Intravitreal anti-VEGF monotherapy for thick submacular hemorrhage of less than 1 week duration secondary to neovascular age-related macular degeneration.

AIM: To investigate the role of anti-VEGF monotherapy in patients with thick submacular hemorrhage (SMH) of ≤ 1 week duration secondary to neovascular age-related macular degeneration (N-AMD).
MATERIALS AND METHODS: A retrospective chart review of 14 eyes of 14 patients presenting with acute decrease in central vision of ≤ 1 week duration secondary to a thick SMH measuring ≥ 2 MPS disk areas from N-AMD was performed. Intravitreal injections of bevacizumab 1.25 mg (13 eyes) or ranibizumab 0.5 mg (1 eye) were given monthly until resolution of SMH and less frequently thereafter, based on treat-and-extend approach utilizing spectral domain optical coherence tomography (SDOCT). Patients with follow-up of ≥ 6 months were included.
RESULTS: Patients presented after a median of 4 (range 1-7) days from the onset of SMH. Mean lesion size was 27.9 mm² (range 5.47-100, median 15), with blood comprising 77-98% of the lesion. Presenting visual acuity (VA) ranged from 20/60 to hand motions (median 20/200). Patients received a mean of 11.4 (range 5-20) injections over 18.4 (range 7-50) months. SMH resolved in all eyes in a mean of 4.8 (range 2-8) months. At 6 months follow-up, mean VA gain was -0.54 logMAR (range: -1.5 to +1, Snellen range 20/25-20/400, median 20/100, P = 0.0037), with 11 gaining ≥ 0.2 logMAR. Mean change in VA from baseline at final follow-up was -0.58 logMAR (range -1.6 to +1, Snellen range 20/30-20/400, median 20/60; P = 0.0022).
CONCLUSION: A good anatomical and visual outcome can be accomplished in patients with thick SMH secondary to N-AMD treated with anti-VEGF monotherapy within 1 week.

Age-related macular degeneration (AMD) is the most common cause of visual loss among adults in the United States.[1] Predominantly, hemorrhagic choroidal neovascular lesions have a poor prognosis,[23] due to underlying neovascular process,[3] iron toxicity,[4] diffusion barrier, and clot retraction.[5] The natural history arm of Submacular Surgery Trial (SST) showed that at 1 year follow-up, median visual acuity (VA) in patients with hemorrhagic N-AMD declined to 20/500 from the baseline of 20/200 with 65% patients having <20/400 vision.[6] These patients were excluded from the pivotal trials that led to the FDA approval of ranibizumab.[78]

Several treatments such as pneumatic displacement with[910] or without intravitreal tissue plasminogen activator (tPA),[1112] intravitreal injection of tPA, gas and anti-VEGF agent,[131415] removal of hematoma with[616] or without[17] removal of choroidal neovascular membrane (CNVM), vitrectomy with tPA-assisted pneumatic displacement,[1618] and subretinal tPA and an anti-VEGF agent,[1920] have been proposed with variable results. Visual outcome in such eyes is related to the extent and thickness of the blood as well as VA at presentation.[21] Surgical approaches are generally considered for patients with thick hemorrhage of recent duration,[22] partly because of the belief that organization of the blood clot results in disciform scar.[23]

Anti-VEGF monotherapy has been evaluated previously in patients with submacular hemorrhage (SMH) from AMD with encouraging results.[2425262728] None of them specifically studied patients with large and thick SMH of recent onset. Berrocal et al. noted that patients with SMH not associated with a CNVM had an excellent natural history, with most patients regaining VA of ≥20/40 regardless of the size or thickness of the hemorrhage.[2]

We reviewed charts of patients with thick SMH ≥ 2 MPS disc areas of less than 1 week duration treated with anti-VEGF monotherapy alone to evaluate whether prompt control of neovascular process might prevent disciform scar formation and improve VA.

Materials and Methods

A retrospective chart review was conducted of patients from September 2006 through September 2010 who presented with an acute central scotoma or defective vision secondary to SMH from N-AMD. Inclusion criteria for the study included: Symptoms of ≤7 days duration, thick (causing visible elevation of the fovea by slit lamp biomicroscopy according to criteria used in previously published studies),[1826] ≥2 disc areas hemorrhage, hemorrhagic component ≥50% of total lesion size,[6] and follow-up of ≥6 months. Exclusion criteria included absence of blood beneath the fovea, indeterminate duration, SMH secondary to any condition other than N-AMD, pre-existing sub-foveal fibrosis, vision ≤20/400 prior to the occurrence of SMH, co-existent vitreous hemorrhage, any other ocular condition responsible for decrease in vision, or recent (≤3 months) thrombo-embolic event.

