Long-Term Outcome of Anti-Vascular Endothelial Growth Factor Therapy Following Vitrectomy for Submacular and/or Vitreous Hemorrhage in Neovascular Age-Related Macular Degeneration

Purpose: To study long-term clinical outcomes in patients with submacular hemorrhage (SMH) and/or vitreous hemorrhage (VH) associated with neovascular age-related macular degeneration (nAMD), who received pars plana vitrectomy (PPV) followed by anti-vascular endothelial growth factor (VEGF) therapy. Methods: Results: Mean best-corrected visual acuity (BCVA) of all patients improved signicantly at 1, 3, 6, 12, 18 and 24 months (P<0.01) post-PPV and at the nal visit (P<0.05). Mean of 13 eyes with anti-VEGF therapy improved signicantly at 1 (P<0.05), 3, 6, 12 (P<0.01), 18 and 24 months (P<0.05), while 12 eyes without anti-VEGF therapy improved at 1, 3 and 6 months (P<0.05) only. Average duration from initial PPV to anti-VEGF therapy initiation was 7.54±9.9 months. Five of 13 eyes (38.5%) with anti-VEGF therapy maintained dry macula for more than 1 year after the last injection.

This study was a retrospective case series of consecutive patients with nAMD complicated by SMH and/or VH, who were treated with PPV and followed for at least 24 months after PPV at the Department of Ophthalmology of Tokyo Medical University Hospital between April 2009 and March 2017. The study adhered to the tenets of the Declaration of Helsinki and was approved by the institutional review board.
All subjects provided written informed consent for the treatment and for this study. Inclusion criteria were nAMD with thick SMH of at least 1.5 disc diameter involving the fovea or VH. Exclusion criteria were other concurrent intraocular diseases. Intravitreal injection of anti-VEGF agent was performed after PPV for unresolved or recurred exudative changes detected by ophthalmoscope or optical coherence tomography (OCT). In some cases, the treat and extend (TAE) regimen [6,7] was used for subsequent injections depending on the activity of nAMD while receiving anti-VEGF therapy.
From 6 months or more after PPV, these examinations were performed every 2-4 months in less active cases. In the case of TAE, the examinations were performed as per visit interval.
The surgical procedures were as follows. In all eyes, a standard 3-port 25-gauge PPV was performed and posterior vitreous detachment was created if not already present. At the time of PPV, phacoemulsi cation, aspiration and intraocular lens implantation (PEA-IOL), injection of 32,000 IU/0.2 ml of tissue plasminogen activator (tPA) (Monteplase; Eizai Co, Tokyo, Japan) solution into the subretinal space using a 38-gauge exible cannula (Alcon), and intravitreal injection of anti-VEGF agents were performed as needed. Fluid-air exchange and tamponade with air, SF 6 or silicon oil was also conducted when necessary.
Statistical analysis was performed using MedCalc Statistical Software version 16.2.0 (MedCalc Software Bvba, Ostend, Belgium; https://www.medcalc.org; 2016). Visual acuity was converted to logarithm of the minimal angle resolution (logMAR), and counting ngers, hand motion and light perception were also converted to logMAR as reported previously [14]. Wilcoxon signed-rank test was used for comparison of preoperative and postoperative visual acuity, and Fisher's exact probability test and Mann-Whitney's U test were used to compare two groups with and without anti-VEGF therapy following PPV. P values less than 0.05 were considered to be statistically signi cant.

