Fundus Autofluorescence in Age-Related Macular Degeneration: An Epiphenomenon?

Over the past decade, there has been an upsurge of interest in studying autofluorescence (AF) in the fundus of the eye. AF is an intrinsic property of certain materials that is characterized by the transient emission of light when the substance is illuminated by an exogenous source. Many tissues and structures in the eye, such as the cornea, the crystalline lens and the retinal pigment epithelium are composed of biological molecules that have autofluorescent properties. ¹ Imaging of autofluorescent tissues has been exploited, particularly with regard to the fundus, where its relevance to disease is of considerable interest. The early imaging systems that were modified to detect fundus AF have since evolved into sophisticated, commercially available instruments such as the confocal laser scanning ophthalmoscope (cSLO). With increasing clinical use of such instruments, information on the topographic distribution and intensity of AF has accrued in both normal ^{2 3} and diseased ^{4 5 6} eyes. It is now time to examine critically the relevance of this information to the understanding of the biological significance of AF and its alterations with physiological ageing and in pathologic states. 

Fundus Autofluorescence and Lipofuscin

Spectrophotometric studies have suggested that lipofuscin and melanolipofuscin are the dominant sources of fundus AF. ¹ Notably, studies have demonstrated striking similarities between observed in vivo fluorescence patterns from AF imaging and the spectroscopic signals from chloroform extracts and flat preparations of human RPE cells that have been biochemically characterized as lipofuscin and other related compounds. ^{7 8 9} Lipofuscin accumulates in the RPE as a result of incomplete degradation of photoreceptor outer segments, and recent studies show that it is a potent source of oxidative stress. ¹⁰ Although circumstantial, the evidence that lipofuscin is involved in mediating cell damage and senescence is compelling. Therefore, it has been proposed that fundus AF may be used in vivo as a noninvasive surrogate marker for monitoring the status of the RPE, particularly in age-related macular degeneration (AMD), in which the aging of the RPE factors strongly in the pathogenesis. 

Lipofuscin and AMD

The exact pathogenic role of lipofuscin in AMD itself, however, has not been satisfactorily resolved to date. Kopitz et al. ¹¹ review several lines of evidence suggesting that aging changes in the RPE, in particular the accumulation of autofluorescent lipofuscin granules in the lysosomal compartment of postmitotic RPE cells, play a key role in the pathogenesis of AMD. They highlight recent studies that indicate that lipidic compounds of lipofuscin and lipid peroxidation of proteins induce lysosomal dysfunction and lipofuscinogenesis in the RPE. 

Despite the foregoing, work by Feher et al. ¹² calls into question the exact pathogenic relevance of lipofuscin itself in AMD. This group noted a significant decrease in number and area of mitochondria, as well as loss of cristae and matrix density in both AMD and control specimens. Although these decreases were greater in AMD than in normal aging, alterations of mitochondria were accompanied by proliferation of peroxisomes and lipofuscin granules in both AMD and control specimens. Quantitative morphometric analysis confirmed that the RPE alterations seen in AMD also develop in normal aging, 10 to 15 years after appearing in patients with AMD. Thus, other mechanisms may be at play, leaving much yet to be determined before attribution of a direct causal relationship between lipofuscin and AMD can be made. 

Katz ¹³ points out that “although a correlation between RPE lipofuscin content and AMD has been reported, a cause-and-effect relationship between RPE lipofuscin accumulation and the development of this disease has not been established. The lack of a definitive link between RPE lipofuscin accumulation and AMD illustrates one of the biggest challenges remaining in lipofuscin research—determining whether lipofuscin accumulation per se has an impact on cell function.” 

Clinical Relevance of AF

Despite the controversy over the role of lipofuscin in the pathogenesis of AMD, fundus AF remains an ideal tool for the in vivo examination of the normal and diseased macula. It offers information not obtained with standard ophthalmoscopy or angiography and possibly insight into pathogenic mechanisms of RPE cell dysfunction secondary to lipofuscin accumulation. In this context, patterns of AF increase and decrease have been described in both inherited ^{3 14 15 16 17 18} and acquired retinal diseases. ¹⁹ This body of literature has shown that increased AF is a marker of hereditary retinal disease and indeed may predate clinical symptoms. Decreased AF has been shown to be associated with atrophy and RPE cell loss. ^{3 14 15 16 17 18} With the development of methods for better quantitation of fundus AF ²⁰ its potential use in longitudinal monitoring of disease has increased markedly. 

AF and AMD

AF imaging has been increasingly used in studying AMD, as the deposition of lipofuscin with age is a tempting correlate of this disorder. In this issue of IOVS, two groups present their findings on the relevance of AF as a prognostic marker for the development of geographic atrophy, an important and sight-threatening feature of end-stage AMD. Schmitz-Valckenberg et al. ²¹ provide evidence from the longitudinal FAM study to show that fundus AF is a marker for disease progression. By contrast, the work by Hwang et al. ²² did not find any association between abnormal fundus AF patterns and either the development or progression of geographic atrophy, the atrophic form of AMD, thus questioning the relevance of the former to progression of the latter. Clearly, when diverging views are expressed in the literature, it is a matter of interest and concern to the scientific community. 

