Objective analysis of perfluoropropane tamponade area after pars plana vitrectomy using ultra-widefield fundus stereographic projection images

To objectively analyze the gas tamponade area in patients with different concentrations of perfluoropropane (C3F8) after pars plana vitrectomy (PPV), retrospective analysis was performed in patients diagnosed with retinal detachment or macular hole and underwent PPV with C3F8 tamponade of two concentrations (14% and 20%). The ultra-widefield fundus (UWF) images at one and 10 days and four weeks postoperatively were corrected using stereographic projection to adjust peripheral distortion. The gas–fluid interface curvatures were delineated using UWF stereographic projection images, and the gas–fluid area ratio and estimated gas area were calculated at each concentration. Among 65 eyes, 30 were in the 14% group and 35 were in the 20% group. The gas–fluid area ratio was 0.71 in the 14% group and 0.83 in the 20% group at 10 days (p = 0.046) and 0.27 and 0.45, respectively, at four weeks postoperatively (p < 0.001). The estimated gas area was 52.56 and 60.82 mm2 at 10 days (p = 0.025) and 19.83 and 33.86 mm2 at four weeks (p < 0.001). The gas tamponade areas were objectively shown to be greater under the 20% concentration than the 14% concentration of C3F8 at 10 days and 4 weeks postoperatively using UWF stereographic projection images.

Pars plana vitrectomy 1,2 has been a procedure of choice in patients with retinal detachment 3 , macular hole 4 , and other vitreoretinal diseases 5 . And more than 50% of vitrectomies involve a gas tamponade to promote anatomical retinal re-attachment or macular hole closure 2 . With the advantages of high surface tension and a wide contact angle 6 , gases such as sulphahexafluoride, hexafluoroethane, or perfluoropropane (C 3 F 8 ) are commonly used mixed with air during vitrectomy 7 . To attain a successful tamponade effect on retinal breaks including sufficient area and duration with enough surface tension and buoyancy 8 , a gas's expansivity and longevity in the eye are important.
Though the term "nonexpansion concentration" is used, all gases expand after they are inserted into the vitreous due to the transfer of blood gases into the bubbles. Subsequently, the partial pressures of nitrogen, oxygen, and carbon dioxide with the retinal blood gas partial pressures are equalized and absorbed slowly 9,10 . For C 3 F 8 , 10% to 17% concentrations have been suggested as possible concentrations to prevent excessive expansion and the elevation of intraocular pressure (IOP) [11][12][13] . Previously, many studies were conducted on gas behavior and longevity in the eye, but most experiments were completed using animal models [14][15][16] or predicted through experimental models 17,18 , and there is a shortage of reports that objectively analyzed changes in intravitreal gas in the human eye 19,20 . Still, many surgeons choose gas concentration based on subjective preference. According to a survey among surgeons using C 3 F 8 , the concentration choice varied from 12 to 20% 12 .
Ultra-widefield (UWF) fundus imaging, which can obtain wide-angle fundus images of about 180°-200° covering the peripheral retina, became the standard-of-care for diagnosis and screening of retinal disease 21,22 . Most observations of gas behavior in humans have studied estimated gas volume through the gas-height level seen through the dilated pupil 23 or using A-scan ultrasound 20 . Unlike those studies, widefield fundus imaging which have large depth of focus allows the peripheral retina, intraocular gas bubble to be in focus simultaneously in gas-filled eyes and has made it possible to measure the gas-fluid level objectively. The UWF images digitally Gas tamponade area. The estimated gas area in the virtual coronal plane of the eyeball using a UWF stereographic projection image compared between each gas concentration is summarized in Table 2. The proportion of gas area in stereographic projection retinal image at 10 days after surgery was 0.71 ± 0.09 in the 14% group and 0.83 ± 0.07 in the 20% group (p = 0.046). At 4 weeks postoperatively, it was 0.27 ± 0.15 and 0.45 ± 0.15, respectively (p < 0.001). The estimated area of gas in virtual coronal plane image of eyeball was 52.56 ± 7.27 mm 2 in the 14% group and 60.82 ± 4.64 mm 2 in the 20% group at 10 days postoperatively (p = 0.025) and 19.83 ± 11.55 mm 2 and 33.86 ± 11.00 mm 2 , respectively at 4 weeks postoperatively (p < 0.001).