Results
In this study, 25 eyes of 25 patients (9 women and 16 men) were included. All patients were followed for at least 24 months (mean 47.0 ± 20.5; range 24-96 months) after the initial PPV. The mean age was 76.2 ± 9.3 (range, 48-89) years at the initial PPV. Eighteen eyes were phakic and 7 eyes were pseudo-phakic at the time of surgery. The subtypes of nAMD were PCV in 18 eyes, RAP in 2 eyes, and unclassi ed in 5 eyes. The reasons for performing PPV were SMH (8 eyes) and VH (17 eyes) associated with nAMD. The average duration of SMH was 9.3 ± 4.6 (range 4-17) days. The average duration of VH was 45.9 ± 40.0 (range, 13-142) days, with the exception of 4 eyes with unknown duration. The average greatest diameter of the SMH was 4.0 ± 1.6 (range 2.0-6.7) disc diameters. Thirteen eyes had previously been treated with photocoagulation (PC), photodynamic therapy (PDT) or intravitreal injection of anti-VEGF drugs. Baseline characteristics of the patients are listed in Table 1. Of the 17 eyes with VH, SMH was also detected during vitrectomy in 5 eyes. Simultaneous with PPV, the following procedures were performed: PEA-IOL in 17 eyes, injection of tPA into subretinal space in 13 eyes (8 eyes with SMH only, 5 eyes with VH and SMH), intravitreal injection of anti-VEGF agents in 6 eyes (3 eyes with SMH only, one eye with VH and SMH, 2 eyes with VH only), and uid-air exchange with tamponade in 12 eyes (air in 1 eye, SF 6 in 9 eyes, and silicon oil in 2 eyes).
Additional treatments after initial PPV in the 25 eyes are shown in Fig. 1. After the initial PPV, 10 eyes received intravitreal injection of anti-VEGF agent. One eye underwent PDT and subsequently received anti-VEGF therapy. Four eyes underwent repeat PPV for postoperative macular hole, postoperative proliferative vitreoretinopathy, silicon oil removal, and recurrent VH (1 eye each), 2 of which later received anti-VEGF therapy. Two eyes underwent PC. Eight eyes received no additional treatment. In the patient with recurrent VH after the initial PPV, retinal detachment occurred after the second PPV, and the third PPV and scleral buckling were performed. Therefore, a total of 13 eyes were initiated intravitreal anti-VEGF therapy and 12 eyes did not receive intravitreal anti-VEGF therapy. Subsequent analyses were conducted to compare these two groups.
Of the 13 eyes that were given intravitreal injection of anti-VEGF agents, 12 eyes received a ibercept (IVA) and 1 eye received ranibizumab (IVR). There were no signi cant differences in gender ratio, age, presence of polyps, and baseline BCVA between the eyes with and those without anti-VEGF therapy. Regarding the reason for the initial PPV (SMH or VH), 6 of 8 eyes (75.0%) with SMH detected before PPV compared with 7 of 17 eyes (41.2%) with VH were treated with anti-VEGF therapy after vitrectomy, and the rate was apparently higher in eyes with SMH although there was no signi cant difference.
Of the 13 eyes with anti-VEGF therapy, BCVA improved in 9 eyes (69.2%) and worsened in 1 eye (7.7%) for more than 3 lines (15 letters) on the EDTRS chart at 24 months after PPV. Among the 12 eyes without anti-VEGF therapy, BCVA improved in 7 eyes (58.3%) and worsened in 1 eye (8.3%). The causes of vision loss were chronic macular edema in one eye with anti-VEGF therapy and macular atrophy after massive subretinal hemorrhage in one eye without anti-VEGF therapy.
The average duration from the initial PPV to initiation of anti-VEGF therapy was 7.54 ± 9.9 (range 1-32) months. The cumulative percentage of anti-VEGF therapy initiation during follow-up is shown in Fig. 4. In the 13 patients who received anti-VEGF therapy, the cumulative percentage of initiation was 53.8% at 2 months after surgery and 84.6% at 10 months. Anti-VEGF therapy was given by an as-needed regimen in 8 eyes and by a TAE regimen in 5 eyes (2 eyes converted to as-needed regimen). The mean follow-up period from the initiation of anti-VEGF therapy was 32.7 ± 17.0 (range 0-63) months, and the average number of anti-VEGF drug injections until the nal observation was 6.2 ± 3.9 (range, 1-13). Three eyes underwent PDT after PPV. In one eye, PDT was performed before the initiation of anti-VEGF therapy. In this eye, the exudative change was not resolved after PDT, and anti-VEGF therapy was started resulting in dry macula. In the other two eyes, dry macula was not obtained even after the initiation of anti-VEGF therapy, and PDT was performed which achieved dry macula in one eye, but not in the other.
Five of the 13 eyes (38.5%) that started anti-VEGF therapy after PPV and 12 of 25 eyes (48.0%) of eyes that did not receive postoperative anti-VEGF therapy maintained dry macula for more than 1 year after the last injection. Eventually, 17 of 25 eyes (68.0%) with or without anti-VEGF therapy achieved dry macula.

