To report the clinical implications of interval changes in choroidal neovascularization (CNV) size measured by indocyanine green (ICG) angiography in neovascular age-related macular degeneration (AMD) patients undergoing intravitreal bevacizumab therapy.
Retrospective, consecutive chart review.
The charts of neovascular AMD patients who underwent intravitreal bevacizumab therapy using a treat-and-extend dosing schedule were reviewed. ICG angiographic CNV surface areas were measured at baseline, 2 months, 6 months, and 12 months in each subject. The primary outcome was change in CNV size. Secondary outcomes included the correlation of change in CNV surface area with change in best-corrected visual acuity (BCVA), change in central macular thickness on optical coherence tomography (OCT), and the number of injections delivered over the 12-month study interval.
A total of 123 subjects were included in the analysis. The baseline CNV size was 1.9 mm 2 ± 2.5 mm 2 . CNV size was 1.66 mm 2 ± 2.11 mm 2 at 2 months, 1.60 mm 2 ± 2.23 mm 2 at 6 months, and 1.50 mm 2 ± 2.12 mm 2 at 12 months. The change in CNV size from baseline was not statistically significant at any of the follow-up intervals. A decrease in CNV size of 33% or more at 2 months was associated with a significant decrease in CNV size at 12 months ( P = .0096), complete resolution of CNV at 12 months ( P = .0013), and a decrease in the number of injections delivered over the study interval ( P = .0165). Complete resolution of CNV at 12 months occurred in 7.3% of subjects. Subjects that had complete resolution of CNV at 12 months were significantly more likely to gain 3 more lines of BCVA at the end of the study interval ( P = .0131). No significant correlation was found between CNV size and change in central macular thickness on OCT.
Our study suggests that change in CNV size on ICG angiography may help the clinician predict the clinical course of neovascular AMD subjects undergoing intravitreal bevacizumab therapy using a treat-and-extend dosing schedule.
The use of digital indocyanine green (ICG) angiography in an ophthalmic setting was first described by Yannuzzi and associates in 1992. Since this first description, routine clinical use of ICG angiography has steadily been on the rise internationally, particularly in Asia, South America, and most parts of Europe. Previous studies have reported that ICG angiography can improve the ability to detect choroidal neovascularization (CNV) in patients with neovascular age-related macular degeneration (AMD) compared to fluorescein angiography (FA) alone, particularly when occult CNV is present. In fact, ICG angiograms obtained using both the scanning laser ophthalmoscope and digital video-angiography enabled the visualization of 55% and 61% of CNV, respectively, in cases where the CNV could not be determined with FA. Detection of classic CNV on ICG angiography reportedly approaches 100% in patients with neovascular AMD.
The structure and permeability of CNV can vary considerably on ICG angiography, depending on the stage of development or regression. Several researchers have reported on the ICG angiographic features of CNV in neovascular AMD patients, and some have attempted to develop a classification system to correlate visual prognosis with ICG angiographic findings. Most of these studies classified ICG angiographic features of CNV on the basis of borders of hyperfluorescence and the presence of dye leakage, although some focused on the size of the lesion’s hyperfluorescence in the era prior to anti–vascular endothelial growth factor (VEGF) therapy. There is no universally agreed-upon CNV classification based on ICG angiographic findings at this time.
A treat-and-extend regimen for the treatment of neovascular AMD attempts to maintain an exudation-free macula while reducing the number of intravitreal injections. Several studies have supported the effectiveness of treat-and-extend protocols with intravitreal bevacizumab, and at the present time, a majority of North American vitreoretinal specialists favor a treat-and-extend approach over pro re nata (PRN) and monthly anti-VEGF dosing schedules for the treatment of neovascular AMD. In this study, the authors evaluate the 12-month effects of intravitreal bevacizumab therapy on ICG angiography–measured CNV size in neovascular AMD subjects. The study employs a measuring system based on CNV surface area measurements without the subjective determination of angiographic hyperfluorescence and leakage by the observer. Currently, no clinical data are available on ICG angiographic CNV surface area measurements in neovascular AMD subjects undergoing anti-VEGF therapy. It is unknown whether changes in CNV size during the course of anti-VEGF treatment might provide the clinician with valuable information to more effectively manage this patient population.
The Southwest Retina Specialists Institutional Review Board (IORG0007600/IRB00009122) approved this retrospective medical record review of neovascular AMD patients treated with intravitreal bevacizumab from January 2009 to June 2013 at a single private-practice institution with 3 attending vitreoretinal specialists. A total of 48.7% of subjects from the earlier years of this study were included in a report on the influence of baseline characteristics on the effectiveness of a treat-and-extend regimen with bevacizumab for neovascular AMD. All research components adhered to the tenets of the Declaration of Helsinki and were conducted in accordance with human research regulations and standards.