A complete eye examination, including best-corrected Snellen VA, slit lamp examination, and fundus biomicroscopy were performed at presentation. Fundus photos and intraveonus fluorescein angiography (IVFA) were performed to document an active CNVM. Images were imported into OIS Winstation™ Software (Ophthalmic Imaging Systems, Sacramento, CA). The software was used to measure total lesion size and size of CNVM in mid-phase angiogram.[29] Hemorrhagic component was calculated from these measurements. Spectral domain optical coherence tomography (SDOCT) was performed with either the Optovue RTVue FD-OCT™ system (Optovue Corporation, Fremont, CA) using the EMM5 (26 × 803 + 16 × 535 A-Scans) and Raster (17 parallel scans, 17 × 1024 A-Scans) Protocols or Zeiss Cirrus HD-OCT system Model 4000 (Carl Zeiss Meditec, Inc., Dublin, CA) with macular cube (512 × 128 line scans over a 6 mm × 6 mm area) and HD Raster (4096 A scans per line, 5 parallel lines, 0.25 mm apart) protocols. OCT scans were manually checked to avoid quantitative errors. At presentation, retinal thickness (distance between RPE to the ILM), location of hemorrhage (subretinal or subRPE), and thickness of subfoveal hemorrhage were measured manually on a high resolution raster image. For subretinal hemorrhage, thickness was equal to the distance between the RPE and the first high reflectivity line within the retina representing the junction between inner and outer segment of photoreceptors (is/os line). Thickness of subRPE hemorrhage was the distance from the RPE to an imaginary line contiguous with the edge of RPE adjacent to hemorrhage. The same OCT machine that was used for the initial evaluation was utilized for follow-up visits to allow for direct comparison between results.

After obtaining informed consent, an intravitreal injection of 1.25 mg bevacizumab was administered under topical or subconjunctival anesthesia with sterile technique using a lid speculum, 10% povidone iodine scrub, and topical 5% povidone iodine solution. Ranibizumab (0.5 mg) was used in Patient 9 who developed SMH 4 weeks after bevacizumab injection. Topical broad spectrum antibiotics were prescribed for 24 hours after the injection.

The patients were seen monthly and they underwent a complete ophthalmic examination, fundus photos, and SDOCT. IVFA was performed every 6-12 months or when there was a worsening of the patient's condition. Monthly injections were continued until complete resolution of hemorrhage[30] and submacular/intraretinal fluid based on clinical examination and SDOCT. Thereafter, interval between injections was increased by 1-2 weeks,[31] but not more than 2 months to optimize the intravitreal therapy to keep the lesion dry. Patient 4 was switched to ranibizumab due to persistent fluid despite monthly bevacizumab injections. Time for resolution of hemorrhage under the fovea was recorded. VA results were analyzed at 3 and 6 month, at 1 year, and at final follow-up. Snellen acuity was converted into logMAR for statistical analysis. Increase or decrease of 0.2 log MAR was defined as visual improvement or decline. VA of counting fingers and hand motions were assigned logMAR value of 1.6 and 2.0, respectively.[32] A two-tailed paired student's t-test was applied for OCT values and hemorrhage size using Microsoft Excel™ (Microsoft Corporation, Redmond, WA). Wilcoxon Signed Rank test was applied to the visual acuity results using MegaStat plug-in for Excel (J.B.Orris, Butler University). P values less than 0.05 were considered to be statistically significant. At final follow-up, appearance of the macula (presence of any scar, RPE tear, or geographic atrophy) was recorded.

Results

Baseline characteristics

Fourteen eyes of 14 patients (M = 2, F = 12), mean age ± SD 80.1 ± 9.54, range 60-91 years were included in the study [Table 1]. Eight had previously been treated for N-AMD, six with anti-VEGF monotherapy alone, one with thermal laser for an extrafoveal CNVM, followed by three applications of photodynamic therapy for subfoveal recurrence followed by bevacizumab injections and 1 with foveal thermal laser. VA prior to the onset of SMH ranged from 20/20 to 20/400 (Median 20/40). Those with a history of bevacizumab injections (n = 7) had received an average of 7.4 ± 3.8 (median 8, range 2-12) injections prior to the onset of SMH. Of these, five had received their last injection less than 20 weeks prior to the onset of SMH (range 4-104, median 20).

