25-, 23-, and 20-gauge vitrectomy in epiretinal membrane surgery: a comparative study of 553 cases
- First Online:
- Cite this article as:
- Sandali, O., El Sanharawi, M., Lecuen, N. et al. Graefes Arch Clin Exp Ophthalmol (2011) 249: 1811. doi:10.1007/s00417-011-1752-3
- 217 Views
To compare the safety and functional outcomes of 25-gauge and 23-gauge (G) micro-incision vitrectomy surgery (MIVS) instrumentation with the standard 20-G vitrectomy system in the treatment of epiretinal membranes (ERM).
A retrospective comparative study of 553 consecutive cases with epiretinal membrane who underwent pars plana vitrectomy. Twenty-gauge, 25-gauge and 23-gauge vitrectomy was performed respectively in 347, 91, and 115 eyes. Surgery duration, visual acuity improvement, intraocular pressure variation, intraoperative and postoperative complications were analyzed.
The mean surgical time in the 23-G group and in the 25-G group was shorter than in the 20-G group (P < 0.001). Visual improvement was higher 8 days postoperatively in the 25-G group than in the 20-G and 23-G groups (P = 0.035), but not at 6 weeks postoperatively (P = 0.186). In the 20-G group, the IOP increased significantly on the first day postoperatively (P < 0.001), while in the 23-G group, the IOP decreased on the first day postoperatively (P = 0.073). In the 25-G group, the IOP did not change significantly (P = 0.807). The incidence of complications was not statistically significant between the three groups. Retinal breaks were significantly related to the induction of posterior vitreous detachment, independent of the system gauge.
In ERM surgery, 23-G and 25-G (MIVS) systems are as safe and effective as the 20-G system, and significantly reduce surgical time. Although the 25-G system provides an earlier visual improvement, the 23- and 25- gauge systems are comparable, and the selection will depend on the surgeon’s preference.
Keywords20-gauge23-gauge25-gaugeVitrectomyEpiretinal membraneVisual improvementComplication
Currently, small incisions are one of the innovative techniques developed for most ophthalmologic surgeries. Sutureless vitrectomy in vitreoretinal surgery was proposed initially by Chen in 1996 . In 2002, Fujii and associates [2, 3], introduced the 25-gauge (25-G) vitrectomy system with sutureless self-sealing sclerotomies. The increased flexibility of the small 25-G instrument was one of the disadvantages of this technique. Eckardt  introduced sutureless 23-gauge (23-G) vitrectomy in 2005, which offers a compromise between the 20-gauge (20-G) and 25-G systems. Recent studies have reported that 25-G and 23-G provide several advantages over traditional 20-G surgery, including reduced operating time, faster wound healing, improved patient comfort, less postoperative astigmatism and inflammation, and early visual recovery [5, 6]. Complications such as hypotony with choroidal detachment, iatrogenic retinal tears, and endophthalmitis have been reported, and their incidence varies considerably in different studies [7–9]. However, the authors of previous studies have not evaluated the three systems in the same series.
In this current study, we compare in a large series of patients the safety and functional outcomes of transconjunctival microincision vitrectomy surgery (MIVS) with 25-G and 23-G instrumentation, and the conventional 20-G vitrectomy, performed for epiretinal membrane (ERM) surgery.
Materials and methods
We retrospectively reviewed all consecutive patients with either primary or secondary ERM who underwent pars plana vitrectomy (PPV) in our vitreoretinal surgery department between January 2004 and February 2010. These patients were operated on by two experienced vitreoretinal surgeons (C.M. and P-O.B.) using the same surgical technique at the Centre Hospitalier National d’Ophtalmologie des XV–XX (CHNO XV–XX) Paris.
Twenty-gauge vitrectomy was performed from January 2004 to April 2009. We obtained our initial experiences with the 25-G vitrectomy system from March 2005 to October 2006. Twenty three-gauge vitrectomy was performed from April 2006 to February 2010.
Patients with advanced cataracts that may obscure surgical visibility undergo a phacoemulsification 3 months prior to the ERM surgery. These include cataracts with a grade exceeding to 3 in nuclear, 3 in mainly cortical, and 2 in mainly posterior subcapsular opacification, using the modified grading of the Lens Opacities Classification System 3 (LOCS 3) .
