Peau d’orange appears to be the earliest funduscopically visible alteration in patients with PXE, preceding the development of angioid streaks (Gills and Paton, 1965; Krill et al., 1973; Naouri et al., 2009). Peau d’orange is characterized by small dark spots, within an area of a slightly whitish or opaque fundus reflex (Figures 1A–C,M, 2A, and 6K). This pattern may be observed at the posterior pole very early in the disease and more peripheral in later disease stages. In the latter case, the slightly whitish or opaque fundus reflex may have become more uniform posteriorly to the area of peau d’orange. Peripheral to the peau d’orange region, the fundus reflex is usually darker (Figure 1M). The difference between the central and peripheral fundus reflex is more obvious in dark pigmented patients in whom the higher overall fundus pigmentation contrasts better with the more posteriorly located whitish areas. It may remain unnoticed in less pigmented individuals as well as in very late disease stages (Charbel Issa et al., 2010a).

Bruch’s membrane is believed to be the primarily affected anatomic structure of the ocular fundus in patients with PXE. Anatomical differences in thickness and integrity between central and peripheral BM (Chong et al., 2005) may account for a higher vulnerability and thus earlier calcification at the posterior pole, with a subsequent centrifugal disease spread.

Based on clinical observations, it has been hypothesized that peau d’orange represents a visible transition zone of BM calcification (Charbel Issa et al., 2010a). Replacement of peau d’orange by a more uniform area of whitish fundus reflex posteriorly may occur once calcification has become continuous. A centrifugal spread of calcification, with the leading edge representing the transition zone of calcifying BM, appears to be the rational for the observation that peau d’orange is seen increasingly more peripheral with age. Currently, there is little published longitudinal data to confirm such cross sectional clinical observations.

Peau d’orange often is assumed to be most pronounced and widespread in the temporal midperiphery. However, there is good evidence that this characteristic fundus finding simply is the most prominent manifestation of a transition zone that in actual fact is present circumferentially, though not symmetrically (Charbel Issa et al., 2010a; Figures 1M and 5).

Angioid streaks, which are breaks in BM, are the most obvious and most consistently observed funduscopic finding in patients with in PXE (Figures 1D–F,H,M, and also visible in most other figures).

Angioid streaks present ophthalmoscopically as irregular and jagged lines that radiate from a concentric peripapillary ring toward the equator of the eye. Visibility and colour of angioid streaks may depend upon the degree of secondary alterations, such as loss of the choriocapillaris, depigmentation or loss of RPE cells, or fibrosis. In some patients, they may be limited to a few almost imperceptible lines whereas in others they present as a complicated interlacing network or grid. The streaks are most prominent at the posterior pole of the eye and typically taper and fade toward the equator of the eye, often dividing into smaller branches. Occasionally they continue for a short distance as irregular, white, depigmented lines. Pigmented wing-like hyperpigmentations may be found along angioid streaks.

Alterations in the RPE adjacent to streaks commonly occur. Loss of pigment imparts a “feathered” appearance to the streak (De Zaeytijd et al., 2010). Angioid streaks enlarge in length and width over time (Mansour et al., 1993), and usually do not cross areas of peau d’orange (Charbel Issa et al., 2010a). The latter finding supports the notion that calcification of BM predisposing to angioid streak formation is present centrally from peau d’orange. In addition, stress lines on the posterior pole of the eye converge on the optic nerve head, which is both the anchor point for the optic nerve, and the hinge point at which, despite the flexibility of the optic nerve itself, a certain degree of mechanical tilting of the eyeball occurs, relative to the optic nerve, in a direction dependent on the gaze. The combination of these stress lines in a calcified BM probably leads to angioid streak formation.

Angioid streaks are usually obvious on funduscopy. Confocal near-infrared (NIR) reflectance imaging was found to be superior to other imaging modalities (Charbel Issa et al., 2009; De Zaeytijd et al., 2010) to document angioid streaks (Charbel Issa et al., 2009). In areas of widespread chorioretinal atrophy, angioid streaks are usually not visible anymore. However, breaks in BM may still be detected in such cases using spectral domain optical coherence tomography (SD-OCT; Charbel Issa et al., 2009).

Although other ocular diseases such as high myopia and ocular trauma may also lead to breaks in BM, the pattern of break formation is different in those diseases (Pruett et al., 1987). While myopic lacquer cracks are short and usually found in a reticular distribution within a posterior staphyloma, traumatic tears are characteristically curved parallel to the optic disk margin and are usually located temporal to the disk in only one eye.

