To investigate the role of fundus autofluorescence (FAF) imaging in the diagnosis of macular telangiectasia type 2 (MacTel) and to describe disease-associated FAF patterns and their origin.
Cross-sectional multicenter study
FAF images were collected from the multicenter MacTel Natural History Observation and Registry Study. In a first qualitative approach, common FAF phenotypes were defined and correlated with multimodal imaging. We then evaluated how many eyes showed FAF changes, and temporal vs nasal asymmetry of FAF changes was graded. Finally, 100 eyes of MacTel patients and 100 control eyes (50 normal eyes and 50 eyes with other macular diseases) were combined and 2 masked graders assessed the presence of MacTel based on FAF images alone.
The study included 807 eyes of 420 patients (33 eyes were excluded owing to poor image quality). Loss of macular pigment, cystoid spaces, pigment plaques, neovascular membranes, and ectatic vascular changes commonly caused characteristic changes on FAF images. All MacTel patients had macular FAF changes in at least 1 eye. In 95% of eyes, these changes were more pronounced temporally than nasally. Common FAF patterns were increased (60%) and mixed/decreased FAF (38%) and/or visibility of vascular changes such as blunted vessels or ectatic capillaries (79%). Based on those features, high diagnostic performance was achieved for detection of the disease based on FAF alone (Youden index up to 0.91).
The study demonstrates that MacTel is consistently associated with disease-specific changes on FAF imaging. Those changes are typically more pronounced in the temporal parafovea.
M acular telangiectasia type 2 (MacTel) is a usually bilateral retinal condition characterized by progressive neurosensory atrophy with concomitant changes of the macular retinal vasculature. Retinal alterations and dysfunction usually remain limited to an oval area centered on the fovea, with a horizontal diameter not exceeding the distance between the temporal optic disc margin and foveal center, and the vertical diameter not exceeding approximately 0.8 times this distance (“MacTel area”). The earliest changes develop in the temporal sector. Even in late disease stages involving the entire MacTel area, the temporal sector usually remains the epicenter of the disease with the most pronounced alterations.
Previously, fundus fluorescein angiography (FFA) had been the standard imaging modality to confirm the diagnosis of MacTel. However, FFA is increasingly being replaced by noninvasive imaging modalities. Many retina specialists now diagnose MacTel based on funduscopy and optical coherence tomography (OCT) assessment in routine clinical practice, possibly complemented by blue reflectance and fundus autofluorescence (FAF) imaging. Nevertheless, only few studies on MacTel included a detailed analysis of FAF imaging, and these were either limited by a small sample size or simple grading schemes. Hence, the various characteristic FAF patterns and their origin have not yet been investigated in detail and are not widely recognized. ,
The aim of our study was to describe the FAF patterns in more detail, to establish a simple guide for evaluation of FAF images and to evaluate its diagnostic usefulness for the detection of MacTel.
Images from 4 clinical study sites (Bonn, London, Münster, Oxford) of the MacTel Study were collected via the Moorfields Eye Hospital Reading Centre (MEHRC) database. Cases were included if the diagnosis was confirmed by the MEHRC and if a 30-degree FAF image recorded with a confocal scanning laser ophthalmoscope (Spectralis HRA; Heidelberg Engineering, Heidelberg, Germany) of sufficient quality was available. The MacTel study was approved by the local medical ethics committees and adhered to the tenets of the Declaration of Helsinki. Informed consent was given by each subject before the start of the study.
For a detailed characterization of common FAF patterns, the full panel of multimodal imaging included in the MacTel study (color fundus photography, 30-degree FAF, FFA, blue reflectance imaging, and OCT) was analyzed in 100 consecutive eyes for which 30-degree macular pigment optical density (MPOD; recorded with dual-wavelength FAF technique) measurements were also available. Blue reflectance, FAF, and early- and late-phase FFA images were acquired using a scanning laser ophthalmoscope (Spectralis HRA or HRA2; Heidelberg Engineering, Heidelberg, Germany), and MPOD images were recorded as described previously using a modified 2-wavelenght HRA classic (Heidelberg Engineering) 3 : MPOD was calculated from 2 averaged FAF images (each consisting of up to 16 single images) derived from excitation using 488-nm and 514-nm light, respectively, in conjunction with a barrier filter that blocks wavelengths shorter than 560 nm. Because of the maximum absorbance of macular pigment at 460 nm, the AF signal recorded using 488-nm excitation light is strongly dependent on macular pigment density, whereas the AF signal of the 514-nm AF image is not. Subtraction of the 2 averaged images results in an image representing MPOD distribution, which was saved with an intensity resolution of 8 bits/pixel, and a 512 × 512-pixel resolution. Spectral domain OCT macular volume scans (15 degrees horizontally × 10 degrees vertically, ~30 µm distance between individual B-scans, recorded in the high-resolution mode with 768 A-scans per B-scan) and single-line scans through the foveal center (30 degrees in high-speed mode with 768 A-scans per B-scan) were acquired using Spectralis OCT (Heidelberg Engineering). The automatic real-time mode was used to improve the signal-to-noise ratio, averaging 9 frames for volume scans and up to 100 consecutive frames for single-scan recordings.
