Treatment Effects in Retinal Angiomatous Proliferation Imaged with OCT Angiography

Retinal angiomatous proliferation (RAP), also referred to as type 3 neovascularization, is a distinct variant of exudative age-related macular degeneration (AMD) characterized by intraretinal neovascularization (IRN) with connections to either the retinal vasculature, the choroidal vasculature, or both [1]. In more advanced stages, RAP is associated with the formation of a pigment epithelial detachment (PED) and retinal choroidal anastomosis (RCA). RAP represents approximately 15–30% of newly diagnosed patients with exudative AMD [2-4]. The prognosis of RAP is poor with typical rapid progression and, without therapy, ending in a disciform scar and atrophy [5].

Numerous treatment strategies for RAP have been proposed, but there is no consensus on which treatment is optimal for RAP lesions [6]. Monotherapy with intravitreal antivascular endothelial growth factor (anti-VEGF) injections shows favorable short-term results in several studies [7-12], but requires repeated administration and conflicting long-term outcomes have been reported [8, 13-16]. A combination treatment of intravitreal anti-VEGF or triamcinolone with photodynamic therapy (PDT) seems to lead to a rapid resolution of the consequences of RAP [17-22]. However, there is little data available on how the treatment modalities act on the neovascularization itself and whether all stages of RAP should be treated the same.

Along with funduscopy, spectral-domain optical coherence tomography (SD-OCT) [23] is currently the standard tool for monitoring the effect of treatment of exudative AMD, including RAP. SD-OCT scans provide highly detailed anatomical information on retinal changes in RAP lesions [24], but without an extension for angiography it lacks the ability to detect flow, delineate (small) vessels, or image a feeder vessel connecting the intraretinal neovascular complex to the retinal or choroidal vasculature. Active neovascularizations can be revealed by fluorescein angiography (FA) with or without indocyanine green angiography (ICG), but these imaging modalities are invasive and give limited information about the depth location of the intra- and transretinal blood flow. Therefore, they are unsuited for regular follow-up measurements in clinical practice.

OCT angiography (OCT-A) is a new extension to standard OCT and can pinpoint abnormal intra- and trans-retinal blood flow noninvasively and to monitor treatment effects on neovascularizations, including RAP [25-28]. Our group has developed a phase-resolved 1,040-nm swept-source OCT system with an OCT-A modality [29-32]. We have demonstrated that OCT-A can identify abnormal intra- and transretinal blood flow in a small case series of 12 RAP patients [32]. We found that abnormal blood flow in RAP was mostly confined to the intraretinal structures and that in one-third of patients an RCA had developed. We hypothesize that OCT-A is also a useful tool for monitoring treatment effects and for improving “treat and observe”-style management decisions. This study is aimed at exploring OCT-A as a treatment monitoring tool in patients treated for RAP. The secondary objective was to explore the differences between conventional angiographic imaging and OCT-A.

Twelve treatment-naïve patients diagnosed with an RAP lesion were included in this prospective case series, between March 2013 and September 2013. The study was approved by the local internal review board of the Rotterdam Eye Hospital and the Medical Ethics Committee of the Erasmus University Hospital (Rotterdam, The Netherlands). All patients provided written informed consent.

The patients were examined at baseline by slit lamp biomicroscopy, Snellen visual acuity, conventional SD-OCT, fundus photography, FA, and ICG. Inclusion criteria were age ≥65 years, no other active ocular diseases affecting the macula, and treatment-naïve RAP. RAP was diagnosed by the presence of a small intraretinal hemorrhage on fundus examination, a choroidal neovascularization seen on FA, and/or a hyperfluorescent mid- to end-phase hotspot on ICG. Multimodal imaging baseline characteristics of all patients have been previously published by Amarakoon et al. [32]. We characterized the visualization of the retinal feeder vessel(s) at baseline by FA and OCT-A as good, fair, or poor. The visualization was graded as “good” if a feeder vessel was visualized connecting to the RAP lesion, as “fair” if only a faint or interrupted feeder vessel could be recognized, and as “poor” if none of the surrounding retinal vessels seemed to connect to the lesion.

