Incidence and risk factors of massive subretinal hemorrhage in retinal angiomatous proliferation

Objective To evaluate the incidence and associated risk factors of massive subretinal hemorrhage (SRH) in patients with retinal angiomatous proliferation (RAP). Methods A total of 187 eyes of 135 treatment-naıve patients diagnosed with RAP were evaluated retrospectively. Clinical records including the time between the initial visit, last anti-vascular endothelial growth factor (VEGF) treatment, last stable examination, and the date of massive SRH were reviewed. Imaging findings including indocyanine green angiography (ICGA) and optical coherence tomography (OCT) were analyzed. Results Massive SRH developed in 18 eyes (9.6%) a median of 20 months after the initial presentation. Kaplan-Meier survival analysis revealed that the incidence (2.8, 5.8, 13.1, and 21.0% after 1,2,5 and 10 years, respectively) continuously increased. Among 14 eyes with discernable vascular anastomosis on baseline ICGA, 13 (92.8%) showed retinal arteriole involvement. On spectral-domain OCT imaging of the last visit prior to the massive SRH, a layered lamellar tissue complex was noted under the retinal pigment epithelium in 9 of 13 eyes, which was significantly associated with massive SRH[hazard ratio(HR),5.883;P = .010]. The average time between the last stable examination/last injection and the massive SRH was 2 and 5 months, respectively. The patients were treated with anti-VEGF, gas and recombinant tissue plasminogen activator injection; however, all except one eye had visual acuity worse than 20/1000 at the final visit. Conclusions Massive SRH can occur in RAP in the course of anti-VEGF treatment, resulting in severe vision loss. A proactive dosing regimen may be more appropriate for these RAP eyes.


Results
Massive SRH developed in 18 eyes (9.6%) a median of 20 months after the initial presentation. Kaplan-Meier survival analysis revealed that the incidence (2.8, 5.8, 13.1, and 21.0% after 1,2,5 and 10 years, respectively) continuously increased. Among 14 eyes with discernable vascular anastomosis on baseline ICGA, 13 (92.8%) showed retinal arteriole involvement. On spectral-domain OCT imaging of the last visit prior to the massive SRH, a layered lamellar tissue complex was noted under the retinal pigment epithelium in 9 of 13 eyes, which was significantly associated with massive SRH[hazard ratio(HR),5.883;P = .010]. The average time between the last stable examination/last injection and the massive SRH was 2 and 5 months, respectively. The patients were treated with anti-VEGF, gas and recombinant tissue plasminogen activator injection; however, all except one eye had visual acuity worse than 20/1000 at the final visit.

Conclusions
Massive SRH can occur in RAP in the course of anti-VEGF treatment, resulting in severe vision loss. A proactive dosing regimen may be more appropriate for these RAP eyes. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Introduction between the optic disc and fovea, and extending past the major vascular arcade of the retina. [21,22] Information regarding the patient's medical history and types and dates of treatment were collected along with clinical details of the massive SRH. The interval between the last treatment and massive SRH, and between the last stable examination and massive SRH, were analyzed. Snellen best-corrected visual acuity (BCVA), digital color fundus photographs, and OCT images were obtained at baseline and at each follow-up visit for all patients. OCT images were obtained using a Stratus (Carl Zeiss Meditec, Dublin, CA, USA) or Cirrus OCT (Carl Zeiss Meditec) before November 2011, and then with a Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). FA and ICGA (Heidelberg Retina Angiograph; Heidelberg Engineering), with a confocal scanning laser ophthalmoscope system, were performed at baseline and when required. Although Yannuzzi and associates classified the three stages of RAP, it is often difficult to differentiate whether the serous PED is associated with subretinal neovascularization (stage 2 RAP) or CNV (stage 3 RAP). Thus, in this study, RAP was classified as early stage (RAP without PED), which was equivalent to stage 1, or as advanced stage (RAP with PED) that included stages 2 and 3.