Discussion
We evaluated the rate of initiation of anti-VEGF therapy after PPV for SMH and/or VH associated with nAMD. In the case of SMH caused by AMD, the rate of anti-VEGF therapy started after PPV was reported to be 38.6% for a mean follow-up period of 15.3 (range 3-70) months by Chang et al. [15] and 80.8% for a mean follow-up period of 6.3 (range 2-13 months) by Plemel et al. [16]. The rate of initiation of anti-VEGF therapy varies among studies because the follow-up period and the conditions for starting anti-VEGF therapy are different. We initiated anti-VEGF therapy after PPV in 6 of 8 eyes (75.0%) with nAMDassociated SMH detected before PPV.
Anti-VEGF therapy was started within 2 months after PPV in approximately one-half of the cases studied. Cases in which anti-VEGF therapy was started early after PPV tended to show delayed improvement in exudative change after the operation. The rate of initiating anti-VEGF therapy was considerably low later than 11 months after PPV.
In all patients, there was a signi cant improvement in mean BCVA compared to baseline during the whole follow-up period. In patients with anti-VEGF therapy after PPV, the improvement in mean BCVA peaked at 3 to 6 months after surgery but remained signi cantly improved by 24 months. In contrast, in patients without anti-VEGF therapy, the mean BCVA improved signi cantly only during the rst 6 months after PPV, and there was no signi cant difference thereafter.
Chang et al. [15] reported that patients who received anti-VEGF treatment had signi cantly improved visual acuity for 6 months after PPV, whereas those who did not receive anti-VEGF therapy improved only up to 3 months after surgery. In patients without anti-VEGF therapy, although vision is improved during the early stage after PPV due to reduced bleeding and exudative changes, visual acuity tends to decline thereafter when scarring and retinal atrophy gradually progress.
There is little evidence in the literature on the effect of anti-VEGF therapy on nAMD in vitrectomized human eyes. Hahn [17] presented a case of successful treatment with a ibercept following single bevacizumab failure in a patient with recurrent choroidal neovascularization following prior macular translocation vitrectomy surgery for AMD. Jung et al. [18] reported a case series of 4 patients with AMD who were previously vitrectomized for macular pucker or macular hole. In their study, treatment with a ibercept was effective in controlling AMD.
Because of faster clearance of anti-VEGF drug from vitrectomized eyes, injection of anti-VEGF drug in vitrectomized eyes is expected to be less e cient than in non-vitrectomized eyes [3][4][5]. On the other hand, there are also reports that intraocular pharmacokinetic properties of anti-VEGF drug in vitrectomized eyes are similar to those in non-vitrectomized eyes [19].
This study has some limitations. First, the preoperative characteristics of eyes that had previous photocoagulation, PDT and anti-VEGF also affects the visual acuity outcomes as these eyes may have more signi cant disease or refractory disease hence leading to the vitreous hemorrhage and/or submacular hemorrhage. Second, the additional postoperative treatments were variable, including PDT and repeat surgery for other complications. All these variabilities affect the outcome of vision and the timing of initiation of anti-VEGF treatment.
In conclusion, in patients with SMH and/or VH caused by nAMD, initiation of anti-VEGF therapy when exudative changes are unresolved or recur after PPV maintains the improved visual acuity over a long period. Therefore, anti-VEGF therapy may be useful even in vitrectomized eyes.

Declarations
Ethics approval and consent to participate The research was conducted in accordance with the guiding principles of the Declaration of Helsinki and approved by the institutional review board of Tokyo Medical University (No.SH3782). Informed consent to participate was obtained from all individual participants included in this study.

Consent for publication
Not applicable Availability of data and material The datasets during and/or analyzed during the current study available from the corresponding author on reasonable request.