An experimental optical frequency domain imaging system with a phase-resolved OCT-A modality was used to visualize blood flow at baseline and after treatment. The instrument uses a swept-source laser (Axsun Technologies Inc., Billerica, MA, USA) with a central wavelength of 1,040 nm operating at a 100-kHz A-scan rate. We obtained three-dimensional volume scans consisting of 300 single backstitched B-scans with 2,000 A-scans/B-scan over a retinal square area of 3 × 3 mm and with an acquisition time of 6 s per volume. The OCT-A system can detect flow velocities of 0.7 mm/s and higher, which is proven to be sufficient to image the retinal capillaries [30]. Elaborate technical details of this OCT-A instrument have been described previously [29, 30, 32].

OCT-A measurements were processed to produce OCT-A en face images (column 1 in each presented figure) and cross-sectional OCT-A tomograms (columns 2 and 3 in each presented figure). The OCT-A en face images display the phase differences (in white) detected between the vitreoretinal interface and retinal pigment epithelium (RPE). The location of the OCT-A is indicated with a dashed square on FA images. B-scans with significant eye motion artifacts were manually removed in the OCT-A en face images to facilitate interpretation and comparison with follow-up measurements, but some discontinuities in the visualized flow due to eye motion artifacts remained. In the OCT-A tomograms, the inter-B-scan phase differences were overlaid in red on the gray scale structural B-scans. The location of the superimposed OCT-A tomograms is indicated with red dashed lines in the OCT-A en face images. Displayed phase differences are predominantly caused by blood flow, but can also be due to noise, flow shadow artifacts, or eye motion artifacts. Flow shadow artifacts (also referred to as projection artifacts) [33] are caused by blood flow signal in large vessels in the inner retina, which produces phase differences in the signal in deeper layers.

The initial treatment schedule of RAP was determined at the ophthalmologist’s discretion and consisted of a combination of PDT and 2 or 3 intravitreal injections with bevacizumab or a combination of PDT and an intravitreal injection with triamcinolone. The laser light activation protocol used a wavelength of 689 nm, spot size range of 1.2–2.7 mm, with an intensity of 600 mW/cm² and was applied for 83 s. The order of treatment steps and the planning of OCT-A measurements were mainly determined by the hospital’s and the patient’s logistic opportunities. The follow-up period with OCT-A lasted until the first check-up by the ophthalmologist. The presence of abnormal blood flow on OCT-A after treatment was qualitatively categorized as increased, unchanged, decreased, or resolved by visual inspection of the whole volume scan.

Twelve RAP patients were included in this study with a median age of 79 years (range 65–90). Baseline characteristics as well as a comparison of baseline OCT-A with conventional images have been reported previously [32]. All 12 patients were imaged with OCT-A at baseline. Patients 1 and 6 were excluded from follow-up measurements because of severe eye movements on the baseline OCT-A scans. Patient 2 did not participate in follow-up treatment and OCT-A measurements because of hospitalization due to other health problems. In the other 9 patients, OCT-A images of sufficient quality were obtained both at baseline and after the initial treatment steps. The median follow-up period during this study was 10 weeks (range 5–19 weeks). A detailed timeline of OCT-A and treatment is indicated in the top right corner of each figure. Patients 7 and 10 were not treated with PDT because of general health issues not allowing them to come in for treatment. VA at baseline and after the initial treatment scheme are presented in Table 1. Median VA changed from 20/50 (range 20/650–20/22) Snellen at baseline to 20/67 (range 20/650–20/20) after treatment.

Patients 3 and 4 (Fig. 1; online suppl. Fig. 1; see www.karger.com/doi/10.1159/000491798 for all online suppl. material) were diagnosed with RAP based on conventional imaging but classified as choroidal neovascularization based on OCT-A [32]. The abnormal subretinal flow seen in patient 3 responded well to the combination of bevacizumab and PDT (see online suppl. Fig. 1). In patient 4 (Fig. 1) the abnormal sub-RPE flow did not respond to either intravitreal bevacizumab or PDT, while an increase of subretinal fluid was noted at week 1 which was most likely a side effect of the PDT (Fig. 1, row 3).