Discussion
Through this study, we were able to visualize the shape of the gas tamponade in the vitreous cavity that was mathematically presented by Eames et al. 17 and Bahill 27 . The curvature was a relatively flat horizontal surface in the lower hemisphere and changed to a meniscus or lens shape in the upper hemisphere. This not only supports the existence of a good match with previous mathematical models but also has the advantage of being clinically useful for gas shape prediction. Due to spherical shape properties, the contact arc changes rapidly at the inferior and superior positions and, when the gas volume in the vitreous cavity decreases below 50%, the contact surface sharply decreases 28 , so it would not have enough force to cover other retinal breaks that are not located in the superior area or to maintain gas-fovea contact after surgery for macular hole. In particular, as shown in Fig. 1, when the gas area is less than 50%, the gas-fluid interface has a more convex meniscus curve, and the contact arc is reduced more rapidly.
One of the concerns of surgeons when injecting gas after vitrectomy is the ideal concentration that can fill the vitreous cavity as much as possible without causing increased IOP. The retinal adhesive force through laser retinopexy stabilizes after 2 to 3 weeks and the cryotherapy weakens the adhesive force due to local inflammation in the first week after surgery 29,30 . In other words, to induce retina re-attachment, it is necessary to apply a force for a sufficient period of at least 2 to 3 weeks. In vitrectomy for macular hole, gas-fovea contact is essential for hole closure, and a long duration of intraocular gas tamponade showed a higher success rate for macular hole surgery (in comparing 16% vs. 10% C 3 F 8 ) 31,32 . Although non-inferiority of shorter-acting gas has been reported in recent studies 33 , there is still a surgeon preference for long acting gas 4,34 . In both indications, gas is injected to completely fill the vitreous cavity and achieve a larger gas-retinal contact area. But this is seldom achieved because of residual subretinal or preretinal fluid or fluid in the anterior vitreous area. For this reason, Williamson et al. suggested the theoretical gas concentration necessary to achieve 100% fill of the vitreous cavity is 13% to 19%,  www.nature.com/scientificreports/ when assuming 20% to 30% vitreous fluid remains after fluid-gas exchange 9 . Tamponade of slightly expansile gas is based on this theoretical calculation. Previously, Han et al. compared the effects of 12% and 20% C 3 F 8 gas on longevity and IOP in 30 patients and reported that the intraocular longevity was 6.7 weeks for 12% C 3 F 8 and 8.4 weeks for 20% C 3 F 8 , with no significant effect on IOP elevation seen between the two different concentrations 23 . In this study, the gas area occupied 83% of the virtual coronal plane of the eye in the 20% group at 10 days after surgery and 45% at 4 weeks after surgery. As C 3 F 8 is known to produce the maximal expansion on the third day and absorbed thereafter 11 , this suggests that 20% C 3 F 8 nearly filled the vitreous cavity at early postoperative days and remained as a larger tamponade area at 1 month postoperatively to achieve anatomical success. It is reported that the decline in the volume of the gas bubble follows a first-order exponential decay, showing bubble to have a constant "half-life" and be independent of the injected volume 15,28 . "Half-life" shows a linear increase with gas concentration 13 , which can help explain the larger tamponade area in the 20% C 3 F 8 in our study.
The use of expansile concentrations has been reported as a risk for IOP elevation 35 . However, in this study, there was an increase in IOP relative to the preoperative state in the 20% group on the first day after surgery (p = 0.016) but with no statistical difference relative to that in the 14% group (p = 0.143). Perhaps adequate outflow facility in these non-glaucomatous, pseudophakic eyes may have accommodated slow expansion without excessive pressure elevation. In the 20% group, ocular hypertension was often observed on the first day after surgery and, because the anti-glaucomatous eyedrop was applied earlier, the effect on the IOP of 20% C 3 F 8 might be masked. In other words, it is suggested that IOP elevation after fluid-gas exchange with 20% C 3 F 8 gas can be adjusted with a proper anti-glaucomatous eyedrop. These findings also support that IOP monitoring should be performed closely, regardless of the concentration used, as there was a patient with an increase in IOP at both concentrations.
There are some limitations in this study. This was a retrospective study of two indications that could not be compared with the success rate of surgery according to concentration. Both the mixed indications and the retrospective design prevent determination of the optimal concentration through this study. Further research is necessary to compare the surgical success rate within a single indication to find clinical benefit for longer duration of gas tamponade. In addition, the gas concentrations were chosen according to the preference of the operator and the timing of surgery between the two groups was different. It is thought that the difference in the timing of the operation did not affect the main outcome, but this should be considered in interpreting the results. Further, it was not possible to compare the longevity by comparing the fundus image until the gas had been completely absorbed. However, considering that the gas tamponade area at 4 weeks postoperatively was significantly larger in 20% C 3 F 8 than 14%, the longevity could also be estimated to be longer than 14% for 20% C3F8. In conclusion, we have demonstrated the UWF stereographic projection image as a useful methods for evaluation of gas tamponade state and we found that 20% C 3 F 8 in vitrectomy and fluid-gas exchange has advantages over 14% C 3 F 8 in terms of its sufficient retinal lesion cover. In nonglaucomatous, pseudophakic or aphakic eyes, the surgeon can choose the gas concentration by closely monitoring the IOP. www.nature.com/scientificreports/ gas (Teknogases ready-to-use gas; Teknomek Medical, Istanbul, Turkey) diluted with air as 14% and 20% connected to the infusion cannula and opposite port from infusion cannula was opened. The surgeon used only C 3 F 8 in vitrectomy cases which needed gas tamponade and used a 14% concentration of C 3 F 8 from January 2016 to February 2018 and switched to a 20% concentration of C 3 F 8 from March 2018 to October 2019 because the surgeon's preference had changed. A sufficient amount of diluted gas was injected over 40 cc while keeping the other port opened such that the vitreous cavity was fully filled with gas. After the cannula was extruded from the eye, if a bubble was observed at the sclerotomy site, the sclera was sutured.

