of a full-thickness retinal defect at the site of the original hole. Secondary outcomes were visual acuity and safety outcomes (retinal detachment and endophthalmitis). These secondary outcomes are not the subject of the present article and will be reported in a separate publication. If the hole failed to close after the primary procedure and subsequent surgery had been successfully performed, this was still regarded as a failure, and the acuity at the time of the failure was used as the outcome acuity. Retrospective data entry was allowed provided surgeons entered all the cases they had performed in a given period with no selective omission of cases (e.g., omission of unsuccessful cases). Outcomes from 1 week onward were accepted if no 3month outcome data were available, provided the tamponade had resorbed to the point that the surgeon could determine whether surgery was successful or not. The principal research questions were whether it is reasonable to withhold FDP and to use SF6 gas in place of longer-acting gases in the surgical management of macular hole. Sulfur hexafluoride as an intraocular tamponade agent was compared with longer-acting gases as a group (C2F6 and C3F8). Eyes receiving air or oil were excluded from the study. Surgeons were asked to record the
2
“duration of face-down posturing advised (days)” and advised to round up to the nearest day. No FDP (zero days) was compared with FDP for any length of time, using an intention-to-treat approach (i.e., advice given to the patient, whether or not it was followed). No attempt was made to record compliance with postoperative positioning advice. A noninferiority approach to the study was used, evaluating whether no FDP was noninferior to FDP and whether SF6 gas was noninferior to longer-acting gas. The noninferiority margin was set at 5% (i.e., if the effect of a surgical decision altered the surgical success rate by <5%, it was deemed to be noninferior). The effect of FDP was of particular interest to the investigators and used as the basis of a sample size calculation. After data for the first 400 patients were collected, the surgical success rate was noted to be 95%, and the ratio of postured to not-postured patients was 9:1. By using a noninferiority margin of 5%, a power of 90%, and type 1 error rate of 5%, the required sample size was n ¼ 1810 to demonstrate noninferiority. This is different than the sample size calculation performed by Yorston et al31 because our treatment groups were of unequal size and we used higher power. Data analysis was performed using Stata 13 (StataCorp LP, College Station, TX). In the primary analysis, the probability of success was modeled using logistic regression including all baseline covariates. Cases with missing data in any of the variables were excluded in the regression. From this model, the confounderadjusted differences in success rates (average marginal differences e referred to as “adjusted effect”) comparing no FDP with FDP and SF6 gas with longer-acting gases were estimated. The effect of FDP and gas choice was estimated globally and by hole size category. A separate analysis was performed on phakic eyes to explore the effect of combined phacovitrectomy surgery on surgical success. Several sensitivity analyses were performed to assess the robustness of the conclusions of the primary analysis. First, a reduced model without the nonsignificant predictors (P > 0.05) was considered. Second, multiple imputation was used as an alternative approach to handle missing data. Finally, patient and surgeon clustering effects, due to bilateral eyes and repeated cases from each site, respectively, were explored using generalized estimating equations methodology.
Results There were 2689 macular holes in the dataset. Of these, 87 were persistent, 5 were recurrent, 29 were traumatic, and 84 were “other” (including 47 with pathologic myopia). These were excluded from the dataset. Five cases received air as a postoperative tamponade, and 13 cases received silicone oil, and these were also excluded. An additional 6 cases were found to be duplicates and 4 had almost total lack of data, leaving a total of 2456 primary idiopathic holes from 2367 patients (89 bilateral), contributed by 50 surgeons and 3 institutions (i.e., 53 sites). The median number of cases per site was 29 (range, 1e173). Seven sites contributed >100 cases, and 9 sites contributed less than 5 cases. The baseline characteristics of the group are summarized in Table 2, and operative intervention is shown in Table 3 and Figure 1. Numbers of cases with missing data also are presented. Refraction was recorded in only 367 cases (mean þ0.46 diopters) and could not be included in the analysis. The majority of cases were submitted from private practices, receiving surgery from consultant surgeons rather than trainees. The median duration of macular hole at the time of surgery (estimated by the operating surgeon) was 3 months. Median hole size was 281 mm, and less than one quarter were >400 mm in diameter. Two thirds received SF6 gas tamponade. Vitreous was attached to the hole margin in
Essex et al
Macular Hole Surgery Outcomes
Table 2. Baseline Demographic and Clinical Features Missing/2456 Mean age
69.1 yrs (SD, 7.6; range, 37e95)
Gender n (%) Male Female Side Right Left Duration of hole (months) <9 mos 9 mos Median hole size (mm) 250 mm 251e400 mm >400 mm Vitreous attached to hole edge Lens status Phakic Pseudophakic Visual acuity (letters)
844 (34.5%) 1606 (65.6%) 1242 (50.8%) 1205 (49.2%) 3 median (range, 1e240) 2059 (87.4%) 298 (12.6%) 281 (IQR, 200e400) 1021 (44.5%) 741 (32.1%) 535 (23.4%) 1158/2326 (49.8%)
13
6
9 99
159
130 21
1909 (78.4%) 526 (21.6%) 48.1 (SD, 15.3)
46
IQR ¼ interquartile range; SD ¼ standard deviation.
