Long-Term Outcomes in Macular Telangiectasia Type 2 With Subretinal Neovascularization

Abstract

Purpose:

This work reports long-term outcomes in macular telangiectasia type 2 (MacTel) with subretinal neovascularization (SRNV).

Methods:

A retrospective, single-center review of medical records was performed on all patients with a diagnosis of MacTel presenting between May 2004 and October 2019. Medical and ocular history, best-corrected visual acuity (BCVA) at baseline and final visit, optical coherence tomography data, and treatment history of SRNV secondary to MacTel were recorded.

Results:

A total of 471 eyes were diagnosed with MacTel. SRNV was present in 44 eyes (9.3%), of which 38 eyes met inclusion criteria for SRNV. Average follow-up duration in the SRNV group was 78.4 months. All SRNV patients underwent antivascular endothelial growth factor (anti-VEGF) therapy. There was no significant change from mean baseline (0.59 ± 0.45) to final (0.70 ± 0.49) BCVA in the SRNV group as a whole (P = .13). Subgroup analysis revealed 17 of 38 eyes had SRNV at diagnosis and received immediate anti-VEGF treatment. In this subgroup mean pretreatment BCVA was 0.89 ± 0.43 and the mean final BCVA was 0.87 ± 0.61 (P = .84). The remainder (21 of 38 eyes) developed SRNV during follow-up. In this subgroup, final BCVA after initiation of treatment was 0.56 ± 0.32, an improvement in BCVA from SRNV onset (P = .04) and a decrease from pre-SRNV onset baseline BCVA (P = .008).

Conclusions:

Visual acuity is maintained, not improved, in long-term follow-up of MacTel with SRNV treated with anti-VEGF. Patients presenting with SRNV have a worse prognosis than those who develop SRNV during follow-up.

Introduction

Macular telangiectasia type 2 (MacTel), or idiopathic juxtafoveal telangiectasia, is a rare, bilateral, neurodegenerative disease of the macula that presents most frequently in the sixth or seventh decade of life with metamorphopsia or impaired reading ability secondary to a macular pathology resulting in a paracentral scotoma. ¹ Clinical manifestations include parafoveal telangiectasias, loss of retinal transparency, right-angle retinal venules, crystalline deposits, and retinal pigment hyperplasia. ^2,3 Diagnosis is often aided by characteristic findings on multimodal imaging: retinal cavitations and internal limiting membrane (ILM) draping on optical coherence tomography (OCT), loss of foveal masking on fundus autofluorescence, and parafoveal telangiectasias and leakage on fluorescein angiography (FA).

Subretinal neovascularization (SRNV) is an uncommon complication for patients with MacTel that is associated with poor visual outcomes. ^2,4,5 Various treatment modalities have aimed to address SRNV in MacTel, including submacular surgery to remove neovascular membranes that historically has yielded poor visual outcomes. ⁶ Transpupillary thermotherapy and photodynamic therapy have demonstrated more favorable results, but both have been largely supplanted by the use of antivascular endothelial growth factor (anti-VEGF) agents. ^{7 -9} Initially thought to be efficacious in patients with MacTel either with or without SRNV, anti-VEGF agents have been shown to have efficacy limited to patients with MacTel and SRNV. ^{10 -14} Among patients with SRNV, anti-VEGF therapy has demonstrated improved visual acuity (VA) and decreased central macular thickness (CMT) in small cohorts with limited follow-up. ^{10 -13}

Data on the long-term outcomes of patients with MacTel and SRNV treated with anti-VEGF agents are scarce. ¹⁴ This study reports the findings of long-term outcomes in the largest cohort to our knowledge of patients with MacTel and SRNV. The long-term outcomes of a large cohort of patients with MacTel without SRNV are also reported.

Methods

A retrospective review of medical records was conducted with the approval of the institutional review board of St. Luke’s Hospital (Chesterfield, Missouri). This review did not influence patient care, thus no consent was required or obtained. Standard treatment consent was obtained for all patients undergoing anti-VEGF therapy.