Patient 5 (see online suppl. Fig. 2) was firstly treated with bevacizumab after which OCT-A was performed, capturing the effect of only bevacizumab. The RAP lesion initially responded poorly to the bevacizumab injection, but a supplementary PDT resolved the lesion (see online suppl. Fig. 2, row 3).

Patient 7 (Fig. 2) was only treated with bevacizumab during the study period. The baseline OCT-A revealed a clearly delineated RCA, which remained present even after two injections of bevacizumab (Fig. 2, row 3, column 2). However, the subretinal neovascular component had disappeared after treatment (Fig. 2, row 3, column 3) as well as most of the intra- and subretinal fluid.

In patient 8 (Fig. 3) resolution of the abnormal blood flow at the site of the former RAP lesion was observed after PDT alone (Fig. 3, row 3). After intravitreal injection with triamcinolone and additionally bevacizumab, the subretinal fluid disappeared as well.

In patient 9, the abnormal blood flow as detected with OCT-A resolved after a combination treatment of bevacizumab and PDT (see online suppl. Fig. 3). In patient 10 the sub-RPE component of the RAP disappeared after one injection of bevacizumab (see online suppl. Fig. 4).

Patient 11 (Fig. 4) was first treated with bevacizumab and 1 week after the injection OCT-A indicated that the RAP lesion was still present, but a reduction in PED height was seen (Fig. 4, row 3). The combination of bevacizumab with PDT resolved the RAP lesion and the sub-RPE fluid disappeared. (Fig. 4, row 4).

Patient 12 (Fig. 5) was imaged with OCT-A after a combination of intravitreal bevacizumab and PDT. The diameter of the applied PDT laser was 2.7 mm and was centered at the hotspot seen on ICG at baseline (Fig. 5, row 1, column 3, red dashed arrow). The abnormal blood flow on OCT-A had resolved after this combination treatment (Fig. 5, row 3). However, local areas of nonperfusion in the choroid were detected at week 6 (Fig. 5, row 3) in the region where the PDT was applied. Because the OCT-A volume scan covers an area of 3 × 3 mm, the major part of the retina and choroid in the scanned area was affected by PDT laser. The reflectivity of the choriocapillaris and choroid was normal in the whole volume scan, indicating that the acquisition of this volume scan was of sufficient quality. A normal density of flow in the choriocapillaris was detected at the edges of the volume scan (see online suppl. Video 1). Our findings were confirmed by a consecutive OCT-A scan acquired at the same visit at week 6 (see online suppl. Video 2). A partial recovery of choroidal perfusion was seen at week 19 (Fig. 5, row 4).

In 7 out of 7 patients with intraretinal fluid before treatment, the intraretinal fluid had disappeared at the superimposed cross-sectional structural OCT of the last OCT-A (see Table 1). In 4 out of 5 patients with subretinal fluid, the subretinal fluid was resolved at the structural OCT. In 1 out of 3 patients with a PED, the PED had disappeared (see online suppl. Fig. 3), although the size of the other 2 PEDs did decline after treatment (Fig. 4; online suppl. Fig. 4). Although the decrease of abnormal blood flow in general corresponded well to the decrease of intra- and subretinal fluid, the timing was not always similar. The OCT-A of patient 7 (Fig. 2) showed the presence of abnormal blood flow after treatment, while the intra- and subretinal fluid had already disappeared. On the other hand, the OCT-A images of patient 8 (Fig. 3, row 4) and patient 11 (Fig. 4, row 3) did not show any suspected blood flow after treatment, but the subretinal or sub-RPE fluid persisted.

The retinal vasculature detected by OCT-A en face images showed a good correspondence to the vasculature revealed by FA (Fig. 1-5) [32]. However, the retinal capillaries are detected in more detail with OCT-A than with early FA in most patients and seem to suffer less from media opacities like cataract (Fig. 3-5; online suppl. Fig. 4). A good or fair visualization of the retinal feeder vessel was seen in 4 out of 9 patients on FA images and in 9 out of 9 patients on OCT-A enface images (see Table 1). In patients 7 and 9 (Fig. 2; online suppl. Fig. 3), an intraretinal hemorrhage obscured the visualization of the RCA on FA, while the OCT-A signal was not affected by the presence of the hemorrhage. The OCT-A cross-sections were able to display the depth location of the RAP lesions, which is not possible with FA or ICG. The hypercyanescent hotspots seen on ICG showed an accurate correspondence to the location of abnormal blood flow on OCT-A in 8 out of 9 patients. A hotspot on ICG was not seen in patient 7 (Fig. 2), where the OCT-A showed transretinal blood flow representing an RCA.

