OCTA-Based Identification of Different Vascular Patterns in Stargardt Disease

Abstract

Purpose

The aim of the present study was to analyze quantitative optical coherence tomography (OCT) and OCT angiography (OCTA) parameters to identify clinically relevant cutoff values able to detect clinically different Stargardt's disease (STGD) subgroups.

Methods

Consecutive STGD patients were recruited and underwent complete ophthalmologic examination, including multimodal imaging. Several quantitative parameters were extracted both from structural OCT and OCTA images and were statistically analyzed. A post hoc analysis was performed to identify a quantitative cutoff able to distinguish two clinically different STGD subgroups. Main outcome measures were total retinal thickness, central macular thickness (CMT), retinal layers thickness, retinal and choroidal hyperreflective foci (HF) number, vessel density (VD), vessel tortuosity (VT), vessel dispersion (Vdisp), and vessel rarefaction (VR) of macular and optic nerve head plexa.

Results

Overall, 54 eyes of 54 STGD patients (18 males) and 54 eyes of 54 healthy age- and sex-matched controls were included in the analysis. All quantitative parameters resulted significantly worse in STGD than controls (P < 0.01). Moreover, a VT cutoff of 5 allowed to distinguish the following two categories: a functionally and anatomically better STGD group and a worse group. BCVA resulted 0.42 ± 0.28 logMAR in the best group versus 1.09 ± 0.36 logMAR in the worst (P < 0.01). Structural OCT and OCTA parameters significantly differed between the two STGD groups.

Conclusions

Quantitative OCTA was able to detect different morphofunctional STGD phenotypes.

Translational Relevance

OCTA-based classification of STGD patients detected different patients' subgroups, differing in terms of morphologic and functional features, with a potential impact on clinical and research settings.

Introduction

Stargardt's disease (STGD) is a common hereditary retinal dystrophy leading to irreversible bilateral visual acuity loss.1 It is characterized by a diffusely compromised retinal pigmented epithelium (RPE)/photoreceptor complex, with heterogeneous manifestations detectable both at the macular level and in the middle and extreme periphery.2,3 Recently developed noninvasive imaging technology, principally optical coherence tomography (OCT) and OCT angiography (OCTA), has led to a better understanding of the underlying pathologic mechanisms governing the onset and progression of the atrophic process, revealing a connection with the damage to the retinal and choroidal vascular network.4–6

The aim of the present study was to analyze quantitative OCTA parameters in order to identify clinically relevant cutoff values.

Materials and Methods

The study was designed as an observational cross-sectional analysis. Consecutive patients affected by STGD were recruited in the Heredodystrophies Unit of the Department of Ophthalmology of San Raffaele Hospital in Milan. Each patient provided signed informed consent before the examination, in accordance with the Declaration of Helsinki. The study was approved by the Ethical Committee of the Vita-Salute San Raffaele University of Milan. The inclusion criteria were as follows: ABCA4 genetically confirmed diagnosis of STGD and age from 18 to 70 years. The following exclusion criteria were applied: refractive errors greater than ±3 diopters (D), increased media opacity, glaucoma and other types of retinal and/or optic nerve diseases (e.g., diabetic retinopathy), any ophthalmic surgery in the last 6 months, any systemic conditions potentially affecting the results of the study.

All the patients underwent a complete ophthalmologic examination, including best-corrected visual acuity (BCVA) using standard Early Treatment Diabetic Retinopathy Study charts, slit-lamp evaluation of anterior and posterior segments, intraocular pressure measurement with a Goldmann applanation tonometer, color fundus photography, and fundus autofluorescence. In addition, structural OCT and OCTA images were acquired by means of Heidelberg Spectralis HRA+OCT (Heidelberg Engineering, Heidelberg, Germany) and SS-OCT DRI Topcon Triton (Topcon Corporation, Tokyo, Japan), respectively. The OCT acquisition protocol included high-resolution radial and raster scans, while OCTA analysis included 3 × 3- and 4.5 × 4.5-mm scans, centered at the level of both the macula and the optic nerve head. Eye tracking was used to assess possible fixation loss. Only high-quality images, evaluated by means of Topcon index quality (>70), were considered.

