1 Introduction

Macular Telangiectasia Type 2 (MacTel2) [3] is a disease of the macula, causing loss of central vision. Typically, in MacTel2, perifoveal vessels leak and/or become dilated. Fluid from leakage causes macular swelling [17], which can decrease visual acuity.

MacTel2 can be present without noticeable symptoms in the early stages. As the disease progresses, blurring, distorted vision, and loss of central vision can be detected and lead to a loss of central vision over a period of 10 to 20 years. MacTel2 does not affect side vision, nor does it usually lead to total blindness. Due to the lack of early symptoms and an adequate therapy for MacTel2, it is important to obtain early and regular eye exams. It may be an inherited disease and, especially in its later stage, it could be mistaken for age-related macular degeneration (AMD) [6], due to similar patterns of neovascularization. But all existing knowledge is restricted to advanced stages of the disease, while its beginnings and early stages are still unclear [3].

MacTel2 is diagnosed with the help of optical coherence tomography (OCT) [12] and specified in detail using fluorescein angiography (FA) [3]. There is no therapy of MacTel2. However, in the case of neovascular vessel growth in the course of MacTel2, the injection of anti-vascular growth agents has been shown to be effective.

Fig. 1.
figure 1

(a) OCT B-scan with segmented layers corresponding to the white dotted line in (b). Inner Limiting Membrane (ILM), Bruch’ s Membrane (BM) and Choroid-Sclera Interface (CSI). IS/OS/EZ denotes the Inner Segment/Outer Segment/Ellipsoid Zone. (b) Choroidal (BM-CSI) thickness map overlaid color-coded on fundus image. The indicated locations correspond to the central (1) and paracentral (2) choroidal thickness measurements, respectively.

Full size image

Choroidal thickness (ChT) has been demonstrated to be an important biomarker for MacTel2 [2, 7, 9] and other pathologies such as age-related macular degeneration (AMD), diabetic macular edema (DME) [17], glaucoma [8] and juvenile myopia [14]. Therefore, monitoring of the choroidal thickness delivers insight into the pathogenesis of such diseases and helps in the planning of their treatment. With this feasibility study, we want to evaluate the impact of the choroidal thickness over time in MacTel2.

State-of-the-Art Research

The authors of [11] examined microcystoid spaces in MacTel2 patients. Evidence was found in the pathogenesis of MacTel2 because the Müller cells dysfunction was reported. Not much research has been done to correlate choroidal thickness changes in the course of MacTel2. The authors of [1] examined choroidal thickness changes in eyes and concluded that the thickness of the choroid did not vary between the eyes of MacTel2 patients and those of age-matched healthy subjects. In contrast, the authors of [10, 15] compared subfoveal choroidal thickness in a similarly large group of MacTel2 patients to that of healthy subjects. They concluded that patients with MacTel2 show a clearly thicker choroid in the subfoveal area.

In this paper, we examine if a correlation between changes in the thickness of the choroid and the development of MacTel2 exists. We validate our results with the help of a state-of-the-art method [5] and manual segmentation by two experts.

Fig. 2.
figure 2

(a) Right eye of a 53-year-old woman with MacTel2. The horizontal B-scan shows the increased subfoveal choroidal thickness of 422 \({\upmu }\)m (arrow). Typical findings of MacTel2 include an intraretinal outer retinal cavity and loss of the IS/OS/EZ indicated with the dashed box. (b) Healthy right eye of a 53-year-old woman from the control group.

Full size image

2 Material and Method

The basis of this retrospective study was formed by forty-five MacTel2 patients who were enrolled at Moorfields Eye Hospital in the MacTel Natural History Observation and Registry Study from June 2012 until March 2016.

Their mean age was 57.1 (range 42 to 71). The study was approved by the local institutional review board and conducted according to the tenets of the Declaration of Helsinki. The volunteers were divided into three groups: the first group was measured a second time after 1 year, the second group after 2 years, the third after 4 years. Both eyes were measured.