Angioid streaks are not necessarily associated with a noticeable decrease in retinal function, and visual acuity may remain normal even in presence of a streak crossing the fovea. Angioid streaks may occur with or without overt damage to the RPE (Charbel Issa et al., 2009). Probably, the overlying RPE first needs to be compromised before loss of function occurs. However, high resolution structure–function correlations, e.g., with microperimetry, has only recently become available (Charbel Issa et al., 2010b) and has not yet been performed in patients with PXE. Definite vision loss occurs when angioid streaks are complicated by the development of a CNV or atrophy of the RPE.

Comet lesions with or without comet tail (Figures 1G, 3D, and 4J) are observed as solitary, subretinal, nodular, white bodies with a tapering white tail extending posteriorly of the comet body pointing toward the optic disk. The body may have some pigmentation at its margin. Sometimes a spray of comets and comet tails can be observed, creating an aspect of “comet rain.” Comets and comet tails are found in the (mid)periphery of the fundus and are the only PXE-related finding that may occur peripheral to peau d’orange (Gass, 2003; Charbel Issa et al., 2010a). They have been suggested to be the only pathognomonic characteristic of PXE (Gass, 2003) and may also occur in heterozygous carriers of ABCC6 mutations (De Zaeytijd et al., 2010). Especially in young patients, in whom angioid streaks are often not yet present, they might thus be of significant diagnostic value.

Choroidal neovascularization of the macular region is a frequent complication in patients with PXE and commonly leads to pronounced vision loss (Figures 1D,I and 3F). Often, CNV in PXE are classic membranes (Nakagawa et al., 2013), i.e., they are located between the RPE and the photoreceptor layer. Classic CNV usually occurs in association with angioid streaks and may develop from occult CNV which is situated underneath the RPE (Nakagawa et al., 2013). The predisposition for the posterior pole is consequent upon the higher frequency of angioid streaks in that area. CNV leads to subretinal hemorrhage and exudation, and eventually formation of a fibrovascular scar (Figures 1J,K). Occasionally, eccentric CNV may remain unnoticed due to its lower impact on visual function. Before the development of a scar, CNV is the only ocular PXE complication that is currently eligible for treatment. Intravitreal inhibitors of vascular endothelial growth factor (VEGF) are currently most effective in the attempt to prevent or limit fibrovascular scar formation with consequent visual loss (Gliem et al., 2013). Older treatment options such as photodynamic therapy with verteporfin or argon laser photocoagulation have been largely abandoned. Overall, classic CNV appears to have a worse prognosis with regards to visual function compared to occult CNV (Nakagawa et al., 2013).

Recently, polypoidal choroidal vasculopathy (PCV) was also described to occur in patients with angioid streaks (Baillif-Gostoli et al., 2010; Nakagawa et al., 2013). PCV may occur as an initial vascular change or secondary to CNVs. In contrast to classic CNV, those polyps appear not to be associated with angioid streaks.

Pattern dystrophy-like changes (Figures 4F and 7) are frequently observed in patients with PXE varying between 10% (Finger et al., 2009a) and almost 70% of cases (Agarwal et al., 2005). Based on a classification suggested by Agarwal and Gass, findings can be categorized due to their similarity to pattern dystrophies, into vitelliform, butterfly, and reticular dystrophy, or fundus flavimaculatus or pulverulentus (von Winning and Oosterhuis, 1974; Shiraki et al., 2001; Agarwal et al., 2005; Sawa et al., 2006; Finger et al., 2009a). It has been postulated that the presence of a pattern dystrophy is a prognostic sign for CNV development although further longitudinal data are needed for confirmation (Finger et al., 2009a).

Chorioretinal atrophy may develop secondary to CNVs, usually surrounding a fibrovascular scar. Atrophy can also occur in the absence of CNV (Figures 7A–D) – a process which, however, has not been well characterized yet. It may initially be observed as patches of atrophy mostly between or along the major vascular arcades, frequently originating within areas of pattern dystrophy. During the subseqent course of the disease, such lesions may grow and – if initially multifocal – become confluent.

Optic nerve head (ONH) drusen (Figure 1H) seem to be more common in PXE patients than in the general population. The reported prevalence ranges from 6–8% (Meislik et al., 1979; Finger et al., 2009a) to just over 20% (Pierro et al., 1994) compared to ~0.3% in the general population. To date, it remains unclear why PXE patients are at an increased risk to develop ONH drusen though it may be assumed that a common process of abnormal mineralization might be involved, e.g., through direct calcification or increased rigidity of the lamina cribrosa. Similar to ONH drusen not associated with PXE, ocular ultrasound or fundus autofluorescence (AF) may be required to detect them and may be useful for documentation and follow up (Finger et al., 2009a; De Zaeytijd et al., 2010).

