Article Text


Added value of infrared, red-free and autofluorescence fundus imaging in pseudoxanthoma elasticum
  1. Julie De Zaeytijd1,
  2. Olivier M Vanakker2,3,
  3. Paul J Coucke2,
  4. Anne De Paepe2,
  5. Jean-Jacques De Laey1,
  6. Bart P Leroy1,2
  1. 1Department of Ophthalmology, Ghent University Hospital, Ghent, Belgium
  2. 2Centre for Medical Genetics, Ghent University Hospital, Ghent, Belgium
  3. 3Fund of Scientific Research, Flanders, Belgium
  1. Correspondence to Dr Bart P Leroy, Department of Ophthalmology & Centre for Medical Genetics, Ghent University Hospital, De Pintelaan 185, Ghent 9000, Belgium; bart.leroy{at}


Purpose Pseudoxanthoma elasticum (PXE) is an autosomal recessive disorder caused by mutations in the ABCC6 gene and primarily affects the oculocutaneous and cardiovascular systems. However, the phenotype, including the ophthalmological manifestations, varies in severity. The present study aims to evaluate the added value of novel funduscopic imaging techniques, such as near-infrared reflectance, red-free and autofluorescence imaging in PXE.

Methods In 22 molecularly proven PXE patients and 25 obligate carriers, PXE retinopathy was evaluated using funduscopy, white light, red-free, infrared and autofluorescence imaging.

Results At least one characteristic of PXE retinopathy was evident on funduscopy of all eyes. Angioid streaks could be subdivided in those with (brick red) or without (feathered) adjacent RPE alterations. Infrared imaging showed the brick-red-coloured streaks as well-demarcated dark fissures, even when these passed unnoticed on funduscopy. Feathered types were detected as triangular areas of hypoautofluorescence. The peau d'orange was much more visible and much more widespread on infrared imaging, with extension from the posterior pole towards the whole midperiphery. Comets and comet tails were best seen with red-free imaging.

Conclusions Infrared, red-free and autofluorescence imaging are more sensitive than white light funduscopy and imaging in visualising early retinal signs of PXE. In addition, this specialised imaging allows a better appreciation of the extent of lesions. Hence, such imaging increases the chances of making a correct diagnosis early, and aids in the accurate evaluation of evolution of disease in the ophthalmic follow-up of PXE patients.

  • Pseudoxanthoma elasticum
  • ocular signs
  • infrared
  • red-free and autofluorescence imaging
  • retina
  • macula
  • genetics
  • imaging
  • diagnostic tests/investigation
View Full Text

Statistics from


Pseudoxanthoma elasticum (PXE; OMIM# 264800) is an autosomal recessive multisystem disorder, mainly affecting the oculocutaneous and cardiovascular systems.1–3 It is caused by mutations in the ABCC6 gene (chrom. 16p13.1; OMIM# 603234), encoding a transmembrane transporter of which the substrate remains as yet unknown.4 5 The histology of PXE is characterised by ectopic mineralisation and fragmentation of elastic fibres in the reticular dermis, Bruch membrane (BM) and the elastic laminae of blood vessels.1 2

Skin symptoms manifest first in childhood as yellowish discoloured papules or increased skin laxity in flexural areas, although there is significant variability in clinical severity.1–3 Cardiovascular complications are those of accelerated atherosclerosis with occlusive vascular disease.1–3 The PXE retinopathy is characterised by a mottled aspect of the retinal midperiphery (called peau d'orange) most prominent temporal to the macula, and ruptures in BM known as angioid streaks. The latter often, if not invariably, lead to subretinal neovascularisation with a risk for spontaneous and/or post-traumatic haemorrhage with subsequent loss of visual acuity.1–3 Additionally, comets and comet tails, crystalline bodies in the retinal midperiphery and juxta-papillary area, with a variable degree of retinal pigment epithelium (RPE) atrophy have been reported. Although not known to impair visual function, they have been described as the only lesions pathognomonic for PXE.6 The clinical heterogeneity of PXE is also reflected in the retinal phenotype: although invariably present, the retinopathy remains rather limited and uncomplicated in young patients, making an early diagnosis more difficult. Additionally, very similar ophthalmological characteristics have been observed in PXE phenocopies, such as those associated with either beta-thalassaemia or the PXE-like syndrome (OMIM# 610842).7 8 However, ocular signs and symptoms are less severe and uncomplicated in the latter.8