Optical coherence tomography was acquired using the Heidelberg Spectralis, Human Reliability Analysis 2 (HRA2) system (Heidelberg Engineering, Heidelberg, Germany). The scan pattern for MacTel2 patients was between \(3.8 \times 2.5 \times {1.9}\,\mathrm{mm{^3}}\) and \(4.4 \times 2.9 \times {1.9}\,\mathrm{mm{^3}}\) including 49–261 B-scans per volume, imaging averaging 8–12 scans, interslice distance was 11–30\(\,{\upmu }\)m. The scan pattern for all healthy subjects was defined by raster lines of \(4.5 \times 3.0 \times {1.9}\,\mathrm{mm{^3}}\) (261 B-scans, averaging 8–12 scans, interslice distance was 11 \({\upmu }\)m).

The thickness of the choroid was calculated for each A-scan, defined as the Euclidean distance between the Bruch’ s Membrane (BM) and the posterior surface of the choroid, delimited by the Choroid-Sclera Interface (CSI, see Fig. 1(a)). Choroidal thickness maps for the \(6 \times 6\) \(\mathrm{mm{^2}}\) macula-centered region imaged by spectral-domain (SD)-OCT scans were created (see Fig. 1(b)) and the average thicknesses were reported in micrometers. The changes in the choroidal thickness, measured between two visits, were automatically detected using CRAR [14], a registration-based algorithm, especially developed for longitudinal studies, which tackles the problems that occur in using image segmentation (i.e. low contrast, loss of signal and artifacts), the common approach to localize the exact position of the CSI.

Instead of segmentation, CRAR uses a piecewise rigid image registration-based approach, in which the changes in thickness in the region around the CSI are examined. In such a way the exact position of the CSI is no longer needed. In CRAR the uniform smoothness of the results is supported by a regularization which matches the anatomic structure of the eye [14].

In the next step, CRAR’ s results were compared with those obtained using a graph search-based state-of-the-art segmentation method [5] and manual expert segmentation by two independent experts. Finally, the subfoveal choroidal thickness was determined. Since MacTel2 manifests itself in the juxtafoveal region, it is reasonable to assume that the subfoveal choroidal thickness is reliable for this kind of analysis (see Fig. 2(a)).

Fig. 3.
figure 3

(a) Bland-Altman plot representing the agreement between the choroidal thickness measured by the two experts. (b) Scatter plot comparing the relationship of subfoveal choroidal thickness and age in MacTel2 eyes (red squares) vs healthy control eyes (black dots). The dashed lines indicate 95% normal tolerance limits for choroidal thickness per year.

Full size image

3 Result

Subfoveal choroidal thickness was manually measured independly by two experts, who showed an agreement with an interobserver correlation \(R^2\) of 0.96 for both the MacTel2 affected and the healthy eyes. A Bland-Altman diagram in Fig. 3(a) shows a mean interobserver difference of 2.32 \({\upmu }\)m, while the 95% limits of agreement are −18 and 19 \({\upmu }\)m, respectively.

There was a predominance of females among both the MacTel2 subjects (32 \(\widehat{=}\) 71%) and healthy subjects (28 \(\widehat{=}\) 62%). However, there was no difference in the results between the groups with respect to gender (\(P=0.068\), Fisher’ s exact test). Mean ages in the study and control groups were 58.4 ± 9.3 (\(\widehat{=} \) 49–68 years) and 52 ± 15.8 (\(\widehat{=} \) 36–68 years), respectively.

The relationship between changes in the subfoveal choroidal thickness and age in the MacTel2 and control group are shown in Fig. 3(b). All changes in the subfoveal choroidal thickness of both the MacTel2 and control group subdivided into time intervals between the two measurements are illustrated in Fig. 4(a).

