Giuseppe Querques Department of Ophthalmology University of Paris Est Creteil Centre Hospitalier Intercommunal de Creteil 40 Avenue de Verdun 94000 Creteil France Tel: + 33 0 1 45 17 52 22 Fax: + 33 0 1 45 17 52 27 Email: firstname.lastname@example.org
Purpose: The purpose of this study was to understand clinical significance of near-infrared reflectance (NIR), blue fundus autofluorescence (FAF) and near-infrared autofluorescence (NIA) in dry age-related macular degeneration (AMD), by correlation with fluorescein angiography (FA) and cross-sectional spectral domain optical coherence tomography (SD OCT).
Methods: We evaluated 110 eyes (62 patients, mean age: 64 ± 8 years) diagnosed with dry AMD between January 2010 and December 2010, which underwent NIR (λ = 830 nm), FAF and FA (excitation λ = 488 nm; emission λ > 500 nm), NIA (excitation λ = 787 nm; emission λ > 800 nm), and simultaneous SD OCT scanning using a combined confocal scanning laser ophthalmoscope/SD OCT device (Spectralis HRA + OCT; Heidelberg Engineering, Heidelberg, Germany).
Results: Drusen showed variable increased/decreased NIR, FAF, NIA and FA, which corresponded to variable increased/decreased thickness of the retinal pigment epithelium (RPE) and possible presence of subretinal deposits on SD OCT. Geographic atrophy (GA) was present in 43/110 eyes (39.0%) and showed increased NIR and fluorescence (FA), absent FAF and NIA, and loss of RPE on SD OCT. The hyperautofluorescence of the GA margin was never larger in FAF than that in NIA, while in 16.2% of cases, it was larger in NIA than that in FAF and corresponded to mild choroidal hyperreflectivity on SD OCT.
Conclusions: Simultaneous recording of SD OCT scans provided ultrastructural data for the evaluation of NIR, FAF, NIA and FA in dry AMD. Near-infrared autofluorescence might detect earlier than FAF areas of RPE cell loss at the GA margin.
Age-related macular degeneration (AMD) is the leading cause of irreversible blindness among individuals older than 60 years in industrialized countries (Klein et al. 1992). Most patients with AMD are affected with the dry form of the disease, characterized by the presence of drusen, pigment abnormalities [focal hyperpigmentation or hypopigmentation of the retinal pigment epithelium (RPE)] and geographic atrophy (GA) of the macula (Klein et al. 1992). Patients with dry AMD, even in the late atrophic stages, can maintain good central visual acuity until the disease progresses to involve the foveal centre or the preferred retinal locus for fixation. Diagnostic procedures used for evaluation and follow-up of dry AMD include colour fundus photography, blue fundus autofluorescence (FAF), fluorescein angiography (FA) and spectral domain optical coherence tomography (SD OCT).
High-quality, stereoscopic colour fundus photography has represented the gold standard in evaluating the severity and progression of AMD in major epidemiologic eye disease studies, such as Age-Related Eye Disease Study (AREDS) (Bartlett & Eperjesi 2007). Limitations of colour fundus photography are the interpatient variability of fundus pigmentation, media opacities, the variability of drusen appearance, the presence of small satellites of atrophy and the impossibility to detect three-dimensional anatomic information of retinal abnormalities (Pirbhai et al. 2005; Sunness et al. 1999). Recently, retro-mode imaging has showed a greater sensitivity than colour fundus photography in drusen quantification (Acton et al. 2011).