The medical records of a retina-only, private practice institution were reviewed for all patients with a diagnosis of MacTel from May 2004 to October 2019. Age, sex, ocular history, and diagnosis of diabetes and/or hypertension were recorded. Snellen best-corrected VA (BCVA) and CMT were recorded at baseline and final visits. Patients with SRNV secondary to MacTel were identified by medical-record review and underwent confirmatory review of OCT and, where available, OCT angiography and FA. In this group the number of injections and BCVA at time of each injection were recorded. For both groups, eyes were eligible for OCT CMT measurement analysis if the eye was imaged on a Heidelberg Spectralis device at baseline and final visit. Intravitreal injection was performed per the protocol of the treating physician. Bevacizumab, ranibizumab, and aflibercept were used in this study. SRNV and complications thereof were treated at the discretion of the treating physician.

Two cohorts were identified for this study. Eyes were included in the MacTel without SRNV group if a minimum of 3 months of follow-up was available. Eyes were included in the MacTel with SRNV group if SRNV was documented and a minimum of 3 months of follow-up were available. Exclusion criteria for the MacTel without SRNV group were the presence of other significant maculopathy, prior vitrectomy, prior macular laser therapy, prior anti-VEGF therapy, and less than 3 months of follow-up. Exclusion criteria for the SRNV group were macular edema or neovascularization secondary to other retinopathy, prior vitrectomy, prior macular laser therapy, and less than 3 months of follow-up.

The primary outcome measures were mean change in BCVA and CMT from baseline in the non-SRNV cohort and from initiation of anti-VEGF therapy in the SRNV cohort. Data collection and analysis were performed using Microsoft Excel. Snellen VA was converted to logarithm of the minimum angle of resolution acuity to facilitate statistical analysis. Statistical significance was determined using a 2-tailed paired t test.

Results

During the study period 471 eyes were diagnosed with MacTel, of which 44 (9.3%) were diagnosed with SRNV secondary to MacTel. A total of 328 eyes of 184 patients met the inclusion criteria for the non-SRNV group and 38 eyes of 32 patients met the inclusion criteria for the SRNV group. Diagnosis of SRNV was established by examination, OCT, and multimodal imaging, with 71% of SRNV confirmed by FA and the remainder by OCT angiography. OCT was obtained at all follow-up visits for patients with or without SRNV. Data on age, sex, and medical history for both groups are presented in Table 1. A female preponderance was observed in both groups (71% in the non-SRNV group and 69% in the SRNV group).

Table 1.

Demographic, Visual Acuity, and Optical Coherence Tomography Data.

	MacTel without SRNV (n = 328)	MacTel with SRNV (n = 38)	P
Male	29%	31%	.94
Female	71%	69%	–
Age, y	65.3 ± 9.5	67.4 ± 8.5	.17
Diabetes	38%	45%	.40
Hypertension	64%	82%	.01
Follow-up, mo	44.0 ± 38.9	78.4 ± 40.0	<.001
Presenting BCVA, Snellen
> 20/40	54%	3%
20/40 to 20/200	45%	76%
< 20/200	1%	21%
Mean	0.26 ± 0.24	0.59 ± 0.45	<.001
Final BCVA, Snellen
> 20/40	47%	17%
20/40 to 20/200	51%	60%
< 20/200	2%	13%
Mean	0.31 ± 0.27	0.70 ± 0.49	<.001
Baseline OCT features
EZ loss	36%	84%	<.001
Subfoveal EZ loss	26%	57%	<.001
ILM draping	42%	42%	.97
Final OCT features
EZ loss	47%	100%	<.001
Subfoveal EZ loss	35%	76%	<.001
ILM draping	50%	26%	.51

Abbreviations: BCVA, best-corrected visual acuity; EZ, ellipsoid zone; ILM, internal limiting membrane; MacTel, macular telangiectasia type 2; OCT, optical coherence tomography; SRNV, subretinal neovascularization.