Dilated fundus photography using white light, whether or not combined with fluorescein angiography, has proven to be a useful technique to evaluate the retinal phenotype. More recently, novel non-invasive imaging techniques, such as near-infrared reflectance imaging (IRI) and autofluorescence imaging (AFI), have emerged.9–14 The former allows evaluation of the integrity and health of the RPE and the underlying choroidal structures. Near-infrared light penetrates more easily through the optical media and has a reduced absorption and increased reflection by melanin and haemoglobins when compared with light of wavelengths shorter than 585 nm. Since, at 787 nm, near-infrared light is barely visible, the technique is very patient-friendly. The major chromophore of AFI is lipofuscin, excessive accumulation of which in lysosomes of RPE cells is a common finding in a variety of hereditary and degenerative retinal disorders.9–11 RPE-lipofuscin seems to derive predominantly from incomplete digestion of photoreceptor outer segments, and there is accumulating evidence suggesting an important role of lipofuscin in RPE cell dysfunction and cell loss.15–23 Both techniques are now often used in a whole range of retinal diseases, including inherited retinal dystrophies, both to detect early signs of disease and for a detailed description of the phenotype.9–14

We provide data on 22 molecularly proven PXE patients and 25 PXE carriers, which illustrate that infrared and autofluorescence imaging are more sensitive than white light funduscopy and imaging in visualising both early and more evolved signs of PXE retinopathy. These techniques thus represent an important additional tool in the diagnosis and ophthalmic follow-up of PXE families.

Patients and methods

Twenty-two patients with a clinically (dermatological and ophthalmological examination, skin biopsy) and molecularly (ABCC6 analysis) confirmed diagnosis of PXE and 25 obligate carriers with one proven ABCC6 mutation were studied. As such, phenocopies of PXE associated with beta-thalassaemia and PXE-like syndrome with identical retinal characteristics compared with classic PXE were excluded.7 8 The respective age of patients and carriers ranged from 11 to 68 years (mean 43 years) and from 15 to 76 years (mean 44 years). All underwent a complete ophthalmic examination, including assessment of best-corrected visual acuity (BCVA), slit-lamp examination, funduscopy and extensive fundus imaging. Fluorescein and indocyanin green angiography and optical coherence tomography (Stratus OCT, Carl Zeiss Meditec, Dublin, California) were performed when and where required. Ocular ultrasonography (OTI scan Ophthalmic Ultrasound, Ophthalmic Technologies, Toronto, Canada) was used to evaluate (potential) optic disc drusen. Fundus imaging consisted of white light digital photography with a Topcon TRC-50EX fundus camera (Topcon Corporation, Tokyo). Fundus autofluorescence, red-free and near-infrared reflectance monochromatic images were obtained using a confocal scanning laser ophthalmoscope (cSLO; Heidelberg Retina Angiograph HRA2; Heidelberg Engineering, Dossenheim, Germany), the optical and technical principles of which have been described previously.9–14 The HRA2 cSLO has a small pinhole aperture, suppressing light originating from outside the focal plane to enhance image contrast compared with non-confocal images. It uses an argon blue laser light with a wavelength of 488 nm for excitation and a barrier filter with a cut-off at 500 nm to record fundus autofluorescence and red-free imaging. For near-infrared imaging, the excitation wavelength is 787 nm, and a barrier filter allows light passage above 810 nm. Full-emission spectra are recorded via a polarisation filter to obtain red-free and near-infrared images. Before acquisition of the autofluorescence sequence, illumination and focus level were adjusted to individual requirements at the near-infrared mode of the device, in order to generate high-quality images. Near-infrared, autofluorescence and red-free images, in series of approximately 50 single pictures per eye, were generated using a 30° field-overview mode. The images encompassed the entire macular area, retinal midperiphery and periphery. Following acquisition, any images containing eye movements, blinks or uneven illumination were discarded. After automated alignment and averaging to improve signal-to-noise ratio by image-analysis software (Heidelberg Eye Explorer, Heidelberg Engineering, Dossenheim, Germany), mean images were used for further analysis.9 Finally, HRA 2 software was applied on selected individual and mean images of excellent quality to generate a seamless montage of the entire fundus in AF, RF and near-infrared mode.