Fundus autofluorescence using a fundus camera or a confocal scanning laser ophthalmoscope (cSLO) has been used to image GA (Schmitz-Valckenberg et al. 2008a). Fundus autofluorescence generated with short-wavelength excitation is dominated by RPE lipofuscin (Delori et al. 1995), a complex mixture of fluorophores that are by-products of the visual cycle (Sparrow et al. 2003), and accumulates in the RPE after phagocytosis. Lipofuscin fluorescence occurs between 500 and 750 nm with a peak emission of approximately 630 nm. Geographic atrophy is characterized by a loss of FAF, because of the loss of lipofuscin contained within the RPE. Limitations to the use of FAF are the difficulty of detecting GA and its boundaries in the presence of advanced cataracts, and, when using the cSLO-based imaging system, the difficulty in identifying the boundaries of GA in close proximity to the centre of the macula because of the retinal xanthophylls that absorb the excitation light and block FAF from the underlying RPE (Schmitz-Valckenberg et al. 2008a,b).
Spectral domain optical coherence tomography is a noninvasive, high-resolution imaging modality capable of producing high-speed, high-resolution, high-density three-dimensional cross-sectional images covering the central macula (Khanifar et al. 2008). Spectral domain optical coherence tomography provides both qualitative information, such as ultrastructural changes during follow-up (Helb et al. 2010), and quantitative parameters, such as area and volume of the studied lesions (Khanifar et al. 2008). A good reproducibility has been showed by three-dimensional SD OCT in dry AMD (Menke et al. 2011). Systems combining SD OCT with a cSLO allow simultaneous recordings of cross-sectional OCT images with various topographic imaging modes such as NIR, FAF, NIA and cSLO FA. Exact alignment of the SD OCT scans with the topographic cSLO images allows correlating quasi in vivo histology with pathologic features observed on reflectance, autofluorescence and angiography images.
Here, we evaluated the multimodal visualization of dry AMD and correlated NIR, FAF, FA and NIA of study lesions with SD OCT to assess their ultrastructural characteristics.
We reviewed the charts of 62 consecutive patients with dry AMD, with and without GA, visited at the University Eye Clinic of Creteil between January 2010 and December 2010. Only eyes having previously undergone multimodal imaging analyses (high-resolution digital colour fundus photographs, NIR, FAF, NIA, FA and SD OCT) were included. All patients underwent a complete ophthalmologic examination, including measurement of best-corrected visual acuity (BCVA) using standard Early Treatment of Diabetic Retinopathy Study (ETDRS) charts, fundus biomicroscopy and multimodal imaging. The nuclear lens status was graded into one of seven severity categories using the AREDS classification system for cataracts (The Age-Related Eye Disease Study 2001). Indocyanine green angiography (ICGA, Spectralis HRA + OCT; Heidelberg Engineering, Heidelberg, Germany) was performed at physician discretion. Informed consent was obtained, as required by the French bioethical legislation, in agreement with the Declaration of Helsinki for research involving human subjects. University Paris Est Creteil Institutional Review Board approval was obtained for this study. Exclusion criteria were age <50 years, presence of choroidal neovascularization, any prior treatment (such as laser photocoagulation, photodynamic therapy, intravitreal injections of steroids or anti-vascular endothelial growth factor), nuclear cataract greater than mild according to AREDS classification (lens grade, ≥4) and high myopia (>6 dioptres).
Multimodal fundus imaging acquisition
Colour fundus photographs were obtained using a high-resolution digital retinal camera (Topcon TRC-50 retinal camera; Tokyo, Japan). For high-resolution multimodal fundus imaging analysis, we used a combined instrument (Spectralis HRA + OCT) that allows for simultaneous recording of cSLO and SD OCT images (Helb et al. 2010). A standardized imaging protocol was performed in all patients, which included acquisition of NIR (λ = 830 nm; field of view, 30° × 30°; image resolution, 768 × 768 pixels), blue FAF and FA (excitation λ = 488 nm; emission λ > 500 nm; field of view, 30° × 30°; image resolution 768 × 768 pixels), NIA (excitation λ = 787 nm; emission λ > 800 nm; field of view, 30° × 30°; image resolution, 768 × 768 pixels) and simultaneous SD OCT scanning using a second, independent pair of scanning mirrors (λ = 870 nm; acquisition speed, 40 000 A-scans per seconds; scan depth, 1.8 mm; digital depth resolution, approximately 3.5 μm per pixel; optical depth resolution, 7 μm; lateral optical resolution, 14 μm) (Fleckenstein et al. 2008). With cSLO acquisition, light from a conjugate plane of interest is detected by the image sensor, permitting suppression of light from planes anterior and posterior to the plane of interest and resulting in high-contrast fundus images. Because of two independent pairs of scanning mirrors, eye movements are registered and corrected automatically. This allows for quasi-pixel-to-pixel correlation of cSLO and OCT findings. Using automated eye tracking and image alignment based on cSLO images, the software allows averaging a variable number of single images in real time [Automatic Real Time (ART) Module; Heidelberg Engineering]. The OCT B-scan is then repositioned in the moving eye and thus stabilized and frozen at the selected retinal location.