We measured retinal layer thickness by means of a high-resolution horizontal structural enhanced-depth imaging (EDI)-OCT scan,7 including total retinal thickness, central macular thickness (CMT), retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL) and ellipsoid zone–retinal pigment epithelium layer (EZ-RPE). Choroidal thickness was also measured. For each measure, four samplings were obtained on the left and four on the right of the foveal, the mean value being used in the analysis. Moreover, hyperreflective foci (HF) (defined as discrete, punctiform, hyperreflective lesions) were measured in the same structural OCT scan, within both the retina and the choroid, but using inverted colors to improve the imaging. HF were analyzed in a region selected on a horizontal structural OCT, including 750 μm on the right and left sides of a vertical line passing through the fovea. Fundus autofluorescence (FAF) images were also used to assess the presence of foveal sparing, defined as the absence of autofluorescence surrounding the fovea by least 180°, not including the fovea.8 A control group of healthy age- and sex-matched subjects underwent the same ophthalmologic assessment.

OCTA images were analyzed by the Topcon full-spectrum, amplitude-decorrelation, angiography algorithm. Each patient contributed with a single eye, which was randomly selected.

The superficial capillary plexuses (SCP), deep capillary plexuses (DCP), and choriocapillaris (CC) were automatically segmented both in the macula and in the optic nerve head; additionally, the radial peripapillary capillary (RPC) plexus was automatically segmented at the level of the optic nerve. For convenience, macular parameters will henceforth be referred to using the prefix “m,” with the prefix “n” applying to the optic nerve head (e.g., mSCP and nSCP for macular and nerve superficial capillary plexuses, respectively).

Each reconstruction was carefully inspected by an expert ophthalmologist (MBP). All reconstructions were exported in .tiff format and loaded in ImageJ software (Bethesda, MD). In-house scripts were built in order to calculate the following parameters: vessel density (VD), vessel tortuosity (VT), vessel dispersion (Vdisp), and vessel rarefaction (VR), with the same approach already adopted.9

The statistical analysis of the quantitative parameters considered, comparing with control values, was performed using the one-way ANOVA test, with Bonferroni correction for multiple comparisons (SPSS, Chicago, IL). Moreover, statistically significant correlations between all considered parameters were assessed by means of Kendall's tau correlation coefficient measurement. We also performed a receiver operating curve (ROC) analysis in order to extract a parameter that could distinguish STGD subgroups differing significantly in terms of BCVA. The statistical significance was set at P < 0.05.

Results

Overall, 54 eyes of 54 STGD patients (18 males) were included in the analysis, and all data were compared with 54 eyes of 54 healthy age- and sex-matched controls. All clinical and imaging findings are extensively reported in Table 1. Duration of STGD, as assessed on the patients' report, was 12.8 ± 10.8 years. The standout finding was STGD patients showed statistically significant worse BCVA, structural OCT parameters, and OCTA parameters. Moreover, we detected significantly higher HF numbers, both within the retina and the choroid in our patients' cohort, with respect to the control group. Partial or complete foveal sparing was detected in 23 of 54 patients (43%). Correlation analyses revealed several statistically significant correlations (Table 2). In more detail, BCVA was influenced by the integrity both of the retinal layers and vascular network. The presence of HF was found to correlate with worse BCVA. Furthermore, both retinal layer thickness and HF number significantly correlated with OCTA parameters. In particular, high VD and VT values were associated with better structural features and lower HF number, whereas high VDisp and VR values were related to worse retinal status and higher HF number. It is worth noting the higher significant correlations with both VD and VT, compared with VDisp and VR parameters. Statistically significant correlations were found when considering macular OCTA parameters alone, optic nerve head parameters showing unremarkable effects (Figs. 1 and 2). Interestingly, retinal layer thickness showed a stronger correlation with VT values than with VD. In addition, as expected, OCTA parameters showed statistically significant correlations between them (P < 0.01). No significant age or duration effect were detected. ROC analysis established a cutoff value based on the VT mean, to separate two subgroups with different BCVA, with both sensitivity and specificity of 0.9. A VT cutoff of 5 created two categories, a functionally and anatomically better STGD group (STGD1) and a worse group (STGD2) (Table 3). Foveal sparing was present in 15 of 35 patients of STGD1 (43%) and in 8 of 19 patients of STGD2 (42%) (P > 0.05), showing no effects on the analyses.

Table 1.