In the non-SRNV group, mean follow-up was 44.0 ± 38.9 months (range, 3-189 months). Mean baseline VA was 0.26 ± 0.24 and mean final VA was 0.31 ± 0.27 (P = .0012). Baseline and final CMT data were available for 190 patients in the non-SRNV group. Mean baseline CMT was 258 ± 43.1 µm and mean final CMT was 259 ± 43.0 µm (P = .94). Ellipsoid zone (EZ) disruption was present in 36% of non-SRNV eyes at baseline and 47% of eyes at follow-up (P = .0037). Subfoveal EZ disruption was present in 26% of non-SRNV eyes at baseline and 35% at follow-up (P = .015) (Figure 1). Eyes with subfoveal EZ disruption at final visit had significantly worse average BCVA (0.45) than those without subfoveal EZ disruption (0.23) (P < .0001).

Figure 1.

Long-term follow-up in a case of macular telangiectasia type 2 without subretinal neovascularization in a 47-year-old woman. (A and B) On presentation, best-corrected visual acuity was Snellen 20/25 OD and 20/30 OS. Optical coherence tomography revealed (A) a small cavitation in the right eye and (B) a foveal cavitation with ellipsoid zone attenuation in the left eye. (C and D) At follow-up 5 years later, best-corrected visual acuity was Snellen 20/30 OD and 20/50 OS. Optical coherence tomography demonstrated (C) internal limiting membrane draping over an enlarged cavitation in the right eye and (D) progressive subfoveal ellipsoid zone loss in the left eye.

In the SRNV group, mean follow-up was 78.4 ± 40.0 months (range, 3-164 months). Mean baseline VA was 0.59 ± 0.45 and mean final VA was 0.70 ± 0.49 (P = .13). Baseline and final CMT data were available for 26 patients in the SRNV group. Mean baseline CMT was 276 ± 77.9 µm and mean final CMT was 252 ± 67.2 µm (P = .23). Additional data on OCT features are presented in Table 1; at the final visit OCT demonstrated an area of EZ disruption in all SRNV eyes, subfoveal EZ disruption in 76% of eyes, ILM draping with cavitations in 26% of eyes, and subretinal fluid in 10% of eyes. As with non-SRNV eyes, the average final BCVA in SRNV eyes with subfoveal EZ disruption (0.81) was worse than in those without subfoveal EZ disruption (0.41) (P = .02).

All 38 eyes in the SRNV group underwent treatment with anti-VEGF agents: 34 with bevacizumab monotherapy, 2 with aflibercept monotherapy, and 2 with bevacizumab and ranibizumab combination therapy (Figures 2 and 3). Ten retina specialists administered the intravitreal injections. In the majority of patients, an as-needed treatment approach was applied by the treating physicians. Three patients received a series of 3 scheduled monthly injections before being switched to an as-needed treatment strategy. The mean number of anti-VEGF injections was 4.2 ± 3.5 (range, 1-18 injections). Within the SRNV group, 17 eyes presented with SRNV and were treated with anti-VEGF injection at presentation. The remaining 21 eyes developed SRNV in the course of follow-up. In this latter group, the mean time from presentation to developing SRNV was 22.2 ± 31.4 months (range, 1-128 months).

Figure 2.

A case of a 53-year-old woman with macular telangiectasia type 2 complicated by bilateral subretinal neovascularization (SRNV) in both eyes. (A-D) On presentation, best-corrected visual acuity (BCVA) was Snellen 20/40 OD with parafoveal ellipsoid zone disruption, (A) small foveal cavitations on optical coherence tomography (OCT), and (C) late parafoveal leakage on fluorescein angiography (FA). SRNV was present in the left eye, for which BCVA was Snellen 5/200. (B) OCT of the left eye demonstrated subretinal fluid and subretinal hyperreflective material correlating to hemorrhage on clinical examination. (D) FA of the left eye demonstrated blockage from the hemorrhage and neovascular membranes. The patient received 3 as-needed injections of bevacizumab in the right eye over the following 4 months. (E-H) At 10-month follow-up, vision improved to Snellen 20/50 OS. (F) OCT of the left eye demonstrated resolution of subretinal fluid and hemorrhage, and (H) the neovascular membranes were regressed on FA. (E and G) In the interim, SRNV developed in the right eye. The patient subsequently received as-needed bevacizumab treatment in the right eye. (I and J) At 5-year follow-up the patient had received 5 as-needed injections of bevacizumab and no further injections in the left eye. There was no active SRNV in either eye, and BCVA was Snellen 20/50 OD and 20/60 OS.