Confocal scanning laser ophthalmoscope and SD OCT images were evaluated on a computer monitor by two independent examiners (R.F. and G.Q.). Geographic atrophy was defined as one or more discrete areas of loss of RPE, whose largest diameter measured 500 μm or more, with a colour and thickness change relative to the surrounding retina, and more prominent visualization of the choroidal vessels (Sunness et al. 2008). Subretinal drusenoid deposits were defined as presence of ≥5 definite drusenoid accumulations above the RPE in >1 SD OCT B-scan (Zweifel et al. 2010). Drusen were classified as small/intermediate (largest diameter <125 μm) and large (largest diameter ≥125 μm). Abnormalities of NIR, FAF, NIA and FA distribution were classified as increased, reduced or absent NIR/FAF/NIA/FA compared with the surrounding areas in the same image. Moreover, a topographic correlation was performed between SD OCT scans and colour fundus photographs, NIR, FAF, NIA, FA images recorded within the macular area.
A total of 110 eyes of 62 patients (36 women and 26 men; mean age: 64 ± 8 years; range, 56–72 years) met the inclusion criteria and had multimodal imaging analyses. Mean BCVA was 55.2 ± 75.3 ETDRS letters. All patients underwent NIR, FAF, NIA, FA and SD OCT. In 12 study subjects, we also performed ICGA. Fundus autofluorescence and NIA intensity patterns according to presence of GA or drusen are summarized in Table 1.
Table 1. Combinations of fundus autofluorescence (FAF) and near-infrared autoflorescence (NIA) intensity patterns according to the presence of geographic atrophy (GA) or drusen in 110 eyes (62 patients) with dry age-related macular degeneration.
FAF and NIA increased
FAF and NIA reduced
FAF and NIA absent
Increased FAF and reduced NIA
Increased FAF and normal NIA
Normal FAF and reduced NIA
Increased NIA and normal FAF
Increased NIA and reduced FAF
Normal NIA and reduced FAF
Drusen were present in all eyes. In the absence of central atrophy, the foveal area always showed normal NIR, a reduced FAF and FA fluorescence, and an increased NIA.
Large drusen showed varying degrees of NIR, FAF and NIA and increased FA fluorescence. Large drusen showed either increased or reduced NIR. Near-infrared autofluorescence of large drusen was either normal or reduced (Figs 1 and 2), or diffusely increased (Fig. 3). Fundus autofluorescence of large drusen was either normal or increased and well defined (Fig. 2). In 2.1% of cases, large drusen showed reduced FAF and increased NIA, while in all other cases, large drusen showed either an increase of both FAF and NIA, or increased FAF and reduced NIA (Table 1; Fig. 2). On SD OCT, large drusen appeared as elevations of the RPE (Fig. 3).
Small/intermediate drusen always showed an increased well-defined NIR, FAF, NIA and FA fluorescence. On SD OCT, small/intermediate drusen appeared as focal thickening of the RPE.
Subretinal drusenoid deposits were present in 32/110 eyes (29.0%). Subretinal drusenoid deposits were bright on NIR, appeared darker than the uninvolved surrounding areas in FAF and NIA, ranged from no demonstrable change to minimal hypofluorescence on FA and were limited to the subretinal space on SD OCT (Fig. 4).