Next, simplified categories of alterations on FAF images were defined to analyze the frequency of specific features in a large cohort of more than 400 patients. The following alterations were not mutually exclusive and could coexist in the same eye:
Presence of FAF changes, and whether or not these were limited to the “MacTel area”
Asymmetry of FAF changes with more pronounced alterations in the temporal MacTel area (ie, temporal to a vertical line through the foveal center)
The pattern of FAF changes: increased compared to normal ± decreased signal (mixed AF)
Visibility of vascular abnormalities (blunted venules and/or ectatic capillaries)
Changes were considered to be more pronounced temporally than nasally, if either there was (1) any abnormal FAF signal temporal and a normal signal nasally, (2) an increased signal nasal to the fovea and a mixed signal temporally, (3) the size of the area with FAF changes was larger temporal than nasal to the fovea, or (4) there were vascular changes only or mainly affecting temporal vessels. Images were deemed gradable if at least the first 3 categories could be assessed. Grading was performed by L.P. and ambiguous cases were discussed with P.C.I.
Youden index was used as performance measure for evaluating the ability of AF images to avoid misclassification. Youden index evaluates the ability of a classifier to avoid misclassifications but putting equal weights on a classifier’s performance on both the positive and negative cases. The Youden index as a performance index that provides the maximum differentiability of classifier when equal weight is given to sensitivity and specificity, thus minimizing the total misclassification rates. For this purpose, 100 FAF images from 100 consecutive MacTel patients from one center (London) were mixed with FAF images of 50 normal controls and of 50 patients with different macular diseases that may mimic MacTel, which were obtained as part of standard care in medical retina clinics between January 1, 2016, and January 31, 2019. The latter included cases with central serous retinopathy (n=11), age-related macular degeneration (n=9), vitelliform lesion (n=7), epiretinal membrane (n=5), pathologic myopia (n=5), pigment epithelial detachment of various aetiology (n=4), retinal vein occlusion (n=3), diabetic maculopathy (n=1), macular hole (n=1), PRPH2 -related macular dystrophy (n=1), vitreomacular traction (n=1), perifoveal exudative vascular anomalous complex (n=1), and acute macular neuroretinopathy (n=1). Two masked retinal specialists (P.C.I., M.G.) independently, separately assessed FAF images in random sequence without additional information from multimodal imaging and to classify them as either having MacTel or not. Only one eye from each patient was assessed without information on the phenotype of the fellow eye.
|Phenomenon||Phenotype||Shown in Figure|
|Macular pigment-related||Wedge-shaped increase temporal to the fovea or an oval area of increased signal||1, A through D|
|Foveal cystoid cavities||Varialbe forms and sizes possible||1, B|
|Photoreceptor atrophy||Larger circular area of increased autofluorescence (usually higher than that due to loss of macular pigment)||1, C|
|“Yellow spot”||Central, symmetrical increase in FAF||1, D|
|Full-thickness macular hole||Round, central increase in FAF||1, E|
|Pigment plaque||Irregular shape next to abnormal vessels, sharp boarders||3, A|
|Neovascular membrane||Area of irregular, alternating increased and decreased FAF||3, B|
|Retinal pigment epithelial atrophy||Small areas of reduced FAF|
|Blot hemorrhages||Different sized areas of reduced FAF|
|Blunted venules||Vessels of a slightly larger caliber than would be expected at that location||3, C|
|Ectatic macular capillaries||Small vessels can be seen adjacent to the macula, which could usually not be displayed||3, D|
Fundus Autofluorescence Features and Multimodal Imaging
Features on FAF images and the associated findings on multimodal imaging were investigated in 100 eyes with available full retinal imaging data set, including fundus photography, macular pigment assessment using MPOD maps, OCT, and blue reflectance imaging. Table 1 provides a summary of the FAF changes as described below.
Changes on FAF images were usually limited to the MacTel area. In mild disease without obvious loss of the photoreceptor layer, a relatively increased AF signal may be limited to a wedge-shaped area temporal to the foveal center ( Figure 1 , A) or may involve the entire MacTel area ( Figure 1 , B). The abnormal AF signal was usually not higher than outside the MacTel area and was associated with reduced macular pigment on MPOD maps and/or hyporeflective cavities on OCT images ( Figure 1 , A and B). Macular pigment loss was usually associated with increased reflectivity on confocal blue reflectance images. Increased FAF associated with hyporeflective cavities is usually limited to the fovea ( Figure 1 , B).