In this explorative case series OCT-A images before and after treatment were obtained in 9 RAP patients. At the end of the follow-up period and after various treatment sequences, the RAP lesion was resolved in 7 patients (Fig. 1, 3, 4; online suppl. Fig. 1–4). In 1 patient the OCT-A depicted only decreased flow in the lesion (see Fig. 2) and 1 patient showed unchanged abnormal blood flow (see Fig. 1). Monotherapy with intravitreal bevacizumab injections resolved the lesion in 1 out of 2 patients. Combined therapy of bevacizumab with PDT showed resolution in 6 out of 7 patients (Fig. 1, 3-5; online suppl. Fig. 1–3).

This study demonstrates the potential of OCT-A to visualize treatment effects on the angiographic features of RAP. The repeated use of FA to determine the activity of the neovascular lesion is restricted by its invasiveness as well as acquisition time. Therefore, in clinical practice OCT has replaced FA as a standard monitoring tool for exudative AMD treatment, withholding the ophthalmologist information on the blood flow in the neovascular network. We showed that retinal capillaries and feeder vessels are detected in more detail with OCT-A than with early FA in most of the patients of this small case series (see Table 1) and that OCT-A reveals the depth localization of RAP lesions in the cross-sectional images. We demonstrated that OCT-A can monitor the effect of treatment on abnormal blood flow in RAP without the use of an invasive dye, on en face and cross-sectional representations, and with a short acquisition time. This broadens the possibilities for the ophthalmologist to evaluate the effect of treatment on retinal and choroidal neovascularizations.

OCT-A might also support the search for an optimal treatment strategy for RAP. Anti-VEGF monotherapy yields promising results for early-stage RAP [7-12], but long-term outcome shows the necessity for repeated injections [8, 13-16]. Combination therapy seems to result in better visual outcome [17-22] and a lower retreatment rate [34, 35]. Tsai et al. [6] reported that the results of anti-VEGF monotherapy for RAP are encouraging, but that concerns remain regarding the development of geographic atrophy and long-term visual outcome. The presented OCT-A data in this study seems to support the hypothesis that a combination of anti-VEGF treatment with PDT leads to rapid resolution of RAP. On the other hand, OCT-A revealed that patient 4 had a small type 1 choroidal neovascularization which did not respond to the combination therapy. If OCT-A had been used in the diagnostic workup of this patient, PDT probably would have been omitted.

Although PDT appears to be efficacious, it has been reported that this treatment may cause thrombus formation in the choroid, leading to a reduction in choroidal perfusion [36]. Patient 12 (Fig. 5) demonstrates that OCT-A can visualize choroidal nonperfusion (Fig. 5, row 3; online suppl. Videos 1, 2), which was most likely caused by PDT treatment. The large spot size of 2.7 mm used in this patient might have increased the risk of this collateral damage. Reperfusion was detected after 19 weeks (Fig. 5, row 4). The range of blood flow velocity in the choriocapillaris was previously reported to be 0.3–3.6 mm/s, while the lower limit of detectable flow velocity in our phase-resolved OCT-A system is 0.7 mm/s (see Braaf et al. [30], Fig. 4). However, the structural OCT shows normal intensity (Fig. 5, row 3, columns 2, 3) and we could confirm our findings in two different volume scans (online suppl. Videos 1, 2). Therefore, we believe that the risk of not having detected persistent flow in this region of the choriocapillaris is very small. In future studies, OCT-A might be used to further elucidate the occurrence rate of this severe adverse event. The OCT-A system used in this study detected choroidal flow which might be explained by sufficient signal penetration depth of the 1,040-nm swept-source laser in our study [37].