All images were analysed for various ocular manifestations of PXE, including angioid streaks, peau d'orange, comets, comet tails and optic disc drusen, in early, neovascular and advanced atrophic stages of the disease. Autofluorescence, red-free and near-infrared were compared with the colour pictures. AF was defined as low (hypoautofluorescence; AF signal lower than background), high (hyperautofluorescence; AF signal higher than background) or unchanged compared with normal background autofluorescence. On near-infrared images the choroidal vessel contrast is usually negative (dark vessels on light fundus background), and the retinal vessels and the optic disc rim appear dark or have dark borders with respect to the surrounding fundus.13

Informed consent was obtained from all patients and carriers, and all procedures were in accordance with the Declaration of Helsinki protocol. The study was approved by the Ethical Committee of the Ghent University Hospital.


Eight male and 14 female PXE patients were included in this study. BCVA ranged from 12/10 to 1/300. All patient fundi examined revealed at least one characteristic of the PXE retinopathy, which was always present in both eyes of each individual (table 1).

Table 1

Ophthalmological characteristics of patients

All patients except the youngest patient, a boy of 11 years old, had angioid streaks in both eyes, which could be visualised with colour imaging red-free imaging (RFI), AFI and IRI.

On standard digital colour imaging, streaks could be divided in those with or without adjacent RPE alterations. Streaks lacking bordering RPE abnormalities were seen as the typical brick-red-coloured streaks, presenting as irregular and jagged lines radiating from a concentric peripapillary ring towards the equator of the eye. In some patients, these were limited to a few almost imperceptible lines (figure 1), while in other patients they presented as thick jagged lines (figure 2) with an occasionally complex interlacing network (figure 3). Angioid streaks were always most prominent at the posterior pole and typically tapered and faded while approaching the equator of the eye, with an arborescent pattern of progressively smaller branches. In rare cases (2/22) they continued beyond the equator as irregular, white and depigmented lines, possibly representing calcium deposition in and around the streak (figure 4).

Figure 1

Right eye of female patient 18, with angioid streaks limited to a few almost imperceptible lines on fundus image radiating from a concentric peripapillary ring towards the equator (A); fine streaks better visible on near-infrared reflectance imaging (B) but not on autofluorescence imaging (C); however, autofluorescence imaging shows a reticular pattern of hyperautofluorescence in the macular area, reminiscent of patterned hypercalcification of the Bruch membrane in the posterior pole.

Figure 2

Right eye of female patient 2 with angioid streaks shown thick jagged lines.

Figure 3

Angioid streaks in the left eye of male patient 3 forming a complex interlacing network; in addition, atrophy and scar of the subretinal neovascular membrane after consecutive treatment with photodynamic therapy and intravitreal bevacizumab are visible in the macular area; white lesions around some angioid streaks may represent chorioretinal calcification.

Figure 4

Angioid streaks in periphery may become (partially) calcified with a whitish aspect, as seen in the nasal periphery of the left eye of female patient 6.

With near-infrared reflectance imaging, angioid streaks appeared as uniform, well-demarcated dark fissures against a lighter background, even when they passed unnoticed on colour imaging. This classic type of streaks had no visually obvious correlate on AFI, although traces could sometimes be discovered when comparing with IRI (figures 1, 6). However, autofluorescence imaging sometimes showed a reticular pattern of hyperautofluorescence in the macular area, reminiscent of patterned hypercalcification of the Bruch membrane in the posterior pole (figure 1C). Alternatively, hyperautofluorescence of this reticular pattern only, or at least partially due to increased lipofuscin at the level of the RPE, is not impossible, albeit less likely. Indeed, blockage of light reflected from choroidal vessels on indocyanin green angiography, seen in such patients, supports a hypothesis of calcification of BM (data not shown). Some PXE patients will also develop avascular detachments of the retinal pigment epithelium in such areas, suggesting disturbed fluid permeability of BM. This is also likely to be due to calcification (data not shown). Only the youngest of 22 patients did not have angioid streaks (figure 8).