Clinical and Imaging Findings

Parameter	STGD Patients		Controls		P Value
	Mean	SD	Mean	SD
Age	43.52	16.39	45.37	17.93	P > 0.05
BCVA, logMAR	0.68	0.44	0.00	0.00	P < 0.01
RNFL	97.86	9.45	106.53	9.63
EZ-RPE	43.87	21.37	73.01	5.55
ONL	26.19	14.10	92.34	7.25
OPL	20.12	9.90	41.51	5.98
INL	33.15	7.71	46.17	6.46
IPL	34.84	8.85	48.21	7.12
GCL	44.21	11.21	62.49	7.48
CMT	103.05	59.03	241.03	11.16
Retinal thickness	224.23	54.24	400.66	19.06
Choroidal thickness	257.01	94.71	322.53	44.37
HF_retina	7.76	7.35	0.00	0.00
HF_choroid	18.26	17.86	0.00	0.00
VD mSCP	0.40	0.02	0.41	0.01
VD mDCP	0.26	0.06	0.43	0.01
VD mCC	0.32	0.16	0.50	0.01
VDisp mSCP	24.62	8.72	10.72	4.15
VDisp mDCP	27.51	8.75	11.45	3.48
VT mSCP	5.31	0.15	7.20	0.31
VT mDCP	4.80	0.50	7.84	0.34
VR mSCP	0.44	0.03	0.39	0.01
VR mDCP	0.47	0.03	0.41	0.01
VD RPC	0.42	0.02	0.44	0.01
VD nSCP	0.41	0.03	0.43	0.01
VD nDCP	0.30	0.02	0.40	0.02
VD nCC	0.51	0.03	0.54	0.03
VDisp RPC	24.78	8.82	10.61	3.70
VDisp nSCP	25.80	11.56	10.35	2.88
VDisp nDCP	27.43	9.72	10.37	3.36
VT RPC	5.64	0.51	7.73	0.30
VT nSCP	4.45	0.26	8.42	0.33
VT nDCP	5.20	0.27	7.06	0.25
VR RPC	0.47	0.02	0.45	0.01
VR nSCP	0.49	0.02	0.44	0.01
VR nDCP	0.43	0.02	0.40	0.01

Table 2.

Correlation Analysis

Parameter	RNFL	EZ−RPE	ONL	INL	IPL	GCL	CMT	Full_Retina
BCVA logMAR
Kendall's tau coefficient	−0.248	−0.939	−0.938	−0.939	−0.941	−0.94	−0.938	−0.938
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	ONL	INL	IPL	GCL	CMT	Full_retina	HF_retina	HF_choroid
EZ-RPE
Kendall's tau coefficient	0.996	0.995	0.993	0.994	0.998	0.998	−0.548	−0.658
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	INL	IPL	GCL	CMT	Full_retina	HF_retina	HF_choroid	VD mCC
ONL
Kendall's tau coefficient	0.994	0.994	0.993	0.996	0.997	−0.545	−0.66	0.963
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	IPL	GCL	CMT	Full_retina	HF_retina	HF_choroid	VD mCC	VT mSCP
INL
Kendall's tau coefficient	0.992	0.993	0.995	0.996	−0.551	−0.661	0.962	0.99
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	GCL	CMT	Full_retina	HF_retina	HF_choroid	VD mCC	VT mSCP	VT mDCP
IPL
Kendall's tau coefficient	0.991	0.993	0.994	−0.545	−0.663	0.959	0.987	0.992
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	CMT	Full_retina	HF_retina	HF_choroid	VD mCC	VT mSCP	VT mDCP	VT RPC
GCL
Kendall's tau coefficient	0.994	0.994	−0.548	−0.662	0.961	0.989	0.993	0.993
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	Full_retina	HF_retina	HF_choroid	VD mCC	VT mSCP	VT mDCP	VT RPC	VR mSCP
CMT
Kendall's tau coefficient	0.998	−0.546	−0.658	0.964	0.991	0.996	0.995	−0.951
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	HF_retina	HF_choroid	VD mCC	VT mSCP	VT mDCP	VT RPC	VR mSCP	VR mDCP
Full_retina
Kendall's tau coefficient	−0.547	−0.658	0.965	0.991	0.998	0.997	−0.951	−0.961
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	HF_choroid	VD mCC	VT mSCP	VT mDCP	VT RPC	VR mSCP	VR mDCP	VR RPC
HF_retina
Kendall's tau coefficient	0.665	−0.516	−0.554	−0.547	−0.547	0.535	0.549	0.526
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	VD mCC	VT mSCP	VT mDCP	VT RPC	VR mSCP	VR mDCP	VR RPC
HF_choroid
Kendall's tau coefficient	−0.625	−0.657	−0.657	−0.657	0.628	0.663	0.679
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01

Figure 1.