Table 1

Baseline characteristics

Patients presented after a median of 4 (range 1-7) days after the onset of acute central scotoma or defective vision. Ten were on systemic anticoagulation, seven on aspirin alone, two on aspirin and an antiplatelet agent, and one on warfarin. Reason for anticoagulant use included ischemic heart disease in 6 patients and a history of deep vein thrombosis in 2 patients. Two patients who were taking over-the-counter aspirin did not have any past history of systemic thrombo-embolic event. Eleven eyes were pseudophakic and 3 phakic eyes remained phakic throughout the period of the study.

Lesion and treatment details

Mean lesion size was 27.9 ± 24.5 mm2 (range 5.47-100, median 15) and mean SMH size was 25.1 ± 24.3 mm2 (range 4.22-90, median 13.1) [Table 2]. Hemorrhage comprised a mean of 87.7 ± 6.3% (median 90.0%, range 77-98%) of the lesion.

Table 2

Treatment details

At 3 months, the mean size of the hemorrhage decreased to 10.3 ± 11.8 mm2 (median 6.08, range 0-36, P = 0.0025). At 6 months, the mean size of the hemorrhage decreased further to 4.3 ± 9 mm2 (median 0.37, range 0-31), which was statistically significant as compared to baseline (P = 0.0037) and 3 months (P = 0.019). Subfoveal hemorrhage resolved in all eyes in a mean of 4.8 ± 1.6 (range 2-8, median 5) months. There was no statistically significant difference in time to resolution between predominantly subretinal hemorrhage and predominantly subRPE hemorrhages [Figs. 1a and b, 2, 3a–c, 4a–c, 5, 6a–c].

Figure 1

(a) Patient 6. Large submacular hemorrhage of 4 days duration. (b) IVFA at presentation showing blocked fluorescence and poorly defined leakage indicating a CNVM

Figure 2

Patient 6 at 1 month after presentation showing significant decrease in thickness of blood

Figure 3

(a) Patient 6 at 7 months follow-up. Note resolution of subfoveal hemorrhage. (b) IVFA of Patient 6 at 7 months showing a large occult subfoveal choroidal neovascular membrane. (c) OCT of Patient 6 at 7 months showing a dry macula

Figure 4

(a) Patient 5. Color fundus photograph at presentation. (b) Patient 5. IVFA at presentation showing blocked fluorescence and poorly defined leakage. (c) Patient 5. OCT at presentation

Figure 5

Patient 5. Two months after presentation showing marked decrease in hemorrhage

Figure 6

(a) Patient 5. Color fundus photograph at 4 months follow-up showing resolution of submacular hemorrhage. (b) Patient 5. IVFA at 4 months showing an occult subfoveal CNVM. (c) Patient 5. OCT at 4 months showing persistent PED, but no intra or subretinal fluid

During the first 6 months, the patients received a mean of 5.4 (median 5, range 5-6) injections. Those with ≥12 months follow-up (n = 10) received a mean of 9.6 (median 9, range 9-12) injections during the first year. Over a mean follow-up of 18.4 ± 11 (range 7-50) months, patients received a mean of 11.4 ± 4.4 (range 5-20) injections.

VA results

Presenting VA ranged from 20/60 to HM (mean 20/400, median 20/200) [Table 3]. At 3 months, visual acuity ranged from 20/30 to counting fingers (median logMAR 1.15; Snellen equivalent 20/280]). Compared to baseline, change in VA at 3 months was not statistically significant (P = 0.104).

Table 3

Visual acuity results

At 6 months (n = 14), VA ranged from 20/25-20/400 with a mean and median of 20/100. Mean VA gain at 6 months compared to baseline was −0.54 ± 0.57 logMAR (range: −1.5 to 1, P = 0.0037). Eleven patients experienced VA gain of ≥0.2 logMAR and 7 gained ≥0.6 logMAR with 5 eyes having vision ≥20/50. Patient 11, who presented with visual acuity of 20/60, remained at the 20/60 level at 6 months despite resolution of subfoveal hemorrhage after 4 months. Patient 12, who had previously received foveal thermal laser, presented with a visual acuity of 20/400 that remained at that level throughout follow-up despite complete clearance of SMH after 3 months. Patient 2 suffered visual decline from 20/80 at presentation to 20/400 6 weeks later due to an RPE tear. She remained at 20/400 level throughout the duration of follow-up of 50 months.