A drusenoid pigment epithelium detachment (PED) was present in 17 of 110 eyes (15.4%). Drusenoid PED showed a patchy hyper-/hypopattern on NIR, FAF and NIA and was hyperfluorescent on late FA phases. The areas of increased/reduced signal on NIR and NIA appeared on SD OCT as a multiple elevations of the RPE band.
Geographic atrophy was present in 43/110 eyes (39.0%). In all cases, areas of RPE loss were bright on NIR and FA, and dark on both FAF and NIA (Figs 3–5, S1 and S2). The GA margin showed increased NIA and normal FAF in 10 cases (23.2%) (Fig. 5), increased FAF and normal NIA in seven cases (16.2%), increased NIA and FAF in 15 cases (34.9%), and normal FAF and NIA in six cases (13.9%). In areas showing either increased FAF or NIA, SD OCT showed thickening of the RPE layer (Fig. S1). In areas showing a reduced FAF and NIA, FA and SD OCT confirmed a RPE loss (Figs 5, S1 and S2). In no case, the hyperautofluorescence of the GA margin was larger in FAF than that in NIA, while in seven of 43 cases (16.2%), the hyperautofluorescence of the GA margin was larger in NIA than that in FAF (Fig. 1). These cases showed a slight increase in FA signal and mild choroidal hyperreflectivity on SD OCT, which, however, was less intense than that in GA area.
This study reported a comparison between NIR, FAF, NIA, FA and OCT for imaging of dry AMD. Comparison of NIA with FAF has been reported by Kellner et al. (2010), but no correlation with FA and the quasi histologic in vivo assessment of the posterior ocular fundus provided by SD OCT was done.
In our study patterns of NIR, FAF and NIA in patients with AMD were similar to those reported previously (Bindewald et al. 2005; Vaclavik et al. 2007; McBain et al. 2007). In the absence of GA, foveal NIR and NIA were always increased, while foveal FAF was reduced. Increased foveal NIR and NIA are attributable to the melanin content of foveal RPE cells, while foveal FAF reduction is because of masking from foveal xanthophylls. In GA, SD OCT has shown central choroidal hyperreflectivity from absence of RPE blockage, determining absent FAF and NIA, and increased FA from window defect. On the other hand, SD OCT has shown RPE thickening in areas of increased FAF or NIA, which could be due to the migration of RPE cells or subretinal deposits. Subretinal deposits have been also referred to as ‘reticular pseudodrusen’ and ‘reticular macular disease’ (Zweifel et al. 2010; Smith et al. 2009; Pumariega et al. 2011). Fundus autofluorescence likely reflects accumulation of lipofuscin, which depends on the phagocytosis of photoreceptor outer segments (Delori et al. 1995). The major contribution to NIA is supposed to derive from ocular melanin, as suggested by the correspondence between NIA distribution and RPE melanin distribution (Weiter et al. 1986), by the contribution of both RPE and the choroid to NIA, and by the markedly increased NIA of choroidal naevi (Keilhauer & Delori 2006). Near-infrared autofluorescence has been shown to derive predominantly from melanosomes in the RPE and has shown to increase in case of increased melanin levels and to reduce after blockage or deficient RPE cells in chloroquine retinopathy or AMD (Weinberger et al. 2006; Keilhauer & Delori 2006; Kellner et al. 2008).