Macular OCTA in STGD. SCP (A) and DCP (B) appear evidently rarefied in STGD. Furthermore, CC (C) results almost absent, with clear exposure of choroidal vessels. SCP, DCP, and CC plexa of a healthy control are shown in (D–F), respectively.

Figure 2.

Optic nerve head OCTA in STGD. If comparing a STGD patient with a healthy control, RPC (A), SCP (B), and DCP (C) appear evidently rarefied in STGD. Moreover, also CC plexus shows signs of perfusion lacks (D). RPC, SCP, DCP, and CC plexa of a healthy control are shown in (E–H), respectively.

Table 3.

OCTA VT Cutoff Analysis (VT = 5)

Parameter	Mean ± SD	Comparison	P Values
Age
STGD1	44 ± 15	STGD1 vs. STGD2	>0.05
STGD2	44 ± 19	STGD1 vs. Controls	>0.05
Controls	45 ± 17	STGD2 vs. Controls	>0.05
BCVA, logMAR
STGD1	0.42 ± 0.28	STGD1 vs. STGD2	<0.01
STGD2	1.09 ± 0.36	STGD1 vs. Controls	<0.01
Controls	0.0 ± 0.0	STGD2 vs. Controls	<0.01
RNFL
STGD1	98 ± 9	STGD1 vs. STGD2	>0.05
STGD2	98 ± 10	STGD1 vs. controls	<0.01
Controls	107 ± 10	STGD2 vs. Controls	<0.01
CMT
STGD1	129 ± 58	STGD1 vs. STGD2	<0.01
STGD2	65 ± 48	STGD1 vs. Controls	<0.01
Controls	240 ± 11	STGD2 vs. Controls	<0.01
Full retinal thickness
STGD1	248 ± 52	STGD1 vs. STGD2	<0.01
STGD2	189 ± 42	STGD1 vs. Controls	<0.01
Controls	403 ± 16	STGD2 vs. Controls	<0.01
CT
STGD1	268 ± 92	STGD1 vs. STGD2	>0.05
STGD2	242 ± 99	STGD1 vs. controls	<0.05
Controls	322 ± 45	STGD2 vs. controls	<0.01
EZ-RPE
STGD1	55 ± 19	STGD1 vs. STGD2	<0.01
STGD2	26 ± 13	STGD1 vs. controls	<0.01
Controls	73 ± 6	STGD2 vs. controls	<0.01
ONL
STGD1	33 ± 12	STGD1 vs. STGD2	<0.01
STGD2	17 ± 15	STGD1 vs. controls	<0.01
Controls	93 ± 7	STGD2 vs. controls	<0.01
OPL
STGD1	21 ± 12	STGD1 vs. STGD2	<0.01
STGD2	19 ± 7	STGD1 vs. controls	<0.01
Controls	42 ± 6	STGD2 vs. controls	<0.01
INL
STGD1	37 ± 7	STGD1 vs. STGD2	<0.01
STGD2	27 ± 4	STGD1 vs. controls	<0.01
Controls	47 ± 6	STGD2 vs. controls	<0.01
IPL
STGD1	39 ± 8	STGD1 vs. STGD2	<0.01
STGD2	27 ± 4	STGD1 vs. controls	<0.01
Controls	49 ± 7	STGD2 vs. controls	<0.01
GCL
STGD1	50 ± 8	STGD1 vs. STGD2	<0.01
STGD2	34 ± 7	STGD1 vs. controls	<0.01
Controls	63 ± 7	STGD2 vs. controls	<0.01
HF_retina
STGD1	7 ± 7	STGD1 vs. STGD2	<0.05
STGD2	10 ± 7	STGD1 vs. controls	<0.01
Controls	0 ± 0	STGD2 vs. controls	<0.01
HF_choroid
STGD1	11 ± 11	STGD1 vs. STGD2	<0.01
STGD2	30 ± 20	STGD1 vs. controls	<0.01
Controls	0 ± 0	STGD2 vs. controls	<0.01
VD mSCP
STGD1	0.40 ± 0.02	STGD1 vs. STGD2	>0.05
STGD2	0.40 ± 0.02	STGD1 vs. controls	<0.01
Controls	0.41 ± 0.01	STGD2 vs. controls	<0.01
VD mDCP
STGD1	0.26 ± 0.06	STGD1 vs. STGD2	>0.05
STGD2	0.28 ± 0.07	STGD1 vs. controls	<0.01
Controls	0.43 ± 0.01	STGD2 vs. controls	<0.01
VD mCC
STGD1	0.42 ± 0.03	STGD1 vs. STGD2	<0.01
STGD2	0.17 ± 0.17	STGD1 vs. controls	<0.01
Controls	0.50 ± 0.01	STGD2 vs. controls	<0.01
VD RPC
STGD1	0.42 ± 0.02	STGD1 vs. STGD2	>0.05
STGD2	0.42 ± 0.02	STGD1 vs. controls	<0.01
Controls	0.44 ± 0.01	STGD2 vs. controls	<0.01
VD nSCP
STGD1	0.41 ± 0.03	STGD1 vs. STGD2	>0.05
