Methods

In this retrospective interventional case series, we reviewed the medical records of all patients with neovascular AMD complicated by SMH with a maximum history of 2 weeks treated at the Department of Ophthalmology of the University Medical Center Schleswig-Holstein, Kiel, Germany between January 2008 and January 2009. Inclusion criteria were neovascular AMD complicated by SMH involving the fovea, a minimum age of 18 years and a maximum history of symptoms of 2 weeks. Exclusion criteria were other aetiologies of SMH, massive SMH extending beyond the equator and pre-existing macular scar.

Preoperative evaluation included standard ophthalmological examination, best-corrected Snellen visual acuity (BCVA) and fundus colour photograph. Surgery was performed within 24 to 36 h from diagnosis. Anticoagulation therapy was discontinued at admission until 5–7 days after surgery. The fovea was defined as the circular area with a diameter of 1.5 mm or approximately one disc diameter in the centre of the macula. Outcome measures were (1) complete displacement of submacular haemorrhage from the fovea, (2) change in BCVA and (3) signs of retinal or RPE toxicity and complications. In all patients, the type of surgery and alternative options were explained in detail. Informed consent was obtained before surgery from all patients. Ethics committee approval was obtained.

Surgical technique

All patients underwent standard three-port pars-plana vitrectomy (ppV) with induction of posterior hyaloid detachment if not already present, subretinal injection of 10–20 μg of rtPA (Actilyse, Boehringer Ingelheim, Germany) dissolved in 0.05–0.1 ml of BSS followed by subretinal injection of 1.25 mg of bevacizumab (Avastin, Roche, Basel, Switzerland) dissolved in 0.05 ml of BSS through a 41-gauge subretinal flexible cannula (DORC, Zuidland, The Netherlands) and fluid–gas exchange with 20% SF6 gas to a complete intravitreal fill. Phakic patients underwent concomitant standard small-incision cataract surgery. Patients were instructed to keep a prone position for at least 1 day postoperatively. All patients were operated by the same surgeon (JH). Bevacizumab (1.25 mg) was administered intravitreally 4 weeks and 8 weeks postoperatively. No other AMD treatments were applied.

Follow-up

Short-term follow-up 4 weeks postop and long-term follow-up 12 weeks postop included BCVA, fundus colour photograph and fluorescein angiography.

Statistical methods

In the statistical analysis of the outcome measures, the type I error (alpha) was set at 0.05. Statistical significance of differences before and after surgery were calculated using the Wilcoxon test. The results were analysed using SPSS 15.0 software (SPSS, Chicago) and Statistica 7 software (StatSoft, Tulsa, Oklahoma).

Results

A total number of 17 patients with neovascular AMD complicated by SMH were treated during the observation period (January 2008 to January 2009). Five patients were excluded from the analysis because of massive SMH extending beyond the equator. Of the excluded five patients, two patients were treated with ppV with CNV extraction, and three patients were treated with ppV, CNV extraction and autologous RPE-choroid sheet transplantation as previously described.23 Hence, a total number of 12 patients, nine female and three male, were included in the analysis. The mean age was 81.5 years (SD = 5.4, range 71 to 90). The mean history of symptoms was 4.8 days (SD = 3.8, range 1 to 14). Six patients had anticoagulation therapy. The mean maximal diameter of SMH was 4.3 disc diameters (SD = 3.2, range 1.5 to 10). Patient 9 presented with vitreous haemorrhage. Patient 2 had been treated with one intravitreal injection of bevacizumab several months prior to the occurrence of SMH. Patients 7 and 9 had been treated with two intravitreal injections of bevacizumab and ranibizumab several months prior to the occurrence of SMH. All other patients had not received any treatment for AMD prior to the occurrence of SMH. At presentation, the degree of lens opacity was described as mild in all six phakic patients. Patient 9 refused concomitant cataract surgery (table 1). Intraoperatively, we observed minimal or no reflux of the subretinally injected solutions through the small retinotomy created by the 41-gauge cannula. The subretinally injected volume produced a dome-shaped retinal elevation which remained unchanged intraoperatively after fluid–air exchange.