Initially, treatment-naïve RAP eyes were treated with a combination of photodynamic therapy (PDT) and bevacizumab (Avastin; Genentech, South San Francisco, CA, USA) followed by anti-VEGF injections with as-needed dosing [pro re nata (PRN)]) or anti-VEGF monotherapy from the beginning, based on PRN dosing with or without three loading injections. Both ranibizumab (Lucentis; Genentech) and bevacizumab were used, and selection between the two drugs was determined mainly by considering the national health insurance coverage and reimbursement schedules. All patients were followed regularly at 1-3-month intervals, depending on the lesion activity, and retreatment was given whenever evidence of persistent/ recurrent fluid, involving or threatening the fovea, was detected on follow-up OCT. For the treatment of massive SRH, patients received anti-VEGF injections with or without gas/recombinant tissue plasminogen activator (rt-PA) injection, and all decisions were at the discretion of a single physician (WK Lee). When gas/rt-PA injection was performed, 0.1 mL (50 μg) of rt-PA and 0.2 mL of 100% C 3 F 8 gas were intravitreally injected, followed by an anti-VEGF injection within 1 week. Patients injected with gas were instructed to remain in the prone position for as long as possible (i.e., for at least 24 h and up to 5 days). A pars plana vitrectomy was performed when a breakthrough vitreous hemorrhage occurred.
The proportion of patients who developed a massive SRH at each time point was analyzed with a Kaplan-Meier survival curve model using SPSS software (ver. 20.0; SPSS Inc, Chicago, IL). The date of initial treatment visit was designated as the index date. If a patient was lost to follow-up or did not present with a massive SRH, the latest visit was considered the last visit and it was regarded as censored data in the survival analysis. Cox regression analysis was performed to investigate the risk factors for massive SRH with clinical variables, such as age, hypertension, initial BCVA, use of PDT, and other funduscopic/OCT findings.

Results
During the study period, in total, 187 eyes of 135 patients were diagnosed with RAP and 104 patients were females (77.0%). The mean age at onset was 75.1 ± 6.3 (range, 58-95) years and baseline logMAR BCVA was 0.51 ± 0.46 (range, 0.0-1.7). The mean follow-up period for all patients was 47.6 ± 29.0 (range, 6-133) months. During the follow-up period, 39 (20.9%) eyes received a combination of PDT and bevacizumab as the initial treatment and 148 (79.1%) eyes received anti-VEGF monotherapy. Baseline characteristics of the 187 enrolled eyes are shown in Table 1.
Among these eyes, RAP as a cause of massive SRH was diagnosed in 18 eyes of 17 patients; thus, the overall incidence of massive SRH was 9.6%. Massive SRH appeared after a median follow-up period of 20 (range, 6-98) months. Kaplan-Meier survival analysis revealed that 2.8% of eyes had developed massive SRH within the first year of the follow-up period. Thereafter, 5.8%, 13.1%, and 21.0% suffered massive SRH within 2, 5, and 10 years, respectively, during the follow-up period (Fig 1). Non-massive SRH (larger than 1 disc diameter, but smaller than defined as massive SRH in our method) occurred in eight eyes during the follow-up period.
The median age of the 17 patients with massive SRH was 71 (range, 64-84) years. Of them, 3 patients were males and 14 were females. Six (35.3%) patients were on aspirin therapy and no patient was using any other kind of anticoagulation therapy. The median follow-up duration was 59 (range, 24-104) months for these patients. Bilateral involvement of RAP was noted in 14 eyes during the follow-up period, and GA without neovascularization was diagnosed in the contralateral eyes of the other four eyes. Macular soft drusen within the major vascular arcade were seen in all eyes and reticular pseudodrusen in 15 (83.3%) eyes. Based on the baseline examinations, all 18 eyes were in the advanced stages (RAP with PED).
Although retinal-retinal anastomosis or retinal-choroidal anastomosis was found in all RAP lesions, their venous or arterial origin was detectable on baseline ICGA in 158 of 187 (84.5%) eyes. In the other 29 eyes, we failed to obtain enough images of the early filling sequence and/or the image quality was too poor to identify the origin of vascular anastomosis. Among 158 eyes, 75 (47.5%) showed retinal arteriole involvement on baseline ICGA and 83 (52.5%) showed involvement of retinal venules alone. The vascular component was detectable in 14 of 18 eyes with massive SRH. Among them, 13 (92.8%) eyes showed retinal arteriole involvement on baseline ICGA and one showed involvement of retinal venules only. The rate of retinal arteriole involvement was significantly higher in eyes with massive SRH (P < .001, Chi-square test).
