Outcomes of a postoperative perfluorocarbon liquid tamponade for complex retinal detachments: 12 years of experience in southern Thailand

Background The study evaluates both functional and anatomical outcomes of retinal detachment (RD) repair by vitrectomy and perfluorocarbon liquid (PFCL) tamponade. Methods A retrospective chart review of patients who underwent vitrectomy using PFCL tamponade for RD repair from causes such as giant tear, chronic RD, or RD with previously failed surgery. Results This study included 122 eyes from 121 patients. One-hundred fourteen eyes (93.5%) had baseline vision worse than 20/200. The median duration of intraocular PFCL retainment was 14 days before gas or silicone oil replacement. The retinal reattachment rate was 80.3%. At 1 year, the retention probability of retinal reattachment was 0.84 (95% confidence interval, 0.77–0.91). Although visual improvement was found in 45.9% of patients, the median of final vision was not different between baseline and the last visit. Conclusion The rate of retinal reattachment operated with a short- to medium-term PFCL tamponade achieved a high satisfaction rate. However, postoperative hypotony was a predictor for unfavorable visual and anatomical outcomes.


Background
Perfluorocarbon liquid (PFCL) is a synthetic, fluorinated, carbon-containing compound. It is clear, colorless, odorless, and heavier than water. The PFCL does not affect the intraoperative view. The physical properties of PFCL make it an ideal tool for flattening the retina during vitreoretinal surgery for many types of pathologies, such as retinal detachment (RD) with severe proliferative vitreoretinopathy (PVR) [1][2][3][4], giant retinal tear [5][6][7][8][9][10], and pediatric RD [11]. Patients with chronic RD or those who underwent previously failed surgery are frequently accompanied by severe PVR, which sometimes produces severe peripheral retinal contraction and shortening and may require a retinectomy. During injection, the high specificity gravity of PFCL allows the detached retina to be smoothly flattened from the posterior pole and displaces unwanted subretinal fluid anteriorly passing through the presenting retinal breaks or peripheral retinectomy. In contrast, an air-fluid exchange flattens the retina from the periphery and pushes subretinal fluid toward the posterior pole; therefore, a retinotomy is required. Using PFCL can also be an alternative or adjunctive to the standard scleral buckling procedure.
We conducted a retrospective study of patients who underwent vitreoretinal surgeries with a short-to medium-term PFCL tamponade. This study primarily aims to demonstrate both anatomical and functional results after PFCL placement and removal. The secondary aim was to evaluate factors that affect the outcomes.

Data collection
This retrospective study was approved by the Ethics Committee of the Faculty of Medicine, Prince of Songkla University, Songkhla, Thailand (REC number 58-364-02-3). The study included patients with RD who underwent pars plana vitrectomy (PPV) using postoperative PFCL tamponade (with or without scleral buckling procedure) in the tertiary referral university hospital. Data collection was performed using the hospital's electronic database to identify patients who had RD and were treated by standard three-port PPV. The search terms used were the following International Statistical Classification of Diseases-10 and International Statistical Classification of Diseases-9-Clinical Modification codes: (1) H33.0 RD with retinal break, (2) H33.4 Traction detachment of retina, (3) H33.5 Other RDs, and (4) 14.74 Other mechanical vitrectomy by posterior approach.
Eligible patients were determined by manual search. The inclusion criteria included rhegmatogenous RD (RRD) from any cause, tractional RD (TRD), and any RRD or TRD having a previously failed treatment. The exclusion criteria were any concomitant ocular diseases, such as macular choroidal neovascularization, dense cataract, advanced or uncontrolled ocular disease (e.g., glaucoma, uveitis, uncontrolled systemic disease), pregnancy, or loss of critical data records.
The patients received 20-or 23-gauged PPV with PFCL (Perfluoron; Alcon Laboratories, Inc., Fort Worth, TX) tamponade by any of three vitreoretinal surgeons in our unit. The surgical techniques of all surgeons were performed in a similar way. A core vitrectomy was done, followed by posterior vitreous detachment induction and peripheral vitreous trimming. As much PVR as possible was removed before flattening the retina by PFCL. All patients were instructed to limit activities and lie in bed in a supine position during the postoperative period in the ward. Patients were allowed to ambulate in upright position only for meals and toileting. After 1 to 2 weeks, either perfluoropropane gas (Ispan C 3 F 8 ; Alcon Laboratories, Inc.) or silicone oil (Oxane 5700; Bausch and Lomb, Inc., Rochester, NY) tamponade was chosen for replacing PFCL in the second operation. Patients who required additional RD repair by reinfusion of PFCL were also excluded from the study. Any further operations to fix the residual RD, as well as cataract surgery, to achieve patient's maximal visual potential were allowed.