Spectral domain OCT, which allows for quasi histologic assessment of the posterior ocular fundus in vivo, has been used to study PXE-related fundus features (Figure 2). On OCT, the calcification of BM may appear as increased reflectivity, and the transition zone from calcified to un-calcified areas correlates well with peau d’orange (Charbel Issa et al., 2009; Figures 2A–E). Angioid streaks have consistently been found to be associated with breaks in BM (Figures 2F–H, arrows). There may be differences with regards to the width of the gap, presence or absence of fibrovascular tissue extending through the breaks (Figures 2I,J), and alterations or preservations of the overlying RPE layer (Charbel Issa et al., 2009). SD-OCT imaging provided the first direct evidence that breaks in BM are indeed the underlying pathology of angioid streaks (Charbel Issa et al., 2009). Disruption and undulation (inward and outward deformation) of BM on OCT images are much more frequent in eyes of (older) PXE patients than in eyes of AMD patients (Ellabban et al., 2012a), facilitating somewhat the differentiation between those two causes for CNV and chorioretinal atrophy.

Comet tail lesions, the peripheral chorioretinal atrophic spots, may be difficult to record using OCT due to their preferably peripheral localization. The few available scans show hyporeflective spaces involving the outer neurosensory retina with a slightly hyperreflective inner lining and focal debris-like deposits just above the RPE level (Charbel Issa et al., 2009).

Pattern dystrophy-like fundus changes are associated with material deposited below the neurosensory retina, and this may be located either within the RPE layer or just below or above the RPE (Charbel Issa et al., 2009; Zweifel et al., 2011).

Spectral domain OCT has also revealed presence of subretinal fluid in the absence of CNV (Figures 2N,Q) in a subset of patients (Zweifel et al., 2011), which may remain undetected by funduscopy. Such fluid accumulation may appear similar to that observed in chronic serous chorioretinopathy. It is probably due to either abnormalities of the RPE pump function, increased hydrophobicity of BM, or a combination thereof. It does not respond to intravitreally applied VEGF inhibitors (Zweifel et al., 2011 and unpublished own observations) or systemic acetazolamide (unpublished own observation). Longstanding cases might develop a vitelliform lesion characterized by deposition of yellowish hyperautofluorescent material at the bottom of the lesion (Figures 2O–Q).

In daily practice, SD-OCT is a very efficient means to identify neovascular leakage (Figure 2K) and monitor treatment efficacy of intravitreal VEGF inhibition. This is of particular importance because it may be difficult in patients with PXE to distinguish between low-grade leakage, which would mean CNV activity, and staining on fluorescein angiography. However, at late stages, atrophic lesions may occur with or without intraretinal cysts (Figures 2L,M), which sometimes makes it difficult to distinguish between atrophic retinal cystic alterations and true leakage on OCT, too.

So far there is only little known about pathologic alterations of the choroid in PXE. Histopathologic works reported atrophic changes and disruptions of the choroid in areas of angioid streaks as well as calcification of choroidal vessels (McWilliam, 1951; Dreyer and Green, 1978). Recently, one study measured choroidal thickness in eyes affected by PXE using enhanced depth imaging-OCT and found a reduced choroidal thickness in the subset of eyes that presented with CNV (Ellabban et al., 2012b).

Fluorescein angiography highlights several features of PXE (Figure 3) and, despite the importance of novel techniques such as SD-OCT, still remains the gold standard for detecting and documenting leakage from a CNV in patients with angioid streaks.

Several authors have presented and discussed the fluorescein angiographic findings of angioid streaks (Smith et al., 1964; Patnaik and Malik, 1971; Federman et al., 1975). In the absence of other signs of CNV, however, fluorescein angiography does not usually add clinically relevant information and therefore may be refrained from in asymptomatic patients.

Indocyanine green angiography uses NIR light for excitation of the chromophore, which is superior to fluorescein angiography in detecting abnormalities under the RPE. ICG differs from fluorescein amongst others in terms of its much more extensive blood-protein binding, and its limited vascular leakage. ICG angiography is capable to outline angioid streaks much better than fluorescein angiography in the majority of cases (Lafaut et al., 1998). Angioid streaks are usually not visible in the early phase, but can be delineated with high sensitivity in the angiographic late phase (Figure 4). In the latter, angioid streaks usually are hyperfluorescent within the area of reduced late phase ICG-fluorescence (see below). Outside this area, they may also appear hypofluorescent or invisible in some cases.