Figure 3.

A case of a 56-year-old woman with macular telangiectasia type 2 complicated by unilateral subretinal neovascularization (SRNV) in the right eye. At presentation, best-corrected visual acuity (BCVA) was Snellen 20/70, and optical coherence tomography (OCT) revealed (A) a large foveal cavitation with ellipsoid zone loss and internal limiting membrane draping. Two years after presentation, (B-E) BCVA dropped to 20/400, and (B) OCT and (D and E) OCT angiography revealed SRNV in the superior fovea. The patient received 1 injection of bevacizumab, and at follow-up 2 years later (F) OCT showed regression of the SRNV with BCVA of Snellen 20/80. The prior foveal cavitation was filled with regressed fibrovascular tissue (not shown).

Subgroup analysis was performed within the SRNV group, differentiating eyes that presented with SRNV from eyes that developed SRNV during the course of follow-up at the study institution (Table 2). Among eyes presenting with SRNV, mean baseline VA was 0.89 ± 0.43 and mean final VA was 0.87 ± 0.61 (P = .84). Among eyes with late-onset SRNV, mean baseline VA was 0.38 ± 0.32 and mean VA at time of subsequent diagnosis of SRNV was 0.71 ± 0.41 (P  <  .001). Mean final VA was 0.56 ± 0.32; this mean final VA was significantly decreased from baseline mean VA (P = .008) and significantly increased from the initiation of anti-VEGF therapy (P = .04) (Figure 4).

Table 2.

Subgroup Analysis of MacTel Eyes by Onset of Subretinal Neovascularization.

	SRNV at Presentation (n = 17)	SRNV Onset During Follow-up (n = 21)	P
Baseline BCVA	0.89 ± 0.43	0.38 ± 0.32	<.001
Conversion BCVA	–	0.71 ± 0.41	–
Final BCVA	0.87 ± 0.61	0.56 ± 0.32	.05
Months to onset of SRNV	–	22.2 ± 31.4	–
Anti-VEGF treatments	4.4 ± 3.0	4.0 ± 4.0	.73

Abbreviations: anti-VEGF, antivascular endothelial growth factor; BCVA, best-corrected visual acuity; MacTel, macular telangiectasia type 2; SRNV, subretinal neovascularization.

Figure 4.

Subgroup analysis of late-onset subretinal neovascularization (SRNV). BCVA indicates best-corrected visual acuity; logMAR, logarithm of the minimum angle of resolution.

Conclusions

In this study, which reports to the best of our knowledge the largest cohort of MacTel with SNRV with the longest mean follow-up, there was no significant long-term change in VA in patients treated with anti-VEGF therapy. This represents a divergence from prior studies, in which anti-VEGF therapy yielded significant improvements in BCVA and CMT for SRNV in MacTel. ^{10 -14} The majority of these studies had 12 or fewer eyes with a mean follow-up of 17 months or less. ^{10 -13} Toygar et al reported the largest long-term cohort prior to this study, describing the outcomes of 25 eyes treated with bevacizumab with a mean follow-up of 41 months. ¹⁴ Statistically significant improvements in BCVA and CMT were present in their cohort; however, a subgroup analysis of 13 patients with more than 2 years of follow-up reported no significant change in CMT and a less significant improvement in VA (P < .0001 in the full cohort vs P = .02 in the longer–follow-up subgroup). ¹⁴ These data, suggesting a dampening of the gains of anti-VEGF therapy with time, presaged the findings of this study.