Subjects were included in the study if: (1) their age was ≥55 years; (2) they had a diagnosis of subfoveal neovascular AMD demonstrated by clinical examination, FA, spectral-domain optical coherence tomography (OCT), and ICG angiography; (3) their pretreatment Snellen best-corrected visual acuity (BCVA) was between 20/25 and 20/200; (4) they were treatment naïve for neovascular AMD at enrollment (no history of intravitreal anti-VEGF treatment, intravitreal steroid treatment, photodynamic therapy, or macular photocoagulation); (5) a 12-month follow-up period was fully documented with interval BCVA, examination details, OCT, FA, and ICG angiography, with ICG angiography having been performed at baseline and at 2 months (±1 month), 6 months (±2 months), and 12 months (±2 months) follow-up; and (6) a treat-and-extend regimen was adhered to as described below.
Subjects were excluded from the study if: (1) they had received photodynamic therapy, macular photocoagulation, or intravitreal injections other than intravitreal bevacizumab to the study eye during the study period; (2) they had media opacity (corneal scar, cataract, etc) that, in the opinion of the examiner, was significantly affecting BCVA or hindering macular examination and imaging acquisition; (3) they had undergone intraocular surgery for any indication during the study period or within 3 months prior to enrollment; (4) a visually significant retinal disease other than neovascular AMD (epiretinal membrane, diabetic retinopathy, macular hole, polypoidal choroidal vasculopathy, etc), active uveitis, or advanced/poorly controlled glaucoma was present at enrollment or developed during the study period; (5) an uncontrolled systemic medical condition was present (ie, hypertension, diabetes, malignancy with metastasis, etc); (6) the subject had a history of a thromboembolic event within 6 months prior to intravitreal bevacizumab initiation; (7) a previous pars plana vitrectomy or anterior vitrectomy was performed at any time to the study eye (before or during the study period); or (8) the CNV margins were indistinct and therefore unable to be sufficiently determined for measurement by ICG angiography (lack of a distinguishable CNV complex or “hot” spots without determinable margins).
All patients had baseline examinations that included BCVA, blood pressure measurement, slit-lamp examination of the anterior and posterior segments, FA, OCT, and ICG angiography. Patients received the initial intravitreal bevacizumab injection during the baseline evaluation. The follow-up examination intervals were determined according to a treat-and-extend protocol. The schedule consisted of an initial intravitreal bevacizumab loading dose of 3 monthly injections, followed by monthly intravitreal bevacizumab injections until the macula was dry (without intraretinal or subretinal fluid) on OCT, without macular hemorrhage on fundus examination, and without leakage of the CNV complex on FA. The interval between treatments was extended by fixed 2-week increments until a maximum inter-visit interval was reached. If fluid recurred on OCT, macular hemorrhage developed, the BCVA dropped by 2 or more lines, or leakage recurred on FA after treatment extension, the interval was reduced by 2 weeks. Patients were told to perform Amsler grid monitoring 2–3 times per week and instructed to contact the treating physician immediately if visual changes occurred. Patients underwent BCVA measurements, slit-lamp ophthalmoscopy, and OCT at each visit. FA was performed at each visit when treatment extension or reduction was considered, as well as at the 8-week, 24-week, and 52-week follow-up examinations. ICG angiography was routinely performed at the same time as when FA was performed. The established study length was 12 months from each subject’s baseline evaluation and initial treatment. Therefore according to the treat-and-extend protocol followed, subjects could receive a minimum of 7 intravitreal bevacizumab injections and a maximum of 12 intravitreal bevacizumab injections during the study interval.
The Heidelberg Spectralis (Heidelberg Engineering, Heidelberg, Germany) system was used to perform all OCT, FA, and ICG angiography procedures. OCT was used to determine the presence or absence of intraretinal and subretinal fluid and to assess central macular thickness. The central macular thickness was obtained from the center subfield of the macular thickness map and only well-centered scans were selected for analysis. FA was used to evaluate for CNV leakage. The size of the CNV on ICG angiography was determined by using the Heidelberg Spectralis surface area–measuring software. Several early to mid-frame (within the first 3 minutes) ICG angiography images were viewed and a focused, centered test image was selected for analysis. The “Inlay” function was then selected on the test image menu and the CNV margins were manually encircled, followed by surface area computation with the software. Side-by-side FA and OCT images of the ICG angiography test image were allowed to confirm the location of the CNV. Two masked vitreoretinal specialists (R.B.R. and A.V.A.) evaluated the ICG angiographies for all subjects independently. CNV measurement discordances were evaluated by a third observer (S.W.R.). If reconciliation was not established within 10% of the lesion size between the 2 masked observers, the subject was excluded from analysis by the third observer.