The pre-, per-, and postoperative regimens were the same for all of the groups. Patients’ anticoagulant treatment was switched for low-molecular-weight heparin. Aspirin and clopidogrel were continued. Picloxydine 0.05 % was instilled topically three times a day, 5 days before surgery. Patients received a biantibiotherapy, associating 4 g of piperacilline intravenously (fosfomycine 4 g for allergic ones) and levofloxacin 500 mg orally, systematically 1 hour before surgery. All patients had a peribulbar block with a monitored anaesthesia care. Before surgery, eyelids and periorbital skin were scrubbed with 5% povidone–iodine solution, first by the nurse and then by the surgeon’s assistant, for 5 minutes each time. The eyelid was draped to keep everting lashes from the operation field. Povidone–iodine 5% was again poured directly on the eye by the surgeon at the beginning and at the end of surgery.
The surgeons had a right dominant hand, and the sclerotomy sites were made superiorly between the 10 and 11 o’clock positions for the vitreotome probe, and between the 1 and 2 o’clock positions for the endoillumination probe. The infusion terminal was made between the 2 and 3 o’clock positions for left eyes and between the 9 and 10 o’clock positions for right eyes. Thus, the infusion terminal was positioned in the upper temporal quadrant near the working port (one and a half hours between the infusion line and the working port). All surgeries were performed using the ALCON Accurus® surgical system.
Twenty-gauge Alcon laboratory system kit: the conjunctiva was opened from 3 o’clock to 9 o’clock, and incisions were made perpendicularly to the sclera. All incisions were sutured at the end of the procedure with a vicryl 7–0. The conjunctiva was closed by vicryl 8–0 sutures or by using a bipolar cautery.
Twenty-five-gauge Alcon laboratory system kit: the conjunctiva was displaced 2 mm, and cannulas were inserted perpendicularly to the sclera with non-tunnel incision.
Twenty-three-gauge single-step Alcon laboratory system kit: the conjunctiva was displaced 2 mm, and the sclera was penetrated by a trocar 3.5 or 4 mm posterior to the limbus depending on the lens status, at an angle between 20° and 30° with the bevel up. Once reaching the trocar sleeve, the cannula was rotated 90° perpendicular to the globe toward the midvitreous cavity. The cuff of the cannula was held in place by forceps, and the trocars were removed from the eye. At the end of surgery, the cannulas were withdrawn from the sclera, and the conjunctiva was pushed laterally with a cotton tip. Pressure was applied over the sclera for a few seconds for wound closure. The infusion line was clamped at the time the cannulas were removed, and unclamped afterwards. If any significant leakage was noted, a vicryl 7–0 suture was performed systematically to close the scleral wound after a small opening of the conjunctiva.
In all of the patients, a central vitrectomy was performed, with triamcinolone used only in high myopic eyes. Then, we used a micro vitreoretinal (MVR) blade as a pik to create a defect in the ERM and the posterior hyaloid membrane, if not detached. In most cases, at the first peeling we removed the attached posterior hyaloid membrane, the ERM, and the internal limiting membrane (ILM) at the same time, without the assistance of staining dyes. The resected tissue induced a whitening of the retinal reflex, with thin, small haemorrhages denoting successful removal of the ILM.
In the other cases, a double peeling was required, using indocyanine green or brilliant blue dyes, to stain and then remove the ILM. This peeling was made in a capsulorhexis fashion in an area of approximately 3 optic disc diameters.
Post-surgical treatment consisted of corticosteroid, tobramycine, and non-steroidal anti-inflammatory eye drops applied topically three times daily for 6 weeks.
All data were assessed by a chart review, including the age and the sex of patients, refractive error, induction of posterior vitreous detachment (IPVD), required sclerotomy suture sites after 25-G and 23-G cannula removal, surgery duration, intraocular pressure (IOP) variation and best-corrected visual acuity (BCVA) at 1 and 6 weeks after surgery. Postoperative complications included severe hyper- or hypotony, suprachoroidal haemorrhage, retinal tear, retinal detachment, vitreous haemorrhage and endophthalmitis.