STGD2	0.40 ± 0.04	STGD1 vs. controls	>0.05
Controls	0.43 ± 0.01	STGD2 vs. controls	<0.05
VD nDCP
STGD1	0.30 ± 0.02	STGD1 vs. STGD2	>0.05
STGD2	0.30 ± 0.03	STGD1 vs. controls	<0.01
Controls	0.40 ± 0.02	STGD2 vs. controls	<0.01
VD nCC
STGD1	0.50 ± 0.02	STGD1 vs. STGD2	<0.01
STGD2	0.52 ± 0.04	STGD1 vs. controls	<0.01
Controls	0.54 ± 0.03	STGD2 vs. controls	<0.01
VD mean
STGD1	0.39 ± 0.01	STGD1 vs. STGD2	<0.01
STGD2	0.36 ± 0.03	STGD1 vs. controls	<0.01
Controls	0.45 ± 0.01	STGD2 vs. controls	<0.01
Vdisp mSCP
STGD1	23.86 ± 6.43	STGD1 vs. STGD2	>0.05
STGD2	25.54 ± 11.83	STGD1 vs. controls	<0.01
Controls	10.53 ± 4.08	STGD2 vs. controls	<0.01
Vdisp mDCP
STGD1	27.71 ± 9.61	STGD1 vs STGD2	>0.05
STGD2	26.50 ± 7.54	STGD1 vs. controls	<0.01
Controls	11.36 ± 3.50	STGD2 vs. controls	<0.01
Vdisp RPC
STGD1	26.71 ± 9.46	STGD1 vs. STGD2	<0.01
STGD2	20.62 ± 6.74	STGD1 vs. controls	<0.01
Controls	10.67 ± 3.75	STGD2 vs. controls	<0.01
Vdisp nSCP
STGD1	28.41 ± 12.61	STGD1 vs. STGD2	<0.01
STGD2	20.40 ± 7.72	STGD1 vs. controls	<0.01
Controls	10.40 ± 2.92	STGD2 vs. controls	<0.01
Vdisp nDCP
STGD1	28.37 ± 9.49	STGD1 vs. STGD2	>0.05
STGD2	24.75 ± 10.80	STGD1 vs. controls	<0.01
Controls	10.50 ± 3.34	STGD2 vs. controls	<0.01
VDisp mean
STGD1	27.01 ± 6.58	STGD1 vs. STGD2	<0.05
STGD2	23.56 ± 5.61	STGD1 vs. controls	<0.01
Controls	10.69 ± 1.37	STGD2 vs. controls	<0.01
VT mSCP
STGD1	5.39 ± 0.11	STGD1 vs. STGD2	>0.05
STGD2	5.27 ± 0.44	STGD1 vs. controls	<0.01
Controls	7.20 ± 0.31	STGD2 vs. controls	<0.01
VT mDCP
STGD1	5.08 ± 0.31	STGD1 vs. STGD2	>0.05
STGD2	4.46 ± 0.84	STGD1 vs. controls	<0.01
Controls	7.85 ± 0.35	STGD2 vs. controls	<0.01
VT RPC
STGD1	5.93 ± 0.30	STGD1 vs. STGD2	>0.05
STGD2	5.24 ± 0.66	STGD1 vs. controls	<0.01
Controls	7.73 ± 0.30	STGD2 vs. controls	<0.01
VT nSCP
STGD1	4.49 ± 0.30	STGD1 vs. STGD2	>0.05
STGD2	4.56 ± 0.87	STGD1 vs. controls	<0.01
Controls	8.43 ± 0.33	STGD2 vs. controls	<0.01
VT nDCP
STGD1	5.24 ± 0.30	STGD1 vs. STGD2	>0.05
STGD2	5.19 ± 0.42	STGD1 vs. controls	<0.01
Controls	7.07 ± 0.25	STGD2 vs. controls	<0.01
VT mean
STGD1	5.23 ± 0.15	STGD1 vs. STGD2	<0.01
STGD2	4.94 ± 0.61	STGD1 vs. controls	<0.01
Controls	7.66 ± 0.23	STGD2 vs. controls	<0.01
VR mSCP
STGD1	0.43 ± 0.02	STGD1 vs. STGD2	<0.01
STGD2	0.47 ± 0.03	STGD1 vs. controls	<0.01
Controls	0.39 ± 0.01	STGD2 vs. controls	<0.01
VR mDCP
STGD1	0.45 ± 0.02	STGD1 vs. STGD2	<0.01
STGD2	0.50 ± 0.03	STGD1 vs. controls	<0.01
Controls	0.41 ± 0.01	STGD2 vs. controls	<0.01
VR RPC
STGD1	0.46 ± 0.01	STGD1 vs. STGD2	<0.01
STGD2	0.49 ± 0.01	STGD1 vs. controls	<0.01
Controls	0.44 ± 0.01	STGD2 vs. controls	<0.01
VR nSCP
STGD1	0.49 ± 0.02	STGD1 vs. STGD2	<0.01
STGD2	0.48 ± 0.02	STGD1 vs. controls	<0.01
Controls	0.44 ± 0.01	STGD2 vs. controls	<0.01
VR nDCP
STGD1	0.43 ± 0.02	STGD1 vs. STGD2	<0.01
STGD2	0.42 ± 0.02	STGD1 vs. controls	<0.01
Controls	0.40 ± 0.01	STGD2 vs. controls	<0.01
VR mean
STGD1	0.45 ± 0.01	STGD1 vs. STGD2	<0.01
STGD2	0.47 ± 0.02	STGD1 vs. controls	<0.01
Controls	0.42 ± 0.01	STGD2 vs. controls	<0.01

