Fundus autofluorescence imaging

Abstract

Fundus autofluorescence (FAF) imaging is an in vivo imaging method that allows for topographic mapping of naturally or pathologically occurring intrinsic fluorophores of the ocular fundus. The dominant sources are fluorophores accumulating as lipofuscin in lysosomal storage bodies in postmitotic retinal pigment epithelium cells as well as other fluorophores that may occur with disease in the outer retina and subretinal space. Photopigments of the photoreceptor outer segments as well as macular pigment and melanin at the fovea and parafovea may act as filters of the excitation light. FAF imaging has been shown to be useful with regard to understanding of pathophysiological mechanisms, diagnostics, phenotype-genotype correlation, identification of prognostic markers for disease progression, and novel outcome parameters to assess efficacy of interventional strategies in chorio-retinal diseases. More recently, the spectrum of FAF imaging has been expanded with increasing use of green in addition to blue FAF, introduction of spectrally-resolved FAF, near-infrared FAF, quantitative FAF imaging and fluorescence life time imaging (FLIO). This article gives an overview of basic principles, FAF findings in various retinal diseases and an update on recent developments.

1. Introduction

Fundus autofluorescence (FAF) imaging allows for non-invasive metabolic mapping of naturally and pathologically occurring fluorophores of the ocular fundus (Fig. 1) (Delori et al., 1995; Holz et al., 2007b; Schmitz-Valckenberg et al., 2008b; Solbach et al., 1997; von Rückmann et al., 1995). The physical principle is identical to that of fluorescein or indocyanine green angiography (i.e., molecules are brought to glow by excitation of light within a certain wavelength range, while the wavelength of the emission is longer as compared to the excitation light). The main difference is that FAF imaging is based on visualization of endogenous fluorophores, while dye-based angiography is based on detection of exogenous molecules. Hence, no intravenously injected agent is required for FAF imaging – the intrinsic molecules are already present. The second important difference between FAF and angiography is that the FAF signal intensity is much weaker at about 2 orders of magnitude lower than the background of a fluorescein angiogram at the most intense part of the dye transit (Delori et al., 1995). These characteristics require sensitive imaging systems.

Fig. 1.

Colour fundus camera photography in normal subject (A) with corresponding confocal scanning laser ophthalmoscopy fundus autofluorescence (FAF), shown with intensities in gray value (B) and as pseudo-3D-colour coding illustration (C). Of note, photopigment of the outer photoreceptor segments act as filter of the signal in the dark-adapted state (D). Following light adaptation and/or with increasing exposure to the excitation light during image acquisition, photoreceptor outer segments are bleached, the filter effects of photopigments are reduced and thus the FAF signal becomes stronger.

2. Basic principles

2.1. Origin of the signal

The FAF signal mainly originates at the level of the retinal pigment epithelium (RPE)/photoreceptor complex (Delori et al., 1995; Schmitz-Valckenberg et al., 2011c). The dominant source are fluorophores accumulating as lipofuscin (LF) in lysosomal storage bodies in postmitotic RPE cells, while photopigments of the outer photoreceptor segments act as a filter of the excitation light (Fig. 2). Further, any fluorophore or filter within the light path may interfere with the detected signal, either in the normal state or as part of pathological conditions. The lens is an important confounder of the signal, particularly with progressive nuclear cataract and when using short-wavelength light (i.e. blue light) for excitation. In fact, the brownish-yellowing of the lens, which typically occurs with increasing age, represents a strong absorber of blue excitation light, not only blocking the excitation light on its way to the posterior pole but also generating its own fluorescence light that is then scattered in the light path further interfering with the FAF signal detection from the ocular fundus. Haemoglobin within retinal blood vessels physiologically blocks both the excitation light and the resultant autofluorescence wavelengths from RPE fluorophores. With retinal haemorrhages, blocking may also be observed, but the signal at the site of the haemorrhages may evolve over time into markedly increased levels as part of metabolic degradation of blood compounds (Piccolino et al., 1996; Sawa et al., 2006b) (see 6.1.3 Choroidal Neovascularization). At the fovea and parafovea, the FAF signal is reduced by increased melanin optical density and, when using blue-light excitation, by macular pigment deposition in inner retinal layers (Fig. 3) (see 2.4 Macular Pigment) (Delori et al., 1995; Paavo et al., 2018). In RPE degeneration and thus loss of RPE fluorophores, additional fluorophores at the level of the choroid may contribute to the FAF signal, either due to melanolipofuscin at this anatomic level or due to other minor fluorophores including elastin and collagen in choroidal blood vessel walls and sclera. Of note, bleaching phenomena and loss of photopigment result in increased FAF by reduced absorbance anterior to the RPE level (Fig. 1) (Staurenghi et al., 2004; Theelen et al., 2007). Finally, so called “pseudofluorescence” phenomena have to be taken into consideration in case of increased FAF intensities. These can occur due to filter leakage in case of strong blue reflectance intensities (e. g. crystalline deposits). Through fluorescence of other structures in the eye, notably the crystalline lens, light may enter the imaging path directly or by scatter, to reach the detector. This extraneous light causes a loss of contrast of the image. These phenomena need to be particularly considered if nuclear lens opacities are present. In this scenario, the excitation light generates fluorescence from the lens. This long-wavelength fluorescence light is scattered inside the vitreous cavity and may be reflected by the posterior pole as “secondary reflectance light” back through the lens, passing the barrier filter and then reaching the detector. Particularly, signal detection at or nearby the optic nerve head, which represents a highly reflective tissue, is prone to this artifact.

Fig. 2.

Identified lipofuscin fluorophores are a mixture of bisretinoid compounds that form as a byproduct of the visual cycle. This ‘biococktail’ of fluorophores forms in photoreceptor outer segments and are transferred to the retinal pigment epithelium in phagocytosed outer segment membrane.

Fig. 3.

Schematic drawing for causes of increased and decreased fundus autofluorescence. For details, see text and also Table 2.

2.2. Retinal pigment epithelium and lipofuscin

The RPE consists of polygonal-shaped cells forming a monolayer between the neurosensory retina and the choroid. Given multiple essential physiologic functions of the RPE, it is not surprising that RPE dysfunction has been implicated in a variety of retinal diseases (Schraermeyer and Heimann, 1999; Sparrow et al., 2010a).

A hallmark of aging is the gradual accumulation of LF granules in the cytoplasm of RPE cells (Fig. 2) (Feeney-Burns et al., 1980; Sparrow and Boulton, 2005; Weiter et al., 1986). The LF of the retina can be accounted for by non-enzymatic reaction of vitamin A aldehyde (retinaldehyde) with phosphatidylethanolamine in photoreceptor outer segments (Fig. 2) (Sparrow et al., 2012). These bisretinoid fluorophores are transferred to RPE secondarily within phagocytosed outer segment membrane. The bisretinoids of LF are a complex family of more than 20 compounds with A2E, the best known, being only one of these (Fishkin et al., 2005; Kim and Sparrow, 2018; Kim et al., 2007; Parish et al., 1998; Wu et al., 2009; Yamamoto et al., 2011). The bisretinoids are subject to photo-oxidative alteration and can cross-link with other molecules, creating complex daughter molecules. Several characteristics of bisretinoids provide evidence that these fluorophores are the constituents of RPE LF responsible for FAF. (Eldred and Katz, 1988). They are chemically characterized by extensive systems of conjugated double bonds. When stimulated with light in the blue range, they typically emit a golden yellow fluorescence (Feeney, 1978; Katz et al., 1987; Sparrow et al., 2010b).

The distribution of LF in postmitotic human RPE cells and its accumulation with age have been extensively studied in vitro, applying fluorescence microscopic techniques (Greenberg et al., 2013a; Sparrow et al., 2012). For example FAF, bisretinoids and native LF all have emission maxima at 590–620 nm and exhibit a characteristic red shift with increasing excitation wavelength (Sparrow, 2007). Furthermore, disease relationships indicate that short-wavelength autofluorescence originates in bisretinoid fluorophores that are generated as byproducts of the visual cycle. Thus conditions that impair the chemical reduction and removal of retinaldehyde such as deficiencies in ABCA4 and retinol dehydrogenases result in increased bisretinoid formation (Burke et al., 2014; Kim et al., 2004), while conditions that disable the regeneration of the 11-cis retinaldehyde chromophore, such as RPE65 mutations tend to confer an absence or profound shortage of short-wavelength autofluorescence (Lorenz et al., 2004; Sparrow et al., 2008).

Lipid fragments such as 2-(ω-carboxyethyl)pyrrole (CEP), 4-hydroxynonenal (HNE) and malondialdehyde (MDA) that have been described in RPE LF may form from photooxidative processes within outer segments or photooxidative processes ongoing in the bisretinoid mixture (Ng et al., 2008; Schutt et al., 2003).

Post-mortem analyses reported that redistribution and loss of autofluorescent granules were among the earliest changes in AMD donor eyes (Ach et al., 2015; Dorey et al., 1993; Wagner et al., 2019). If excessive formation occurs, the RPE cell has no means of either safely degrading bisretinoids in lysosomes or transporting the LF material and granules into the extracellular space via exocytosis. Subsequently, these granules are trapped in the lysosomal organelles of the postmitotic RPE cells. Previous studies have shown that various LF bisretinoids such as A2-E possess toxic properties that may interfere with normal cell function via various molecular mechanisms. For instance, photooxidative processes involving bisretinoid release damaging fragments carrying carbonyls and aldehydes that are detected in drusen (Crabb et al., 2002; Wu et al., 2010; Yoon et al., 2012). Other suggested mechanisms include impairment of lysosomal degradation due to inhibition of the lysosomal adenosine triphosphate-dependent proton pump and detergent-like activity (Bergmann et al., 2004; Hammer et al., 2006; Schütt et al., 2000; Sparrow et al., 2003). The molecular composition of LF may possibly depend on specific underlying molecular mechanisms. Zhou and associates demonstrated with an in vitro assay a link between inflammation, activation of the complement system, oxidative damage, drusen, and RPE LF (Zhou et al., 2006). They suggested that products of the photo-oxidation of RPE LF components could serve as a trigger for the complement system which could predispose the macular area to a chronic, low-grade inflammatory process over time. Another study demonstrated that accumulation of LF-like material in vitro renders RPE cells susceptible to phototoxic destabilization of lysosomes, resulting in inflammasome activation and secretion of inflammatory cytokines (Brandstetter et al., 2016). The proposed mechanism of inflammasome activation links photo-oxidative damage and innate immune activation in RPE pathology.

The fluorophores that ultimately accumulate in the RPE, have their origins in the retina, and specifically in the outer segments of photoreceptor cells. In diseases causing an accumulation of unphagocytized outer segments, intense autofluorescence not originating from the RPE can be detected. A common feature of these conditions is separation of the photoreceptor outer segments from the underlying RPE by fluid. These conditions include tractional detachment of the retina, central serous chorioretinopathy, optic nerve pit maculopathy, vitelliform macular dystrophy, ocular nevi and melanomas, adult vitelliform lesions, slowly absorbing fluid after rhegmatogenous retinal detachment repair, and acute exudative polymorphous vitelliform maculopathy (Chung and Spaide, 2004; Kaga et al., 2001; Laud et al., 2007; Spaide, 2004b; Spaide and Klancnik, 2005; Spaide et al., 2006).

2.3. Macular pigment

Macular pigment, composed of lutein and zeaxanthin, extensively accumulates along the axons of the cone photoreceptors in the central retina (Fig. 3) (Davies and Morland, 2004; Snodderly et al., 1984; Whitehead et al., 2006). Various physiological functions have been proposed for macular pigment (Davies and Morland, 2004; Wolf and Wolf-Schnurrbusch, 2007), including filtering of blue light, which may reduce photooxidative damage and glare, minimization of the effects of chromatic aberration on visual acuity, improvement in fine-detail discrimination, and enhancement of contrast sensitivity. Neutralization of reactive oxygen species by macular pigment may have a protective effect on the neurosensory retina. Although there may be a large variation with regard to the concentration of macular pigment, the pattern of distribution is relatively uniform in the normal population. It generally shows a peak concentration at the foveal center and rapidly decreases with eccentricity, with very little present at about 8° of eccentricity. A subset of individuals exhibits a double-peaked (peaks at 0° and 1°) central-to-peripheral gradient of macular pigment.