49.8%, and the breakdown of attachment by hole size is shown in Table 4. The baseline characteristics of cases with no FDP are compared with those who were advised any duration of FDP in Table 5 (available at www.aaojournal.org). The groups were well balanced with respect to all baseline features except vitreous attachment to hole margin, which was slightly less common in the FDP group (47% vs. 56%, P < 0.001). Patients who were
Table 3. Operative Intervention Missing/2456 ILM peel Yes No Tamponade used SF6 C2F6 C3F8 Postoperative FDP (any duration) Yes No Combined cataract and vitrectomy surgery Grade of surgeon Consultant Trainee Gauge of instruments used 20G 23G 25G
2392 (97.7%) 57 (2.3%)
7
1653 (67.7%) 87 (3.6%) 702 (28.8%)
14
1806 (74.2%) 628 (25.8%) 250 (13.3% of 1880 phakic eyes)
22
2282 (94.9%) 124 (5.2%)
50
79 (3.3%) 1353 (55.6%) 1001 (41.1%)
23
29
Figure 1. Duration of postoperative face-down positioning (FDP) advised (days, rounded up to nearest day).
advised not to posture face-down postoperatively were more likely to receive SF6 gas. They were also more likely to have small-gauge surgery, be operated on by a trainee, and to have the ILM peeled. The statistical analysis for the primary outcomes (i.e., the adjusted effects of no FDP and of SF6 gas use) corrects for any imbalances in baseline covariates.
Primary Outcome Outcome was available in 94.9% of cases (2330/2456), and the rate of primary macular hole closure among these was 95.0% (2214/ 2330). There was complete baseline and outcome data in 82.2% of cases (2019/2456), and this group was used for the primary analysis. The 95% confidence interval (CI) for the adjusted effect of no FDP was a 4.21% worse success rate to a 0.64% better success rate compared with FDP. This does not cross the predetermined 5% noninferiority margin, demonstrating that withholding FDP was noninferior to FDP for the study population as a whole. When the effect of posturing was estimated by hole size category (<250 mm, 251e400 mm, and >400 mm), the 95% CI straddled the noninferiority margin in holes >400 mm (Fig 2). Noninferiority in larger holes cannot be concluded. We did not observe FDP to be superior to no FDP at any hole size (all CIs also straddled zero effect). Sulfur hexafluoride gas use was noninferior to longer-acting gas tamponades (95% CI for adjusted effect on success, 1.76
Table 4. Preoperative Optical Coherence Tomography Grading of Macular Holes Hole Size N (%*)
C3F8 ¼ perfluoropropane; FDP ¼ face-down positioning; ILM ¼ internal limiting membrane; SF6 ¼ sulfur hexafluoride.
Vitreous attachment to hole margin
Not attached Attached
250 mm
251e400 mm
>400 mm
420 (18.8%)
383 (17.2%)
336 (15.1%)
588 (26.4%)
337 (15.1%)
167 (7.5%)
*Percent of number with both hole size and vitreous attachment recorded in dataset (¼2231).