Displacement of haemorrhage

Complete displacement of SMH from the fovea was achieved in nine of 12 patients (table, fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Fundus photographs and fluorescein angiography. Patient 3: fundus photograph preop (A) showing submacular haemorrhage (SMH). Fundus photograph 4 weeks postop (B) showing complete displacement of SMH from the fovea. Fluorescein angiography 4 weeks postop (C, early phase; D, late phase) showing occult choroidal neovascularisation (CNV). Patient 12: fundus photograph preop (E) showing large SMH. Fundus photograph 4 weeks postop (F) showing complete displacement of SMH from the fovea, scarred CNV and drusen. Fluorescein angiography 4 weeks postop (G, early phase; H, late phase) showing scarred occult CNV. Patient 5: fundus photograph preop (I) showing SMH and a small area of geographic atrophy. Fundus photograph 4 weeks postop (J) showing complete displacement of SMH from the fovea, a circumscribed semicircle shaped rip of the retinal pigment epithelium (RPE), and the small area of geographic atrophy also visible preop (I). Fluorescein angiography 4 weeks postop (K, early phase; L, late phase) showing the circumscribed semicircle shaped rip of the RPE and the small area of geographic atrophy. Late-phase fluorescein angiography showing small occult CNV (L). Patient 10: fundus photographs preop (M) and 4 weeks postop (N) showing incomplete displacement of SMH from the fovea. Fluorescein angiography 4 weeks postop showing a secondary macular hole and occult CNV (O, early phase; P, late phase).

BCVA

The mean BCVA improved significantly from preop logMAR 1.9 (range 3.0 to 0.7) to logMAR 1.2 (range 3.0 to 0.3) at 4 weeks postop (p = 0.01) and to logMAR 0.9 (range 1.6 to 0.2) at 12 weeks postop (p = 0.006).

The mean improvement of BCVA 4 weeks postop as compared with preop was logMAR 0.7 (range −0.2 to 2.3). The mean improvement of BCVA 12 weeks postop as compared with preop was logMAR 0.96 (range −0.3 to 2.8).

As compared with preop at 4 weeks postop, BCVA had improved in eight patients, remained unchanged in three patients and worsened in one patient. As compared with preop at 12 weeks postop, BCVA had improved in 10 patients, remained unchanged in one patient and worsened in one patient (table, fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Scatter plot of best-corrected visual acuity (BCVA) (logMAR) preop, 4 weeks postop and 12 weeks postop. The difference in BCVA between preop and 4 weeks postop, as well as between preop and 12 weeks postop, was statistically significant (p = 0.01 and p = 0.006, respectively, Wilcoxon test).

Complications

Intraoperative

In patient 5, a circumscribed semicircular rip in the RPE was caused by the inadvertent injection of the rtPA and bevacizumab solution into the subRPE space. It seems likely that the SMH was located at least partially under the RPE as a haemorrhagic RPE detachment adjacent to a small area of geographic atrophy which is visible on the pre- and postoperative fundus photographs (fig 1). In patient 10, a secondary macular hole developed at the fovea during subretinal injection, apparently caused by the mechanical stress applied to the retinal tissue by the volume of the solution (fig 1).

Postoperative

In patients 7 and 8, SMH recurred. In patient 7, the recurrence of SMH 3 weeks postop was associated with dense vitreous haemorrhage. The patient underwent repeat ppV with CNV extraction and silicone oil tamponade. No further intravitreal bevacizumab injections were administered. In patient 8, SMH recurred 5 weeks postop. The patient underwent repeat ppV with subretinal coapplication of rtPA and bevacizumab, and fluid–gas exchange. No further intravitreal bevacizumab injections were administered until the follow-up visit at 12 weeks after the first operation when the occult CNV initially detected by fluorescein angiography 4 weeks after the first operation was inactive (table 1).

Discussion

The main findings of this study are (1) successful complete displacement of small and large SMH from the fovea in nine of 12 patients, (2) significant mean functional improvement of logMAR 0.96 at 12 weeks’ follow-up and (3) tolerability of subretinal coapplication of rtPA and bevacizumab suggested by the mean functional improvement and the findings of postoperative funduscopy and fluorescein angiography.