Peak absorption of luteal pigment is at 460 nm. These absorption properties can be readily recorded in vivo by blue-light autofluorescence imaging (Wolf and Wolf-Schnurrbusch, 2007). Therefore, blue FAF imaging can also be used to determine the topographic distribution of macular pigment. Compared to other methods, including heterochromatic flicker photometry, the advantage of FAF imaging is its objective acquisition technique independent of psychophysical cooperation by the examined individual (Delori et al., 2001).

3. Techniques of fundus autofluorescence imaging

3.1. Historical background

The imaging of FAF phenomena in vivo was already demonstrated in the early days of fluorescein angiography (i.e. in the early 1960s), when pre-injection fluorescence of optic disc drusen and optic disc hamartomas in tuberous sclerosis, and within lesions of Best vitelliform macular dystrophy was detected (Fig. 4) (Bonnin et al., 1976; Mustonen and Nieminen, 1982; Neetens and Burvenich, 1977; Schatz et al., 1978; Wessing and Meyer-Schwickerath, 1968). However, these observations were limited to a few patients with pathological accumulations of highly fluorescent material. When using the excitation and emission filters as applied for fluorescein angiography, the conventional fundus camera produces autofluorescence images with low contrast and high background noise in young persons. In the elderly, the quality of the images drops even further and it becomes practically impossible to evaluate the FAF distribution. To overcome the limitations of a relatively low FAF signal and to reduce the confounding effects of progressive lens opacifications with age, novel hardware and software enhancements for FAF imaging had to be developed.

Fig. 4.

Visualization of fundus autofluorescence phenomena in the pre-injection phase was already described in the early days of fluorescein angiography, here shown for optic drusen (A) with the Zeiss fundus camera (B). Reprinted from Wessing A (1968). Fluoreszenzangiographie der Retina. Lehrbuch und Atlas, Georg Thieme Verlag. Stuttgart. (C) Schematic drawing of the set-up of modern confocal scanning laser ophthalmoscopy as used for fundus autofluorescence imaging.

Delori and co-workers pioneered in-vivo FAF imaging by the introduction of the fundus spectrometer. By reducing the angle of view to small retinal areas (2° diameter) and by incorporating several additional hardware adjustments (image intensifier diode array as a detector, beam separation in the pupil, confocal detection), they were able to systematically analyze the excitation and emission FAF spectra. These important studies demonstrated that fundus fluorescence is emitted across a broad band from 500 to 800 nm. Both at the center of the fovea and at 7° temporally, optimal excitation occurred at 510 nm, with peak emission at approximately 630 nm, indicating the predominance of a fluorophore at these excitation and emission spectra. There was a significant increase in fluorescence with age and the recording along a horizontal line through the fovea showed a minimum fluorescence at the fovea, a maximum intensity at 7–15° from the fovea, and a decrease toward the periphery, most likely reflecting the concomitant absorption of macular pigment and melanin interfering with the emission of the dominant fluorophore. The optic disc was characterized by a less intense signal, presumably due to the absence of fluorophores. The relationship with age and the topographic distribution of the dominant fundus fluorophore were consistent with those of RPE bisretinoid LF as measured in the RPE of human donor eyes (Sparrow et al., 2010b; Weiter et al., 1986; Wing et al., 1978). Along with autofluorescence recordings in patients with several pathologic conditions, the initial work by Delori et al. demonstrated that LF is the dominant source of intrinsic fluorescence of the ocular fundus (Delori et al., 1995). However, the small area sampled by the fundus spectrometer as well as the customized relatively complex instrumentation and techniques were not practical for recording FAF from patients in a clinical setting.

3.2. Confocal scanning laser ophthalmoscopy

Image acquisition using the confocal scanning laser ophthalmoscopy (SLO) optimally addresses the limitations of the low intensity of the autofluorescence signal and the interference of the crystalline lens. It was used initially by von Rückmann and coworkers in a clinical imaging system, parallel to the work by Delori and co-workers (Fig. 4C) (von Rückmann et al., 1995). The confocal SLO projects a low-power laser beam on the retina which is swept across the fundus in a raster pattern (Webb et al., 1987). The intensity of the reflected light at each point, after passing through the confocal pinhole, is registered by means of a detector, and a two-dimensional image is subsequently generated. Confocal optics ensure that out-of-focus light (i.e., light originating outside the adjusted focal plane, but within the light beam) is suppressed and, thus, the image contrast is enhanced. This suppression increases with the distance from the focal plane and signals from sources anterior to the retina, i.e., the lens or the cornea, are effectively reduced.

In contrast to the 2° retinal field of the fundus spectrophotometer, the confocal SLO allows imaging over larger retinal areas. To reduce background noise and to enhance image contrast, a series of several single images is usually recorded (reviewed by (Schmitz-Valckenberg et al., 2008b)). For the final FAF image, a number of these frames (usually out of 4–32) are typically averaged to improve signal to noise. Last, the pixel values are normalized to increase the perceived contrast (Fig. 5). Given the high sensitivity of the confocal SLO and the high frame rate of up to 16 frames per second, FAF imaging can be performed within seconds and at low excitation energies which are well below the maximum retinal irradiance limits of lasers established by the American National Standards Institute (AINSI) and other international standards (referring to normal eyes) (American National Standards Institute, 2014).

Fig. 5.

Acquisition of fundus autofluorescence (FAF) images with the confocal scanning laser ophthalmoscope (Frame size 30 ° × 30 °). Without the barrier filter in place, blue reflectance (A) is imaged. When the barrier filter is activated, a low signal (“noisy”) FAF image (B) is recorded. In order to improve the signal-to-noise ratio, several FAF images (so-called “frames”) are averaged to obtain a mean image (C), following registration to each other for corrections of eye movements and blinking. To further improve the topographic representation of the FAF signal intensities, the calculated averaged image is then normalized (D), i.e. the pixel distribution is expanded over the digital spectrum of intensities from 0 to 255. This averaged and normalized FAF image (D) is usually the image that is available for FAF assessment in clinical routine management.

With confocal SLO FAF imaging, excitation is usually induced in the blue range (λ = 488 nm), and an emission filter with a bandwidth between 500 and 700 nm is used to detect emission of the autofluorescence signal. At this blue excitation wavelength – compared to an excitation in the green range – there is reduced FAF intensity in the central retina due to absorption by macular pigment (Chen et al., 2019). By applying an excitation in the green range, there is a markedly less significant reduction of the FAF signal in the macular center due to less absorption of the excitation light by macular pigment (Fig. 7). Thus, application of this wavelength is advantageous to detect foveal pathological lesions exhibiting a decreased FAF signal, e.g. geographic atrophy in AMD or Stargardt macular dystrophy (Muller et al., 2018a; Pfau et al., 2017; Wolf-Schnurrbusch et al., 2011).

Fig. 7.

Imaging the right eye of a 76-year-old male subject with geographic atrophy secondary to AMD in the presence of beginning nuclear lens opacities. Confocal scanning laser ophthalmoscopy with near-infrared reflectance (A), blue (B) and green (C) fundus autofluorescence imaging allows for high-contrast visualization of atrophic areas. These show a severely reduced signal due to loss of lipofuscin in the degenerated retinal pigment epithelium cell. Note the different in macula pigment absorption in the central macula between blue (B) and green (C) fundus autofluorescence. Lower row: Fundus camera wide-field photography. Colour photography (D) permits assessment of central and midperiphery changes at high-resolution (e. g. visualization of reticular pseudrusen in the midperiphery, particularly nasally). By contrast, blue (E) and green (F) fundus autofluorescence are characterized by low signal-noise ratio, while the blue image is inferior compared to the green, probably due lens opacities.

The most widely used confocal SLO system for FAF imaging to date is the Heidelberg retina angiograph (HRA)/Spectralis HRA (Heidelberg Engineering GmbH, Germany) (Fig. 6). One key advantage of this device is the simultaneous acquisition of spectral-domain optical coherence tomography (OCT) recordings that allows for both averaging of several OCT B-scans in order to enhance the signal-to-noise ratio and the synchronous topographic alignment of FAF features with OCT findings (Helb et al., 2010). Other previous instruments, such as the Rodenstock confocal SLO and the Zeiss prototype SM 30 4024 for FAF imaging, are no longer commercially available. Nidek has introduced the F-10 and Mirante confocal SLO platforms that also allow for blue and green FAF imaging with a variety of confocal pinholes. Finally, Centervue has introduced the EIDON device, a confocal SLO with a larger pinhole, for FAF imaging (see 5.3 Spectrally resolved FAF).

Fig. 6.

Overview of fundus autofluorescence imaging of the same right normal eye with different devices. For each system, the excitation and emission range are shown at the bottom. The four examples with light blue background colour operate with blue excitation light, the remaining with green background colour operate with excitation in the green-light range. Please note the variability in macular pigment absorption in the fovea and parafovea between blue and green light excitation.

3.3. Fundus camera

Delori and coworkers subsequently described a modified fundus camera for FAF imaging using an interference filter for excitation and a glass absorbance filter to serve as the barrier filter (Delori et al., 2000). Their design also included the insertion of an aperture in the illumination optics of the camera in order to minimize the loss of contrast caused by light scattering and fluorescence from the crystalline lens. The system also employed a charge-coupled device sensor cooled to −20 °C. The modification also resulted in the restriction of the field of view to a 13° diameter circle; this, together with the complex design, is the likely reason why this configuration has not been further pursued. In 2003, Spaide reported the modification of a commercially available fundus camera system by shifting the excitation and emission wavelengths for FAF imaging towards the red end of the spectrum in order to suppress the fluorescence originating from the lens (Fig. 6) (Spaide, 2003, 2008, 2007). This system used interference filters for both the excitation and barrier filters and a scientific grade charge-coupled device at room temperature. The relatively inexpensive purchase of an additional filter set, together with the broad availability of the flash fundus camera, has made this an attractive alternative. These operate with excitation in the green spectrum and emission is recorded in the yellow–orange spectrum. The field of view of the image is defined by the fundus camera and is typically up to 50°. Today, fundus camera manufactures typically offer these extra filter sets for the application of FAF imaging in the clinical setting.

3.4. Wide-field imaging

The standard image field of the typical confocal SLO encompasses a retinal field of 30 ° × 30 °. Additional lenses allow for imaging of a 55° field or, using the composite mode, imaging over even larger retinal areas. The 102° lens for confocal SLO (Heidelberg Engineering) that was originally designed for fluorescein and indocyanine green angiography, usually does not allow for meaningful FAF imaging as the signal yield for this wide-field lens is too low in most subjects.

Using the fundus camera, so-called montage images can be manually generated using image analysis software on the basis of a seven-field panorama survey. Using wide-field lenses, the reduced signal yield is also apparent with fundus camera systems (Fig. 7).

Peripheral FAF images can also be recorded with a wide-field SLO (P200Tx, Optos) (Fig. 8). Instead of pinhole optics, this system operates with an ellipsoidal mirror to create a virtual point of focus behind the pupillary plane (which can be considered “low confocal”). It allows for FAF acquisition in less than 2 s by using green-light excitation (532 nm) (Duisdieker et al., 2015; Oishi et al., 2014; Tan et al., 2013; Witmer et al., 2012a, 2012b). Overall, FAF imaging beyond the vascular arcades is helpful for assessment of the peripheral manifestations of retinal diseases. Therefore, wide-field SLO FAF is also useful to evaluate the longitudinal evolution of diseases affecting the peripheral retina.

Fig. 8.

Wide-field pseudo-colour (left) and fundus autofluorescence (right) imaging using non-confocal scanning laser ophthalmoscopy in right eye of a patient Stargardt retinal dystrophy. In addition, to areas of central atrophy, peripheral changes and their extent are clearly visualized by fundus autofluorescence.

3.5. Comparison between different imaging systems

For comparison of different imaging systems that are available for the clinical setting, it is not only important to review the general principles of FAF imaging, but also to consider the technical differences between the devices. These may result in variable susceptibilities to artefacts, in variable accuracy with regard to detection of signal alterations and also in different measurements of lesion sizes. Overall, fundus camera, confocal SLO and non-confocal SLO systems are not entirely comparable and findings may not be entirely equivalent to each other. Aspects to consider – already from a technical perspective - are excitation and emission spectra, the modes of excitation and detection, the image acquisition and the image processing post acquisition (Table 1).