Data management
All eyes underwent at least two operations: PFCL placement and PFCL removal. The date of surgery with PFCL placement was defined as the starting date of the study. Data were recorded at 1 week, 1 month, 3 months, and 6 months after PFCL removal, as well as at the last visit. These included best-corrected visual acuity (BCVA), postoperative retinal status, and postoperative complications.
The primary outcome was reattachment of the retina (anatomical outcome). The secondary outcomes were vision improvement (functional outcome) and complications during the follow-up period. Vision improvement was determined by an increase of at least one line of BCVA by Early Treatment Diabetic Retinopathy Study chart or a one-step gain of nonnumerical vision test (e.g., hand motion improved to counting of fingers) at the last visit compared with the baseline BCVA. The final visual outcome was then classified into two groups: "better or improved VA" and "stable or worsened VA". A logarithm of the minimum angle of resolution (logMAR) VA was also used for vision calculation and comparison. All defined complications were elevated IOP greater than 25 mmHg, hypotony (IOP, < 5 mmHg), cataract progression, optic disc atrophy, ERM, corneal decompensation or edema, and retained intraocular PFCL bubble.

Statistical analysis
Baseline characteristics and important variables were analyzed in an R program version 3.4.0 (The R Group, Vienna, Austria) with Epicalc software (Brixton Health, www.brixtonhealth.com). The continuous variables were described as mean ± standard deviation (SD) or as median with interquartile range (IQR), depending on data distribution. The relationships between the status of either reattachment or final VA change were evaluated by Student's t test or the Wilcoxon rank-sum test for continuous data and by chisquare test or Fisher's exact test for categorical data. The recurrence time of RD after PFCL removal was used to plot a Kaplan-Meier survival graph. To identify a correlation of variables to the final outcomes, significant variables in a univariate analysis were introduced into a multivariate analysis. A logistic regression was applied to predict factors on retinal reattachment and final visual outcome. A result was considered statistically significant when there was a P value less than 0.05. In 30 eyes with previously failed RD repair, 64.5% had received silicone oil endotamponade, and the rest had received gas endotamponade before PFCL placement. A retinectomy was performed in 36.6% of these eyes. A retinectomy with PFCL placement was also performed in 31 (53.1%) eyes with chronic RD and 13 (76.9%) eyes with severe TRD.
After the operation with PFCL placement, 97.5% of the retinas were reattached at the time of PFCL removal. When PFCL was removed, 101 eyes (82.8%) were replaced by silicone oil and 21 eyes (17.2%) were replaced by gas. Of the 80 eyes (65.6%), 38 had additional operation for silicone oil removal, 33 had combined surgery for cataract extraction and silicone oil removal, 4 required additional vitrectomy for residual RD repair, 3 had cataract extraction only, 1 had an ERM peeling, and  1 had partial silicone oil release to reduce elevated postoperative IOP.
Of the 11 eyes (9%), 3 had cataract surgery or secondary intraocular lens insertion, 3 had a recurrent detachment repair with silicone oil tamponade, 2 had combined surgery for cataract extraction and silicone oil removal, 1 had four operations for silicone oil removal, 1 had an intravitreal antibiotic injection for suspected endophthalmitis, and 1 had an evisceration for phthisis bulbus.
Three eyes (2.5%) had fifth operations, one each for silicone oil removal, intraocular lens reposition, and band keratopathy removal. Only one patient, who had undergone a band keratopathy removal, underwent a sixth operation for silicone oil removal. A residual PFCL droplet in the anterior chamber or in the vitreous cavity was noted in six eyes (4.9%). However, none of them were confronted with subretinal PFCL.
Retinal reattachment was achieved in all cases during PFCL retention in the vitreous cavity. The reattachment rate at the last visit was 80.3%. Figure 1 demonstrates that the recurrent RD after PFCL removal was more likely to occur in the early phase. At 1 year, the probability of   Table 2 demonstrates groups of postoperative BCVA at different time points during the follow-up period. The vision of most eyes tended to improve over time. The final BCVA improved from baseline in 56 eyes (45.9%), remained stable in 46 eyes (37.7%), and worsened in 20 eyes (16.4%).
The most common complication was an immediate postoperative IOP elevation (51.2%), which usually occurred the next day after PFCL placement. IOP elevation and hypotony occurring after PFCL removal was found in 25.4 and 23% of eyes, respectively. Optic nerve atrophy and cataract progression were equally found in 22.1% of eyes. ERM occurred in 13.1% of eyes. A case of suspected late postoperative endophthalmitis, occurring at 5 years after the patient's third operation of combined phacoemulsification and silicone oil removal, was successfully treated with a single injection of intravitreal antibiotics.
The univariate analyses of the effect of pre-and postoperative variables on retinal reattachment and visual outcome are demonstrated in Tables 3 and 4. Although male sex was predominant in the study, sex had no effect on either anatomical or visual outcome (P = 0.828 and 0.405, respectively). Table 5 summarizes the results of multivariate analyses of characteristics regarding anatomical and functional successes. The significant predicting factors for retinal reattachment were right eye, lens status, and the absence of postoperative hypotony. Pseudophakic eyes appeared to have a better anatomical outcome than aphakic and phakic eyes. The significant predicting factors for better final BCVA were RD caused by giant tear and the absence of postoperative hypotony.