In two eyes, massive SRH occurred within 1 month after the last anti-VEGF injection (3 days and 3 weeks, respectively). The other 16 eyes showed no sign of disease activity, such as macular hemorrhage or fluid, on the most recent examination and OCT before the occurrence of , and the vertical axis shows the proportion of eyes that developed SRH. The incidence rates of massive SRH were 2.8%, 5.8%, 13.1%, and 21.0% of eyes at 1, 2, 5, and 10 years, respectively, after the initial treatment. https://doi.org/10.1371/journal.pone.0186272.g001 the massive SRH. The average time between the last stable examination and the massive SRH was 2 (range, 1-3) months in these 16 eyes.
OCT at the last visit prior to the massive SRH showed widespread fibrovascular tissue under the RPE in 17 of 18 eyes, without any intraretinal/subretinal fluid in 16 eyes. Spectralis OCT was performed at this visit in 13 eyes, and interruption of the RPE line (atrophic RPE layer) was noted within an area of widespread fibrovascular tissue in all 13 eyes. A well-organized layered lamellar band was seen within fibrovascular tissue in 9 of 13 eyes. Representative cases are shown in Figs 2 and 3.
In total, 14 eyes received intravitreal anti-VEGF with gas/rt-PA injections, and four eyes were treated with anti-VEGF injections only. Treatment decision was made at the discretion of a single physician (WK Lee) considering the height of SRH under the fovea, visual acuity at the last visit prior to the massive SRH, and whether the patient can maintain a prone position. In 10 (55.6%) eyes, massive SRH resulted in vitreous hemorrhage, and these eyes underwent pars plana vitrectomy. During the follow-up period, RPE tears (a bare choroid with elevation or scrolling of the torn RPE flap on fundus photography and OCT), which were not seen initially with massive SRH, were noted in six eyes after the absorption of massive SRH (Fig 4).
The median BCVA was 20/100 (range: 20/200-20/32) at baseline, decreasing to 20/200 (range: 20/1000-20/40) at the last visit prior to the massive SRH. All except one patient (case 10) had vision worse than 20/1000 at the final visit after the massive SRH. The clinical characteristics of the patients and the treatment results are summarized in the Table 2.
Among 187 eyes of 135 patients, 169 (92.0%), 4 (2.1%), and 11 (5.9%) eyes were analyzed with Spectralis, Cirrus, and Stratus OCT, respectively, until their final visit. Cox regression analysis of 169 eyes analyzed with Spectralis OCT revealed a statistically significant association of massive SRH with the layered lamellar tissue complex on OCT (hazard ratio, HR = 5.883; 95% confidence interval, CI = 1.524-22.713; P = 0.010). No other variable, such as age, hypertension, baseline BCVA, use of PDT, reticular pseudodrusen, or fibrovascular tissue under the RPE at final visit, showed a significant association with the occurrence of massive SRH (Table 3).

Discussion
This is the first study to report the incidence of massive SRH in eyes with RAP lesions (9.6%). The incidence increased steadily, from 2.8% at year 1 to >10% after the 5-year period of treatment and follow-up. In the 18 eyes of our study, massive SRH occurred at the advanced stages of the disease. All except one eye showed widespread fibrovascular tissue under the RPE at the visit prior to the massive SRH, indicating disease progression despite treatment primarily with anti-VEGF drugs (Fig 5). The median BCVA also worsened from baseline until the occurrence of the massive SRH. No eye presented with massive SRH at initial presentation, and the median interval between baseline and the massive SRH was approximately 2 years. This observation somewhat differs from the clinical characteristics of SRH in PCV, which is often found at the initial presentation.
The proposed mechanism of SRH in PCV involves the rupture of polyps, which may occur before any treatment by an increased tension within the polyps during the natural exudative course of the disease. [23] In contrast, massive SRH occurred in RAP eyes that were being treated with chronic and repeated anti-VEGF treatments. Nine of the 13 eyes that underwent Spectralis OCT before the massive SRH showed well-organized layered lamellar bands within the fibrovascular tissue under the RPE, which showed a significant association with the occurrence of massive SRH in a Cox regression analysis. This structure was termed "multilayered PED" by Rahimy and associates. [24] They assumed that this hyperreflective band develops within a chronic fibrovascular PED after repeated anti-VEGF suppression, and consisted of fibrocellular tissue. This fibrous tissue may exert contractile force on the fragile neovascular vessels, which could lead to rupture of these vessels, resulting in massive SRH. RPE tears observed after absorption of the hemorrhage in about one-third of our patients support this. The mechanism of RPE tearing is postulated to involve rapid involution and contraction of neovascular tissue adherent to the undersurface of the RPE. [25] It is noteworthy that retinal arterioles were involved in the RAP lesion in 13 of 14 (92.8%) eyes with discernable vascular anastomosis on ICGA. A limited number of studies have investigated the nature of the vascular anastomosis found in RAP. [26][27][28] The involvement of retinal veins ranged from 60 to 82.1%, while the involvement of retinal arteries ranged from 17.6 to 40%, which was 47.5% in eyes without massive SRH in our data. The percentage of arteriole involvement of vascular anastomosis in eyes with massive SRH was notably higher than in eyes without massive SRH and previous studies. The hydrostatic pressure of retinal arteries entering the eye is approximately 60-70 mmHg, and the pressure of the retinal vein is equal to the intraocular pressure (IOP). The hydrostatic pressure of choriocapillaris is 7-8 mmHg higher than the IOP [29], and the pressure gradient between the retinal and choroidal vessels in retinal-choroidal anastomosis may be higher when retinal arterioles are involved in the anastomosis, compared with the involvement of retinal venules. It is therefore possible that the large pressure gradient between the retinal arteriole and neovascular lesion may be one of the factors contributing to the development of massive SRH in RAP.