Table 1.

Summary of technical differences between the fundus camera and scanning laser ophthalmoscope for fundus autofluorescence Imaging.

Fundus Camera	Scanning Laser Ophthalmoscopy
Bandwidth filters for excitation and emission	One excitation wavelength (laser source) Large emission spectrum (cutoff filter)
One single flash at maximum intensities	Continuous scanning at low light intensities in a raster pattern (bleaching phenomena need to be considered)
Entire cone of light (prone to light scattering)	Confocal with variable pinhole size or ellipsoid mirror (non-confocal) system
Flash light intensity, gain and gamma of detector adjustable	Laser power fixed by manufacturer, detector sensitivity adjustable
Manual contrast and brightness	Imaging processing with averaging of single frames and pixel normalization

4. Interpretation of fundus autofluorescence images

The FAF image shows the spatial distribution of the intensity of the FAF signal for each pixel in gray values (arbitrary values from 0 to 255) (Fig. 1). Per definition, low pixel values (dark) illustrate low intensities and high pixel values (bright) denote high intensities. The topographical distribution of FAF in normal eyes demonstrates a consistent pattern, as illustrated in Fig. 1 (Schmitz-Valckenberg et al., 2008b). A diffuse FAF signal over the posterior pole can be seen, while retinal vessels (due to an absorption phenomenon by blood contents, i.e., hemoglobin) and the optic nerve head (absence of autofluorescent material) are characterized by a very low signal and appear dark. Showing a high degree of interindividual variability, decreased FAF intensities in the macular area with a minimum in the fovea are observed; these are caused by absorption of short-wavelength light due to luteal pigment (lutein and zeaxanthin) and higher optical density of melanin (Fig. 3). Other contributing factors may be an increased accumulation of melanin and a reduced deposition of LF granules.¹

Using pixel gray values, typical ratios between the intensity of the fovea and perifoveal macula have been established in normal subjects. Based on these findings, qualitative descriptions of localized FAF changes are widely used. Usually, the FAF signal over a certain retinal location is categorized in decreased, normal, or increased intensities in comparison to the background signal of the same image. Table 2 lists the major causes and pathophysiological categories for increased or reduced FAF signals.

Table 2.

Sources of abnormal fundus autofluorescence signal intensities

Causes for a reduced FAF signal	Causes for an increased FAF signal
Absence/reduction in RPE lipofuscin density RPE atrophy Increased RPE melanin content Local changes in retinal dystrophies Absence of RPE/lipofuscin at optic nerve head	Excessive RPE lipofuscin accumulation Lipofuscinopathies incl. Stargardt’s disease, Best’s disease and adult vitelliform macular dystrophy Age-related macular degeneration, e.g. RPE in the junctional zone preceding enlargement of occurrence of geographic atrophy
Absorption from extracellular material/cells/fluid anterior to the RPE	Lack of absorbing material, loss of photopigment, pro-longed exposure to excitation light (“bleaching”)
Migrated melanin-containing cells	Occurrence of fluorophores anterior or posterior to the RPE cell monolayer (e. g. vitelliform material in the subretinal space)
Intersection of petaloid configuration in macula oedema due to macular pigment displacement	Cystoid cavities of the petaloid configuration in macula oedema due to displacement of macular pigment
Fresh Intra- and subretinal haemorrhages	Subretinal fluid leading to separation of the outer segments of the photoreceptors from the underlying RPE leading to improper outer segment turnover
Intra- and Subretinal lipid	Macrophages containing lipofuscin in the subretinal space (choroidal tumours such as nevi and melanomas)
Fibrosis, scar tissue, borders of laser scars	Migrated RPE cells or macrophages containing lipofuscin or melanolipofuscin (seen as pigment clumping or hyperpigmentation on funduscopy)
Retinal vessels	Older intra- and subretinal haemorrhages (typically corresponding to ochre appearance on biomicroscopy)
Luteal pigment (lutein and zeaxanthin)	Choroidal vessel in the presence of RPE and choriocapillaris atrophy, e.g. in the center of laser scars or within patches of RPE atrophy
Media opacities (vitreous, lens, anterior chamber, cornea)	Depletion of luteal pigment, e.g. in idiopathic macular telangiectasia type 2 Displacement of luteal pigment, e.g. cystoid macular oedema
Artefacts (e. g. low laser power, low yield of fluorescence signal by setup)	Optic nerve head drusen
	Artefacts (e.g. filter leakage in case of high reflective material or detection of secondary reflectance light caused by generation of fluorescence by the lens/cataract)

The quantification of absolute intensities and their comparison between subjects or within longitudinal observation in the same subject is more complicated and remain a challenge in FAF imaging. Of note, as the histogram of the pixel grayscale values in the usual available confocal SLO images is normalized in order to optimize the perceived contrast of the topographic distribution of the FAF intensity (see Figs. 5 and 3.2 Scanning Laser Ophthalmoscopy), the pixel values are not absolute and these images must not be used for absolute intensity analyses from the outset. Furthermore, when interpreting FAF images, the digital resolution of the detector in current imaging devices exceeds the maximum spatial resolution of ocular media and the optics of the system, mainly due to high-order aberrations. This also explains why increasing the digital resolution of the detector usually does not improve the resolution of the actual image. For quantitative FAF measurement, specific hardware and software adjustments are required (see 5.2 Quantitative Autofluorescence Imaging).

5. Further developments in autofluorescence imaging

5.1. Spectrally resolved fundus autofluorescence

More recent, a confocal blue-light FAF device (EIDON, CenterVue, Padua, Italy) has been introduced in a clinical setting using a 450 nm wavelength and light-emitting diode (LED) light source (Fig. 10) (Borrelli et al., 2018; Dysli et al., 2019; Muller et al., 2019). A different range of fluorophores are thought to be excited at 450 nm compared with 488 nm (Schweitzer et al., 2007). Therefore, application of 450 nm FAF imaging enables the study of these fluorophores, which may provide further insights into disease pathogenesis. Isolating the signal from these minor fluorophores however, is challenging, because the magnitude of the emission signal may be relatively weak. The new confocal LED blue-light FAF system offers an important potential advantage in that the full-emission spectrum is detected on a colour sensor, providing so-called ‘colour FAF’ imaging. This enables the emission spectrum to be divided into long-wave and short-wave emission fluorescence components (‘red’/REFC (560–700 nm) and ‘green’/GEFC (510–560 nm). Thus, colour FAF imaging has the potential to provide additional ‘emission maxima’-based contrast to reveal different substructures in fundus lesions (Borrelli et al., 2018).

Fig. 10.

Some of the autofluorescence imaging devices allow acquisition of autofluorescence images with stratified emission channels (Heidelberg Engineering FLIO device; CenterVue EIDON). The images above were acquired with the EIDON device with a green channel (green emission fluorescent component [GEFC, 500–560 nm]) and a red channel (red emission fluorescent component [REFC, 500–560 nm]. Please note, some of the flecks towards the very periphery tend to show longer emission wavelengths (i. e., more orange colored) as opposed to the central flecks with rather short emission wavelengths (green colour).

5.2. Near-infrared autofluorescence

Near-infrared autofluorescence (NIR-FAF) images can also be obtained in vivo, most commonly and easily by using the indocyanine green angiography (ICGA) mode of the SLO, i.e., without dye injection (excitation wavelength at 795 nm, emission with barrier filter > 800 nm) (Fig. 9) (Keilhauer and Delori, 2006; Weinberger et al., 2006). Due to the excitation and emission beyond the red end of the visible spectrum, the topographic distribution of fluorophores other than LF may be studied by this technique. It has been suggested that the NIR-FAF signal is largely melanin-derived (Keilhauer and Delori, 2006; Kellner et al.; Weinberger et al., 2006). As such, Keilhauer and Delori further speculated that, to varying degrees, choroidal sources contributed to this signal. Gibbs et al. investigated NIR-FAF in humans and mice and suggested that melanosomes in the RPE and choroid were likely the dominant origin of the signal (Gibbs et al., 2009). Except for measurements in cell cultures at low magnification, their analyses were limited to excitation at 633 nm, in contrast to in vivo NIR-FAF, which is generated at 795 nm. Using a customized magnification lens attached to the front of the confocal SLO, Schmitz-Valckenberg and coworkers studied the distribution of the NIR-FAF signal in retinal cross-sections of a human donor eye and correlated ex vivo autofluorescence measurements to in vivo findings in a rat animal model (Schmitz-Valckenberg et al., 2011c). They observed that the NIR-FAF signal was spatially confined to the RPE monolayer and melanin in the choroid.

Fig. 9.

Upper row: The amount of abnormal signal intensities in the right eye of a 35-year-old male subject with Stargardt retinal dystrophy are less pronounced with blue (A) as compared to near-infrared (B) fundus autofluorescence. Lower row: Quantitative autofluorescence (blue = low, white = intense signal intensity, see colour scale). Please note the reference standard at the upper part of both images. In the normal (C) eye, the quantitative autofluorescence signal is similar to the reference standard (“bluish”). In Stargardt retinal dystrophy (D, another patient as compared to upper row), the absolute intensities are markedly increased, showing the highest level at about 8° from the foveal (“redish ring-shaped pattern”).

Multi-modal imaging that includes NIR-FAF has shown that the loss of melanin signal from RPE exhibits better correspondence with ellipsoid zone loss than does the signal from short-wavelength FAF (Duncker et al., 2014b); the measurement of disease progression in patients with retinitis pigmentosa by constriction of the ring with increased intensities by short-wavelength-FAF can also be tracked using NIR-FAF images (Jauregui et al., 2018); and reduced or absent NIR-FAF signal in association with fundus flecks in recessive Stargardt disease indicates that the underlying RPE cells are likely atrophied and the short-wavelength FAF signal is probably generated directly from degenerating photoreceptor cells (Sparrow et al., 2015).

5.3. Quantitative fundus autofluorescence imaging

To combine measurements of FAF intensity with spatial information, an approach to measuring the intensity of short-wavelength FAF (quantitative fundus autofluorescence; qAF) has been developed and applied (Fig. 9) (Delori et al., 2011a; Sparrow et al., 2020). Briefly, the method employs cSLO images, an internal fluorescent reference in the optical pathway to compensate for differences in detector sensitivity; together with correction for magnification and anterior media transmission. To preclude histogram stretch, the images are also saved in non-normalized mode. Accordingly, this method enables comparisons amongst serial images in the same or different subjects.

Attaining reliable qAF measurements is critically dependent upon good image quality. The operator must be experienced and skilled and must follow established protocols. Key requirements for images suitable for qAF measurement are uniform and maximal signal intensity, fine-tuned focus, central alignment of the camera with the eye to avoid obstruction by the iris, and exposure within the range of linearity of the detector (Greenberg et al., 2013b). Lens and vitreous opacities represent a challenge here too, as absorption confounds measured FAF intensity levels originating from the retina and RPE.

Regarding qAF values, the coefficient of agreement between right and left eyes is ±14% a(Greenberg et al., 2013a). Between session repeatability of qAF (95% confidence limits; method of Bland-Altmann) calculated in multiple studies has ranged from 7 to 10%(Burke et al., 2014; Delori et al., 2011b; Duncker et al., 2014a, 2015a, 2015b) indicating that qAF will only differ by more than 7–10%, 5% of the time.

Quantitative autofluorescence levels exhibit a significant increase with age. In healthy eyes, qAF increases with increasing eccentricity up to 10–15° from the fovea with highest values superotemporally (Greenberg et al., 2013b). Furthermore, qAF values have been shown to be higher in females. Finally, there may be ethnic differences: compared with Hispanics, qAF is significantly higher in whites and lower in Blacks and Asians which is likely related to a higher melanin content in the RPE and choroidea compared to Caucasians.

