Indications and outcomes for intravitreal injection of C3F8 gas for symptomatic vitreomacular traction

To evaluate the indications and outcomes of perfluoropropane (C3F8) gas injection for symptomatic vitreomacular traction (VMT). A retrospective analysis of eyes with VMT treated with 0.3 mL of C3F8 gas was performed. Patients were not asked to posture after gas injection. In phakic patients, cataract surgery was performed simultaneously. Patients were examined after one week and one month postoperatively. Twenty-nine consecutive eyes of 26 patients with symptomatic VMT who underwent pneumatic vitreolysis were included. A complete posterior vitreous detachment was achieved in 18 eyes (62.1%) after a single gas injection at the final visit. The rate of posterior vitreous detachment was reduced significantly with the presence of epiretinal membrane (ERM) (p = 0.003). Three eyes formed a macular hole (MH) postoperatively and another eye developed a retinal detachment. Mean visual acuity increased significantly after one month (p < 0.008). Pneumatic vitreolysis is a viable option for treating VMT with few adverse events. Patient with concomitant ERM had a significantly lower success rate.


Patients and methods
The study was approved by the Ethikkommission Ostschweiz (EKOS Nr. 19.015) and was performed as part of departmental quality control. All data were anonymized, therefore patient consent was waived by the Ethikkommission Ostschweiz (EKOS Nr. 19.015). The study adhered to the tenets of the Declaration of Helsinki.
In this retrospective study, all patients who underwent a single C3F8 gas injection for symptomatic vitreomacular traction between January 2017 and June 2020 at the Eye Clinic of Sankt Gallen Cantonal Hospital (KSSG) were included. Exclusion criteria were history of vitrectomy or injections, history of laser photocoagulation to the macula, high myopia, aphakia, history of retinal detachment, retinal breaks or vitreous hemorrhage and obvious lattice areas. Each patient received a complete ophthalmological examination including fundus examination, best corrected distance visual acuity using Snellen charts (BCVA, in decimal) and spectral domain optical coherence tomography (SD-OCT, Spectralis HRA OCT, Heidelberg Engineering, Heidelberg, Germany) before surgery, at 1 week and 4 weeks after treatment. In general, every patient had a follow-up after 1 week and a second followup after 1 month. In case that the VMT release did not occur additional follow-ups usually took place. Maximal central retinal thickness (CRT) and maximal VMT adhesion diameter prior treatment was digitally measured using SD-OCT scans. Postoperative outcome was graded in complete VMT release, partial VMT release or no release at all using SD-OCT images. Partial release was defined as persistent VMA without traction (Fig. 1).
Data collection included patients' demographics (gender, age, and laterality), best-corrected visual acuity (BCVA; in decimal), maximal central retinal thickness as well as maximal VMT adhesion diameters. Primary outcome was the release rate of VMT at 1 and 4 weeks and the final visit. Secondary outcomes were the change in BCVA and postoperative complications. After the cataract operation, the gas injection was carried out.
Surgical technique. The patient is placed in rear supine position. The procedure is performed in topical anesthesia only using lidocaine eyedrops. The eye and the eyelids are cleaned with Betadine and a drape is used to cover the face. An eyelid speculum holds the eyelids open. Anterior chamber paracentesis with a MVR blade is performed to reduce intraocular pressure by controlled drainage of aqueous humor from the anterior chamber. The location of injection is marked using a caliper: 3-3.5 mm from the limbus. The injection of 0.3 ml of C3F8 gas is performed with a short 30-gauge needle tip usually in the supero-temporal quadrant. At the end of surgery topical antibiotic is administered into the eye. Postoperatively, the patient is asked not to lie flat on the back for 5 days. In phakic patients phacoemulsification was performed first followed by intravitreal gas injection. The procedures were addressed simultaneously by the same surgeon. Statistical analysis. Data was summarized with descriptive statistics. For numeric variables, medians and ranges are reported, and for categorical variables, number and percentages of patients in each category are reported. For outcome categories, Wilson 95% confidence intervals for the percentages are also given.
Associations between outcome categories and other factors were analyzed with Fisher's exact tests, those with numeric variables were analyzed using Kruskal-Wallis tests (three outcome categories) or Wilcoxon rank sum tests (two outcome categories).
Ethics approval and consent to participate. The study was approved by the Ethikkommission Ostschweiz (EKOS Nr. 19.015) and was performed as part of departmental quality control. All data were anonymized,