The presence of slow abnormal blood flow in small capillaries can be visualized before and after treatment using OCT-A, but fluid leakage is associated with far lower velocities and cannot be detected with OCT-A in the currently available systems. It would be interesting to know whether the absence of abnormal blood flow corresponds to the inactivity of the RAP lesion. In this small case series, we found that the absorption of intra- and subretinal fluid on structural OCT mostly corresponds to the disappearance of abnormal blood flow (see Table 1). In some cases, however, the disappearance of abnormal blood flow on OCT-A preceded (Fig. 2) or followed (Fig. 3, 4) the absorption of intra- and subretinal fluid on structural OCT. This demonstrates that OCT-A provides additional information compared to standard OCT, which could lead to improved monitoring of treatment outcome and thereby optimizing (re-)treatment decisions. It would be interesting to follow up patients with OCT-A after discontinuation of treatment to evaluate the role of OCT-A in detecting early reactivations. Although the amount of detail captured by FA seems to be lower than that by OCT-A, FA and ICG could play an important role in validating OCT-A changes after treatment by revealing fluid leakage. However, in daily clinical practice, the amount of residual leakage is monitored by structural OCT alone. Therefore, the additional value of OCT-A in retreatment decisions must be compared primarily to structural OCT.

Strengths of this study include the combination of en face images and high-resolution cross-sectional images which enabled us to pinpoint the depth of the lesions, and the penetration depth of the 1,040-nm swept-source laser which enabled visualization of blood flow in the choroid. Limitations of this explorative study are the small sample size, the short follow-up period, the multiple treatment strategies, the small field of view, and the amount of eye motions. A longer follow-up period would be needed to evaluate long-term effects of anti-VEGF monotherapy and to establish the recurrence rate of neovascularization after successful closure of RAP. A larger field of view would be beneficial to monitor several neovascularizations simultaneously or to screen for new neovascularizations when treatment appears to be ineffective. Real-time eye motion correction can result in higher-quality flow information [38] and combined with structural information this allows to search for more subtle treatment effects.

In conclusion, OCT-A detected a decrease in abnormal blood flow after the initial treatment sequence of RAP in 8 out of 9 patients. We have shown that RAP short-term anti-VEGF treatment combined with PDT rapidly results in closure of RAP in this small study. Because of its noninvasiveness, short acquisition time, and simultaneous imaging of structural changes, OCT-A appears to be a promising treatment monitoring tool for macular vascular pathologies such as RAP. A larger trial is warranted using OCT-A to examine treatment effects of combination therapy and monotherapy separately. Interestingly, such a trial would be more feasible at the present time, as the capabilities of our experimental prototype OCT-A are now operative in several commercially available OCT-As.

We thank José Martinez, MD (Rotterdam Eye Hospital, The Netherlands) for his assistance in evaluating fluorescein and indocyanine green angiograms in all patients. Fluorescein angiography and indocyanine green angiography images are reprinted from Amarakoon et al. [32], with permission from Elsevier.

Jan H. de Jong, Boy Braaf, Sankha Amarakoon, Maximilian Gräfe, Suzanne Yzer, Koenraad A. Vermeer, and Tom Missotten have no financial disclosures.

Johannes F. de Boer received the following research funding: government and government sponsored foundations – Vici-ZonMW (grant 918.10.628), STW (grant No. 12822, 13936), ZonMW (grant No. 91212061), and LaserLaB Europe (grant agreement No. 654148); foundations – Kika retinablastoom and ISAO; commercial company – Heidelberg Engineering; personal fees (income from lecture) – Heidelberg Engineering; intellectual property rights, royalties, patents, licenses – Massachusetts General Hospital.

Mirjam E.J. van Velthoven has commercial relationships with Novartis Netherlands (lecture fee), Allergan Europe (travel expenses), Bayer Netherlands (lecture fee), and AbbVie Netherlands (lecture fee).

Funding for this study was received from the following: Stichting Combined Ophthalmic Research Rotterdam (project No. 2.0.0), Rotterdam, The Netherlands; MaculaFonds (project No. 2012.8), Utrecht, The Netherlands; and Stichting Life Sciences Health TKI (project No. LSHM16001), Heidelberg Engineering, Heidelberg, Germany.

Meeting presentations: NOG Jaarcongres, Groningen, The Netherlands, April 6, 2016; ARVO Annual Meeting, Seattle, WA, USA, May 1, 2016.