The surgical duration was defined as the time period between insertion and removal of the lid speculum. It was assessed by the surgeon and his assistant, and was recorded in the patients’ operative reports.
The posterior vitreous detachment (PVD) status was established by observing the presence of the Weiss ring in preoperative biomocroscopic examination, and according to the intraoperative findings.
Snellen visual acuity was converted into logarithm of the minimum angle of resolution (LogMAR) for statistical analysis. We excluded, for the visual outcomes comparison only, patients who had a severe myopia with an axial length >26 mm, or who had macular disorders such as age-related degenerative maculopathy or diabetic maculopathy, and patients who had a history of retinal detachment with macular involvement, which could influence the visual improvement.
Intraocular pressure was measured with the Goldmann applanation tonometer. Severe postoperative hypotony and hypertony were respectively defined as IOP <6 mmHg and >30 mmHg. The mean IOP and the postoperative IOP variation >10 mmHg were analyzed.
For retinal periphery examination, we used a contact lens (Three-Mirror Glass or Volk quadraspheric lens, depending on the lens status) intraoperatively, and a non-contact lens (Volk digital wide-field lens and Volk digital superfield lens) postoperatively.
We distinguished two types of definite breaks: breaks located 1 hour from sclerotomies (sclerotomy-related breaks), and breaks located away from sclerotomies (“breaks elsewhere”). In an effort to determine whether the induction of a PVD influenced the development of retinal breaks, all breaks in the three groups were analyzed.
Criteria for the diagnosis of endophthalmitis were: decreased vision, progressive eye pain, anterior chamber cell with fibrin, hypopion, and vitreous cells with fibrin within 6 weeks of the PPV.
Preoperative data were obtained from the preoperative examination. Postoperative data were collected on day 1, day 8, and at 6 weeks after surgery. OCT scans were carried out preoperatively and 6 weeks postoperatively.
Results were presented as mean ± standard deviation (SD). The pattern of distribution of the variables included was compared with a theoretical normal distribution using a Kolmogorov–Smirnov test. A paired Student’s t-test was used to statistically evaluate comparisons between preoperative and postoperative logMAR VA and IOP. For binary outcomes, the stratified Cochran chi-2 test and the Fisher exact test were used for inter-group comparisons of proportions when appropriate. Analysis of variance or analysis of covariance models were used to analyze continuous outcomes. P values of 0.05 or less were considered statistically significant. Statistical analysis was carried out using SAS version 9.2 for Windows (SAS Institure, Cary, NC, USA).
In accordance with French law, institutional review board and ethics committee approval were not required for this study because no modifications to French standards of treatment or follow-up were made.
Characteristics of patients
20-gauge (n = 347)
23-gauge (n = 115)
25-gauge (n = 91)
Age, years [mean (SD)]
Preoperative best-corrected visual acuity, logMAR [mean (SD)]
Age macular disease
In the 20-G group, the mean age was 68.4 ± 9.6 years, and 187 (54%) were female. ILM was peeled in 223 cases (64%), with Blue vision used in 64 cases (29%) and indocyanine green in 27 cases (12.1%). In the 23-G group, the mean age was 69.7 ± 9.3 years, and 70 (61%) were female. ILM was peeled in 84 cases (73%), with Blue vision used in 43 cases (51.2%) and indocyanine green in five cases (6%). In the 25-G group, the mean age was 68.9 ± 9.2 years, and 48 (53%) were women. ILM was peeled in 40 cases (44.0%), with Blue vision used in ten cases (11.0%). All ERM were successfully removed from macula.
Almost all patients (545 out of 553 patients) had OCT scans preoperatively and 6 weeks postoperatively. The OCT scans confirmed the anatomic removal of the ERM in all patients. Operating time: The mean surgical time (±SD) was 30.15 ± 7.34 minutes in the 23-G group and 31.53 ± 5.76 minutes in the 25-G group, which were significantly shorter than in the 20-G group ( 42.15 ± 9.08 minutes) (P < 0.001, ANOVA).
No patients in the 25-G group needed suturing of sclerotomy, whereas 23 patients required suturing in the 23-G group which could affect the duration of the surgery. Compared with the 25-gauge group, the mean surgical time was shorter (29.50 ± 7.30 minutes) in the 23-G group without suture (92 patients) and longer (32.75 ± 7.53 minutes) in the 23-G group with suture (23 patients); however, the difference was not significant (P = 0.18, ANOVA).
