Age-related macular degeneration (AMD) is one of the main causes of visual impairment worldwide, and its prevalence increases substantially with aging populations in many countries.1 Among patients with neovascular AMD (nAMD), outer retinal tubulation (ORT) can be found and is described as branching tubules in spectral-domain optical coherence tomography (SD-OCT) C-scans (“en face”) and as ovoid or round hyporeflective area surrounded by hyperreflective borders. It has been reported in 56% of patients with nAMD and 21% with atrophic AMD.2,3 They have also been identified in patients with neovascular angioid streaks, myopic choroidal neovascularization (CNV), acute zonal occult outer retinopathy, retinitis pigmentosa, Stargardt disease, gyrate atrophy, choroideremia, and various other degenerative conditions.4,5
ORT is typically observed in areas where there was considerable disruption of outer retinal architecture with relative preservation of the photoreceptor layer (preserved inner and outer segment [IS/OS] junction), often overlying retinal pigment epithelium (RPE) damage or subretinal fibrosis. In eyes treated with intravitreal anti-vascular endothelial growth factor (VEGF) agents, ORT was typically found in areas where, before treatment, there had been substantial intraretinal fluid. It can be found in advanced stages of retinal diseases affecting the outer retina and retinal pigment epithelium, especially in nAMD.2
It was hypothesized that ORT could result from sublethal damage to the photoreceptors through loss of interdigitation of the outer segments with the RPE or through a direct degeneration of the RPE itself.2 Several studies have been investigating possible pathogeneses for ORT.6 It is important to rule out ORT in nAMD since it can mimic findings of intraretinal and subretinal fluid (intraretinal fluid [IRF] and subretinal fluid [SRF], respectively), possibly leading to unnecessary interventions.2
This review aims to describe the current knowledge about the prevalence and pathogenesis of ORT in AMD patients and to discuss its clinical implications on the prognosis and treatment of nAMD.
Patients and Methods
A search of the Pubmed, Medline, and Embase database was performed employing keywords that included: “Outer retinal tubulation,” “outer retinal tubulations,” and “outer retinal cysts.” The present review includes studies published through June 2017.
This review considered both prospective and retrospective studies that reported the presence of ORT in patients with AMD. The search was limited to peer-reviewed publications and those published in English.
The exclusion criteria included articles with no subjects with AMD, ORT, or outer retinal cysts; case reports; and letters.
The authors independently screened abstracts, excluded ineligible publications, reviewed full-text studies, and extracted data. The selected articles were evaluated for the following: age, gender, VA, type of CNV, presence of IRF and subretinal fibrosis, prevalence of ORT, anti-VEGF response, SD-OCT (B-scans and en face) and fluorescein angiography.
All visual acuity (VA) values were converted to ETDRS employing previously published formulas to facilitate the comparison of the data between the studies.7
In order to evaluate the anti-VEGF response, the type of medication (ranibizumab [Lucentis; Genentech, South San Francisco, CA], bevacizumab [Avastin; Genentech, South San Francisco, CA], and/or aflibercept [Eylea; Regeneron, Tarrytown, NY]), duration of treatment, and the treatment regimen were noted. Application of the inclusion and exclusion criteria resulted in a total of 18 articles grouped by subject as follows: ORT in patients with AMD (both within non-nAMD with geographic atrophy [GA] and nAMD treated with anti-VEGF therapies), anatomical outcomes on SD-OCT, and pathogenesis.
Studies on ORT in AMD (GA and nAMD) Treated With Anti-VEGF
Retrospective studies: A total of six retrospective studies evaluating ORT in patients with nAMD treated with anti-VEGF were identified (Table 1). In these retrospective studies, ORT showed either similar or worse outcomes to eyes without ORT in the setting of nAMD.