3
Ophthalmology Volume -, Number -, Month 2016
Figure 2. The adjusted effects of withholding face-down positioning (FDP) and of sulfur hexafluoride (SF6) gas use on the probability of hole closure. The adjusted effect for the entire study population and the effect in the 3 hole size subgroups for each of the 2 interventions are shown. *Noninferiority of withholding FDP cannot be concluded in the subgroup of holes >400 mm. CI ¼ confidence interval.
to þ2.25). The noninferiority of SF6 gas was observed at all macular hole sizes (Fig 2). The effects of the other covariates on the odds of success are presented in Table 6. Lack of ILM peel, increasing hole size, hole duration 9 months, increasing age, and 20-gauge surgery all were associated with lower odds of success. The number of eyes in which the ILM was not peeled (n ¼ 57, 2.3%) and with 20-gauge surgery (n ¼ 79, 3.3%) was small. Vitreous attachment to the hole
margin was not associated with outcome when corrected for hole size. Removing nonsignificant predictors from the model had no effect on these conclusions, nor did accounting for intrasurgeon correlations (which were found to be negligible: þ0.011) or intrapatient correlation (also low: 0.077). Analyses using multiple imputation as an alternative method for managing missing data also did not affect the conclusions.
Combined Phacovitrectomy Surgery Table 6. Effect of Baseline and Other Surgical Features on the Odds of Surgical Success
No ILM peel Increasing age by 10 yrs Female gender Left eye Increase hole size by 100 mm Vitreous attachment to hole margin Hole duration >9 mos Pseudophakic lens status Trainee surgeon Small-gauge instruments used (23G or 25G vs. 20G)
Odds Ratio (of Success)
95% CI
P Value
0.22 0.69 1.06 1.27 0.75 1.12
0.092e0.54 0.51e0.94 0.67e1.67 0.83e1.93 0.69e0.82 0.72e1.75
0.001 0.018 0.808 0.274 <0.001 0.604
0.53 1.24 0.68 2.53
0.32e0.87 0.72e2.14 0.26e1.82 1.08e5.84
0.012 0.437 0.445 0.031
Boldface indicates P < 0.05. CI ¼ confidence interval; ILM ¼ internal limiting membrane.
4
A separate analysis was performed on phakic eyes to investigate the effect of combined phacovitrectomy surgery on closure rate. The numbers undergoing combined surgery were relatively small (n ¼ 250/1880 phakic eyes). Combined surgery was not observed to have any effect on the odds of hole closure overall (odds ratio, 0.95; 95% CI, 0.48e1.86; P ¼ 0.875). There was no interaction observed between combined surgery and gas choice (P ¼ 0.20) or FDP (P ¼ 0.12), although we had low power to estimate these interactions with confidence. A further analysis was performed on the entire study group using “lens status at the end of the procedure” as a covariate. Again, pseudophakia and combined surgery were not associated with anatomic outcome, and the significance of the other covariates was not affected (results not shown).
Discussion The present study has demonstrated the noninferiority of SF6 gas in macular hole surgery (when compared with longer-acting gases) and that withholding FDP
Essex et al
Macular Hole Surgery Outcomes
postoperatively is noninferior to FDP in holes 400 mm in diameter. Noninferiority of no FDP could not be concluded in holes >400 mm in diameter. A noninferiority approach is the appropriate way to investigate a treatment if “.it is expected to cause fewer side effects, or lead to improved quality of life.” when compared with the standard treatment.32 This is the case for withholding FDP (compared with FDP) and for SF6 gas (compared with longer-acting gas). Only 1 previous study33 has used a noninferiority design to investigate whether withholding posturing is noninferior to a more intrusive FDP regimen. However, the investigators chose a noninferiority margin of 15%, which in our opinion was far too generous. At the commencement of the present study, the study steering committee determined that a 5% noninferiority margin was most appropriate, and Yorston et al31 have subsequently independently published sample size calculations using the same 5% noninferiority margin. This margin balances the desire for hole closure with the onerous nature of FDP, which itself has potential adverse outcomes (reviewed by Gupta26). Attempts to study the effect of withholding FDP have been made difficult by the generally high success rates of surgery, the fact that most studies have been small and lacked sufficient power, and the confounding effects of hole size and concomitant cataract surgery.31,33e41 The majority of published series state that no differences could be observed between the posture and nonposture groups then state or imply that there is no difference. The lack of a noninferiority