At 1 year (n = 9), VA ranged from 20/30-20/400 with a mean and median of 20/100 and 20/50, respectively. All 6 eyes that gained ≥0.2 logMAR VA at 6 months maintained that gain at 1 year. Mean change in VA from baseline was −0.43 ± 0.6 logMAR (median −0.60, range −1.42 to 1). There was no significant difference between the VA at 6 months and at 1 year (P = 0.38).

At final follow-up, (mean 18.4, range 7-50 months), VA ranged from 20/30-20/400, mean 20/100, median 20/60 for the 14 eyes. Eleven achieved a VA gain of ≥0.2 logMAR from baseline. The mean change in VA from baseline in logMAR was −0.58 ± 0.57 (median − 0.6, range −1.6 to ± 1, P = 0.0022). No statistical significance was found between the VA at 6 months and at final follow-up (P = 0.23).

SDOCT results

OCT results were available at presentation in 8 patients and showed a mean foveal thickness of 506 ± 218 μ (range 371-978, median 405). Mean thickness of hemorrhage was 322 μ (range 127-637, median 301). Five had predominantly subretinal hemorrhage, while 3 had sub-RPE hemorrhage. OCT results at 6 months showed a mean foveal thickness of 243.5 ± 41 μ (range 174-330, median 231, P = 0.023 as compared to baseline). At 1 year, the mean foveal thickness was 256.5 ± 37 μ (range 206-313, median 263, P = 0.89 as compared to foveal thickness at 6 months). At final follow-up, mean foveal thickness was 246.8 ± 43 μ (range 187-329, median 233, P = 0.81 as compared to foveal thickness at 6 months). SDOCT showed a dry macula in all eyes at final follow-up.

Fundoscopic and angiographic findings

FA after resolution of SMH showed an occult CNVM in 10 eyes, peripapillary classic CNVM in 2 eyes, regressed classic CNVM in 1 eye, and an RPE tear in 1 eye [Table 2]. Thin subretinal scar was noted in 2 eyes, laser scar in 1 eye, and no scar in the remaining 10 eyes.

No ocular or systemic adverse effects secondary to the use of the anti-VEGF medications were noted during the study.

Discussion

Our study showed excellent anatomical and visual results with anti-VEGF monotherapy alone in patients with thick SMH secondary to N-AMD who presented in ≤7 days. Visual gain was usually observed between 3-6 months. At 6 months, 11/14 (79%) experienced VA gain equivalent of ≥2 lines and 7/14 (50%) gained equivalent of ≥6 lines on an ETDRS chart. Of the remaining 3 patients, 1 presented with good VA (20/60), and 1 had limited potential for improvement due to prior foveal thermal laser. Only 1 patient suffered visual decline due to an RPE tear. Visual gain was accompanied by resolution of subfoveal hemorrhage in all eyes in a mean of 4.8 months with monthly anti-VEGF injections. Anatomical and visual improvement was maintained for 1 year and beyond with ongoing anti-VEGF injections. Although numbers were small, there was no apparent difference in outcome between subRPE and subretinal hemorrhages.

Stifter et al. demonstrated that, in 21 eyes with hemorrhagic N-AMD treated with intravitreal bevacizumab, 48% had improvement in VA by >1 letter, 9.5% demonstrated no change, while 43% lost <3 lines at 4 months, although median VA remained unchanged.[24] However, only 9 (43%) eyes had subfoveal hemorrhage, thickness of hemorrhage was not determined, 7 (33%) had symptoms for >30 days, presenting VA was relatively better (mean 20/80 versus 20/400 in our study), patients received fewer injections (median 3 compared to 9 in our study) over a 1-year period, visual gain was modest, and only 25% gaining approximately 2 lines of vision compared to 67% in our study at 1 year.

McKibbin et al. reported outcomes with intravitreal ranibizumab injections in 11 eyes with SMH secondary to N-AMD.[26] At 6 months, 3 (28%) gained >15 ETDRS letters, 4 (36%) gained 1-14 letters, and 4 (36%) lost 1-14 letters. Mean VA gain was 7.6 letters. All eyes had subfoveal hemorrhage, but three had < 50% hemorrhagic component. Patients were treated with 3 monthly loading injections, followed by PRN dosing; 7 eyes followed for 1 year received a mean of 6 injections. Mean duration of SMH was 6 (range 1-16) weeks. Patients were included in the study regardless of the size or thickness of hemorrhage (range: 0.5-45.4 mm2). They found no difference in outcome for duration ≤ 4 weeks compared to >4 weeks, but duration of symptoms exceeded 7 days in all eyes.

