Alex Walsh

　　Doheny Eye Institute，Keck School of Medicine，USC

　　Background：Fundus autofluorescence (FAF) has emerged as a non-invasive, in vivo method of imaging the health and metabolism of RPE and photoreceptor cells. FAF is based on excitation of naturally-occurring fluorophores, such as lipofuscin (LF), in the tissues of the eye. Lipofuscin is a complex structure consisting of highly cross-linked proteins, lipids and carbohydrates which are not normally amenable to intracellular proteolytic degradation. For this reason, LF granules accumulate with age in RPE cells1 due, in large part, to incomplete breakdown of constantly-supplied photoreceptor outer segment discs.2-4 For example, by the age of 70, nearly 25% of the cytoplasmic space of an RPE cell is occupied by lipofuscin or melano-lipofuscin.3 This excessive accumulation of LF may be a common downstream effect of many degenerative retinal diseases, but is now receiving much attention in the study of AMD.

　　FAF has been found to correlate with LF content in RPE and photoreceptor cells. LF granules absorb incoming blue light and emit green light. A barrier filter blocks the reflected blue light so that only the emitted green light is detected. This constitutes the autofluorescence (FAF) signal. In normal FAF, the optic nerve head is dark since there are no RPE cells present and no LF accumulation. Blood vessels are dark due to absorption by blood and a decrease in autofluorescence may occur towards the fovea due to absorption by luteal pigment. Abnormal FAF can be caused by changes in the amount or composition of LF in RPE and/or photoreceptor cells or the presence of either absorbing or autofluorescent material anterior to the LF. In general, FAF is increased with RPE dysfunction, decreased or absent with RPE cell loss and decreased with photoreceptor loss。

　　Due to the large number of fluorophores that contribute to FAF, unique spectral features of LF have the potential to act as robust biomarkers of disease activity that can be quantitatively monitored to assess lesion development and/or therapeutic efficacy. On the other hand, absorption of blue light by macular pigment (MP) introduces a degree of variability that may interfere with accurate quantification of FAF signals. Other tissue components that can influence the FAF signal include variations in lipofuscin content, melanin, unknown chromophores and anatomic features such as the foveal depression or the presence or absence of a cataract. Instrument parameters that can affect the FAF signal include variations in detector gain, illumination irregularities and electronic noise.

　　Comparison of Two Dominant Clinical Methods：

　　At present, the two dominant methods of clinical FAF imaging are 1) confocal SLO (cSLO) or 2) modified fundus camera (mFC). In cSLO FAF imaging, a point source of light is rapidly swept across the fundus in a raster pattern. Confocal optics ensure that the reflectance and fluorescence signals come from the same approximate "conjugate" plane. This reduces scattered AF signals from sources anterior to the retina, such as the crystalline lens, and increases contrast while reducing background noise.5 In mFC FAF imaging, a standard fundus camera equipped with appropriate filters illuminates the entire fundus with flash illumination. Since these systems are not confocal, AF signals originating outside of the retina may contribute to the final FAF image.

　　One major challenge in comparing FAF images from these two systems is that they tend to image slightly different fluorophores. Therefore, results from these two modalities are not expected to be identical. However, several aspects of FAF images from cSLO and mFC systems can be compared。

　　FAF images inherently have a low signal-to-noise ratio (SNR) because fundus autofluorescence is a relatively weak signal. To account for this, the most common cSLO system can average multiple FAF frames together to increase the SNR. Although this could theoretically also be done with mFC devices, it is not currently available on commercial mFC systems. The higher SNR of cSLO systems may further lead to enhanced image contrast to aid in assessments of subtle FAF changes.

　　As discussed above, macular pigment (MP) may block AF in the fovea and parafovea due to the concentration of pigments in this area. Current commercial cSLO systems are especially prone to this irregularity and may give variable AF results in this region whereas mFC filter sets have been designed to eliminate AF variability due to MP and may provide more consistent results in the fovea.

　　The confocal optics of cSLO systems is felt to be an advantage since it decreases scatter from out-of-plane sources of AF. Special filter sets have been developed for mFC systems to greatly reduce AF signals from one of the largest sources of out-of-plane AF, the crystalline lens. However, out-of-plane AF can still sometimes be seen with mFC systems by looking carefully at the optic nerve which, instead of being entirely dark, may contain bright AF signals that were scattered from elsewhere in the fundus.

　　Although both systems are capable of measuring the area of FAF abnormalities, neither system can quantify abnormalities in FAF intensity that could enable in vivo measurements of LF concentrations. Beyond the myriad of instrument and subject variables that interfere with this quantification, cSLO systems also typically normalize AF intensities after capture to improve visual assessments which harms absolute quantitative assessments. mFC systems are subject to lighting irregularities inherent in flash fundus imaging systems and currently, are also unable to objectively quantify FAF.

　　The final differentiator between these systems is cost. mFC FAF is typically a simple, relatively low-cost add-on to an existing fundus camera system while cSLO FAF requires a separate, relatively expensive instrument. Therefore, the right FAF system for a given practice will ultimately depend on the practice's needs, their existing equipment, and the expected return on investment from an FAF instrument。

　　References：

　　1.Terman A, Brunk UT. Oxidative stress, accumulation of biological 'garbage', and aging. Antioxid Redox Signal. 2006 Jan-Feb;8(1-2):197-204。

　　2.Wing GL, Blanchard GC, Weiter JJ. The topography and age relationship of lipofuscin concentration in the retinal pigment epithelium. Invest Ophthalmol Vis Sci. 1978 Jul;17(7):601-7。

　　3.Feeney-Burns L, Hilderbrand ES, Eldridge S. Aging human RPE: morphometric analysis of macular, equatorial, and peripheral cells. Invest Ophthalmol Vis Sci. 1984 Feb;25(2):195-200。

　　4.Weiter JJ, Delori FC, Wing GL, Fitch KA. Retinal pigment epithelial lipofuscin and melanin and choroidal melanin in human eyes. Invest Ophthalmol Vis Sci. 1986 Feb;27(2):145-52。

　　5.Lois N, Halfyard AS, Bird AC, and Fitzke FW. Quantitative evaluation of fundus autofluorescence imaged "in vivo" in eyes with retinal disease. Br J Ophthalmol. 2000;84:741-745