In five of six phakic patients, the improvement in BCVA may have been caused to some degree by concomitant cataract surgery. Systematic preoperative cataract grading was not performed, but the contribution of cataract surgery to the overall change in BCVA was probably small because the degree of lens opacity was described as mild in all phakic patients.

In neovascular AMD with SMH, the functional outcome largely depends on the extent of the underlying CNV, which could be a reason for the scatter of BCVA in this study (fig 2) as well as in other studies.11 17 18 19 24 25 Successful displacement of SMH in neovascular AMD and significant functional improvement has been reported after intravitreal injection of gas without rtPA,25 after intravitreal injection of rtPA and gas,10 11 12 after ppV with subretinal injection of rtPA and fluid–gas exchange17 18 19 and after intravitreal injection of rtPA, bevacizumab and gas.24 To date, there is no consensus regarding an optimal treatment of SMH or the key factors determining the outcome. Since studies evaluating different treatments of SMH have been either retrospective uncontrolled case series with a limited number of patients6 7 9 10 11 12 17 18 24 25 or non-randomised, retrospective comparative case series19 with a weak level of evidence, the results cannot be directly compared with the present case series. A comparison of the efficacy of subretinal versus intravitreal applications of rtPA and an anti-VEGF agent as a treatment of neovascular AMD with SMH will require a prospective randomised patient study. Furthermore, an adequate prospective study is needed to evaluate the outcome of small SMH in neovascular AMD treated with intravitreal injections of an anti-VEGF agent with or without ppV, rtPA and gas.

Because of the size of the rtPA molecule, it is unclear whether or not intravitreally injected rtPA penetrates the retina to reach a subretinal clot.13 14 15 16 The rtPA molecule exceeds the experimentally determined molecular exclusion limit of human retina.15 Indeed, in the experiments of Kamei et al, rtPA injected into the vitreous of rabbits failed to pass through the intact retina.13 On the other hand, molecules with similar molecular weight (eg, albumin) have been shown to penetrate the diseased retina.14 Heiduschka et al demonstrated that bevacizumab, which also exeeds the molecular exclusion limit of the retina, passed through the intact retina of cynomolgus monkeys.16 However, although bevacizumab traverses the retina, it probably does so only at a slow rate because of its size. While intravitreal application of bevacizumab may suffice to produce a therapeutic response, subretinal application ensures the delivery of the drug directly at the site of the CNV and may enhance its effect. Furthermore, we chose subretinal instead of intravitreal application because the pharmacokinetics of bevacizumab in a vitrectomised, gas-filled eye are unknown and because retinal damage from haemorrhage may alter permeability.

Fifty per cent of the patients in this study were on anticoagulation treatment. Although anticoagulation is probably a risk factor for the development of SMH, and anticoagulation was discontinued at admission and restarted 5–7 days after surgery, SMH recurred in only two patients. We also found a low rate of SMH recurrence in our previous patient study while the rate of patients treated with anticoagulation was lower (seven of 47 patients).19 The low rate of SMH recurrence in the present study inspite of many patients treated with anticoagulation may have been caused by early CNV regression induced by the simultaneous application of bevacizumab. However, although recurrence of SMH did not occur often, the procedure was associated with a risk of other complications. It is recommended that patients with RPE detachment be excluded in order to avoid RPE rip. Subretinal injection should be performed slowly with a maximal total volume of 0.1 ml in order to avoid rupture of retinal tissue at the fovea, the point of least mechanical resistance. To further reduce the risk of recurrence of SMH and vitreous haemorrhage, it may be advantageous to discontinue anticoagulation therapy until three doses of an anti-VEGF agent have been applied and the underlying CNV has regressed.

In conclusion, ppV with subretinal coapplication of rtPA and bevacizumab followed by fluid–gas exchange for neovascular AMD complicated by SMH is a promising and effective treatment. Some complications can be avoided in part by careful patient selection (eg, excluding patients with pre-existing RPE detachment). Weaknesses of the present study include its retrospective character, a small number of patients and the absence of a control group. A prospective randomised patient study is required to compare the efficacy and safety of subretinal versus intravitreal applications of rtPA and anti-VEGF agents in order to establish guidelines for the management of neovascular AMD with SMH.