Several studies have demonstrated the potential of qAF to guide clinical diagnosis and genetic testing (see 6.2 Macular and Diffuse Retinal Dystrophies). Furthermore, this approach enhances the understanding of disease processes and may serve as a diagnostic aid, as a more sensitive marker of natural disease progression, and as a tool to monitor the effects of therapeutic interventions targeting LF accumulation. The qAF approach has, at this time, not been widely incorporated into clinical routines probably because of limitations that include the requirement of a trained and committed operator, the installment of an appropriate internal fluorescent reference and software availability (Sparrow et al., 2020). Nonetheless, studies of qAF have expanded our understanding of retinal disorders that involve changes in LF distribution and concentration.

5.4. Fluorescence lifetime imaging ophthalmoscopy

In contrast to commonly used autofluorescence-based imaging modalities, the information obtained with fluorescence lifetime imaging ophthalmoscopy (FLIO) is not based only on the intensity of individual fluorophores (Fig. 11). Light energy is absorbed by an atom or molecule and energy state of an electron is elevated to a higher state. When the electron relaxes to its ground state it releases a photon. The released photon is usually at a longer wavelength, called the Stokes shift. The wavelength of the fluorescence and the delay between excitation and release of the fluorescent photon varies by with the entity fluorescing and its milieu. Conventional ophthalmic autofluorescence imaging detects the fluorescence over a period of micro-to milliseconds in broad ranges of wavelengths. It is possible to quantify the exact time from excitation to fluorescence either by directly measuring the time or by modulating light and measuring the phase delay in the emission. The data can be fitted to exponential or multiexponential decay plots. The data obtained can be analyzed using a phasor plot, which is a 2D graphical representation of each point in the image as vectors in a polar plot (Digman et al., 2008). The advantages of a phasor approach are that molecular species can be identified and the proportions in a mixture of components can be determined by linear combination of the vectors of the components. In clinical ophthalmic FLIO, a research instrument has been constructed that reports photons detected in two spectral channels (short, 498–560 nm and long, 560–700 nm). The resultant lifetimes were reported as short or long, while they were fitted as a double exponential (Dysli et al., 2017b; Sauer et al., 2018a). In healthy eyes, an increase of lifetimes with age is observable (Dysli et al., 2014). The shortest lifetimes are observed for the fovea and peri-fovea, possibly due to macular pigment (Sauer et al., 2018b). Although the technique is relatively new, many diseases have already been investigated and very specific disease-related changes have been discovered, e.g. characteristic patterns in patients with Macular telangiectasia Type 2 (MacTel) (Sauer et al., 2018d), AMD (Sauer et al., 2018c), and Stargardt retinal dystrophy (Dysli et al., 2016a). Further, FLIO has been also discussed in another article in this journal (Dysli et al., 2017b). The results of these studies suggest that FLIO may serve as a tool for early diagnosis of retinal diseases and may play a role in monitoring disease progression in natural history studies and interventional trials (Bernstein and Sauer, 2019). There are many potential fluorophores being measured simultaneously and by summarizing the fluorescence lifetime characteristics into broad groups, the potential granularity of the information may be lost. In addition, the depth where the fluorescence originates cannot be determined with current implementations.

Fig. 11.

Fluorescence lifetime imaging ophthalmoscopy (FLIO) from the short (498–560 nm, SSC) and the long (560–720 nm, LSC) spectral channel as well as blue-light fundus autofluorescence intensity images of two healthy eyes and two eyes with nonexudative AMD. Copyright: (Patterns of Fundus Autofluorescence Lifetimes In Eyes of Individuals With Nonexudative Age-Related Macular Degeneration. Sauer L, Gensure RH, Andersen KM, Kreilkamp L, Hageman GS, Hammer M, Bernstein PS. Invest Ophthalmol Vis Sci. 2018 Mar 20; 59 (4):AMD65-AMD77.).

6. Clinical applications

Two important aspects should be considered when applying FAF imaging in the clinical setting. Compared to OCT, the spectrum of application for FAF is smaller. Therefore, a prerequisite for using FAF imaging is knowing which situation warrants its use and may allow identification of key findings for clinical management. Second, FAF imaging should not be regarded as a stand-alone diagnostic tool but should always be used in conjunction with clinical presentation, and particularly with clinical findings by ophthalmoscopy and other diagnostic measures. Additional value is achieved by combining FAF with other high-resolution imaging modalities, so-called “multimodal imaging”. In this context, a typical example is the assessment of foveal sparing in geographic atrophy (Lindner et al., 2015). While blue-light excitation FAF imaging is usually very helpful for detecting atrophic areas, the use of blue FAF imaging alone to assess foveal sparing by atrophy can be challenging because of interference from macular pigment (see 6.1.2 Geographic Atrophy). In this case, the use of corresponding near-infrared-reflectance SLO images, green instead of blue-light excitation and/or OCT is particularly helpful.

6.1. Age-Related Macular Degeneration

6.1.1. Early and intermediate age-related macular degeneration

Early manifestations of AMD include focal hypo- and hyperpigmentation at the level of the RPE as well as drusen with extracellular material accumulating in the inner aspects of Bruch’s membrane (Bird, 1996). Increased FAF signal adjacent to drusen, which corresponds to focal hyperpigmentation and pigment figures on biomicroscopy, has been attributed to the presence of melanolipofuscin or changes in the metabolic activity of the RPE. Areas of hypopigmentation on colour photographs tend to be associated with a corresponding decreased FAF signal, suggesting the absence of RPE cells or degenerating RPE cells with reduced LF granule content (reviewed by (Schmitz-Valckenberg et al., 2009)). Drusen by themselves are not necessarily correlated with notable FAF changes (Bindewald et al., 2005a).

Overall, larger drusen are more frequently associated with significant FAF alterations than smaller ones, with the exception of basal laminar drusen. Increased FAF signals at the site of drusen may be typically associated with vitelliform material in the subretinal space, as detected by corresponding OCT imaging (Gobel et al., 2015).

Crystalline drusen, also known as refractile drusen, typically demonstrate a corresponding decreased FAF signal. These drusen may have a myriad of punctate refractile elements, the histologic correlate is assumed to be hydroxyapatite spherules that can be present in drusen (Suzuki et al., 2015). One explanation for the decreased FAF signal could be that the overlying degenerated RPE has lost its autofluorescent properties and blocks at the same time the visualization of the refractile hyperreflective foci. In case of increased levels at the site of crystalline deposits, filter leakage should also be considered, caused by the strong reflectivity of the deposits that might be not entirely blocked by the barrier filter in front of the detector (see 2.1 Origin of the Signal).

In contrast to drusen, reticular pseudodrusen (also called subretinal drusenoid deposits (SDD)) are located in the subretinal space and have a different composition compared to drusen. In FAF images, reticular pseudodrusen exhibit a characteristic reticular pattern, further subdivided into “dot,” “target,” and “ribbon” configurations (Alten et al., 2014; Arnold et al., 1995; Schmitz-Valckenberg et al., 2011a; Steinberg et al., 2015). They are mostly associated with AMD but may occur in the context of various retinal diseases (Wightman and Guymer, 2019).

The variability of the FAF phenotype of drusen in AMD contrasts with young patients with monogenic disorders in whom drusen typically autofluoresce brightly, presumably reflecting a distinctly different composition of the accumulating material from age-related drusen (von Ruckmann et al., 1997a, b).

Applying a multimodal imaging approach including spectral domain optical coherence tomography (SD-OCT) and FAF revealed that focal hyperreflectivity overlying drusen was most frequently spatially confined to increased FAF, while outer nuclear layer thinning and choroidal hyperreflectivity were associated with decreased FAF (Gobel et al., 2015; Terheyden et al., 2019). Gliem and coworkers, in contrast to their underlying hypothesis, demonstrated using quantitative autofluorescence that the qAF levels in patients with early and intermediate AMD are lower compared to age-matched controls (Gliem et al., 2016). Thus, despite localized lesions with increased autofluorescence, the overall LF levels in the RPE outside focal lesions might be rather sub-normal in eyes with AMD. In terms of fluorescence lifetimes, eyes with AMD were shown to exhibited overall elongated mean lifetimes using FLIO (Dysli et al., 2017a). Localized long fluorescence lifetimes were shown to correspond to intraretinal hyperreflective foci as seen by SD-OCT. Localized short fluorescence lifetimes were occasionally also observed and co-localized with deposits in the area of the photoreceptor outer segments. Later on, Sauer and coworkers demonstrated that FLIO detects a clear ring-shaped pattern of changes within the fundus, which appears to be AMD-associated (Fig. 11) (Sauer et al., 2018c). These changes are already visible in early AMD stages and not masked by the presence of other coexisting retinal diseases. These findings may be useful for the early diagnosis of AMD and to distinguish AMD from other retinal diseases.

6.1.2. Geographic atrophy

FAF imaging is one of the most reliable imaging modalities to detect, delineate, quantify, and monitor progression of outer retinal atrophy (Holz et al., 2017). The loss of RPE and its inherent fluorophores in GA correlates with well-defined areas of decreased autofluorescence, allowing for precise manual, semi-automatic or automatic GA segmentation methods based on FAF imaging (Figs. 7 and 12). Hereby, the semi-automatic region-growing image analysis approach has been integrated in the RegionFinder^™ software (Heidelberg Engineering). Schmitz-Valckenberg et al. demonstrated that the software allows for reproducible delineation of atrophy with high inter-rater agreement (Schmitz-Valckenberg et al., 2011b). A new update, first described by Lindner et al. allows for combined grading of FAF and near-infrared reflectance confocal SLO images to differentiate between foveal atrophy and foveal sparing (Fig. 7) (Lindner et al., 2015). In eyes with foveal sparing GA, Lindner et al. were able to demonstrate that centrifugal progression of perifoveal outer retinal atrophy is significantly faster than centripetal progression in the context of AMD. While the underlying pathogenic mechanisms for differential GA progression remain unknown, local factors may be operative that appear to protect the foveal retina. The RegionFinder^™ software has been used to quantify the primary outcome measure in (ongoing) clinical trials (NCT02247531, NCT02247479, NCT02087085, http://clinicaltrials.gov). In this context, FAF imaging has been recommended among other imaging modalities for the detection and measurement of atrophy by the CAM group (Classification of Atrophy Meeting) (Holz et al., 2017; Sadda et al., 2018). Later, Pfau et al. demonstrated that green-light FAF imaging provides even better inter-rater agreement suggesting that its use may be an alternative in clinical trials examining the change in lesion size as a clinical endpoint (Fig. 7) (Pfau et al., 2017). It must be noted, that very early atrophic lesions in presence of drusen (nascent geographic atrophy) may be characterized by both increased and decreased autofluorescence impeding the quantification of these small lesions (Wu et al., 2014). Recently, the CAM group developed a new classification system of early atrophy manifestation in which OCT imaging is considered the base imaging modality, while other imaging modalities including FAF may support findings made by OCT (Guymer et al., 2019; Sadda et al., 2018).

Fig. 12.

Semi-automated assessment of atrophic lesion size over two year with 6 month intervals (from left to right), showing at each time point colour fundus photography (first row), fundus autofluorescence (second row), annotated lesion boundaries based on fundus autofluorescence (third row), actual absolute area size for each individual lesion and the total size (fourth row). Note individual lesion expansion with increasing number of spots, coalescence of existing spots and overall increase of total lesion size.

A striking finding of FAF imaging in GA patients is the frequent presence of areas of relative increased FAF in the junctional zone surrounding the patch of atrophy (Holz et al., 2001). Distinct patterns of abnormal FAF in the junctional zone of atrophy and a high degree of intraindividual symmetry between fellow eyes have been described (reviewed (Schmitz-Valckenberg et al., 2009)). A classification system of FAF patterns in the junctional zone of atrophy in GA patients has been proposed (Bindewald et al., 2005b). Studies of retinal sensitivity have underscored the importance of increased FAF surrounding areas of GA (Schmitz-Valckenberg et al., 2004; Scholl et al., 2004a). Applying so-called patient-tailored fundus-controlled perimetry grid (also called “microperimetry”), Pfau et al. introduced an innovative approach to maximize the number of test points in disease relevant regions while – at the same time – limiting the overall number of test points (Pfau et al., 2018). The results demonstrate a relative decrease of retinal sensitivity at the border zone of atrophy, continuously increasing away from the atrophic edge.