approach in many of these studies invalidates these conclusions. Attempted (overlapping) meta-analyses have been challenging because of differing study designs and lack of randomization, and have yielded conflicting results regarding the benefits of FDP.42e44 The present study failed to demonstrate noninferiority of withholding FDP in holes >400 mm in diameter. Two small randomized studies have concluded that FDP is associated with better hole closure rates in holes >400 mm in diameter.37,38 On the basis of the results of these 2 studies37,38 and this study, the authors would advise caution in withholding postoperative FDP for eyes with holes >400 mm. Although others have suggested that FDP may have a differential effect depending on whether or not combined phacovitrectomy surgery is performed, the present study failed to identify any interaction between combined surgery and the effect of FDP on outcome.31,37 In addition, the present study failed to demonstrate a beneficial effect of combined surgery itself or pseudophakic lens status on hole closure. Less attention has been given in the literature to the choice of gas tamponade. Most studies are not randomized, all are small, and several confound the question by using different surgical techniques in the treatment groups (e.g., cataract surgery or differing posturing advice).13,38,45e51 Most studies fail to demonstrate any difference between SF6 and longer-acting gas, although they are again flawed by the lack of a noninferiority design. Tognetto et al7 found a higher success rate (uncorrected) with SF6 gas in their large retrospective study. The present study presents strong evidence supporting the noninferiority of SF6 gas.
Although we did not find SF6 gas to be superior to longer-acting gases, it is logical to prefer the shorteracting SF6 gas in light of our results. We found that lack of ILM peel and increasing hole size were associated with a lower odds of success. The importance of ILM peel has been well demonstrated by other investigators,7e10 as has the effect of hole size.5,52,53 Longer hole duration was observed to be associated with a lower probability of success by Wendel et al2 in their 1993 article, and Tognetto et al7 observed longer duration of symptoms to be associated with a lower odds of success. Our results support these earlier findings. The effect of hole duration was not consistently recorded in the UK National Ophthalmology Database.13 However, good success rates are still reported for chronic holes.54 Increasing age was reported by Tognetto et al7 to be associated with a lower odds of success, and our results are consistent with this finding. Smaller studies have failed to observe an effect with age,5 but these lacked the power to detect a small effect. It is unclear whether this association is present at the extremes of the age range, particularly in younger patients in whom pathogenesis may differ from the more typical idiopathic macular holes. It is important to emphasize that increasing patient age and duration of hole are not contraindications for surgery: Surgeons and patients simply should be aware that the probability of a successful outcome is moderately reduced. We observed patients receiving 20-gauge surgery to have a lower odds of success; however, the numbers of patients receiving 20-gauge surgery were small (3.3%), and this result may be anomalous or confounded by other unmeasured factors (e.g., socioeconomic). We are not aware of other published studies documenting better macular hole outcomes with small-gauge surgery. The international vitreomacular traction study group has recommended that vitreous attachment to hole margin be included in the optical coherence tomographyebased grading system of macular hole.30 This is particularly relevant to pharmacologic vitreolysis with ocriplasmin, but we failed to observe any association between vitreomacular attachment and hole closure postvitrectomy.55 In this study, 41.5% of eyes in the dataset had vitreomacular attachment and hole size less than 400 mm and would have been eligible for the Microplasmin for Intravitreous InjectiondTraction Release without Surgical Treatment (MIVI-TRUST) studies. Our observed (unadjusted) surgical success rate was 97.0% in this subgroup (data not shown). Study Limitations The present study is not randomized. It is possible that there is selection bias in the dataset and that patients with a higher risk of failure preferentially received longer-acting gases or were allocated to FDP. If this were the case, it could mask a beneficial effect of these interventions. However, the dataset includes many baseline covariates that are all included in the logistic regression. Logistic regression corrects for any imbalance providing there are no unmeasured determinants of outcome that are also imbalanced between the study groups. This cannot be known and would be addressed only