Discussion
Our study included 122 eyes from 121 patients with varied retinal pathologies. Difficult cases such as chronic RRD with PVR and previously failed surgery for retinal attachment comprised more than half of the patients included. The duration from the patient's symptoms to the starting date of this study was 125 ± 156.1 days. Therefore, the chronicity of the RD would be at least 4 or more months on average. The longer the duration of RD, the higher the risk of surgical failure in these patients. Therefore, silicone oil was a preferred choice of vitreous substitution to hold the retina in place after the retina was successfully reattached by PFCL (81.1%, oil; 17.2%, gas) and to lessen the risk of postoperative hypotony.
The result showed that a short-and medium-term PFCL retainment (≤20 days) was safe enough for complex RD repair in real-life practice. The success rate of reattachment was 80.3% of eyes. The causes of RD had an effect on the final visual outcome, but not on the retinal reattachment rate. A TRD, which was mostly caused by proliferative diabetic retinopathy, had the worst visual prognosis. The eyes with giant tear achieved better final vision than eyes with RD from other causes. This may be because of the fact that 84.4% of eyes with giant tear underwent a PFCL-assisted surgery on the first attempt and PVR was not involved much in these surgeries. The success of using PFCL in eyes with chronic RD and with reoperation was compromised from the preoperative PVR. An interval from the onset of symptoms to the time of PFCL placement did not affect the reattachment rate.
The majority of patients started with baseline VA worse than 20/200. The visual level at the last visit had improved in 45.9% of eyes. Eyes with giant tear accounted for the highest percentage of postoperative vision improvement. Mean logMAR VA increased from 1.8 ± 0.4 at baseline to 1.6 ± 0.7 at the last visit. However, the median logMAR VA was not different between baseline (2, 0.4-2.3) and last visit (2, 0.1-3.0). Other studies have reported a visual improvement that varied from 56.8 to 60% [1,4].
Our results showed a high rate of early postoperative IOP elevation in the next day after PFCL placement, which could be controlled by medication in all patients. Other reports found a transient IOP elevation that varied from none to 35.9% at any time postoperatively [8,13]. We also found that an IOP fluctuation was the main problem associated with either PFCL placement or PFCL removal. However, this did not affect the reattachment rate. Postoperative hypotony was noted to be a strong predictor for both anatomical and unfavorable visual outcomes.
Retinal toxicities and foreign body reaction from PFCL were occasionally reported [15,16]. There was no sign of retinal toxicity from the PFCL reported in our study. However, optic nerve atrophy was found in 22.1% of the overall cases in the late postoperative period, and this compromised a visual recovery in successful cases. Either using PFCL or repeated surgeries could account for late optic nerve atrophy. There were some limitations in this study. First, our center is a tertiary referral hospital; therefore, selection bias might have occurred. Second, the study was conducted retrospectively. Third, a small number of patients with TRD made it difficult to strongly predict the final visual prognosis. Finally, some patients still had their eyes filled with silicone oil at the end of the data collection period; therefore, the recorded BCVA might not reflect their true, final visual outcomes.

Conclusion
Using short-to medium-term PFCL tamponade for complex RD was anatomically successful in 80.3% of patients in this study. At least one-line or one-level vision gain was achieved in 45.9% of eyes. Postoperative hypotony was a strong predictor for both anatomical and unfavorable visual outcomes.