Outer retinal atrophy in the context of AMD is a dynamic process with gradual enlargement of atrophic areas over time. Initial natural history studies on atrophy progression in GA patients using FAF imaging demonstrated the occurrence of new atrophic patches and the spread of preexisting atrophy in areas with abnormally high levels of FAF at baseline (Fig. 13) (Holz et al., 2001). Looking at larger patient groups with longer review periods, the significance of increased junctional FAF for foreshadowing atrophy enlargement has been demonstrated (Holz et al., 2007a; Schmitz-Valckenberg et al., 2006, 2016). Overall, GA progression rates reported in the literature for total study populations ranging from 0.53 to 2.6 mm²/year (median, ~1.78 mm²/year), assessed primarily by colour fundus photography or FAF imaging. Several factors that could inform an individual’s disease prognosis have been replicated in multiple cohorts: baseline lesion size, lesion location, multifocality, fellow eye status, and patterns of changes in FAF in the junctional zone of GA (reviewed by (Fleckenstein et al., 2018)). These findings underscore the importance of abnormal FAF intensities around atrophy.

Fig. 13.

Monitoring of atrophic progression over time with fundus autofluorescence imaging, showing the natural course of the disease over 11 years. Enlargement of existing and occurrence of new atrophic spots is observed in areas with previously high levels of fundus autofluorescence.

6.1.3. Macular neovascularization

Theoretical considerations would suggest that FAF imaging may provide important clues to our understanding of macular neovascularization secondary to AMD. For example, it may be helpful to assess the integrity of the RPE which may influence the development and behavior of new vascular complexes as well as photoreceptor viability and potential therapeutic success. In the presence of exudation and neovascularization, the FAF signal may be also masked by haemorrhages or attenuated by fluid (Signal attenuation by fluid is usually not observed using fundus camera systems with green-light excitation).

Patients with early neovascular membranes secondary to AMD tend to have patches of “continuous” or “normal” autofluorescence corresponding to areas of hyperfluorescence on the companion fluorescein angiograms, implying that RPE viability is preserved at least initially in neovascular development (McBain et al., 2007; Vaclavik et al., 2006). By contrast, eyes with longstanding macular neovascularization typically exhibit more areas of decreased FAF signal, which could be explained by scar formation with increased melanin deposition, loss of the RPE, and potentially through epithelial mesenchymal transition as part of what is called fibrosis. Heimes and coworkers analyzed the prognostic value of RPE autofluorescence with respect to the therapeutic outcome of anti-vascular endothelial growth factor therapy in exudative AMD (Heimes et al., 2008). The analysis of 95 eyes showed a significant difference in visual acuity outcomes in eyes with changes in FAF within the central 500 and 1000 μm.

One other important finding in eyes with macular neovascularization is that abnormal FAF intensities typically extend beyond the edge of the angiographically defined lesion, indicating a more widespread involvement than is apparent from conventional imaging studies. Increased FAF signal has also been described around the edge of lesions. It has been speculated that this observation may reflect the proliferation of RPE cells around the neovascular membrane (Sawa et al., 2006b). As in other exudative retinal disease, such as central serous chorioretinopathy, areas with increased FAF are commonly found inferior to the leakage on fluorescein angiography, most likely representing gravitational effects of fluid tracking in conjunction with partial loss of photoreceptor segments resulting in less photopigments (Spaide, 2003, 2008; Spaide and Klancnik, 2005). In contrast to fluid, haemorrhages and intraretinal exudates typically show a decreased FAF signal because of light absorption obscuring the underlying retinal details. When retinal haemorrhages undergo organization and evolve into an ochre colour on fundoscopy, they may become intensely autofluorescent (Fig. 14) (Sawa et al., 2006b). Later, with disappearance of the yellowish material seen on biomicroscopy, a large RPE scar and atrophy with decreased autofluorescence may be visible (Heimes et al., 2008). Strong pre-injection fluorescence using indocyanine green angiography (i.e. NIR-FAF) in subretinal haemorrhages in which the colour of blood was no longer red, but rather greenish, gray or black, has been known for decades. This NIR-FAF signal has also been explained by degradation products of hemoglobin (porphyrins) (Piccolino et al., 1996).

Fig. 14.

Evolution of pre-retinal foveal haemorrhage caused by Valsalva-maneuver at first presentation (baseline, first column), at three weeks (second column) and at 10 months (third column). At each timepoint, colour fundus photography (first row), representative B-scan through the foveal center (second row) and fundus autofluorescence (FAF) image (third row) are shown, respectively. Note the initial several decreased signal by FAF which turns into increased levels over time.

Detection of atrophy in eyes with exudative AMD is becoming more important since there is increasing evidence that atrophy development in eyes treated with anti-VEGF therapy is an important cause for severe visual loss in the long term (Bhisitkul et al., 2015). FAF imaging herein represents an important tool, although, delineation of “pure” atrophy by FAF imaging is challenging, since retinal changes associated with exudation also exhibit a decreased FAF signal (Kuehlewein et al., 2016; Kumar et al., 2013). Hence, a multimodal imaging approach appears to be most appropriate to detect, evaluate, and quantify atrophy in eyes with exudative AMD.

Multimodal testing of eyes with macular neovascularization showed that the area of confluent decreased FAF and involvement of the foveal center with confluent decreased FAF were independent predictors of visual acuity. Contrast sensitivity was inversely associated with the size of confluent decreased FAF as was reading speed (Sato et al., 2015). In a follow-up study of 116 patients, confluent loss of autofluorescence was seen in 58.6% of eyes; this variable was a significant predictor of visual acuity. Over a mean of 2.9 years of follow-up, more eyes developed confluent decreased FAF and those with absent autofluorescence had expansion of the areas involved. The best predictor of final visual acuity was the area of absent autofluorescence at the final follow-up. The change in visual acuity over the follow-up was inversely correlated with the area change of absent autofluorescence. These findings suggest application of a treatment strategy to limit RPE cell loss may prove useful in eyes with neovascular AMD (Kumar et al., 2013).

6.1.4. Retinal pigment epithelial tears

Vascularized pigment epithelial detachments (PED), a subtype of neovascular AMD, may be complicated spontaneously or following photodynamic therapy or anti-VEGF therapy by tears of the RPE. In FAF imaging, an RPE tear is characterized by well-demarcated decreased autofluorescence due to absence of RPE, with an adjacent region of increased autofluorescence corresponding to the retracted RPE. Interestingly, reappearance of the fluorescence within the area of the tear has been observed, which might potentially represent repopulation of the area of “RPE resurfacing” (Chuang and Bird, 1988; Orini et al., 2012; Sarraf et al., 2014). This increased autofluorescence appears when the RPE cells are attached again to photoreceptors and start to build up LF material. This signal does not correspond to a decreased visualization of denudated choroidal vessel, since these RPE cells do not have melanin granules (Orini et al., 2012).

Finally, atypical RPE defects with retained photoreceptor layers have been described in addition to geographic atrophy and characteristics RPE tears (Giannakaki-Zimmermann et al., 2017). These eyes appear to have reasonable visual function and survival of photoreceptor layer integrity over time. It has been speculated that repair mechanisms such as ingrowth of the RPE/drusenoid material and persistent subretinal fluid or RPE-independent visual cycle for cone photopigment within the neurosensory retina may contribute to the preservation of photoreceptors and function.

6.2. Macular and diffuse retinal dystrophies

In macular and diffuse retinal dystrophies, various associated abnormalities in FAF have been described (reviewed by (Birtel et al., 2018; Boon et al., 2008; von Ruckmann et al., 1997b)). The extent and pattern of abnormal FAF may show characteristic abnormal distributions in retinal dystrophy disease entities and therefore aids in the differential diagnosis (Figs. 8 and 15). In particular, in late-onset macular dystrophies (e.g., late-onset Stargardt disease and central areolar choroidal dystrophy), FAF imaging is an important technique to differentiate such masquerading disease entities from AMD (Saksens et al., 2014).⁶⁹ Moreover, FAF imaging has recently become an important outcome measure in clinical trials of retinal dystrophies (Safety and Efficacy of Emixustat in Stargardt Disease NCT03772665, Choroideremia Gene Therapy Clinical Trial NCT02553135). It is well established that autofluorescent material excessively accumulates in the RPE in association with various genetically determined retinal diseases.

Fig. 15.

Detection and longitudinal monitoring of disease manifestation and progression by wide-field imaging, showing at each column pseudocolour (first row) and fundus autofluorescence (second row) non-confocal scanning laser ophthalmoscopy. The third and fourth row show the central area and one peripheral area at high magnification as outlined in the second row, respectively. In a 67-year-old female with sectorial rod-dystrophy, both central and peripheral abnormalities and their extent are clearly visible by fundus autofluorescence at baseline (A–D). After 3.3 year of follow-up, progression of both central and peripheral changes is shown (E–H). In a 19-year-old female patient with unilateral pigmented paravenous retinochoroidal atrophy, no changes in both central and peripheral lesion morphology over time are detected (Baseline I-L, follow-up after 4.8 years M-P).

6.2.1. Stargardt disease

In recessive Stargardt disease, bisretinoid production is accelerated in photoreceptor outer segments due to dysfunctioning of the ATP-binding cassette transporter (ABCA4) in outer segments (Ahn et al., 2000; Illing et al., 1997; Molday et al., 2000; Papermaster et al., 1978; Quazi and Molday, 2014; Sun et al., 1999; Sun and Nathans, 1997). The function of ABCA4 is to deliver retinaldehyde across the lipid bilayer so as to aid in the detoxification of the aldehyde by reduction to all-trans-retinol. Deficiency in ABCA4 enables elevated bisretinoid formation and thus higher intensities of short-wavelength FAF (Burke et al., 2014). In the animal model, accumulation of fluorophores can be reduced by reduction of serum retinol and (Radu et al., 2005). Charbel Issa et al. demonstrated that FAF imaging can be applied in the animal model to monitor lipofuscin accumulation and melanin-related changes by using both blue and NIR FAF imaging (Charbel Issa et al., 2013a). Indeed, studies of patients diagnosed with ABCA4 mutations and Abca4−/− mice have shown that bisretinoid lipofuscin at elevated levels can modulate NIR-FAF signal from melanin (Paavo et al., 2018). While it has not been demonstrated that histological changes in the RPE correspond spatially to flecks in short-wavelength FAF images, it was shown that in NIR-FAF images flecks are commonly hypoautofluorescent, the NIR-FAF signal originating primarily in RPE melanin (Fig. 9) (Sparrow et al., 2015). This observation is consistent with the finding that the increased FAF signal in Stargardt macular dystrophy and fundus flavimaculatus fades over time, with subsequent atrophy development. Sparrow and coworkers further evaluated this incongruous observation, and in 2015 reported that in NIR-FAF, flecks are predominantly less autofluorescent and larger, and that NIR-FAF darkening occurs prior to development of an increased FAF signal (Sparrow et al., 2015). They concluded that these observations indicate that RPE cells associated with flecks in recessive Stargardt disease are considerably changed or lost, and that the bright FAF signal of flecks likely originates from augmented LF formation in degenerating photoreceptor cells impaired by the failure of RPE. Recently, Fang et al. demonstrated that lipofuscin and melanolipfofuscin granules contributed to NIR FAF in the mouse model. The flecks which showed increased intensities in both blue and NIR FAF imaging corresponded to subretinal macrophages fully packed with pigment granules (Fang et al., 2020). Application of FLIO has demonstrated that recent onset flecks in Stargardt disease display short fluorescence lifetimes and convert into longer fluorescence lifetime flecks over time (Solberg et al., 2019a). This transition may represent a change in the composition of retinal deposits with accumulation of LF and retinoid by-products from the visual cycle.

In regards to monitoring disease progression, the ProgStar study group has demonstrated that FAF imaging may serve as a reliable tool and definitely decreased autofluorescence and total area as outcome measures for interventional clinical trials that aim to slow disease progression in patients with Stargardt disease (Strauss et al., 2019). Moreover, lesion-shape descriptors as assessed by FAF imaging were shown to be prognostic of future disease progression rates in Stargardt disease (Muller et al., 2020).

6.2.2. Best disease

The ophthalmoscopically visible pale/yellowish lesions in Best macular dystrophy and adult vitelliform macular dystrophy are also associated with an intense locally increased FAF signal (Duncker et al., 2014a). This FAF may originate in impaired outer segments that in SD-OCT scans can be seen to project posteriorly into the domain-shaped lesion.