Figure 5

Aspect of feathered type of angioid streaks in the right eye of young female patient 8 on white light (A), infrared (B), autofluorescence (C) and red-free imaging (D); note the better visibility of the streaks as whitish areas on near-infrared reflectance imaging and red-free imaging than on white light fundus imaging; streaks appear as dark areas on the autofluorescence image.

Figure 6

Parallel white light, infrared and autofluorescence images of infrapapillar angioid streak in the left eye of female patient 15: whereas the streak is better visible on the infrared reflectance imaging (B) than on the fundus image (A), it is difficult to see on the autofluorescence imaging (C).

In all fundi, RPE alterations, mostly loss of pigment and more rarely hyperpigmentation, were observed adjacent to some of the streaks.

RPE hypopigmentation, seen in 26 fundi of 13 patients, yielded a ‘feathered’ appearance to the streak on colour imaging (figure 5). With AFI, the feathered streaks and surrounding area depleted of pigment appeared as a triangular area of hypoautofluorescence, with the base of the triangle oriented towards the optic disc. Near-infrared imaging depicted these feathered streaks in a similar way to how colour photography did. In older and larger streaks, small islands of normal autofluorescence could be observed within the streak. Such islands in the older, larger streaks are more visible in IRI than on colour pictures (figures 7, 11).

Figure 7

Right eye of female patient 20: small islands of normal retinal tissue within older, broader streaks showed normal autofluorescence, but are better visible in infrared reflectance imaging (B) than on colour pictures (A), autofluorescence imaging (C) or red-free imaging (not shown).

Figure 8

Peau d'orange, much more extensively visible on infrared reflectance imaging (B) than on the colour fundus image (A) in 11-year old male patient 5; note the extension of peau d'orange changes around the whole posterior pole on infrared reflectance imaging; no angioid streaks are present at this early age, although comets and comet tails are visible.

Figure 9

Hyperpigmentation due to retinal pigment epithelium hyperplasia in macular scar after photodynamic therapytreatment (A); hyperpigmentation and atrophy appear as dark areas on autofluorescence imaging (C); hyperpigmentation appears as a light zone on infrared reflectance imaging (B), and black on red-free imaging (D).

Hyperpigmented RPE alterations were seen as dark brown, round or oval zones of variable size along areas of streaks or in atrophic macular scars in four patients. Equal in aspect to atrophic areas, on AFI these appear as hypoautofluorescent lesions with the same size (figure 9). On IRI, a limited hyperintensity compared with the background was detected. However, this would pass unnoticed on the basis of IR evaluation alone (figure 9).

Figure 10

Burst of comets and comet tails inferior to optic disc of female patient 19; note that the comets are best visible on red-free imaging (D) compared with colour fundus imaging (A), infrared reflectance imaging (B) and autofluorescence imaging (C); in addition, broad angioid streaks emanating from the optic disc are best seen on infrared reflectance imaging (B) and the colour fundus image (A).

Peau d'orange, which, according to some investigators, represents a widespread and patchy increase in pigment in the RPE cells,1 2 could potentially also be due to changes in BM architecture. Indeed, calcification and fragmentation of elastic fibres, which lead to dermal peau d'orange lesions, may also underlie the fundus lesions. The speckled or mottled appearance from the macula to the midperiphery was seen in 36 fundi. On colour pictures, this fine mottling was visible in the nasal retina and temporal fundus between the macula and the equator of the eye. On AFI, no difference in autofluorescence was observed in regions of peau d'orange. However, IR imaging always revealed a diffuse speckled pattern of peau d'orange, extending beyond the temporal midperiphery (figure 8). In 55% of fundi (24/44), the peau d'orange area covered the entire midperiphery to the equator and the posterior pole, including the entire macula. Extensive macular atrophy and/or scarring precluded evaluation of peau d'orange in 10 fundi. The overall extent of peau d'orange was best appreciated on composite near-infrared images of the fundus, with often a far larger area considered affected than that visible on white light digital fundus images.