The long-term follow-up period in this study yielded views into the natural history of SRNV in MacTel. While 17 eyes presented with SRNV secondary to MacTel, 21 eyes developed SRNV during the course of follow-up. In this latter group, which was under observation during the onset of SRNV, a significant and expected drop in BCVA occurred from baseline (mean Snellen equivalent, 20/48) to the development of SRNV (mean Snellen, 20/103). With the initiation of anti-VEGF therapy, improvement was subsequently noted at final follow-up (mean Snellen, 20/73). Although this represents a notable improvement from the onset of SRNV, it is nevertheless a significant decline in BCVA from pre-SRNV baseline. This stands in contrast to the 17 eyes that developed SRNV prior to presentation, in which BCVA at presentation (mean Snellen, 20/155) and final visit (mean Snellen, 20/148) demonstrated no significant change. Notably, there was no significant difference in the number of anti-VEGF treatments between these 2 subgroups. These data suggest that timely detection of SRNV complicating MacTel can improve visual outcomes, but even despite prompt treatment a decline in BCVA is likely.

OCT analysis of the SRNV group revealed a decrease in mean CMT from baseline to final visit that did not reach statistical significance (P = .23). Toygar and colleagues reported similar findings in their subgroup analysis of SRNV eyes with long-term follow-up. ¹⁴ The utility of OCT CMT data in MacTel with SRNV is perhaps unclear given the distinct processes of neovascularization and neurodegeneration contributing to changes in CMT.

EZ disruption on OCT offers another promising means of assessing disease burden and photoreceptor degeneration in MacTel. ¹⁵ In the period of long-term follow-up in this study, we found over time an increased prevalence in EZ disruption both in the SRNV and non-SRNV groups. We also report, as expected, that eyes with subfoveal EZ disruption demonstrated significantly worse average BCVA than those without subfoveal EZ disruption in both groups. The proportion of eyes with EZ disruption was significantly higher in the SRNV group, although notably a quarter of eyes in the non-SRNV group demonstrated subfoveal EZ disruption at final follow-up. In contrast to these EZ data, the prevalence of ILM draping and inner retinal cavitations did not significantly progress over time, did not vary significantly between the SRNV and non-SRNV groups, nor was associated with worse BCVA.

This study, reporting on SRNV secondary to MacTel, also describes the findings of a review of a large cohort of patients with MacTel (471 eyes) with or without SRNV. A preponderance of female cases was noted in our cohort, in contrast to the early cohort of Gass and Blodi ² but corresponding with more recent reports from the MacTel Research Group. ^4,16 The mean age of diagnosis in our study, in the seventh decade of life, is consistent with prior large cohorts as well. ¹⁴ The prevalence of diabetes (39%) and hypertension (68%) in our combined cohort is higher than in some prior studies. ^{2 -4} The incidence of SRNV in our study (9.3%) falls within the range of 2% to 14% reported in prior large cohorts. ^2,4,17,18

Among MacTel patients without SRNV, we report a small but statistically significant decline in BCVA from baseline (mean Snellen, 20/36) to final visit (mean Snellen, 20/41). Despite this gradual decline, a large portion of patients without SRNV retain Snellen 20/40 or better vision, which is consistent with prior large cohorts. ^2,17,18 BCVA outcomes were significantly worse in patients with SRNV than those without SRNV; this disagrees with early theories of the natural course of SRNV in MacTel but corresponds with more recent studies. ^5,17,19

The limitations of this study include its reliance on VA as a main outcome measure. Recent literature has demonstrated that VA often does not correlate to the functional visual deficits experienced by patients with MacTel. ^1,15 Microperimetry to monitor the size of paracentral scotomas and OCT analysis to measure the area of EZ loss can yield a more intricate understanding of the progression of disease and response to treatment in MacTel. ^1,15,20 The treatment protocols and goals of the physicians treating SRNV with anti-VEGF agents were not standardized given the retrospective nature of this study. Nevertheless, this study adds a large cohort to the literature on this rare maculopathy and offers a new understanding of the expectations for long-term outcomes in MacTel with SRNV.