6.2.3. Choroideremia

Choroideremia (CHM), a rare hereditary retinal disease due to a mutation of the CHM gene located on the X chromosome, is characterized by an onset of night blindness during the first decade followed by progressive loss of peripheral vision (Dimopoulos et al., 2017). CHM may result in total blindness when all remaining photoreceptors-islands have degenerated (Zinkernagel and MacLaren, 2015), the remaining islands with preserved RPE appear as regions of normal or increased autofluorescence intensity due to LF accumulation, whereas areas with RPE atrophy are characterized by decreased autofluorescence. FAF imaging is especially of interest in CHM, since it allows for reproducible quantification of the residual autofluorescence and represents therefore a useful outcome measure (Hariri et al., 2019; Jolly et al., 2016). In some patients, a distinct phenotype with areas of central smooth and more peripherally mottled regions of FAF intensities surrounded by peripheral areas of severely decreased FAF intensities has been described (Stevanovic et al., 2020). As the mottled areas with irregular FAF intensities were also associated with more ellipsoid zone disruption as seen by OCT, Stevanovic et al. concluded that these RPE changes reflected more advanced stages of the disease. Dysli et al. demonstrated that autofluorescence lifetimes may additionally identify areas of remaining photoreceptors even in the absence of the RPE (Dysli et al., 2016b). Thus, the state of photoreceptors in patients with CHM may be assessed using FLIO, in addition to the state of the residual RPE.

In studies combining blue FAF with NIR-FAF and SD-OCT it was revealed that in CHM-affected males, the blue FAF signal could coexist with absent or pronounced reduction of NIR-FAF and hypertransmission of SD-OCT signal into the choroid suggesting that the blue FAF emission originated in impaired photoreceptor outer segments. At other locations no blue FAF signal was observed yet the melanin signal conveyed by NIR-FAF, confirmed the presence of RPE. Importantly, reduced blue FAF intensities detected by qAF analysis were also exhibited by female carriers of CHM (Paavo et al., 2019).

6.2.4. Other macular and retinal dystrophies

Certain retinal degenerations may also exhibit an overall low FAF signal due to an enzymatic dysfunction of the visual cycle (e.g., RPE65 Leber congenital amaurosis) or a barrier at the level of Bruch’s membrane impeding the delivery of retinoids (e.g., Sorsby fundus dystrophy). Lorenz and coworkers described absent or minimal FAF intensities in patients with early-onset severe retinal dystrophy associated with mutations on both alleles of RPE65 (Lorenz et al., 2004). The lack or severe decrease of FAF signal would be consistent with the biochemical defect and could be used as a clinical marker of this genotype. Another study demonstrated that patients with Leber congenital amaurosis having vision reduced to light perception and undetectable electroretinograms (ERGs) may still exhibit normal or minimally decreased FAF intensities (Scholl et al., 2004b). The authors concluded that the RPE–photoreceptor complex is, at least in part, functionally and anatomically intact. This finding would have implications for future treatment, suggesting that photoreceptor function may still be rescuable in such patients. Examples of monogenic diseases with a reduced overall FAF signal in association with pathological alterations of Bruch’s membrane are Sorsby fundus dystrophy and pseudoxanthoma elasticum (Gliem et al., 2017).

Discrete, well-defined lines of increased FAF may occur in various forms of retinal dystrophies (Fig. 15) (Fleckenstein et al., 2009; Robson et al., 2007; Von Rückmann et al., 2007). These lines have no prominent correlate on fundus biomicroscopy, although there is evidence that these lines precisely reflect the border of the regions of retinal dysfunction (Fleckenstein et al., 2009; Popovic et al., 2005; Robson et al., 2007). Despite the variable orientation of this line in different entities, e.g., orientation along the retinal veins in pigmented paravenous chorioretinal atrophy (PPCRA) or as a ring-like structure in retinitis pigmentosa (RP) or macular dystrophies, the similar appearance on FAF images and the concordance of functional findings indicate that these lines in heterogeneous diseases share a common underlying pathophysiologic mechanism. Fleckenstein and coworkers first described the SD-OCT correlate of these lines of increased FAF (Fleckenstein et al., 2008). Specifically, these corresponded with a discrete junctional zone between an area with preserved OCT layers and an area where the outer aspects of the retina are lost and the external limiting membrane band appeared to rest directly on the RPE. The same SD-OCT correlate has been demonstrated in patients with RP (Lima et al., 2009). This zone might be characterized by progressively altered photoreceptor outer and inner segments. While the pathophysiologic mechanism is unknown, it may be hypothesized that, since the fluorophores responsible for FAF form in photoreceptor cells, the increased FAF signal observed in various retinal dystrophies at sites of photoreceptor cell impairment reflects dysfunctional photoreceptor cells. For instance, one of the energy-demanding processes with which photoreceptor cells are burdened is the detoxification of all-trans-retinaldehyde by NADPH-dependent chemical reduction to all-trans-retinol by retinol dehydrogenase (RDH) activity. Insufficient clearance of all-trans-retinal leads to bisretinoid formation that would also explain the elevated FAF. Changes in absorption of the FAF signal due to loss of photoreceptor outer segments may also contribute to this phenomenon (Schuerch et al., 2017). In the zone with a normal FAF signal but impaired retinal sensitivity, the structure of the photoreceptors seems to be severely distorted. A normal FAF signal, therefore, does not necessarily reflect an intact photoreceptor–RPE complex, but may rather correspond to a structurally intact-appearing RPE cell monolayer with or without the presence of intact photoreceptors.

QAF imaging has been shown to be useful for differential diagnosis in retinal dystrophies, e.g., qAF allows for differentiation between ABCA4-associated and non-ABCA4-associated retinal disease (Duncker et al., 2015c). Moreover, PRPH2/RDS- and ABCA4-associated disease exhibiting phenotypic overlap may be partially discriminated when qAF values are corrected for age and race. In general, ABCA4 patients have been shown to exhibit higher qAF values than PRPH2/RDS patients, while most patients without mutations in PRPH2/RDS or ABCA4 have qAF levels within the normal range (Duncker et al., 2015c; Gliem et al., 2020). Interestingly, monoallelic ABCA4 subjects (i.e. genotyped parents of biallelic ABC4 patients) were phenotypically normal regarding qAF levels, while biochemical analysis in the mouse model revealed that one mutation in ABCA4 already resulted in minor increased levels of bisretinoids (as compared to normal levels in wildtype and excessive levels in ABCA4^−/− mice) (Müller et al., 2015). While high qAF levels represent a hallmark of ABCA4-related disease in the absence of atrophy, patients with mutations in BEST1 exhibit mean non-lesion qAF values that are within normal limits for age (Duncker et al., 2014a). Therefore, it has been concluded that, based on qAF, mutations in BEST1 do not cause increased LF levels outside the yellowish, focal lesions. A bull’s eye phenotype can be associated not only with ABCA4-associated disease but also due to mutations in other photoreceptor-associated genes such as RPGR (retinitis pigmentosa GTPase regulator) and PROM1 (Prominin 1). It was found that even when qualitative features of fundus autofluorescence and SD-OCT images did not serve to distinguish ABCA4-positive versus ABCA4-negative patients, ABCA4-positive patients exhibiting a bull’s eye phenotype had higher qAF levels than ABCA4-negative patients (Duncker et al., 2015b). Similarly, patients carrying ABCA4-mutations can present with a pattern dystrophy phenotype similar to that associated with mutations in PRPH2/RDS-associated disease. Again, the presence of PRPH2/RDS mutations qAF values were shown to be lower than in the group of patients with pattern dystrophy-like ABCA4-related disease (Duncker et al., 2015d).

6.3. Macular telangiectasia type 2

Macular telangiectasia type 2 (MacTel) is a bilateral disease of unknown cause with characteristic alterations of the macular capillary network and progressive retinal cell death (Fig. 16) (Charbel Issa et al., 2013b; Gass and Blodi, 1993). The disease typically initially manifests temporal to the fovea, and may later encompass an oval-shaped area centered on the foveola.

Fig. 16.

Multimodal imaging of the right eye of a 60-year-old male with macular telangiectasia type 2 (MacTel), showing colour fundus photography (A), blue-light fundus autofluorescence (B), late-phase fluorescein angiography (C) and spectral-domain optical coherence tomography B-scan through the foveal center (D). Note that there is no blocked fundus autofluorescence signal in the fovea due to lack of macula pigment. Instead abnormally increased intensities are detected in the fovea and particularly temporal to the fovea.

As outlined above (2.4 Macular Pigment), normal eyes show masking of the foveal 488-nm blue-light FAF due to the distribution of luteal pigment. Reduced macular pigment density in MacTel reduces this masking. Eyes with MacTel show an abnormally increased signal in the macular area to a variable degree with blue-light FAF imaging (Helb et al., 2008; Schmitz-Valckenberg et al., 2008a). A loss of luteal pigment may initially occur in the area temporal to the foveal center (Fig. 16) (Charbel Issa et al., 2009). Quantitative analysis confirmed that the loss of luteal pigment was more pronounced in the temporal compared with the nasal parafoveolar area and suggested that zeaxanthin would be more reduced than lutein.

Further proof for the depletion in macular pigment derived from a postmortem analysis of an eye with MacTel (Powner et al., 2010). Macroscopic examination disclosed the absence of the central yellowish spot. A yellow ring of residual macular pigment was present eccentrically in accordance with the in vivo imaging observations. The loss of macular pigment was subsequently divided into three classes based on a cross-sectional analysis of two-wavelength (blue-light and green-light) FAF images (Zeimer et al., 2010): class 1 shows a wedge-shaped loss of macular pigment restricted to an area temporal to the foveal center. In class 2, the area is larger and also involves the foveal center. Class 3 is characterized by loss of luteal pigment within an oval-shaped area centered on the foveola. There was a significant association of these three classes of macular pigment loss with the consecutive disease stages of MacTel described by Gass and Blodi (1993). Correlation studies with microperimetric data revealed a trend towards worse retinal function with increasing class of macular pigment changes. A study by Muller et al. demonstrated that the progressive loss of the ellipsoid zone, a typical phenomenon in MacTel, is confined to the area of reduced macular pigment optical density (Muller et al., 2018b).

Another study succeeded in visualizing even earlier pathologic changes in fellow eyes which otherwise did not meet conventional diagnostic criteria for MacTel (Charbel Issa et al., 2016). While no functional deficits were detected, eyes consistently showed a severely reduced directional cone reflectance (Stiles–Crawford effect). An additional consistent minimal disease manifestation was an asymmetric configuration of the foveal pit with focal temporal thinning, which was most pronounced at 1° eccentricity. Topographically related, macular pigment optical density was reduced in a small wedge-shaped temporal paracentral sector, resulting in an increased signal on fluorescein angiography and FAF imaging. These described alterations may be helpful for early identification of patients and affected family members.

Sauer and colleagues more recently demonstrated that FLIO detects retinal changes in patients with MacTel with high contrast, presenting a distinctive signature that is a characteristic finding of the disease (Sauer et al., 2018d). The non-invasive properties of this novel imaging modality provide a valuable addition to clinical assessment of early changes in the disease that could lead to more accurate diagnosis of MacTel.

6.4. Pseudoxanthoma elasticum

Pseudoxanthoma elasticum (PXE) is caused by a mutation in the ABCC6 gene (Germain, 2017). More than 300 distinct loss-of-function mutations representative of over 1000 mutant alleles in ABCC6 have been found. Many of the missense mutations occur at locations in the protein involving domain–domain interactions in the ABCC6 transporter. Even heterozygotes can show manifestations of disease. FAF abnormalities are common in eyes affected by PXE (Finger et al., 2009; Gliem et al., 2017; Sawa et al., 2006a; Spaide and Jonas, 2015; Zweifel et al., 2011). Typical phenotypic alterations, including angioid streaks and drusen of the optic nerve, have autofluorescence correlates. Peau d’orange is hardly detectable on FAF, whereas comet-tail lesions are typically apparent. RPE atrophy can be widespread and heterogeneous, located mostly adjacent to angioid streaks or neovascular membranes. Furthermore, irregular patterns of increased FAF at the posterior pole with an appearance similar to that of pattern dystrophies can be found in eyes with PXE. In these eyes, areas of yellowish deposits and hyperpigmentation on colour photography corresponded to areas of increased FAF. Agarwal et al. suggested the following classification: a fundus appearance similar to a pattern dystrophy of the fundus flavimaculatus, the reticular, the vitelliform and the fundus pulverulentus types, respectively (Agarwal et al., 2005). The pattern dystrophy-like changes in PXE may occur unilaterally or bilaterally. Eyes with PXE commonly have collections of subretinal fluid, which precede the pattern dystrophy findings by at least 1 year. The pattern dystrophy findings precede the development of atrophy or macular neovascularization (Zweifel et al., 2011).