Results
Twenty-nine eyes of 26 consecutive patients who underwent pneumatic vitreolysis with C3F8 gas were included. The age ranged between 60 and 92 years with a median age of 77.9 years. In 15 of 29 gas injections, the procedure was combined with a phacoemulsification. Eight eyes (27.6%) had concurrent epiretinal membrane and two eyes (6.7%) had a FTMH stage 2. Three eyes (10.3%) suffered from age-related macular degeneration (AMD) and one each eye (3.4%) suffered from mild non-proliferative diabetic retinopathy (NPDRP) and branch retinal vein occlusion (BRVO) with peripheral sectorial retinal laser, respectively. Patients' characteristics are summarized in Table 1. The median duration of follow-up was 75 days (range 7-510 days). In one patient, the VMT release already occurred within 1 week and the visual acuity went from 0.8 preoperative to 1.0 within that week, so no additional follow-up took place. At the final visit, a complete or partial VMT release was achieved in 24 out of the 29 eyes (82.8%). Only in five eyes (17.2%), no release at all was achieved. The exact rates of VMT release within the different follow-up times are displayed in Table 2. Subgroup analysis of VMT release in relation to patient's characteristics and OCT findings did not show any correlation with age, gender, simultaneous phacoemulsification and maximal central retinal thickness or maximal VMT adhesion diameter, respectively. However, analysis of secondary diagnosis showed that patients with an accompanying ERM had a significant lower release rate ( Table 3). The VMT basis threshold of 1500 microns, above which the risk of VMT release failure may be increased, had no impact on the VMT release rate (Fig. 2). Both patients with stage 2 macular holes experienced a successful VMT release with closure of the macular holes. In addition, the two patients with peripheral sectorial laser after BRVO and NPDRP achieved a complete VMT release.
The visual acuity improved at least one line in 23 eyes (79.3%) and remained stable in 5 eyes (17.2%) at the final visit. In one eye (3.4%) visual acuity worsened from 0.4 to 0.1 after developing a macula-involving retinal detachment. In total, adverse events occurred in 4 out of 29 eyes (13.7%). Additionally to the retinal detachment, three eyes (10.3%) developed a macular hole after successful VMT release. However, this was not associated with increased central retinal thickness (CRT) or VMT adhesion diameter (Fig. 3).

Discussion
Pneumatic vitreolysis (PVL) by intravitreal gas injection was introduced by Chan et al. 15 . The gas bubble should liquefy the vitreous body during the gas expansion and induce the cortical vitreous to collapse during the absorption phase of the bubble, leading to PVD. Recent studies have shown PVD rates ranging from 40 to 88.6% depending on the intravitreal gas administered [16][17][18][19][20][21] . In the present study, we analyzed the outcomes after a single 0.3 ml C3F8 gas injection as a treatment for VMT. A complete release could be achieved in 16 out of 29 (55.2%) within 1 month and in 18 out of 29 (60.9%; 95% CI 40.8-77.8) at the final visit. In another 6 eyes (20.7%) a partial release occurred until the final visit. Overall, in 24 out of 29 eyes (82.8%) some sort of VMT release could be achieved. Only in five eyes (17.2%) the injection did not lead to any VMT release. The mean time until VMT resolution is consistent with the majority of previous studies that also reported a VMT release within the first month after treatment in most cases [16][17][18][19] . The mean ages of previous studies included in the meta-analysis by Yu et al. was between 51.4 and 76.6 years whereas our study had a mean age of 77.9 years 20 . This could be a possible reason for our moderate success rate, since Chan et al. reported significantly higher release rates in younger patients. In fact, they even reported age to be the strongest predictor for success when compared to better baseline visual acuity, absence of diabetes mellitus and female gender 19 . However, we could not show any significant association between age and success rate in our study. Furthermore, according to Chan et al. female gender is a negative predictor for PVL success (2017) 19 . We were also not able to show a significant association between gender and success rate in our study.
OCT findings are often used as inclusion or exclusion criteria whether certain treatments should be performed or not. We investigated the prediction value of the VMT adhesion diameter and central retinal thickness for the success rate for VMT release but no significant association between these prediction values and the VMT release rate could be found. However, previous studies suggested that large and diffuse VMT diameter could lead to a lower success rate and are therefore risk factors for failure for VMT resolution 16,18,19 .
In our study, patients with epiretinal membrane had significant lower VMT release rates at one month and at the final visit. Chan et al. showed also a trend of association of success rate and lack of epiretinal membrane, however, the results did not reach statistical significance 19 . In contrast to our findings, Yu et al. were not able to show a significant association between lower success rate and the presence of epiretinal membrane 20 . However,   www.nature.com/scientificreports/ we noticed that patients with VMT and ERM still suffered from metamorphopsia postoperatively despite successful VMT resolution. The two patients with a stage 2 macular hole and VMT were successfully treated with PVL leading to a complete VMT release with macular hole closure. The successful closure of stage 2 macular holes after PVL is consistent with multiple previous studies [15][16][17][18][19] . Some studies suggest that stage 2 macular holes and a lack of diabetes mellitus are predictors of success in VMT release 18,19 .
For other secondary ocular diagnoses such as BRVO, diabetic retinopathy and AMD, no significant difference in the VMT release rate could be detected, likely due to the small number of patients included in this study.
We also analyzed whether or not simultaneous cataract operation had a positive impact on release rates, but surprisingly no significant association was found.
Since all of these studies including ours are of retrospective nature, confounding factors may have led to wrong conclusions about the prediction value of these measurements for the success rate for VMT release.
Adverse events associated with PVL occurred in four eyes (13.7%) in the present study. Macular holes developed in three eyes (10.3%) and retinal detachment occurred in one eye (3.4%). Pars plana vitrectomy was successfully performed in two patients with macular and one patient with retinal detachment. One patient with macular hole denied further treatment because of systemic risk factors.
Similar rates adverse events have been reported in various previous studies [18][19][20] . Most common adverse event is a macular hole formation (10.3%) postoperatively. However, this is comparable with the natural history where 12% of all patients with VMT will develop a macular hole 22 . In addition, postoperative macular hole formation are reported in 9.8% after PPV and in 8.6% % after ocriplasmin injection, respectively 8,11 .
In conclusion our study suggests that PVL is a viable option for treatment of symptomatic VMT with reasonable resolution and similar complication rates when compared with natural history, ocriplasmin injection or vitrectomy. However, VMT associated with epiretinal membrane has a significant lower success rate and pars plana vitrectomy with ERM peel might be considered as the first line option. However, further larger and prospective studies are needed to support our findings.

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.