In large and small/intermediate drusen (varying from normal to hyperautofluorescence) and GA areas (absence of autofluorescence), FAF and NIA patterns showed to correlate. In particular, in large drusen, hyperautofluorescence was well defined in FAF and diffused in NIA. On the opposite, in small/intermediate drusen, similar well-defined increased FAF and NIA were present and correlated with increased FA fluorescence. Spectral domain optical coherence tomography showed RPE elevation in large drusen and small/intermediate drusen. In drusenoid deposits, subretinal hyperreflective material above the RPE was detectable. A sub-RPE localization of deposits (i.e. drusen) would lead to increased FAF or NIA because of the absence of a masking effect, while a subneuroepithelial localization (i.e. subretinal drusenoid deposits) would determine blockage of FAF and NIA originating from RPE cells. A drusenoid PED was present in 15.4% of cases. Drusenoid PED showed a patchy hyper-/hypopattern of NIR, FAF and NIA, an increased FA fluorescence because of staining, and appeared on SD OCT as multiple elevations of the RPE band. A similar increase in NIR, FAF and NIA was present in 88.3% of cases. Increased FAF and NIA most likely indicate areas of increased phagocytosis with lipofuscin formation (Kellner et al. 2010) and increased melanogenesis and melanolipofuscin formation; alternatively, altered melanin autofluorescence characteristics because of oxygenation may be responsible for increased NIA (Smith-Thomas et al. 1996). As drusen can regress spontaneously with restoration of normal retinal anatomy and improvement in visual acuity, RPE cells migration could represent a reversible process because of photoreceptor stress. Fundus autofluorescence and NIA in combination with SD OCT could represent noninvasive methods to monitor this process during follow-up.
In 16.2% of eyes with GA, hyperautofluorescence of the margin was larger in NIA than that in FAF. In these cases, FA fluorescence was slightly increased while SD OCT showed mild choroidal hyperreflectivity, which was, however, less pronounced than that in GA area. Choroidal hyperreflectivity on SD OCT is known to derive from reduction in blocking effect by RPE cells. An increased NIA at the GA margin could be explained by the anterior migration of melanin granules in the RPE cells or by an increase of melanin, melanolysosomes or melanolipofuscin content in the RPE cell before lipofuscin accumulation as a result of active phagocytotic process and oxygenation (Smith-Thomas et al. 1996; Peters et al. 2000). Aetiologic mechanisms for GA onset include oxidative stress (Beatty et al. 2000), inflammation, alterations in blood flow and lipid assembly. Another proposed mechanism is the progressive loss of lipofuscin phagocytotic capacity by RPE cells and their consequent death. Our data suggest a role of NIA in detecting areas of RPE cell loss and ongoing oxidative stress at the GA margin. Early detection with NIA could precede detection with FAF.
Limitations of this study are the cross-sectional design and the lack of a correlation between morphological and functional changes. Both increased and reduced FAF indicate areas with loss of photoreceptor function (Scholl et al. 2004), and from the present findings, a similar correlation to function can be expected for correspondingly increased and reduced NIA. Another limitation is the utilization of Spectralis HRA + OCT to obtain NIA imaging. In Spectralis HRA + OCT, a beam splitter is used at around 850 nm to separate the signal into two channels: one for the cSLO, the other for the SD OCT detector. This results in smaller bandwidths range for NIA of the Spectralis HRA + OCT as compared to Heidelbergh HRA2, resulting in inferior signal-to-noise ratio. In our study to obtain NIA, simultaneous FA + ICG acquisition mode was used, maximum level of illuminance and eye tracking were set to increase signal-to-noise ratio.
In conclusion, in this study, we used a digital retinal camera and a SD OCT system combined with a multimodal cSLO topographic imaging system to obtain colour fundus photographs and simultaneous recordings of cross-sectional SD OCT and c-SLO images, to correlate FA, NIR, FAF and NIA images and to understand NIR, FAF and NIA clinical significance in dry AMD. Increased NIR with absent FAF and NIA would be due to absent RPE band on SD OCT, while variable levels of NIR, FAF and NIA intensity would depend on RPE thickness and presence of subretinal deposits. The understanding of ultrastructural processes at the basis of FAF and NIA are important to evaluate the onset and progression of the disease. A prospective study is planned to correlate the morphological data provided by NIR, FAF NIA and SD OCT with functional changes in dry AMD.