Material and methods
Follow up. Follow-up visits were performed at one, three, and 10 days and one and two months postoperatively. At every visit, all patients underwent IOP measurement, slit lamp examination and dilated fundus examination. UWF fundus images were taken at one and 10 days (± 1 days) and 4 weeks (± 3 days) postoperatively after pupillary mydriasis. Patients without UWF images and patients whose gas-fluid interfaces were observed during UWF imaging at one day after surgery were excluded as from consideration due to insufficient gas tamponade in surgery or postoperative leak of gas. When the amount of gas in the vitreous cavity was sufficient so that the boundary was not visible in the dilated pupil (more than 80% according to the classical definition 19 ), no gas-fluid interface was observed on the UWF image (Fig. 2a,b). Anti-glaucoma eyedrops were added in patients with an IOP elevation of 21 mmHg or more. Patients with sustained ocular hypertension (IOP > 21 mmHg) who were not controlled were referred to a glaucoma specialist.
Gas tamponade area measurement. To avoid peripheral image distortion, UWF stereographical projection images (three-dimensional wrap eye model images) were analyzed using software available from the manufacturer (V2 Vantage Pro version 2.11; Optos Inc., Dunfermline, UK). There was a significant difference in the gas-fluid curvature between the conventional UWF image and stereographical projection image, especially when the gas was less than 50% in the eye (Fig. 3a,b). After the gas was absorbed partially, the image of the retina through the gas (upper ovoid area) and the reflected image of the inferior retina from the gas surface (lower banana shaped area) were observed (Fig. 2c,d). We considered the inferior line of the reflected image as the  found (a,b). Only those patients who showed no gas-fluid margin at one day postoperatively were included in this analysis. As the gas was absorbed, the gas-fluid curvature presented as a banana shape (c,d). The superior oval area originates from the image seen thorough the gas bubble and the inferior banana-shaped area from the reflecting mirror image of the inferior retina. The upper margin of the mirror image (red dotted line) was thought to be the gas boundary that contacted the retina. The lower margin of the mirror image from the reflection of the gas surface (blue dotted line) was thought to be the inferior margin of the gas. www.nature.com/scientificreports/ gas-fluid interface (boundary) 22 . To reduce underestimation of the gas in the vitreous cavity in the UWF stereographical projection images due to artifacts (e.g., patient's eyelids and eyelash), we established a circle centered on the macula that contained both ends of the horizontal gas-fluid interface as the horizontal angle was usually larger than the vertical angle of field due to the patient's eyelid using the open-source GNU Image Manipulation Program (GIMP 2.8.14) (Fig. 3c). Based on the fact that wide-field fundus photography has an angle of about 180° (reported as 180° to 200°), we adopted a circle as the "virtual coronal eyeball plane" that intersected the equator of the eye, with the fovea and cornea as poles 26 . The gas-fluid curvature was delineated using a Bézier curve model (Fig. 3c, blue dotted line), and gas-fluid curvatures from each UWF image were collected and synthesized (Fig. 1). Though the actual eyeball is not a perfect sphere, it is modeled as such, and the virtual coronal plane of the eye was assumed to be a circle whose diameter was the axial length of the patient's eyeball. For this assumption, we excluded myopic eyes (axial length > 25 mm) which might have a more aspherical shaped eyeball contour. Based on this assumption, the diameter of the "virtual coronal plane" was defined as the axial length of the patient's eyeball and the area was calculated. The gas tamponade area in projection image was measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA) as a pixel ratio (Fig. 3d) and calculated to estimate the gas area in the "virtual coronal plane" using axial length.
Statistical analysis. For statistical analysis, SPSS (SPSS version 20.0 for Windows; IBM Corp., Armonk, NY, USA) was used. BCVA was converted to the logarithm of the minimal angle of resolution (logMAR) and analyzed. The independent sample t-test used to compare continuous variables and the chi-squared test were The gas tamponade area (blue painted) in the virtual coronal plane is measured in this circle as a ratio. www.nature.com/scientificreports/ used in comparing for categorical variables. As IOP were repeatedly measured from the same subject over time, a repeated-measures analysis of variance (ANOVA) test was used to compare two groups (14% vs. 20% C 3 F 8 ).

Data availability
The raw data for this study is available upon reasonable request from the corresponding author.