Summary of the Retrospective Studies of Patients (Eyes) With nAMD With ORT Treated With Anti-VEGF
Zweifel et al. performed a retrospective, single-center study of 69 eyes of 63 patients with ORT, of whom 53 (76.8%) had nAMD and three (4.3%) had GA. The remainder of eyes has non-AMD diagnoses with ORT. Data from all patients in whom ORT was identified during a 3-month period were retrospectively evaluated. ORT characteristics, change over time, and response to anti-VEGF treatment were analyzed. Median ETDRS VA was 35 letters with ORT height and width ranging from 40 μm to 140 μm and from 40 μm to 2260 μm, respectively. In the majority of cases, ORT stability was observed in patients receiving anti-VEGF therapy.2
Gildener-Leapman et al., in a retrospective chart review of 543 patients who received anti-VEGF injections for nAMD and underwent SD-OCT evaluation, assessed the correlation between anti-VEGF therapy and the incidence of ORT in patients with nAMD. ORT was identified in 70 patients (12.8%) after a mean number of 12.5 injections (range: three to 34 injections). There was also a great variation in VA change without a noticeable correlation between the number of treatments until the development of ORT and VA. Some patients with ORT maintained a stable VA; however, in other patients, VA significantly declined during ORT development.8
Espina et al. retrospectively evaluated patients with wet AMD with ORTs in at least one eye treated with intravitreal anti-VEGF agents. A total of 51 eyes with nAMD were identified, of which 33 (64.7%) presented with ORT. ORT changes and responses to the anti-VEGF therapy were analyzed, and ORT was found to be stable in 23 eyes and changed in 10 eyes. The mean ETDRS best-corrected VA (BCVA) in eyes with ORT was 41 ETDRS letters when in eyes without ORT was 55 ETDRS letters (P = .071).9
Contrary to findings from Espina et al., Giachetti Filho et al. (P < .01), Dirani et al. (P = .001), and Faria-Correia et al. (P = .042) observed that eyes with ORT had worse VA than eyes without ORT.5,10,11 In a retrospective study of 480 patients (546 eyes) with nAMD treated with ranibizumab in a variable regimen, Dirani et al. identified ORT in 30% of eyes after a mean follow-up period of 12 months ± 11.5 months. Kaplan-Meier survival analysis revealed that the ORT incidence increased over time (2.5%, 9.8%, 17.5%, and 41.6% at presentation, 6 months, 1 year, and 4 years, respectively), despite visually effective anti-VEGF treatment. At various time points, eyes without ORT had significantly better mean BCVA and greater mean BCVA change from baseline than those eyes with ORT. Additionally, lower VA at baseline was associated with a higher risk of developing ORT (log-rank test, P = .001).10
The study by Faria-Correia et al. included 377 eyes (323 patients) with nAMD that had received bevacizumab or ranibizumab therapy in a 1+ pro re nata (PRN) regimen for at least 3 months. Patients were distributed in two groups according to the presence or lack of ORT on SD-OCT. ORT was present in 31 eyes (8.2%) of 31 patients. When comparing the two groups, initial and final ETDRS BCVA were lower in patients with ORT (34.48 vs. 44.53, P = .020; 37.65 vs. 47.18, P = .042). Additionally, ORT regression was not observed after anti-VEGF therapy.11
Giachetti Filho et al. evaluated the prevalence of ORT in 158 eyes (142 patients) with CNV treated with bevacizumab in a PRN regimen between 2012 and 2014. ORT was found in 27.7% of eyes (33 out of 119 eyes) with nAMD, 62.5% of eyes (five out of eight eyes) with neovascular angioid streaks, and 16.7% of eyes (two out of 12 eyes) with myopic CNV. When comparing CNV patients with and without ORT, patients with ORT had a statistically worse ETDRS BCVA (P < .01).5
Prospective studies: Four prospective studies evaluated the presence of ORT in nAMD patients treated with anti-VEGF agents (Table 2). The studies found variable outcomes in those with ORT in the setting of nAMD.