5
Ophthalmology Volume -, Number -, Month 2016 by a randomized study. It was reassuring to observe that the observed baseline covariates including hole size were reasonably well balanced in the FDP and no FDP groups. The initial sample size calculation for this study was based on an observed ratio between the nonposture and posture groups of 1:9. During the course of the study, this evolved to closer to 1:3. This, combined with a push encouraging surgeons to catch up with their missing data and cases, resulted in us collecting a dataset more powerful than was necessary to address the primary study questions. This did not present any ethical issues because no patients were inconvenienced in any way by the study. However, power calculations were made for the study group as a whole and did not consider subgroup analyses. The raw (unadjusted) success rates in holes 250 mm, 251 to 400 mm, and >400 mm were 97.8%, 95.6%, and 88.8%, respectively. The adjusted estimate of the effect of FDP in the large hole group was a 4% improvement in success rate, but with a CI that extended from a 9% benefit to 1% worse. By using this as the basis for a sample size calculation (using a type 1 error rate of 5% and a power of 80%), a study designed to demonstrate this 4% difference would require 2070 participants with holes >400 mm in diameter. Such a study is unlikely ever to occur, and for now we are left with the present study and others on which to base our clinical decisions. Ongoing data collection using this and other registries, such as the IRIS registry and the UK National Ophthalmology Database, may better inform us in time. It is important these registries collect posturing information, gas use, and hole size in their datasets in addition to other baseline covariates known to be relevant to outcome. Ongoing data collection is vital to better inform our decision making. We have demonstrated the noninferiority of SF6 gas tamponade in the surgical management of macular hole. We also have demonstrated the noninferiority of withholding postoperative FDP in holes 400 mm in diameter. We cannot offer definitive advice regarding the efficacy of FDP in holes >400 mm, but would advise caution if posturing is withheld in this group on the basis of the results of this study and others. Measurement of the hole size using optical coherence tomography is an important component of the preoperative assessment of patients with macular hole.
References 1. Kelly NE, Wendel RT. Vitreous surgery for idiopathic macular holes. Results of a pilot study. Arch Ophthalmol 1991;109:654–9. 2. Wendel RT, Patel AC, Kelly NE, et al. Vitreous surgery for macular holes. Ophthalmology 1993;100:1671–6. 3. Kim JW, Freeman WR, Azen SP, et al. Prospective randomized trial of vitrectomy or observation for stage 2 macular holes. Vitrectomy for Macular Hole Study Group. Am J Ophthalmol 1996;121:605–14. 4. Freeman WR, Azen SP, Kim JW, et al. Vitrectomy for the treatment of full-thickness stage 3 or 4 macular holes. Results of a multicentered randomized clinical trial. The Vitrectomy for Treatment of Macular Hole Study Group. Arch Ophthalmol 1997;115:11–21.
6
5. Ezra E, Gregor ZJ; Morfields Macular Hole Study Group Report No 1. Surgery for idiopathic full-thickness macular hole: two-year results of a randomized clinical trial comparing natural history, vitrectomy, and vitrectomy plus autologous serum: Morfields Macular Hole Study Group Report no. 1. Arch Ophthalmol 2004;122:224–36. 6. Tornambe PE. The evolution of macular hole surgery twenty years after its original description. Am J Ophthalmol 2009;147:954–6. 7. Tognetto D, Grandin R, Sanguinetti G, et al. Internal limiting membrane removal during macular hole surgery: results of a multicenter retrospective study. Ophthalmology 2006;113: 1401–10. 8. Christensen UC, Kroyer K, Sander B, et al. Value of internal limiting membrane peeling in surgery for idiopathic macular hole stage 2 and 3: a randomised clinical trial. Br J Ophthalmol 2009;93:1005–15. 9. Brooks HL Jr. Macular hole surgery with and without internal limiting membrane peeling. Ophthalmology 2000;107: 1939–49. 10. Lois N, Burr J, Norrie J, et al. Internal limiting membrane peeling versus no peeling for idiopathic full-thickness macular hole: a pragmatic randomized controlled trial. Invest Ophthalmol Vis Sci 2011;52:1586–92. 