On average, the subfoveal choroidal thickness of the MacTel2 patients increased in the time interval between the two measurements with 2.47 ± 7.89 \({\upmu }\)m (after 1 year), 4.32 ± 13.12\(\,{\upmu }\)m (2 years) and 10.85 ± 21.43 (4 years). In the case of healthy eyes, the same measurements showed a decrease of the subfoveal choroidal thickness of −0.71 ± 6.35 \({\upmu }\)m, −3.92 ± 10.76 \({\upmu }\)m, −8.37 ± 15.01 \({\upmu }\)m, respectively. The total mean increase of the MacTel2 group (over all time intervals and subjects) was 5.88 ± 14.15 \({\upmu }\)m, while in case of the control group this decrease was −4.33 ± 10.71 \({\upmu }\)m (see Fig. 4(b)). On average, the subfoveal choroidal thickness decreased by 2.55 \({\upmu }\)m per year.

There is a decrease in the subfoveal choroidal thickness in adults ranging from 14–54 \({\upmu }\)m every ten years [4, 16]. On the other hand, the thickness increased in the MacTel2 group by 2.79 \({\upmu }\)m per year. The increase of the MacTel2 slope showed no correlation to the decrease of the control group slope, nor the other way round (\(P = 0.54\), test of interaction).

For the whole MacTel 2 group, the state-of-the-art algorithm provided mean subfoveal choroidal thickness changes of 5.02 ± 10.55 \({\upmu }\)m, while both experts in average 6.84 ± 21.14 \({\upmu }\)m. For the whole control group the detected subfoveal choroidal thickness changes were \(-4.81\) ± 12.87 \({\upmu }\)m and \(-6.32\) ± 19.14 \({\upmu }\)m. After conducting a power analysis, we could attest a superior performance of CRAR vs the state-of-the-art-method (\(P < 0.05\), medium effect size as Cohen’ s d in the range \(0.43-0.51\), paired t-test), as well as vs the manual expert segmentation (\(P < 0.001\), large effect size as Cohen’ s distance \(d >0.8\), paired t-test).

4 Discussion

As mentioned above, our main findings is the increase of the choroidal thickness in MacTel2 patients compared to healthy subjects. Similar results were mentioned in [10, 15], where authors also came to the conclusion that a statistically significant positive correlation was found exist between subfoveal choroidal thickness and age in MacTel2 subjects, while subfoveal choroidal thickness and age are negatively correlated in healthy adults [4, 16]. A thickening of the choroid may be an early manifestation of MacTel2 eyes and therefore be potentially useful for diagnosis and monitoring. In previous studies of MacTel2 eyes, there has been no mention of an abnormal choroid [13]. However, diurnal changes were not considered in this study what can be a limiting factor. Müller cell dysfunction might be behind the pathogenesis of MacTel2, because Müller’ s cell loss is commonly associated with macular pigment depletion [13]. Müller cell dysfunction may potentially lead to changes not only in the retinal but also in the choroidal vessels.

Based on our findings of choroidal involvement, the thesis that MacTel2 is only linked to the retina needs to be reconsidered as there is a reasonable assumption that the choroid may also be involved. Another limitation regarding this study is that we only measured the subfoveal choroidal thickness, whereas the rest of the choroid was not evaluated. But because the disease activity is primarily present in the juxtafoveal region, it is nevertheless reasonable to assume that the subfoveal choroidal thickness is reliable for this kind of analysis.

Finally, the age disparity between patients with MacTel2 and healthy subjects should be taken more into consideration in future studies, although even when adjusted for age, using analysis of covariance as shown in [10], the differences in choroidal thickness measurements between MacTel2 and control eyes persisted.

Fig. 4.
figure 4

(a) Subfoveal ChT changes detected by CRAR in the control (C) and MacTel2 (M) group subdivided into the different time intervals between the two measurements. (b) Total average of subfoveal ChT changes for MacTel2 and control group.

Full size image

5 Conclusion and Outlook

A thickened choroid may be a valuable diagnostic and monitoring clue in identifying MacTel2, and this may lead to better understand its pathogenesis. Further studies are needed to validate the prognostic value of these findings and to determine whether there are quantitative or qualitative changes in the choroid that predict the onset and progression of this disease.