The use of intravitreal injections of anti-VEGF agents has resulted in a significant improvement of visual outcomes for patients with RAP; however, the optimal dosing regimen still remains to be determined. The most widely used dosing strategy in clinical practice involves a variety of as-needed regimens and treat-and-extend regimens. [30,31] An advantage of the treat-and-extend regimen compared with the as-needed regimen is that it can reduce the risk of SRH during the injection-free interval because of a mandated injection at each visit. [32] In contrast, additional injections with this regimen might increase the risk of contraction of fibrovascular tissue and development of hemorrhage; however, only 2 of 18 eyes developed massive SRH within 1 month after the last anti-VEGF injection in our series. Most of the eyes were not receiving anti-VEGF injections because of apparently stable disease status. The median time between the last injection and the massive SRH was 5 months, and all eyes developed sightthreatening SRH within 3 months of a stable clinical examination and OCT, indicating the need for proactive treatment in these cases. Another concern regarding a proactive regimen is that more frequent injections may accelerate the development of GA. RAP has been identified as a risk factor for GA development following anti-VEGF treatment. However, there is a controversy regarding any association between the number of injections given and the GA growth rate, and at present, there are insufficient data to determine this. [33,34] Because massive SRH in RAP resulted in severe vision loss in most of our cases, it would be beneficial to use the treat-and-extend regimen for the prevention of hemorrhages in RAP eyes with a high risk of a massive SRH. Our study had several limitations, including its retrospective nature. The initial treatment regimens differed among patients and two patients received PDT initially. However, the intervals between the last PDT and SRH were too long in these patients to suggest a direct effect of PDT on massive SRH. Since we could not collect that data from all enrolled RAP patients, the use of antiplatelet and/or anticoagulation therapy was not included in Cox regression analysis. Although antiplatelet/anticoagulation drugs could be a risk of ocular hemorrhage, the association between such drugs and SRH in nAMD has not been established in large clinical studies yet. [17,35] Finally, involvement of the retinal arteriole was not included in the Cox regression analysis because vascular anastomosis was not discernable in about 30% of all cases. However, the percentage of retinal arteriole involvement in eyes with massive SRH was higher than in previous reports with larger cohorts and that of eyes without massive SRH in our series, so we consider that it may be a contributing factor to the development of massive SRH in RAP.
To our knowledge, this is the first report on the incidence of massive SRH with a large RAP-only cohort, and we collected a relatively large number of cases with massive SRH which was previously considered as an uncommon complication. We performed a survival analysis using follow-up data, which ranged up to 11 years from the initial presentation. The cumulative incidence rate of massive SRH in RAP increased steadily with time and massive SRH usually occurred during injection-free intervals, in the chronic, advanced stages of RAP. We assume that layered lamellar tissue complexes within fibrovascular tissue under the RPE and involvement of retinal arterioles were contributing factors in the development of massive SRH. .240 Well-organized layered lamellar tissue within fibrovascular tissue on SD OCT at final visit Because the incidence of massive SRH in RAP is not rare, and the clinical course is devastating even after many different treatments, proactive dosing of an anti-VEGF agent may be a more appropriate approach in RAP eyes at high risk. Additionally, identification of the precise mechanisms, risk factors, and management of massive SRH in RAP should be investigated in future studies.

Author Contributions
Conceptualization: Jae Hyung Lee, Won Ki Lee.