Sacu et al. found that eyes treated with pneumatic displacement with intravitreal tPA and SF6 gas and anti-VEGF had superior outcomes as compared to eyes treated with anti-VEGF agents alone.[25] However, the groups were not comparable. Most notably, the mean duration of symptoms in tpa/gas/anti-VEGF group was 7 days as compared to 12.9 days in the anti-VEGF group. All eyes treated with anti-VEGF monotherapy developed subretinal fibrosis. Pneumatic displacement has potential advantage of faster visual rehabilitation and shorter exposure time of the RPE and photoreceptors to toxic degradation products of blood clot, but it may result in shearing injury to photoreceptors, may not be effective in subRPE hemorrhages, has lower success rate for hemorrhages larger than 5 mm diameter, and has potential of displacing the blood into the fovea if significant amount of blood is located superior to fovea.[21]

Combination of pneumatic displacement with an anti-VEGF agent has been evaluated in a few small case series.[131415] Although this approach seems logical, it is currently not known whether it is superior to anti-VEGF injections alone.[21] A recent publication found no evidence that addition of gas injection to displace SMH provided any benefit over anti-VEGF injections alone.[27]

Chang et al. reported a prospective study of 7 eyes treated with ranibizumab for predominantly hemorrhagic N-AMD lesions.[28] Patients were treated with 3 monthly injections followed by PRN dosing, receiving a median of 7 injections in 1 year. VA gain of >2 lines was noted in 43%, and the median VA improved from 20/320 to 20/250. However, duration and thickness of SMH were not specified, and 5/7 eyes had subfoveal fibrosis prior to the onset of SMH.

Although our patients had large and thick lesions and had poor presenting VA, we believe our study demonstrated outcomes better than any of the previously reported studies for three reasons. First, treatment was initiated within 1 week. Earlier treatment in our patients compared to previous studies might have played a significant beneficial role in improving outcome. Second, patients with pre-existing subfoveal fibrosis secondary to N-AMD were not included in this review. Chang showed superior visual outcome in patients without pre-existing subfoveal fibrosis.[28] Third, our patients received more intravitreal injections during the first and second year than in previously reported studies. We attempted to achieve a completely fluid-free macula guided by spectral domain OCT, similar to the criteria used in CATT.[30] Although we utilized treat and extend protocol, we did not extend treatment-free interval beyond 8 weeks. About half of our patients had previously received anti-VEGF injections for N-AMD and presented with acute decrease in central vision due to thick SMH after a median of 20 weeks from their last anti-VEGF injection, similar to that in Levine et al.,[33] While we are aware of potential side effects, expense, and inconvenience associated with multiple intravitreal injections, we were hesitant to extend the treatment-free interval “too much” and risk a re-bleed. We noted that mean injections of 11.4 over a mean follow-up of 18.4 months are not much different from the mean of 14.1 injections of bevacizumab over 2 years in the PRN arm of CATT.[30] In a retrospective study, Dadgostar et al. showed that patients who received more frequent injections had superior visual outcome.[34] Less frequent dosing might have contributed to worse visual results in other studies.

A major goal of therapy for neovascular AMD is prevention of disciform scar. Gehrs et al. showed that organization of blood clot is a major component of disciform scarring.[23] However, no disciform scar forms in patients where etiology of SMH is other than a CNVM.[2] Thus, vascular proliferation is needed for scar formation, and its inhibition by early anti-VEGF monotherapy may prevent scarring.

Our study has several limitations. Indocyanine green angiography (ICG) was not done. Therefore, patients with polypoidal choroidal vasculopathy might have been missed. Snellen, rather than standardized ETDRS VA, was used, leading to lack of precise VA measurements below 20/400. Definition of “thick” hemorrhage was based on clinical criteria and not on SDOCT, which was not available in 6 (43%) at baseline. Other limitations of our study include retrospective and noncomparative nature and a small sample size. However, it seems that disciform scar formation can be prevented and good anatomical and visual outcome can be accomplished in patients treated with anti-VEGF monotherapy within 1 week from the onset of SMH secondary to N-AMD.