Table 2.

Extended

Parameter	HF_retina	HF_choroid	VD mCC	VD nCC	VT mSCP	VT mDCP	VT RPC	VR mSCP
BCVA logMAR
Kendall's tau coefficient	0.56	0.697	−0.903	0.366	−0.943	−0.939	−0.938	0.925
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	VD mCC	VT mSCP	VT mDCP	VT RPC	VR mSCP	VR mDCP	VR RPC	VT Mean
EZ-RPE
Kendall's tau coefficient	0.964	0.992	0.997	0.995	−0.952	−0.962	−0.913	0.735
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	VT mSCP	VT mDCP	VT RPC	VR mSCP	VR mDCP	VR RPC	VT Mean	VR Mean
ONL
Kendall's tau coefficient	0.991	0.997	0.997	−0.951	−0.963	−0.913	0.735	−0.725
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	VT mDCP	VT RPC	VR mSCP	VR mDCP	VR RPC	VT Mean	VR Mean
INL
Kendall's tau coefficient	0.995	0.993	−0.952	−0.961	−0.914	0.736	−0.726
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	VT RPC	VR mSCP	VR mDCP	VR RPC	VT Mean	VR Mean
IPL
Kendall's tau coefficient	0.993	−0.953	−0.961	−0.918	0.738	−0.726
P value	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
	VR mSCP	VR mDCP	VR RPC	VT Mean	VR Mean
GCL
Kendall's tau coefficient	−0.952	−0.962	−0.917	0.731	−0.729
P value	<0.01	<0.01	<0.01	<0.01	<0.01
	VR mDCP	VR RPC	VT Mean	VR Mean
CMT
Kendall's tau coefficient	−0.962	−0.914	0.735	−0.724
P value	<0.01	<0.01	<0.01	<0.01
	VR RPC	VD Mean	VT Mean	VR Mean
Full_retina
Kendall's tau coefficient	−0.913	0.373	0.738	−0.724
P value	<0.01	<0.01	<0.01	<0.01
HF_retina
Kendall's tau coefficient
P value
HF_choroid
Kendall's tau coefficient
P value

Table 2.