Comets and/or comet tails were observed as solitary, subretinal, nodular, white bodies with a tapered white tail of RPE atrophy extending posterior to the body on colour images in 90% (38/42) of the examined fundi. The body may have some pigmentation at its margin. AFI showed small, punctiform hyperautofluorescent lesions but could not detect all comets individually. Near-infrared imaging was able to visualise several comets, which went undetected on colour, as small white hyperintensities. However, the most sensitive technique in detecting those pathognomonic lesions was RFI. With the latter, all of these were seen as whitish flecks on a grey background (figure 10). In some fundi, a spray of comets and comet tails was observed, creating an aspect of ‘comet rain’ (figure 10).

Figure 11

Superficial optic drusen in LE of patient 11 (A) visualised unambiguously on autofluorescence imaging because of hyperautofluorescence of the optic disc (C); disc drusen could be suspected on near-infrared reflectance imaging (B) and red-free imaging (D) albeit they did not stand out; note the areas of normal retinal tissue within broad angioid streaks inferotemporal of fovea.

Optic disc drusen, thought to be composed of hyalin-like calcific material within the substance of the optic nerve head, were seen binocularly in two patients and uniocularly in one patient using colour imaging and AFI. On standard ophthalmoscopy, buried drusen could be suspected due to an elevated disc with a scalloped margin and without a physiological cup. Surface drusen of the disc appear as waxy pearl-like irregularities. Drusen on the surface of the optic nerve head may be visible with autofluorescece, but if they lie deeper the optic disc may look normal. In such cases, ultrasonography is indispensable. On AFI, superficial drusen were visualised unambiguously in all examined fundi because of the hyperautofluorescence of the optic disc without attenuation of the signal by overlying tissue (figure 11). On confirmation B-scan ultrasonography, the drusen were recognised by their high acoustic reflectivity. Optic disc drusen could be suspected on near IR imaging and RF imaging afterwards, albeit they did not stand out in these images.

Fibrovascular scars following choroidal neovascularisation in the macular area were observed in 17 eyes. Autofluorescence in these end-stage fundi was very low in areas of atrophy and high in the junctional zone separating atrophic zones from surrounding, more normal retina. In five patients, additional lobular hypoautofluorescent lesions were noted on AFI, adjacent to but separated from the fibrovascular scars. In two patients, previously treated with photodynamic therapy, a paler zone was observed around the area of choroidal neovascularisation on colour imaging (figure 3). On AFI, the borders of a uniform hypoautofluorescent zone correlate with the borders of the paler zone found around the fibrovascular scar on colour pictures. These paler zones appeared as an atrophic area on near IR imaging.

In eight of the 25 carriers (three males and five females) examined, comets and comet tails were observed. In three carriers, the appearance of the comets was similar to that seen in PXE patients. However, in five carriers, comets were limited to small white dots with neither hyperpigmented borders nor comet tails. No other abnormalities of fundus autofluorescence or near-infrared imaging, such as limited peau d'orange or angioid streaks, were observed.


Visualising the PXE retinopathy with AF and IR imaging

Pseudoxanthoma elasticum manifests with cutaneous, ophthalmological and cardiovascular symptoms, with considerable morbidity.1–3 Because of significant intra- and interfamilial variability in phenotypic severity, making a correct clinical diagnosis of PXE is often difficult, with consequent delay of a potentially essential, early intervention.

This study aimed to evaluate whether novel imaging techniques, such as monochromatic near-infrared reflectance and autofluorescence imaging with cSLO, are able to better visualise the PXE retinopathy compared with standard digital colour fundus pictures, thereby facilitating early diagnosis.

Angioid streaks were observed in all but one fundi, as expected, and could be classified into two types, based on whether or not they were accompanied by surrounding RPE alterations. All eyes contained a mixture of both types. In the absence of RPE alterations, AFI was not able to visualise the streaks—even with multiple contrast settings. In contrast, angioid streaks were very distinct on IR imaging compared with colour fundus photography, and were visible as well-demarcated, dark fissures with high contrast to surrounding tissues. Hence, streaks either absent or initially unnoticed on colour pictures were detected with IRI. Streaks with concomitant RPE alterations showed a feathering pattern, pigmented borders and/or deposits in the middle of the streaks. On AFI, the overall autofluorescence intensity of these feathered streaks appeared lower, which might be explained by loss of lipofuscin in the RPE cells adjacent to the streaks. The pathological substrate of this finding could be absence of photoreceptors, RPE atrophy, increased pigment clumping within RPE cells, a decrease in outer segment phagocytosis by RPE cells or a combination of several of these. Since none of the patients suffered from scotomas in the areas of streaks, which would be expected in case of absent photoreceptors or RPE atrophy, the latter three possibilities seem the most plausible.24 The hyperpigmented borders appeared as uniform dark hypoautofluorescent zones with the same size compared with those in the colour images. The decreased autofluorescence could indeed be due to increased pigment clumping within RPE cells at the borders, blocking AF, and attenuation due to dysfunction of RPE cells overlying the streak which lose their lipofuscin.16 The presence of dotted areas of normal autofluorescence within longstanding angioid streaks visualised on AF pictures may represent normal zones of RPE and potentially BM in the centre of streaks, due to fragmentation of the main body of the RPE at the streak margin. For these streaks associated with RPE alterations, IR imaging did not reveal any significant difference compared with colour photography.