11. Khan RS, Khan KN, Costen MT. Surgical management of macular holes: a national survey of current UK practice. Br J Ophthalmol 2009;93:553–4. 12. Stone T. ASRS 2015 Preferences and Trends Membership Survey. 2015. Chicago, IL: American Society of Retina Specialists; 2015. 13. Jackson TL, Donachie PH, Sparrow JM, Johnston RL. United Kingdom National Ophthalmology Database study of vitreoretinal surgery: report 2, macular hole. Ophthalmology 2013;120:629–34. 14. Iezzi R, Kapoor KG. No face-down positioning and broad internal limiting membrane peeling in the surgical repair of idiopathic macular holes. Ophthalmology 2013;120:1998–2003. 15. Yagi F, Takagi S, Tomita G. Combined idiopathic macular hole vitrectomy with phacoemulsification without face-down positioning. J Ophthalmol 2012;2012:571748. 16. Yagi F, Sato Y, Takagi ST, Tomita G. Idiopathic macular hole vitrectomy without postoperative face-down positioning performed by different surgeons. Asia Pac J Ophthalmol (Phila) 2013;2:20–2. 17. Nadal J, Delas B, Pinero A. Vitrectomy without face-down posturing for idiopathic macular holes. Retina 2012;32: 918–21. 18. Heath G, Rahman R. Combined 23-gauge, sutureless transconjunctival vitrectomy with phacoemulsification without face down posturing for the repair of idiopathic macular holes. Eye (Lond) 2010;24:214–20. quiz 21. 19. Yagi F, Sato Y, Takagi S, Tomita G. Idiopathic macular hole vitrectomy without postoperative face-down positioning. Jpn J Ophthalmol 2009;53:215–8. 20. Rubinstein A, Ang A, Patel CK. Vitrectomy without postoperative posturing for idiopathic macular holes. Clin Experiment Ophthalmol 2007;35:458–61. 21. Merkur AB, Tuli R. Macular hole repair with limited nonsupine positioning. Retina 2007;27:365–9. 22. Simcock PR, Scalia S. Phacovitrectomy without prone posture for full thickness macular holes. Br J Ophthalmol 2001;85: 1316–9. 23. Tornambe PE, Poliner LS, Grote K. Macular hole surgery without face-down positioning. A pilot study. Retina 1997;17: 179–85.
Essex et al
Macular Hole Surgery Outcomes
24. Madgula IM, Costen M. Functional outcome and patient preferences following combined phaco-vitrectomy for macular hole without prone posturing. Eye (Lond) 2008;22:1050–3. 25. Carvounis PE, Kopel AC, Kuhl DP, et al. 25-gauge vitrectomy using sulfur hexafluoride and no prone positioning for repair of macular holes. Retina 2008;28:1188–92. 26. Gupta D. Face-down posturing after macular hole surgery: a review. Retina 2009;29:430–43. 27. Eckardt C, Eckert T, Eckardt U, et al. Macular hole surgery with air tamponade and optical coherence tomographybased duration of face-down positioning. Retina 2008;28: 1087–96. 28. Kikushima W, Imai A, Toriyama Y, et al. Dynamics of macular hole closure in gas-filled eyes within 24 h of surgery observed with swept source optical coherence tomography. Ophthalmic Res 2015;53:48–54. 29. Muqit MM, Akram I, Turner GS, Stanga PE. Fourier-domain optical coherence tomography imaging of gas tamponade following macular hole surgery. Ophthalmic Surg Lasers Imaging 2010;41 Online:e1–6. 30. Duker JS, Kaiser PK, Binder S, et al. The International Vitreomacular Traction Study Group classification of vitreomacular adhesion, traction, and macular hole. Ophthalmology 2013;120:2611–9. 31. Yorston D, Siddiqui MA, Awan MA, et al. Pilot randomised controlled trial of face-down posturing following phacovitrectomy for macular hole. Eye (Lond) 2012;26:267–71. 32. Schumi J, Wittes JT. Through the looking glass: understanding non-inferiority. Trials 2011;12:106. 33. Tadayoni R, Vicaut E, Devin F, et al. A randomized controlled trial of alleviated positioning after small macular hole surgery. Ophthalmology 2011;118:150–5. 34. Feist RM Jr, Pomerleau DL, Feist R, et al. Nonsupine positioning is preferred by patients over face-down positioning and provides an equivalent closure rate in 25- and 23-gauge macular hole surgery. Retin Cases Brief Rep 2014;8:205–8. 35. Alberti M, la Cour M. Face-down positioning versus nonsupine positioning in macular hole surgery. Br J Ophthalmol 2015;99:236–9. 36. Forsaa VA, Raeder S, Hashemi LT, Krohn J. Short-term postoperative non-supine positioning versus strict face-down positioning in macular hole surgery. Acta Ophthalmol 2013;91:547–51. 37. Lange CA, Membrey L, Ahmad N, et al. Pilot randomised controlled trial of face-down positioning following macular hole surgery. Eye (Lond) 2012;26:272–7. 38. Guillaubey A, Malvitte L, Lafontaine PO, et al. Comparison of face-down and seated position after idiopathic macular hole surgery: a randomized clinical trial. Am J Ophthalmol 2008;146:128–34. 39. Tranos PG, Peter NM, Nath R, et al. Macular hole surgery without prone positioning. Eye (Lond) 2007;21:802–6.