Statistical Analysis and Outcome Measures
The primary outcome measure of this study was change in CNV surface area on ICG angiography at 2 months, 6 months, and 12 months follow-up. Secondary outcome measures included the correlation of change in BCVA, change in central macular thickness on OCT, and the number of injections delivered over the 12-month study interval with change in CNV surface area on ICG angiography at 2 months, 6 months, and 12 months. CNV response to treatment was considered to be significant if the ICG angiographic–measured CNV surface area change was 33% or more from the designated follow-up interval. For statistical analysis and later discussion, subjects that received 7–8 injections during the study interval were considered “Prompt Responders,” subjects that received 9–10 injections were considered “Moderate Responders,” and subjects that received 11–12 injections were considered “Inadequate Responders.” Statistical calculations were performed using the JMP software package from the SAS Institute (Version 10; Cary, North Carolina, USA). Snellen BCVA was converted into a logarithm of the minimal angle of resolution (logMAR) for the statistical analysis. Numerical means were compared using 1-way analysis of variance (ANOVA), whereas nominal variables were compared using χ 2 and likelihood ratios where appropriate. A probability of <.05 was considered statistically significant.
There were 144 subjects that met study criteria. Of the 144 subjects, 123 subjects were included in the data analysis after CNV measurements were evaluated for discordances, thereby resulting in an 85.4% agreement between the 2 masked observers within 10% of the lesion size for the study. Of the 21 subjects excluded because of masked observer measurement discordances, 5 subjects had minimally classic leakage, 16 had occult leakage, and none had classic leakage on FA. The means and distributions of the study population characteristics were as follows: the baseline age was 77.5 (±8.1) years. There were 61.7% female subjects and 38.2% male. There were 47.9% phakic subjects and 52.0% pseudophakic subjects. A total of 50.4% of the subjects were on blood thinning medication (aspirin, clopidogrel, or warfarin) at baseline. At baseline, 26.0% of study subjects had classic/predominantly classic, 26.8% had minimally classic, and 47.1% had occult CNV on FA. The baseline BCVA was 0.47 (±0.26) logMAR (Snellen 20/59). The final BCVA at 12 months was 0.33 (±0.28) logMAR (Snellen 20/42). There were 28.4% of subjects that gained 3 or more lines of BCVA at 12 months. There were 5.7% of subjects that lost 3 or more lines of BCVA at 12 months. The baseline central macular thickness on OCT was 383.7 (±110.9) μm. The final central macular thickness on OCT at 12 months was 309.3 (±81.7) μm. The total number of injections delivered over the 12-month study interval was 9.3 (±1.6). According to our classification by injection response during the study interval described above, 33.3% of subjects were “Prompt Responders” (7–8 injections), 40.6% were “Moderate Responders” (9–10 injections), and 26.0% were “Inadequate Responders” (11–12 injections). The change in BCVA and central macular thickness on OCT significantly improved during the 12-month study interval ( P < .0001 for both variables).
Means and distributions of CNV surface area measurements on ICG angiography were as follows. The baseline CNV size was 1.9 (±2.5) mm 2 (median 1.0); 2-month follow-up CNV size was 1.66 (±2.11) mm 2 (median 0.96); 6-month follow-up CNV size was 1.60 (±2.23) mm 2 (median 0.88); and 12-month follow-up CNV size was 1.50 (±2.12) mm 2 (median 0.7). The change in CNV size from baseline to 2 months was −12.6%, from baseline to 6 months was −15.7%, and from baseline to 12 months was −21.0%. CNV size did not significantly change from baseline at any of the time intervals evaluated: 2 months ( P = .40), 6 months ( P = .31), and 12 months ( P = .16). A decrease of 33% or more in CNV size at 2 months follow-up occurred in 28.4% of subjects. An increase of 33% or more in CNV size at 2 months follow-up occurred in 5.7% of subjects. A decrease of 33% or more in CNV size at 12 months occurred in 45.5% of subjects. An increase of 33% or more in CNV size at 12 months occurred in 13.0% of subjects.
The complete resolution of CNV on ICG angiography at 12 months occurred in 7.3% of subjects. All 9 of the subjects that had complete CNV resolution on ICG angiography at 12 months were in the “Prompt Responder” group, which was found to be statistically significant when compared to the “Moderate Responder” and “Inadequate Responder” groups ( P < .0001). Out of the 9 subjects that had complete CNV resolution on ICG angiography at 12 months, 66.7% (6/9) had 3 or more lines of BCVA improvement, which was significantly better than the rest of the study subjects without CNV resolution ( P = .0131). Subjects with complete resolution of CNV on ICG angiography at 12 months were also more likely to demonstrate an overall gain in BCVA from baseline ( P = .0279), but were not more likely to have a significant percentage decrease in central macular thickness on OCT ( P = .76) compared to the rest of the study subjects without CNV resolution. In addition, subjects with a CNV surface area in the lower quartile of the study population (<0.5 mm 2 ) at baseline were more likely to have complete resolution of CNV on ICG angiography ( P = .0050) and require fewer total injections ( P = .0312) at the end of the study period compared to the subjects in the other quartiles. Subjects with a CNV surface area lower than the median of the study population (<1.0 mm 2 ) at baseline were more likely to have a gain of 3 lines or more of BCVA ( P = .0480) at the end of the study period. Figures 1 and 2 show examples of CNV measurement and response to intravitreal bevacizumab during the study interval.