The difference in the logarithm of the minimum angle of resolution of BCVA values between the three groups was not significant at baseline (P = 0.306, ANOVA), or at 8 days (P = 496, ANOVA) or 6 weeks postoperatively (P = 0.795, ANOVA) (Fig. 1).
Visual acuity improvement
Visual acuity improvement (LogMAR)
20-gauge (n = 296)
23-gauge (n = 100)
25-gauge (n = 77)
At 8 days
At 6 weeks
However, at 6 weeks postoperatively, the difference between the three groups with respect to the VA improvement was not significant (P = 0.196, ANOVA; Table 2).
Peroperatively, wound leakage occurred after cannula removal only in the 23-G group, in 23 patients (20%): 13 eyes (11.2%) required suture of one sclerotomy site, nine eyes (7.8%) required suture of two sclerotomy sites, and one eye (0.9%) required suture of the three sclerotomy sites. In relation to all of the 23-gauge sclerotomy sites (n = 345), 34 (9.9%) required closure with a suture. No wound leakage occurred in the 25-G group.
Preoperatively, 33 patients (9.5%) were treated for glaucoma in the 20-G group, as were ten patients (8.7%) in the 23-G group and seven patients (7.7%) in the 25-G group (P = 0.856, chi-square test).
Preoperative and postoperative mean intraocular pressure
Intraocular pressure [mean (SD)]
20-gauge (n = 347)
23-gauge (n = 115)
25-gauge (n = 91)
One day after surgery
One week after surgery
20-gauge (n = 347)
23-gauge (n = 115)
25-gauge (n = 91)
Severe hypotony at 1 day (<6 mmHg)
Severe hypertony at 1 day (≥30 mmHg))
Increase of IOP > 10 mmg at 1 day
The incidence of complications, including severe postoperative hypotony and hypertony, retinal break, retinal detachment, intravitreous haemorrhage, and endophthalmitis, was not significant between the three groups (Table 4).
Severe hypotony and hypertony
Postoperatively, severe hypotony was observed in three eyes (0.9%) in the 20-G group and in one eye (0.9%) in the 23-G group. These cases of hypotony were not complicated by choroidal detachment, and resolved spontaneously in all eyes during the first postoperative week. Severe hypotony was not observed in any patient in the 25-G group.
The incidence of severe hypertony was similar between the three groups (P = 0.278, Fisher exact test).
The incidence of retinal breaks in eyes with or without the induction of posterior vitreous detachment during surgery
20-gauge (n = 347)
23-gauge (n = 115)
25-gauge (n = 91)
Rhegmatogenous retinal detachment
The rate of rhegmatogenous retinal detachment in our series was 0.5% (3/353). These occurred in the 20-G group within 1 month following initial surgery (10 days, 3 weeks and 1 month). In two cases, no breaks were observed during the initial surgery. In one case, a break was found during the initial surgery and was treated by cryotherapy and scleral buckling, but another break was the causative defect in the subsequent retinal detachment.
Mild vitreous haemorrhage occurred in eight cases (2.3%) in the 20-G group (four of them were under anticoagulant therapy), in six cases (5.2%) in the 23-G group (one patient was under anticoagulant therapy), and in one case (1.1%) in the 25-G group.
No cases of endophthalmitis occurred in the three groups.
The advent of MIVS presented several advantages for epiretinal membrane surgery [5, 6]. However, increased complications have been reported [7–9], and there are no studies comparing the efficacy and safety of 25-, 23- and 20-gauge vitrectomy. In this series we demonstrate the safety and efficiency of the 23-G and 25-G (MIVS) systems in comparison to the 20-G system. The former contribute significantly to reduce operative time. The 23-G and 25-G systems were similar, although the 25-G provided an earlier visual improvement. Retinal breaks were significantly related to the IPVD, independent of the system gauge.