40. Szurman P, Di Tizio FM, Lafaut B, et al. Significance of postoperative face-down positioning after surgery for idiopathic macular holes: consecutive case-control study. Klin Monbl Augenheilkd 2000;217:351–5. 41. Jaycock PD, Bunce C, Xing W, et al. Outcomes of macular hole surgery: implications for surgical management and clinical governance. Eye (Lond) 2005;19:879–84. 42. Tatham A, Banerjee S. Face-down posturing after macular hole surgery: a meta-analysis. Br J Ophthalmol 2010;94: 626–31. 43. Solebo AL, Lange CA, Bunce C, Bainbridge JW. Face-down positioning or posturing after macular hole surgery. Cochrane Database Syst Rev 2011;CD008228. 44. Hu Z, Xie P, Ding Y, et al. Face-down or no face-down posturing following macular hole surgery: a meta-analysis. Acta Ophthalmol 2015 Sep 19. doi: 10.1111/aos.12844. [Epub ahead of print] Review. 45. Briand S, Chalifoux E, Tourville E, et al. Prospective randomized trial: outcomes of SF(6) versus C(3)F(8) in macular hole surgery. Can J Ophthalmol 2015;50:95–100. 46. Mulhern MG, Cullinane A, Cleary PE. Visual and anatomical success with short-term macular tamponade and autologous platelet concentrate. Graefes Arch Clin Exp Ophthalmol 2000;238:577–83. 47. Gribomont AC. Surgery for idiopathic macular hole in stage III/IV: is long-acting gas tamponade indispensable? Bull Soc Belge Ophtalmol 2001;101–5. 48. Kumar A, Gogia V, Kumar P, et al. Evaluation of predictors for anatomical success in macular hole surgery in Indian population. Indian J Ophthalmol 2014;62:1141–5. 49. Rahman R, Madgula I, Khan K. Outcomes of sulfur hexafluoride (SF6) versus perfluoroethane (C2F6) gas tamponade for non-posturing macular-hole surgery. Br J Ophthalmol 2012;96:185–8. 50. Xirou T, Theodossiadis PG, Apostolopoulos M, et al. Macular hole surgery with short-acting gas and short-duration facedown positioning. Clin Ophthalmol 2012;6:1107–12. 51. Kim SS, Smiddy WE, Feuer WJ, Shi W. Outcomes of sulfur hexafluoride (SF6) versus perfluoropropane (C3F8) gas tamponade for macular hole surgery. Retina 2008;28:1408–15. 52. Ip MS, Baker BJ, Duker JS, et al. Anatomical outcomes of surgery for idiopathic macular hole as determined by optical coherence tomography. Arch Ophthalmol 2002;120:29–35. 53. Ullrich S, Haritoglou C, Gass C, et al. Macular hole size as a prognostic factor in macular hole surgery. Br J Ophthalmol 2002;86:390–3. 54. Shukla SY, Afshar AR, Kiernan DF, Hariprasad SM. Outcomes of chronic macular hole surgical repair. Indian J Ophthalmol 2014;62:795–8. 55. Stalmans P, Benz MS, Gandorfer A, et al. Enzymatic vitreolysis with ocriplasmin for vitreomacular traction and macular holes. N Engl J Med 2012;367:606–15.