What may be the reasons for this disparity? Intuitively, one could suggest several reasons, even without considering a full analysis of the scientific basis of the described methodologies. The small sample sizes, which can lead to overestimation of effects, variations in nomenclature, and lack of standardization of equipment ²³ are all potential introducers of bias. 

AF Disease Phenotypes

The principles of physics as applied to the imaging of live tissues in the fundus are highly complex and create considerable potential for misinterpretation of findings. Perhaps the greatest challenge facing this potentially valuable tool is the lack of a standardized normative database and descriptors of disease developed from large longitudinal databases. Scrutiny of the evidence reveals not only a lack of a properly constructed and evaluated normative reference database of fundus AF, but also the corresponding absence of a comprehensive library of disease phenotypes. 

It is notable that the terminology used to classify disease varies from one report to another, with different groups invoking different classification systems, often based on a relatively small number of patients. Various authors have developed classification systems and AF has been described by one group as focal, banded, patchy and diffuse, ²⁴ by a second group as normal, minimal change, focal increased, patchy, linear, lacelike, reticular and speckled, ²⁵ and by a third group as focal increased, reticular, combined and homogeneous. ²⁶  

The absence of normative AF imaging databases makes it impossible to make robust age-matched comparisons between studies. This is an important issue because macular pigments vary with age and affect the measurement of fundus AF. Furthermore, none of the studies of AF has systematically examined reproducibility and consistency. Typically, data have been collected using different acquisition systems, and presented in isolation from patterns of disease or correlation with other features and risk factors of AMD. Although one study found that all cSLOs allow clinically useful fundus AF imaging in retinal diseases, gray scale levels and contrast are much lower on the Rodenstock cSLO when compared with the Heidelberg Retinal Tomographer [Heidelberg Engineering, Heidelberg, Germany] or the Zeiss cSLO [Carl Zeiss Meditec, Dublin, CA)]. ²³ In this context, it is also worth noting that no two SLOs have the same output, and thus equipment differences may interfere with the definition of normal or diseased states when data are collected at multiple sites. To compound this problem, differences in photodetector gain and argon laser amplification between SLO devices could introduce even more variability into the data. These parameters are mandatory for an absolute comparison of AF images if data are to be combined. This important problem has not been routinely addressed in the literature in AF imaging. 

Developing true “norms” in AF imaging is likely to be difficult, as the RPE is a highly variable tissue, even within one individual. Although the RPE appears phenotypically regular, Burke and Hjelmeland. ²⁷ note that there is striking cell-to-cell variability in the content of melanin and lipofuscin granules, as well as in the expression of many proteins. “This naturally occurring cell heterogeneity likely arises by normal mechanisms regulating gene expression during development and postnatal aging. The consequence is a tissue in which individual cells may differ in their ability to support adjacent photoreceptors, and which may respond differentially to oxidative stress and other environmental influences that contribute to cell dysfunction during aging.” 

One study ²⁵ described patterns of AF in 100 eyes of 100 individuals older than 55 years in five clinical centers. Although this was a laudable attempt to pool resources, the study had many drawbacks. First, the definitions used would have made it difficult to distinguish between normal and abnormal alterations in AF. Second, eight possible patterns were classified as abnormal, resulting in very small samples in the different phenotypic groups. Third, the data were obtained with a variety of acquisition systems. 

Recommendations

The terminology used to describe pathologic patterns of AF requires harmonization, standardization, and refinement. Inconsistency in image acquisition must be addressed, and differences in the performance of the myriad of clinical devices that image AF must be characterized and understood. The need for database development, disease-stage classification, and algorithms for the correct interpretation of AF patterns cannot be overstated. They are needed to form a framework in which the significance of AF patterns in phenotyping and determining disease progression can be placed. Without a demonstration of reliability and reproducibility, both in data acquisition and interpretation, the value of AF will be diminished. Despite these caveats, there is a pressing need for the prospective development of large, well-designed, normal comparative databases so that patterns of AF in aging and disease can be evaluated within a meaningful frame of reference. Should such a source of reference material be accumulated and described, it would be an important milestone in the use of AF imaging and would aid enormously in the development of clinical practice guidelines. With proper study, spatial and temporal patterns of fundus AF have the potential to provide useful information on progression in diseases such as AMD where the RPE cell plays such a key role. AF clearly has great potential as a tool for providing solutions to many as yet unanswered questions in the field of aging and diseases of the macular retina. 

 This article was written as a collaborative effort during UC’s sabbatical in the Doheny Retina Institute, Los Angeles, California.

 Supported by the Research and Development Office in Northern Ireland and the Doheny Retina Institute.

 Submitted for publication November 20, 2005; accepted April 11, 2006.

 Disclosure: J. Hopkins, None; A. Walsh, None; U. Chakravarthy, None