A highly characteristic PXE-related fundus feature on ICG angiography is a reduced late phase fluorescence centered at the posterior pole (Charbel Issa et al., 2010a), while more eccentric areas exhibit a normal late phase fluorescence. In between these two areas, there is a spotted transition zone, which is most prominent on the temporal side of the fundus in most cases (Figures 4 and 5). There are no related obvious alterations on fundus photography, early phase ICG angiography, or fundus AF. Notably, the spotted pattern of peau d’orange is located more eccentric. It was hypothesized that the reduced late phase fluorescence may be due to BM calcification or a dysfunctional RPE (Charbel Issa et al., 2010a).

Indocyanine green is less well suited to detect leakage from CNV compared to fluorescein angiography. Due to its invasive nature it is therefore not recommended to use routinely for monitoring patients but may be used to confirm suspected PCV or occult CNV and to further investigate ocular pathophysiology in PXE.

Fundus AF imaging with most commonly used blue or green excitation light allows evaluation of the integrity and health of the RPE in vivo. Many fundus alterations commonly found in PXE may present with characteristic fundus AF abnormalities. Angioid streaks can show areas of increased as well as areas of decreased fundus AF. The latter usually suggests more severe damage of the RPE with cell loss. Wing-like focal spots of increased AF alongside angioid streaks, consisting of pigmentations visible on fundus photography, constitute the parastreak phenomenon (Finger et al., 2009a; De Zaeytijd et al., 2010). Patterns of AF in the macular area in PXE patients are often similar to those observed in patients with pattern dystrophies (Figure 7). Comets – and to a lesser extent their tails – typically show a hyperautofluorescent signal. Not all comets can be detected individually on AF imaging (Finger et al., 2009a; De Zaeytijd et al., 2010). Whether this hyperautofluorescence is caused by lipofuscin, or the presence of calcification, or a combination thereof is as yet unclear.

Peau d’orange is usually not highlighted on fundus AF images even when visibility was marked on color photography, irrespective of the severity of changes (Figures 6A-I; Finger et al., 2009a; De Zaeytijd et al., 2010). This seems logical when accepting that the appearance of peau d’orange originates from alterations in Bruch membrane, which is located underneath the RPE (with melanin as strong absorber of short wavelength visible light) and therefore is less effectively imaged using blue light.

Fundus AF illustrates RPE atrophy as areas of hypo-autofluorescence, which are often more extensive than the areas of atrophy seen on funduscopy (Figures 7A–D). Thus, fundus AF is useful as a non-invasive tool to monitor progression of RPE changes, including chorioretinal atrophy.

Confocal NIR reflectance imaging is highly sensitive in detecting peau d’orange and angioid streaks (Charbel Issa et al., 2009; De Zaeytijd et al., 2010). The low absorption rate of 790 nm light by melanin within the RPE, in combination with the high contrast of a confocal imaging system, leads to the superior illustration of such structural alterations underneath the RPE cell layer.

With NIR reflectance imaging, angioid streaks appear as uniform, well-demarcated dark fissures against a lighter background, even when they remain unnoticed on color imaging (Figure 6).

Near-infrared reflectance imaging always revealed a diffuse, speckled pattern of peau d’orange, extending beyond the area considered affected on white light digital fundus images. In the absence of extensive macular atrophy or scarring, the peau d’orange area may cover the entire posterior pole and the midperiphery up to the equator (Figures 5B,C; Charbel Issa et al., 2009; De Zaeytijd et al., 2010).

Near-infrared reflectance imaging is superior to visualize comets, which went undetected on color images. Comets appear as small white hyperintensities, suggesting high reflectivity for light of NIR wavelengths.

Recently, confocal NIR reflectance imaging has been shown to be highly sensitive for detecting reticular drusen in patients with age-related macular degeneration. Calcification of Bruch membrane due to PXE also seems to predispose to the development of reticular drusen (Figures 6H,I). However, those own preliminary observations have not yet been investigated systematically.

There are very little data available on the natural history of visual acuity loss in PXE, and the same is true for visual field testing. Visual acuity frequently drops to 20/200 or less around the fourth to fifth decade of life and it is not uncommon that only hand movements may be seen (Gliem et al., 2013). Dark adaptation and scotopic perimetry over peau d’orange has not shown significant functional alterations (Holz et al., 1994).

Although electrophysiological techniques provide important additional information regarding the underlying causes of visual failure, it has not yet been extensively used to investigate the PXE retinopathy.

Early electrophysiological studies in 15 PXE patients (Francois and De Rouck, 1981) found mild reduced responses on electroretinography (ERG) and electro-oculography (EOG) testing in about 50% of eyes, mostly in those with advanced pathology. Audo et al. (2007) have extended the spectrum of ERG findings in PXE patients suggesting that general retinal dysfunction may occur, which may explain difficulties with night vision, which are frequently reported by PXE patients.