Extended

Parameter	VR mDCP
BCVA logMAR
Kendall's tau coefficient	0.939
P value	<0.01
	VR Mean
EZ-RPE
Kendall's tau coefficient	−0.723
P value	<0.01
ONL
Kendall's tau coefficient
P value
INL
Kendall's tau coefficient
P value
IPL
Kendall's tau coefficient
P value
GCL
Kendall's tau coefficient
P value
CMT
Kendall's tau coefficient
P value
Full_retina
Kendall's tau coefficient
P value
HF_retina
Kendall's tau coefficient
P value
HF_choroid
Kendall's tau coefficient
P value

Discussion

STGD is a complex disease primarily associated with RPE/photoreceptor dysfunction and is accompanied by a compromised vascular network, which can be noninvasively analyzed by OCTA.6 In particular, macular CC alterations represent a key element contributing to outer retinal atrophy.4

The aim of the present study was to analyze both retinal layer and vascular network changes in STGD. A significant thinning of inner and outer retinal layers, together with the presence of relevant alterations of the retinal vascular network were found, providing further confirmation of the CC macular damage. Although significant dysfunction of the optic nerve head plexa was detected, these changes may be interpreted as a consequence of modifications to the retina. Quantitative OCTA parameters, including VT, VDisp, and VR, enabled the involvement of the retinal vascular network in STGD to be described in greater detail. In particular, retinal plexa appeared rarer and more disorganized than in controls, VT reduction representing the consequence of reduced blood flow perfusion.9

The dysfunction present throughout the retinal vascular network may be due to the primary RPE/photoreceptor damage. Indeed, the RPE is essential to the physiologic release of growth factors, including VEGF, which guarantee the trophism of retinal vessels and CC, along with the regulation of metabolic exchanges through the blood–retinal barrier (Le YZ, et al. IOVS. 2008;49:E−Abstract 857).10 With regard to inner retinal thinning, this finding might be explained as inner retinal cell damage caused by anterograde degeneration, taking into account the exclusive expression of the ABCA4 gene in the RPE/photoreceptor complex.11 Alternatively, the loss of trophic stimuli from RPE cells, in addition to the retinal vascular network deficiency, may contribute to inner retinal degeneration and thinning. The results of the correlation analysis, especially the strong correlation revealed between retinal layer thickness and VT, further underline the influence of the impaired blood supply in the progression of the degenerative process occurring in STGD.

The higher number of retinal and choroidal HF in STGD patients has already been described12 and represents an imaging biomarker of the disease's activity, as reported in other conditions.13

The most interesting findings of the present paper are the results of the cutoff analyses of OCTA parameters. A VT cutoff value of 5, obtained from the mean of the values of all retinal plexa, was able to identify two different STGD groups on the basis of the BCVA. These patients' subgroups also differed in terms of retinal layer thickness and OCTA parameters, thus enabling us to divide our patients' cohort into better and worse morphofunctional groups. Interestingly, STGD1 and STGD2 subgroups did not differ significantly in terms of foveal sparing, suggesting that other factors besides the atrophic changes of the fovea contribute to the BCVA loss. We hypothesize that the more extensively dysfunctional vascular network typical of STGD2 is responsible for an overall decrease in macular functionality. In other words, assuming the same degree of foveal involvement, a more trophic vascular network leads to a better retinal function. Although obtained from what is only a pilot study, our findings may have important clinical implications. In particular, patient evaluation in clinical practice might benefit from the information provided by this kind of analysis, enabling different clinical phenotypes to be identified. Furthermore, viewed in the context of clinical trials designed to test potential drugs, this approach might have a role in patient selection and in assessing clinical and anatomic outcomes.

Our study has undoubted limitations. Both OCT and OCTA analyses are known to be potentially affected by a number of artifacts,14,15 so we adopted all possible precautions to minimize their potential effect on data analysis. We are also aware that our imaging-based analysis might certainly benefit from histopathologic validation. Last, our cases were insufficient to draw definitive conclusions and it would therefore be advisable to carry out a longitudinal follow-up of patients in order to confirm our data.

In essence, our study reveals that extensive changes both in the retinal vascular network and in the retinal layers can be detected in STGD. A quantitative cutoff based on VT can separate two different morphofunctional phenotypes.

We believe a quantitative assessment might contribute to more accurate patients' categorization in a future scenario that sees possible new drugs combining with the introduction of artificial intelligence–based techniques to identify and treat STGD. Further studies are warranted to provide further support to the present findings.