Abnormalities of the RPE–photoreceptor complex detected by FAF imaging are more diverse and widespread than expected from conventional fundus imaging. Such extensive alteration of the RPE suggests an important role of pathologic RPE changes in the evolution of visual loss in PXE. Ocular alterations related to PXE typically begin at the central fundus and then spread centrifugally, resulting in the most pronounced phenotype in the papillomacular area with the least abnormalities in the periphery (Charbel Issa et al., 2010; Spaide, 2015). This process also allows the observation of different disease stages from the periphery to the central fundus. Interestingly, similar to observations made in AMD eyes, reticular pseudodrusen (also called subretinal drusenoid deposits (SDD)) have been described in about 50% of PXE eyes by multimodal high-resolution imaging, including using FAF imaging (Gliem et al., 2015a). The appearance of peau d’orange, which marks the transition from calcified to uncalcified Bruch’s membrane, is localized peripheral to areas of reticular pseudodrusen. A similar topographic relationship was identified for the transition zone between normal and centrally reduced late-phase indocyanine green angiography fluorescence, the latter being a result from reduced indocyanine green staining of altered Bruch’s membrane and/or RPE. Thus, specific PXE-related fundus changes appear to be localized eccentric to the reticular pseudodrusen and precede their development (Gliem et al., 2015b). Recently, an acute retinopathy similar to Multiple Evanescent White Dot Syndrome has been described in the context of PXE, characterized by lesions with an increased FAF signal, which may fade with bleaching of the surrounding retina (see 6.8 Uveitis and White Dot Syndromes) (Gliem et al., 2019). OCT angiography has shown that patients with PXE have marked choriocapillaris signal voids suggestive of poor choriocapillaris perfusion (Gliem et al., 2015a; Zweifel et al., 2011). Of interest, reticular pseudodrusen are also associated with poor choriocapillaris flow indices (Spaide, 2016).

6.5. Central serous chorioretinopathy

Central serous chorioretinopathy (CSC) is a condition characterized by idiopathic leaks at the level of the RPE leading to serous pigment epithelial and neurosensory retinal detachments (van Rijssen et al., 2019). In the early phases of the disease, visual acuity may be good despite the presence of the macular detachment, and after resolution visual acuity often shows improvement. More chronic forms of CSC are associated with atrophic and degenerative changes of the retina and RPE and consequently with visual acuity decline (Imamura et al., 2011; Spaide and Klancnik, 2005).

Accordingly, FAF findings in CSC are dependent on the extent of involvement of the RPE and the stage of the disease. Patients with acute leaks imaged within the first month have minimal abnormalities (Fig. 17). A slight decrease in autofluorescence in the center part of the area of the serous retinal detachment may be typically observed (due masking effects of fluid). At the edge of the lesion, a slight increased FAF signal may be additionally detected. Compared to the center of the subsensory elevation, only little fluid is present at the edge, while the effects of photopigment loss become prominent. The loss of photopigment results in less absorption of the excitation light and thus the detection of an increased FAF signal.

Fig. 17.

Fundus autofluorescence (FAF) findings (A) in comparison to optical coherence tomography (OCT) B-scan in the early manifestation of central serous chorioretinopathy using confocal blue FAF imaging. Note the slight decreased of the FAF intensities in the center of the subsensory retinal detachment due to masking effects of fluid. At the edge (arrow heads) where no substantial fluid is present, the FAF signal is slightly increased due to reduced filter effects of photopigments in photoceptors outer segments. By fundus camera based FAF imaging using green excitation light, levels of strongly increased FAF inside the typical detachment (history of two months) are detected (C). Stereo imaging shows the signal originates from the outer portions of the detached retina. OCT imaging shows an accumulation of material that seems to represent photoreceptor outer segments (Arrow, D). This accumulation is referred to as “shaggy photoreceptors”.

Over time, the area of the detachment increasingly exhibits more irregular increased autofluorescence. In some patients, discrete granules with increased intensity within the detachment are observed which correspond with the pinpoint subretinal precipitates seen on ophthalmoscopy. It has been suggested that these dots may represent macrophages engorged with phagocytosed outer segments. Patients with chronic disease have irregular levels of autofluorescence with markedly decreased intensity over areas of atrophy. A typical finding also includes the visualization of fluid tracks in the inferior retina along with a reduction of outer retinal thickness (i.e. loss of photopigments).

The area of the leak also undergoes change in autofluorescence over time. Soon after the development of CSC, little or no change in the autofluorescence pattern in the area around the leak is seen, although the leak site may be somewhat less autofluorescent compared to the normal background signal. Patients with more chronic leaks can have decreased autofluorescence surrounding the known leaks. This area of decreased FAF appears to expand in size with increasing chronicity of the leak. Altered FAF in chronic CSC patients has been shown to have a functional correlation quantified by microperimetry. This study underlines the impact of FAF changes on retinal sensitivity and their value to reflect the functional impairment in chronic CSC (Eandi et al., 2015).

The FAF abnormalities in central serous chorioretinopathy show multiple patterns and are related with the chronicity and visual acuity. FAF patterns in central serous chorioretinopathy are helpful when considering the timing of treatment and predicting the disease status (Han et al., 2019). Interestingly, NIR-FAF tends to show more widespread abnormalities as compared to short-wavelength autofluorescence in central serous chorioretinopathy (Zhang et al., 2015). Ultra-widefield FAF and indocyanine green angiography may reveal peripheral areas of previous or ongoing choroidal hyperpermeability and thereby assist in the diagnosis of CSC (Pang et al., 2014).

6.6. Chloroquine and hydroxychloroquine retinopathy

The use of the anti-inflammatory agent hydroxychloroquine will increase most likely following publication of the LUMINA study, which demonstrated a clear survival benefit of treated patients with systemic lupus erythematosus (Alarcon et al., 2007). Since chloroquine/hydroxychloroquine may also impair the replication of several viruses (including flaviviruses, retroviruses, and coronaviruses), the drug may experience an intermittent revivals in the context of viral diseases (Savarino et al., 2003). However, chloroquine (CQ) and hydroxychloroquine (HCQ) may lead to retinopathy, especially in patients taking a daily HCQ dose greater than 5.0 mg/kg real weight (Marmor et al., 2011). Besides a high daily and cumulative dose, concomitant renal disease and/or the use of tamoxifen represent major risk factors (Marmor et al., 2016). FAF imaging may allow for early detection of CQ/HCQ retinopathy showing a parafoveal ring of increased autofluorescence corresponding to photoreceptor damage (Kellner et al., 2006; Rickmann et al., 2019). As the disease progresses, parafoveally decreased autofluorescence due to RPE atrophy (bull’s eye maculopathy) becomes apparent (Fig. 18). Besides SD-OCT and automated perimetry (10–2), FAF imaging and multifocal electroretinography were recommended as “additional useful screening tests” in the current American Academy of Ophthalmology recommendations on screening of CQ/HCQ retinopathy (Marmor et al., 2016). Noteworthy, patients of Asian heritage may show early damage in a more peripheral pattern. Recent data suggest that FLIO may show even earlier signs of CQ/HCQ retinopathy (Sauer et al., 2019; Solberg et al., 2019b). However, longitudinal evidence regarding this observation is currently lacking.

Fig. 18.

Multimodal imaging of the left eye of a 79-year-old male with hydroxychloroquine maculopathy, showing colour fundus photography (A), blue-light fundus autofluorescence (B) as well as combined near-infrared reflectance and spectral-domain optical coherence tomography (SD-OCT) B-scan through the foveal (C). Note the parafoveal decreased autofluorescence in a bull-shaped pattern, indicating atrophy and corresponding to severe outer retinal thinning and choroidal hypertransmission by SD-OCT. In addition, an epiretinal membrane is present causing a distorted foveal appearance.

6.7. Other toxic retinopathies

Didanosine, a nucleoside reverse transcriptase inhibitor for HIV treatment, inhibits the synthesis of mitochondrial DNA. Didanosine-induced retinal toxicity mirrors features of mitochondrial disorders with foveal sparing and patches decreased autofluorescence corresponding to atrophy in the midperiphery (Gabrielian et al., 2013). Deferoxamine-induced retinal toxicity, which is caused by a chelator used to treat iron overload, may show a variety of funduscopic manifestations including pigmentary abnormalities, vitelliform lesions, and bull’s eye maculopathy. In a prospective study, it was demonstrated, that changes in FAF imaging were more apparent than in fundus photography (Viola et al., 2012). Recently, chronic exposure to pentosan polysulfate sodium (PPS), a therapy for interstitial cystitis (IC), has been linked to a possibly avoidable maculopathy (Pearce et al., 2018). This maculopathy exhibits a rather distinct autofluorescence phenotype with areas of decreased FAF (RPE-atrophy) surrounded by a reticular pattern of increased FAF changes.

6.8. Uveitis and White Dot Syndromes

FAF has been shown to be useful in the diagnosis and monitoring of inflammatory diseases (Yeh et al., 2010). Compared to conventional colour photography, FAF imaging allows for precise observation of RPE changes reflecting choroidal and retinal alterations and highlighting areas of disease activity (reviewed by (Deak et al., 2020)). Particularly widefield FAF images may reveal larger areas of involvement in posterior uveitis and more clearly delineated areas of pathology compared with colour fundus photos (Reznicek et al., 2014).

Exemplarily, FAF abnormalities in eyes with Birdshot chorioretinitis, are widespread, and the relationship between these abnormalities and intraocular inflammation suggests that the RPE has a role in disease pathophysiology (Koizumi et al., 2008). Böni and colleagues demonstrated that FAF abnormalities in Birdshot chorioretinitis are markers of vision dysfunction. They further found some evidence to support the assumption that FAF abnormalities progress from granular decreased FAF to confluent decreased FAF and possibly that disease progresses centripetally toward the fovea, raising the possibility that FAF imaging can be used clinically to monitor patients for disease progression during treatment (Boni et al., 2017).

In Vogt-Koyanagi-Harada (VKH) syndrome, combined use of FAF and SD-OCT imaging allowed non-invasive delineation of RPE/outer retinal changes which were consistent with previous histopathologic reports (Vasconcelos-Santos et al., 2010). Morita and co-workers reported that FAF patterns may be related to functional compromise in the long-term and may be useful for early identification of severe cases (Morita et al., 2016). Wide-field FAF imaging has been shown to allow for detection of peripheral FAF abnormalities in VKH syndrome (Heussen et al., 2011).

For Multiple Evanescent White Dot Syndrome (MEWDS), lesions with increased FAF that fade during autofluorescence imaging have been described (Joseph et al., 2013; Mantovani et al., 2016, 2019; Zicarelli et al., 2019). Presumably, these initially bright lesions show the underlying RPE autofluorescence due to lack of chromophore, while the outer segments of the remaining retina initially mask the RPE autofluorescence partially. Upon bleaching, this masking effect is abolished.

Multifocal choroiditis and panuveitis (MCP) is an inflammatory condition more commonly affecting young myopic women (Fig. 19). A related condition punctate inner choroidopathy is similar, leading some to lump the two conditions together. Visually significant damage is associated with several discrete findings that may occur together. Inflammatory infiltrate in the sub-RPE and subretinal space is the hallmark of the condition. The subRPE collections may expand, leading to separation of the RPE monolayer. With these spots of decreased FAF are seen. Even with resolution of inflammation, these spots may remain even though there are few ophthalmoscopically visible signs, making autofluorescence imaging a useful way to monitor these patients. Macular neovascularization is a frequent complication of MCP. These start as small areas of increased FAF at the border of an inflammatory lesion. OCT angiography is a helpful way to image the neovascularization because there often is leakage during fluorescein angiography related to the inflammation which can obscure visualization of any associated new vessels.