Summary of the Prospecpective Studies of Patients (Eyes) With nAMD With ORT Treated With Anti-VEGF
Wolff et al. performed a prospective case series of 173 eyes (88 patients) with AMD (107 patients with nAMD, 29 patients with atrophic AMD, and 37 patients with early AMD). Patients were examined with SD-OCT and the prevalence of ORT was determined. ORT was present in 60 eyes with nAMD (56%) and in six eyes with atrophic AMD (20.7%). Thirty-nine eyes with nAMD were treated with a mean of 6.4 intravit-real injections of ranibizumab. ORT size varied from 60 μm to 600 μm without changes with anti-VEGF therapy after 6 months of follow-up.3
Iaculli et al. performed a prospective case series of 78 eyes (78 patients) with subfoveal CNV due to nAMD and BCVA of 35 ETDRS letters or better to assess the ORT incidence, its characteristics, and vision in patients with and without ORT. Patients were treated with ranibizumab therapy in a PRN regimen after 3 monthly doses for at least 12 months. No eyes presented ORT at baseline. ORT occurred in 16 eyes (20.5%) after a mean follow-up period of 8.06 months ± 2.67 months (range: 3 months to 12 months). At the end of the follow-up period, mean BCVA of the patients with ORT was statistically worse from those of patients without ORT (54.5 letters ± 6.5 letters vs. 66.5 letters ± 64.5 letters, P < .0001).12
Massamba et al. analyzed 24 eyes (24 patients) with nAMD previously treated with more than six ranibizumab injections that switched to aflibercept due to treatment failure after 6 months of treatment. ORT was present in 23 eyes (97%) prior to treatment switch and in 18 eyes (75%) after (P = .219). No statistically significant improvement in ETDRS BCVA was observed after initiation of aflibercept (P = .12). Additionally, there was no statistically significant difference in ETDRS BCVA between eyes with and without ORT after switching to aflibercept therapy (38 letters ± 15 letters and 35 letters ± 15.5 letters, P = .70).13
In a prospective cohort within the Comparison of Age-Related Macular Degeneration Treatments Trials (CATT), Lee et al. evaluated the prevalence of ORT and its associations with patient-specific and ocular features at baseline and follow-up in eyes with nAMD following anti-VEGF therapy. Patients in the CATT study were randomly assigned to receive ranibizumab or bevacizumab treated in a monthly / PRN dosing regimen. A subset of eyes was imaged with SD-OCT beginning at week 56. ORT was present in seven of 69 eyes (10.1%) with IRF at week 56 and in 64 eyes of 368 eyes (17.4%) with IRF at week 104. When comparing eyes with and without ORT at week 104, eyes with ORT had worse mean ETDRS BCVA than eyes without this structure (58.5 ETDRS letters vs. 68.8 letters; P < .0001). Eyes with worse baseline VA had increased risk of having ORT at 104 weeks (P = .003). Furthermore, in some eyes, ORT appeared to change over the time. Among the seven eyes with ORTs at week 56, only one eye (14.3%) still had ORT at week 104, and among the 52 eyes without ORTs at week 56, eight eyes (15.4%) had new ORTs at week 104.14
Predictive Factors for the Occurrence of ORT
CNV lesions with classic components were the most associated with ORT. Faria-Correia et al, Massamba et al and Dolz-Marco et al identified classic components in CNV lesions in 67.7%, 70.8% and 85%, respectively, of eyes with ORT.6,11,13 Lee et al found that absence of diabetes, poor VA, blocked fluorescence, GA, greater lesion size, and presence of subretinal hyper-reflective material (SRHM) at baseline were independently associated with greater risk of ORT at 104 weeks (P < .05).14 Demographic characteristics and drug and dosing regimen were not significantly associated with the occurrence of ORT.9,14
Anatomical Features of ORT on SD-OCT
ORT has been associated with diverse alterations in retinal layers as seen by SD-OCT images, and various phenotypes of ORT have been described. Dirani et al. described changes in the outer retina prior to the ORT development as flat SRF, small dome-shaped elevations of the ellipsoid layer, and hyperreflective dense nodules in the outer nuclear layer (ONL).10,14
Various shapes of ORTs were identified on SD-OCT: open, closed, forming and, branching (Figure 1).15 Schaal et al. described two subtypes of formed ORT: open ORT, in which the hyperreflective ring was incomplete, and closed ORT, in which the IS of the photoreceptors and external limiting membrane (ELM) were completely encircled. Both subtypes were preceded by the forming ORT. They observed closed ORT in 34 patients (100%), whereas open ORT was observed in 14 patients (41%).16 Likewise, Dolz-Marco et al. identified 89% of the closed form and 11% of the open form.6 Wolff et al. found 11 eyes (73.3%) with “pseudodendritic” or branching ORTs (near fibrotic tissue) and eight eyes (100%) with “perilesional” or singular ORTs (near atrophic tissue).17
Outer retinal tubulation (ORT) types in cross-section for intensity analysis. (A) Closed ORT, (B) open ORT, (C) forming ORT, and (D) branching ORT. Hyperreflective ORT bands were sampled at 0° (red), 90° (yellow), 180° (green), and 270° (cyan). Raw (linear) spectral-domain optical coherence tomography intensity (ranging from 0 to 1 in a 32-bit image) by ORT type and angle was sampled at 5-μm steps from the inner aspect of ORT band along the colored lines. Lines of the cross in the centroid of ORT cross-sections are 20 μm long. (E) Schematic representation of different shapes of ORTs. Pictures modified from Litts et al.15 highlighting the ORTs structure and scars.