Media opacities including lens opacification may result in AF images that are of insufficient quality if they can be obtained at all. Such opacities do not hamper IR imaging, due to better penetration of long-wavelength light.

Together, these observations suggest that a combination of AF and IR imaging is optimal in visualising the different types of angioid streaks and as such is superior to colour imaging for the early detection and follow-up of angioid streaks.

It has been suggested that angioid streaks could disappear in patients with longstanding PXE.25 We observed streaks becoming obscured by extensive chorioretinal scarring and accompanying proliferation of the retinal pigment epithelium. However, on colour, AF and IR imaging, distinctive remnants of streaks persisted at the periphery of such lesions, suggesting that angioid streaks are persistent throughout the natural history of the disorder.

Although peau d'orange has previously been described as granular stippling on AFI,25 the pigment alterations seen on colour imaging did not yield any significant difference in fundus autofluorescence in the present study. On near-infrared reflectance imaging, we observed the peau d'orange to be more widespread than originally described, and present throughout the posterior pole and the whole midperiphery. This might be due to pure technical reasons, since IR imaging is less compromised by macular lutein and zeaxanthin pigment. This observation further strengthened our belief that this technique has a significant higher sensitivity in detecting changes of deeper structures such as the outer retinal layers and the Bruch membrane compared with colour fundus imaging.

We found near IR reflectance imaging to be also superior in visualising comets and comet tails in combination with RF imaging. Even comets barely visible or absent on colour pictures could be seen with IRI and RFI. In contrast, on AFI not all comets were recognised. If comets and comet tails indeed are the only pathognomonic ocular characteristic of PXE,26 they are of significant diagnostic value, especially in young patients in whom angioid streaks are often not yet present (figure 8). Since they can also occur in carriers, a thorough skin evaluation and molecular analysis of the ABCC6 gene are indicated when such lesions are detected.

Drusen of the optic disc seem to be more common in PXE than in the general population.27 AFI proved to be efficient in detecting these optic disc drusen which were confirmed on ultrasound. It may be assumed that a common process of abnormal mineralisation might be the predisposing factor both for drusen of the optic nerve head and for comets and comet tails.

In end-stage fundi, increased autofluorescence in the junctional zone around atrophic areas may represent areas of future RPE atrophy.10 However, these findings are of limited predictive value. The standard examinations to detect subretinal neovascularisation in such patients remain fluorescein and indocyanine green angiography.


Because of extensive variability in clinical severity, the diagnosis of PXE is often delayed until ophthalmological complications occur. Since novel ophthalmic therapeutic agents, such as different agents with anti-VEGF action, have become available, the need for an early diagnosis has become even more important. The present study showed that near-infrared reflectance, red-free and autofluorescence imaging are of significant clinical relevance in PXE, as they contribute to improved imaging quality and hence early diagnosis and more adequate follow-up. Serial, prospective investigations of patients with these techniques can give us important clues as to the natural history of this complex disorder. As such, we consider them part of the standard evaluation of novel and known PXE patients and their family members.


The authors wish to acknowledge the patients for their kind collaboration in this study.


View Abstract


  • Funding Research Foundation-Flanders (FWO Vlaanderen), Egmontstraat 5—1000 Brussels, Belgium.

  • Competing interest None.

  • Ethics approval Ethics approval was provided by the Ethical Committee, Ghent University Hospital.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • At a glance
    Harminder S Dua Arun D Singh