Fig. 19.

Multifocal choroiditis and panuveitis. This 28 year-old myopic woman had a central fibrotic scar from neovascularization and a suggestion of surrounding faint spots (A – colour fundus photography). The autofluorescence image shows increased intensities around the outer border of the neovascularization (B) Eccentric to these changes, particularly superior, levels of decreased fundus autofluorescence intensities are visible indicating additional retinal pigment epithelium atrophy. She was managed with intravitreal anti-vascular endothelial growth factor injections and local corticosteroids. Over a follow-up of 5 years when this photograph was taken, the patient had new chorioretinal scars in the right eye and began to have lesions in the left eye (C). The autofluorescence image show areas of increased autofluorescence in 3 of the lesions (three arrow, D). The inset shows an optical coherence tomography angiography image which demonstrates each of the lesions had macular neovascularization.

Acute zonal occult outer retinopathy (AZOOR) was a disease complex proposed by Gass manifested by outer retinal damage in otherwise normal appearing eyes (Gass, 1993). He proposed the condition prior to the wide-spread use of either autofluorescence imaging or optical coherence tomography. Eyes with AZOOR may have autofluorescence patterns suggestive of RP because both cause damage to or loss of the photoreceptors (Fig. 20). This loss removes some of the ocular barrier to the excitation wavelengths, allowing greater excitation of remaining fluorophore. However concurrently there is a loss of RPE loading by outer segments and there is collateral damage to the RPE in AZOOR (Spaide, 2004a). Expanding areas of granular decreased FAF occur surrounding the optic disc and in the periphery. The outer border of these regions may be festooned with scalloped edges of an increased FAF signal. These areas expand over time and the associated overlying retinal thickness decreases as well (Fujiwara et al., 2010). Autofluorescence findings have been used as a main determinant in a proposed classification system for AZOOR (Mrejen et al., 2014).

Fig. 20.

Acute zonal occult outer retinopathy (AZOOR). This 47 year-old woman had a history of Hashimoto’s thyroiditis and AZOOR. Autofluorescence imaging of the right eye shows a lesion with hypo and hyperautofluorescent regions bounded by a perimeter of increased autofluorescence (A). The left eye showed minimal changes along the superior border of the nerve (B). When examined after 15 years of follow-up, the lesion shows expansion with loss of RPE autofluorescence in the more central portions of the lesion (C). The left eye demonstrated remarkable enlargement of the area of involvement (D).

6.9. Retinal detachment

Nadelmann and coworkers recently suggested that widefield FAF imaging allows for differentiation between retinal detachment and retinoschisis. They found differences with regard to the FAF signal of the posterior border and homogeneity of the area affected. They concluded that widefield FAF should be considered in the era of multimodal imaging, particularly when clinical exam alone is inadequate to differentiate these two entities (Nadelmann et al., 2019). FAF imaging can further be used to detect macular displacement following rhegmatogenous retinal detachment. Hence FAF imaging may reveal this cause for visual disturbance, as distortion of lines or objects appearing smaller or narrower following successful rhegmatogenous retinal detachment surgery (Lee et al., 2013; Shiragami et al., 2009). Moreover, FAF is discussed to serve as predictors for long-term functional outcome in rhegmatogenous retinal detachment (Poulsen et al., 2019).

Interestingly, Jaggi and colleagues found that areas of previously detached retina exhibit significant fluorescence lifetime changes. They found a significant correlation of fluorescence lifetimes within the fovea with visual acuity after successful rhegmatogenous retinal detachment repair. Their data suggest that the prolongation of fluorescence lifetimes in the fovea is mainly driven by loss of macular pigment. Therefore, fluorescence lifetime imaging ophthalmoscopy may be useful in the prediction of long-term functional outcomes after macula-off rhegmatogenous retinal detachment surgery (Jaggi et al., 2019).

6.10. Intraocular tumours

FAF imaging of intraocular tumours is challenging for several reasons. Depending on the elevation and curvature of the lesion, out-of-focus regions might not be imaged properly using confocal SLO imaging devices. Moreover, FAF from deeper layers will not be detected due to absorbing molecules within the optical path ahead of the tumour and due to a limited light penetration depth. Imaging of very peripheral tumours has been difficult or even impossible but is now being facilitated using wide-angle and ultra-widefield imaging technologies (Callaway and Mruthyunjaya, 2019). Most intraocular tumours are located at the level of the choroid and therefore underlie the RPE (e.g., choroidal nevus, choroidal melanoma, choroidal haemangioma, choroidal osteoma, and choroidal metastasis). Thus, FAF imaging seems not to be the preferred imaging method to visualize the tumours themselves as FAF imaging refers to fluorophores at the level of the RPE-photoreceptor complex. In addition, tumours might not contain autofluorescent molecules. Microscopic examinations of Lohmann et al. revealed little autofluorescence of choroidal melanomas (Lohmann et al., 1995). Gündüz et al. did not find autofluorescence within a melanoma itself using fluorescence microscopy (Gunduz et al., 2009). However, choroidal tumours may affect the overlying RPE-photoreceptor complex directly and lead to subsequent alterations. Hence, FAF imaging in presence of choroidal tumours gives additional information regarding the status of the overlying RPE and may additionally be altered by coexisting subretinal fluid. In case of choroidal melanomas, the most common primary intraocular malignancy in Caucasian (Hu et al., 2005), both, a normal overlying FAF signal and FAF changes are found (Almeida et al., 2013; Shields et al., 2008). Here, normal FAF implies lack of RPE involvement, whereas FAF changes may derive from secondary hyperpigmentation, pigment epitheliopathy (e.g., RPE degeneration, RPE clumping, RPE atrophy), LF accumulation within the RPE (orange pigment), and subretinal fluid (Gunduz et al., 2007; Lavinsky et al., 2007; Shields et al., 1976). Ophthalmoscopically visible orange pigment is associated with a markedly increased FAF signal and is a known indicator of malignancy (Shields et al., 2000, 2007). Rather, detection of subclinical orange pigment by FAF imaging is notably helpful for early diagnosis of choroidal melanomas (Cennamo et al., 2018; Gunduz et al., 2007). Reports of FAF imaging after brachytherapy of choroidal melanomas, underline the visualization of irradiation effects beyond ophthalmoscopically visible changes (Bindewald-Wittich et al., 2020; Navaratnam et al., 2019). FAF changes after Ruthenium-106 brachytherapy exceed funduscopically visible changes. They are well-delineated by a rim of increased FAF. An increased FAF signal along with speckled decreased FAF (“FAF mottling”) within the irradiation field is primarily caused by a window defect due to photoreceptor cell loss combined with various stages of RPE degeneration (Bindewald-Wittich et al., 2020). FAF findings due to choroidal metastases include a reduced FAF signal as well as foci of increased FAF in dependence of the status of the RPE cells, LF accumulation and the existence of subretinal fluid (Collet et al., 2008; Ishida et al., 2009). FAF imaging in eyes with retinoblastoma might be limited due to the very young age of the patients and restricted cooperation to obtain FAF images. But if performed, an obviously increased FAF signal visualizes calcified portions of the retinoblastoma (Ramasubramanian et al., 2011; Villegas et al., 2013). In summary, FAF imaging of intraocular tumours complements multimodal imaging methods in respect of diagnostic approaches, documentation, examination of the tumour’s spatial relationship to adjacent ocular tissues, and secondary effects of the tumour on various retinal structures. In combination with other imaging modalities it is also a convenient tool for monitoring patients during follow-up visits after globe-sparing treatment of intraocular tumours.

6.11. Alzheimer’s disease

Based on a pilot study, Jentsch et al. reported in 2015 that FLIO revealed changes in the retina of patients with Alzheimer’s disease in relation to Alzheimer-specific markers (Jentsch et al., 2015). They found correlations with the mini-mental state examination (MMSE) and p-tau181-protein concentration in the cerebrospinal fluid which could not be established by other conventional retinal imaging techniques. In another pilot study, Sadda et al. demonstrated correlation of FLIO-derived parameters with beta-amyloid, tau levels in the cerebrospinal fluid, and ganglion cell layer plus inner plexiform layer thickness on OCT (Sadda et al., 2019). Using dual-wavelength FAF imaging, another study observed significantly less macula pigment and low serum concentration of lutein and zeaxanthin in patients with Alzheimer’s disease (Nolan et al., 2014).

7. Limitations

Finally, limitations and disadvantages of the use of FAF imaging should not remain unmentioned. Of note, it is an additional procedure and therefore implementation in high-frequent clinics as well as costs for purchase and maintenance need to be considered. Further, training in FAF image acquisition is important for obtaining adequate and meaningful results. In most subjects, the pupil must be dilated before meaningful image acquisition is possible. Of note, the short-wave excitation light is not only susceptible to reduced image quality (e. g. nuclear opacities, see 2.1 Origin of the Signal), it is also somewhat uncomfortable for patients, especially compared to long-wave excitation light as used in OCT imaging. Besides the issue of comfort, there is safety. The retinal radiant exposure of blue FAF excitation light is far below AINSI safety thresholds (American National Standards Institute, 2014). Considering these standards, blue-light FAF imaging (using the widely used Spectralis device (Heidelberg Engineering)), is safe for up to 8 h (continuous scanning, 9 J cm⁻²), whereas typical examinations irradiate the retina for less than 5 min (<0.1 J cm⁻²) (Teussink et al., 2017). However, the AINSI thresholds and other guidelines do not specifically evaluate the contribution of endogenous photosensitizers in enhancing a patient’s susceptibility to retinal phototoxicity. These phototoxic reactions are more prominent for shorter wavelengths of light (Hunter et al., 2012). The short-wavelength excitation light (λ = 488 nm) carries significantly more energy per photon than those used in green-light excited autofluorescence (λ > 530 nm). Over many years of clinical use, it does not appear that blue-light autofluorescence causes damage (we are not aware of any evidence of toxic light-damage after FAF imaging in the clinical setting), but there is a lack of a sensitive in vivo measurement for potential damage. Of note, eyes in which autofluorescence imaging provides the most useful information, such as Stargardt disease and AMD, appear to have the most potential for light induced damage. Modeling of photo-oxidative stress showed that the rate of oxygen photoconsumption by LF increases during FAF imaging. This increase is accentuated in older individuals. In healthy persons at age 20, 40, and 60, the rate of oxygen photo consumption by LF increases by a factor of 1.3, 1.7, and 2.4 respectively. In patients with Stargardt disease below the age of 30 the rate was 3.3 times higher as compared with age-matched controls (Teussink et al., 2017). Overall, this simulation determined similar sensitivity to photo-oxidative stress in the RPE for short-wavelength FAF imaging as compared to exposure to indirect (diffuse) sunlight as it occurs in daily life. In conclusion, FAF imaging is a safe procedure for clinical routine as demonstrated by wide spread use for more than two decades. The exposure to the excitation light should not be prolonged. Longer wavelength applications of FAF imaging (green and NIR-FAF instead of blue-light) reduce the potential and so far theoretical risks of light toxicity.

8. Future directions and conclusions

Variations of existing and new innovative application of FAF imaging have been brought forward following its initial introduction for clinical applications. They will continuously increase our understanding of various retinal diseases and our ability to detect and monitor pathological alterations as well as to evaluate therapeutic interventions. Instead of using blue excitation light, we anticipate that the application of green-light FAF will further increase in the clinical setting. The value of FAF findings should be always assessed in a multimodal imaging approach, by appreciating specific advantages and benefits but also disadvantages and limitations of each imaging modality.

An important challenge is the identification and establishment of meaningful clinical endpoints for interventional trials in macular and retinal diseases. The key advantage of FAF is the ability to provide information about the metabolic state of the retina. Quantification of accumulated fluorophores, assessment of the dynamics of bleaching phenomena, multi-spectral (and potentially hyper-spectral) evaluation of fluorescence intensities and the application of FLIO all represent innovative approaches that may be not only promising for detection of early disease changes and precise monitoring of disease progression, but also allow to objectively investigate functional correlates. As the potential benefits of FAF findings are continually assessed, particularly along with the recent advances in artificial intelligence-based analysis of retinal imaging data, we expect a further expansion of FAF in both clinical care and research.