In this study, we demonstrated shorter overall surgery duration for 23- and 25- gauge systems by an average of 10 minutes in comparison to the 20-G. Short duration of surgery and sutureless system vitrectomy contribute to the comfort of both the patient and the surgeon. The bigger diameter of the 20-G system results in higher dynamics flow through the probe and shorter vitrectomy duration . However, this gained time is lost on the conjunctival opening, hemostasis, terminal infusion fixation, and sclerotomy wound closure . Although 23-G vitrectomy duration is shorter than 25-G , a longer time is needed to check the self-sealing sclerotomies, with possible sutures required.
In our initial experiences, retinal manipulations to remove the ERM were slightly longer with the 25-gauge system, which provided a smaller delicate grasp of tissue. Although in ERM surgery, meticulous peripheral vitreous dissection is unnecessary, we prefer the more rigid 23-G to the 25-G instruments. The advent of the new enhanced 25-G probe’s rigidity and dynamics outflows, 25 gauge + Alcon system, which we used recently in some cases of ERM and macular hole surgery, will improve the quality of the 25-G and extend its use to the majority of vitreous and retinal surgeries.
We noted a faster visual recovery in the 25-G group in the first week postoperatively than in the 20-and 23-gauge groups. Several studies have reported an earlier postoperative visual rehabilitation in 23- and 25- gauge system patients in comparison to 20-G system patients [5, 6]. Earlier visual recovery depends on postoperative induced corneal astigmatism, inflammatory reaction and postoperative cataract development. Our results are not influenced by an eventual cataract: advanced cataracts were systematically operated 3 months before surgery, and there was no difference of low grade pre-existing lens opacity between the groups. Postoperative cataract formation in the first month is less common in the absence of an intraoperative lens touch . Postoperative astigmatism is demonstrated to be higher with the 20-G system, and it may be attributed to the sclera cautery and suturing at the entry port [13–16]. Increased postoperative inflammation with the 20-G system has been reported by several studies [5, 6, 17], which may be explained by the increased surgery duration and the required conjunctival opening with bleeding of conjunctival and episcleral vessels.
In our series, we noted an earlier visual improvement related to 25-G in comparison to the 23-G group with or without suture. One prospective study of 85 cases compared the two MIVS systems in patients who were treated for various vitreoretinal diseases, and reported no significant difference in anterior segment change and visual improvement . Postoperative induced corneal astigmatism and inflammatory reaction were not evaluated in our study. Hence, the reason for the difference in early visual improvement between patients who underwent 23-gauge vitrectomy without suture and those who underwent 25-gauge vitrectomy is not apparent.
In the sixth week postoperatively, visual improvement was the same in the three groups. Indeed, the inflammatory reaction is well-controlled, and the corneal astigmatism changes reduce with the suture lyses 1 month after the surgery .
The 23-G and 25-G systems have been reported to be related to more wound leakage complications such hypotony and choroidal detachment [7, 8, 19]. The 23-G and 25-G incisions are respectively 0.72 mm and 0.5 mm wide. Experimental works of Singh et al.  demonstrated that the 23-G requires an oblique incision to be self-sealing, and the 25-G had an acceptable physiologic leakage even when performed straightly. In the current study, for the 25-G system, no wound leakage or hypotony were noted, although we used no shelved sclera incision. Rizzo et al. report the same finding for ERM surgery . Reported suture rates and hypotony for 23-G pars plana vitrectomy in ERM surgery are respectively 0% and 0% by Hikichi T et al. , and 0% and 9.4% by Haas et al. . In our series, only one patient (0.86%) had a transient hypotony without a choroidal detachment, but 23 patients (20%) required a sclerotomy suture. Our explanation for this high suture rate could be our lack of tolerance for any evidence of wound leak. This may explain our low rate of 23-G hypotony, which was similar to that observed with the 20-G and the 25-G systems. According to the literature reports [8, 19, 21], the rates of wound leakage and hypotony in 23- and 25-gauge systems in ERM surgery are lower in comparison to PPV performed for other retinal diseases. ERM surgery duration is relatively short, with less delicate maneuvers and instrument constraint on sclerotomy sites. Moreover, a lower volume of vitreous is removed during ERM surgery which could account for the lower incidence of hypotony in vitrectomy for ERM compared with that for other vitrectomy.