Footnotes and Financial Disclosures Originally received: October 22, 2015. Final revision: December 20, 2015. Accepted: December 29, 2015. Available online: ---.
3
Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Australia.
4
Manuscript no. 2015-1844.
1
Centre for Advances in Epidemiology and IT, Canberra Hospital, Canberra, Australia.
Academic Unit of Ophthalmology, Australian National University, Canberra, Australia.
5
Sydney Eye Hospital, Sydney, Australia.
6
Royal Victorian Eye and Ear Hospital, Melbourne, Australia.
2
7
Mater Misericordiae University Hospital, Dublin, Ireland.
Clinical Epidemiology and Biostatistics Unit, Murdoch Childrens Research Institute, Melbourne, Australia.
8
Lions Eye Institute, Centre for Ophthalmology and Visual Science, University of Western Australia, Perth, Australia.
7
Ophthalmology Volume -, Number -, Month 2016 Members of the Australian and New Zealand Society of Retinal Specialists who contributed cases to the study are as follows: Dr. Penelope Allen, Dr. John Ambler, Dr. Robert Bourke, Dr. Michael Branley, Dr. Robert Buttery, Dr. William Campbell, Dr. Andrew Chang, Dr. Devinder Chauhan, Dr. Fred Chen, Dr. Simon Chen, Dr. Ben Clark, Dr. Mark Donaldson, Dr. John Downie, Dr. Rohan Essex, Dr. Kay Evans, Dr. David Fabinyi, Dr. Ben Fleming, Dr. Adrian Fung, Dr. J.S. Gilhotra, Dr. Mark Gorbatov, Dr. Erwin Groenveld, Dr. Stephen Guest, Dr. Peter Hadden, Dr. Anthony B. Hall, Dr. Wilson Heriot, Dr. I-Van Ho, Dr. Alex Hunyor, Dr. Timothy Isaacs, Dr. Andrew Jones, Dr. Tony Kwan, Dr. Hyong Kwon Kang, Dr. Stewart Lake, Dr. Lawrence Lee, Dr. Alan Luckie, Dr. Ian McAllister, Dr. Mark McCombe, Dr. David McKay, Dr. Mike O’Rourke, Dr. Joseph Park, Dr. Russell Phillips, Dr. Ian Reddie, Dr. Ed Roufail, Dr. Niladri Saha, Dr. Devaraj Subramaniam, Dr. Gina Tsanaktsidis, Dr. Kevin Vandeleur, Dr. Vilacorta-Sandez, Dr. Sarah Welch, Dr. H.C. Wong, Dr. Dimitri Yellachich, Royal Victorian Eye and Ear Hospital (various surgical fellows), Sydney Eye Hospital (various surgical fellows), Royal Perth Hospital (various surgical fellows). Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Supported in part by a grant from the Ophthalmic Research Institute of Australia, 2012. The funding organization had no role in the design or conduct of this research.
8
Preliminary results of this project were presented at: the annual meeting of the American Society of Retinal Specialists, August 9e13, 2014, San Diego, California. Author Contributions: Conception and design: Essex, Hunyor, Campbell, Connell, McAllister Data collection: Essex, Kingston, Moreno-Betancur, Shadbolt, Hunyor, Campbell, Connell, McAllister Analysis and interpretation: Essex, Moreno-Betancur, Shadbolt, Hunyor, Campbell, Connell, McAllister Obtained funding: Essex, Hunyor, Campbell, Connell, McAllister Overall responsibility: Essex, Moreno-Betancur, Shadbolt, Hunyor, Campbell, Connell, McAllister Abbreviations and Acronyms: CI ¼ confidence interval; C3F8 ¼ perfluoropropane; FDP ¼ face-down positioning; ILM ¼ internal limiting membrane; SF6 ¼ sulfur hexafluoride. Correspondence: Rohan W. Essex, MBBS, MBiostat, Canberra Hospital Department of Ophthalmology, Yamba Dr., Garran ACT 2605, Australia. E-mail: rohan. [email protected].