Most studies have focused on ocular hypotony and wound leakage, and there is limited information about IOP variation and severe hypertony after surgery [22, 23]. On the first postoperative day, the mean IOP increased slightly by 0.98 mmHg only in the 20-G group, and did not significantly change in the 23- and 25-gauge groups. Although more cases of severe hypertony and increased IOP >10 mmHg occurred in the 20-G group in comparison to the 23- and 25-gauge groups, the difference was not statistically significant. The inflammatory anterior chamber reaction is reported to be higher in the 20-G system, and can lead to an inflammatory trabecular meshwork obstruction [24, 25].
In our series, the incidence of retinal tears was 8.4% in the 20-G group, 6% in the 23-G group and 2.2% in the 25-G group. The difference was not significant. The main risk factor associated with retinal tears in the 20-G and 23-G groups was the induction of PVD during the vitrectomy, independently of the system gauge. The maneuvers for initiating and propagating induction of a PVD are essentially the same, and the use of MIVS will not change the incidence of retinal tears . Retinal tears related to induction of a PVD were more frequently located inferiorly, away from sclerotomy sites. These findings have been reported in macular hole surgery [27, 28], but not in ERM surgery . We suppose that the use of a cannulated system may minimize intraoperative vitreous incarceration, which may result in a lower incidence of related sclerotomy tears. There were slightly fewer related sclerotomy tears in the 23-G group than in the 20-G group; however, the difference was not significant. Retinal tears occurred near the sclerotomy sites corresponding to the dominant hand of the surgeon in 70% of our cases. The multiple insertions and withdrawals of instruments (vitreotome probe, micro vitreoretinal blade, forceps, needle for dye injection, backflush) from the vitreous cavity will result in vitreous incarceration through the sclerotomy. We discovered intraoperatively 82% of retinal tears and retinal detachments occurred in only three cases in our series (0.5 %). All three cases occurred in the 20-G group. An examination of the peripheral retina should be performed at the end of surgery with the help of a scleral indentation to identify any retinal breaks that would result in a retinal detachment postoperatively.
The incidence of vitreous hemorrhage was not statistically different between the three groups. All haemorrhages were mild, and resolved spontaneously within the first postoperative week. No obvious source of bleeding was detected. A possible source of haemorrhage could have been a rupture of scleral or choroidal vessels after sclerotomy sutures, or after cannula removal if it was not performed in the same tunnel of incision.
No cases of endophthalmitis occurred in any group. In our series, we used a broad spectrum antibiotic regimen (fluoroquinolone orally in combination with piperacilline, or fosfomycine in penicillin-allergic patients, intravenously) to prevent intraoperative contamination. The use of poviodone–iodine disinfection before and at the end of surgery, a systematic vitreous excision if prolapsed out of the incision, and a systematic sclerotomy suture if there is any suspicion of wound leakage were the other measures taken to prevent endophthalmitis. Reported endophtamitis of 20-G vitrectomy has decreased in recent years, ranging from 0.02 % to 0.23 % [29–31]. Kunimoto  and Scott  reported an increased risk of endophthalmitis related to the 25-G system. Thereafter, in a large series of patients no significant difference was demonstrated in endophthalmitis incidence between the 20-G and 25-G systems [29, 34]. With regard to the 23-G system, although there are limited publications in literature, endophthalmitis is a rare event [7, 35–37]. Parolini et al.  reported no increased risk of endophthalmitis in 23-G PPV in comparison to the 20-G system in a large series of patients. Recently, Scott et al. showed in a recent multicentric study no significant difference in the incidence of endophthalmitis between 20-, 23- and 25-G vitrectomy .
The limit of this study is its retrospective design. We could not evaluate some data concerning corneal topography changes, inflammatory anterior chamber, and patient postoperative discomfort. However, it is the first and the largest study that compares the safety and the visual outcome of the three gauge systems in ERM surgery.
Our results demonstrate that in ERM surgery, 23- and 25-(MIVS) gauge systems are as safe and effective as the 20-gauge system, and significantly reduce surgical time. Although the 25-G provides an earlier visual improvement, the 23- and 25-gauge systems are comparable, and the selection will depend on the surgeon’s preference and the availability of appropriate instruments.
Conflicts of interest
The authors have no proprietary, commercial or financial interests in any of the products described in this study. They have no conflict of interest.