Outer retinal layers were analyzed in multiple studies and classified as to the retinal structure impacted upon: the outer retina (photoreceptor and ELM) and RPE integrity, and presence of fibrosis and fluid (IRF, SRF, and sub-RPE fluid). Faria-Correia et al. found loss of subfoveal photoreceptor integrity in all patients with ORT when compared to 63.29% of patients without ORT (P = .0007) and Espina et al. reported disruption in the foveal ELM in almost 93.9 % of eyes with ORT when compared to 66.7% of eyes without ORT (P = .017).9,11 In addition, Schaal et al. observed that the underlying RPE was dysmorphic or absent in patients with ORT, and Wolff et al. reported no deficit in photoreceptors, ELM, RPE, or IS/OS junction in areas where ORTs were present.3,16
Espina et al. observed ORT in 48.5% of eyes with IRF, 42.54% of eyes with combined IRF and SRF, and in 9.1% of eyes with just SRF. It was also noticed that eyes that presented ORT changes during the follow-up had IRF (60%) or combined SRF (40%) at baseline.9 SRF was also present in 80.64% (P = .0000) and 85.93% (P = .012) of eyes with ORT, according to Faria-Correia et al. and Lee et al., respectively.11,14 Larger scars were also noticed in eyes with ORT had when compared to eyes without ORT (2,871.79 μm ± 1,587.130 μm vs. 1,637.89 μm ± 1,830.37 μm, respectively; P = .027).9
Two studies that analyzed the correlation between GA due to AMD and ORT were identified. Hariri et al. compared the enlargement rate of GA in eyes with and without ORT in patients with AMD who completed the MAHALO study (phase 1b/2 multicenter, randomized, single-masked, sham-injection, controlled clinical trial of the safety, tolerability, and evidence of activity of lampalizumab in patients with GA associated with AMD). ORT was present in 24 eyes of a total of 108 eyes included in the study. In eyes with ORT, the enlargement rate of GA was significantly slower when compared to eyes without ORT (1.85 ± 0.78 vs. 2.67 ± 1.61; P = .001).18
Moussa et al., in a post-hoc analysis of 43 eyes of 43 patients with AMD with GA who participated in the GATE study (a randomized trial to evaluate the intravitreal injection of tandospirone in GA secondary to AMD), aimed to identify with SD-OCT morphologic alterations in eyes with GA due to AMD and to correlate them with GA size, enlargement rate, and presence of multifocal patches of atrophy. Twenty-eight eyes (65%) presented ORTs in atrophic regions. There was no association between larger lesion sizes (P = .096) or the presence of multifocal patches of atrophy (P = .723) with ORT. However, ORTs in the atrophic regions had faster enlargement rates (P = .003).19,20
The pathogenesis of ORT is still unclear. To date, four studies investigated this matter through SD-OCT and histology evaluation.
Spectral-Domain Optical Coherence Tomography Findings
Zweifel et al. proposed that ORT is a result of a rearrangement of the photoreceptor layer in response to a sublethal retinal injury caused by IRF and SRF. The loss of the interdigitation of the outer segments with the RPE or the degeneration of the RPE itself and the disruption of tight junctions of neighboring cells possibly lead to outward folding of the photoreceptor layer until opposite sides of this fold establish contact and form new lateral connections through tight junctions, thus reconstituting the IS/OS junction and forming a tubular structure. When ORT develops after treatment with the fluid disappearance, the residual cells reconnect laterally, resulting in the tubulation process. Zweifel et al. also postulated that ORT could communicate with the subretinal space (open ORT), allowing leakage from the CNV to the ORT. This communication could explain the temporary decrease in the ORT size after anti-VEGF therapy.2
Additionally, Dolz-Marco et al. proposed progressive steps in ORT development based on shapes of the ELM descent at the border of outer retinal and RPE atrophy as flat, curved, reflected, or scrolled. As the last step of ELM descent, the ELM scrolls over Bruch's membrane.6 On SD-OCT, ORT borders appear hyperreflective due to a combination of IS mitochondria and ELM.21
Schaal et al. reported four histologic phases of photoreceptor degeneration present in the lumenal walls of ORT. The nascent phase presents both inner and outer segments protruding into the lumen; the mature phase presents inner segments only; the degenerate phase presents IS atrophy; and the end-stage phase, with mitochondrial separation and migration within IS, allowing to remain only an ELM circle made by Müller cells without photoreceptors at the end stage.16 Moreover, Litts et al. observed that during the degeneration process, ORT cone nuclei retracts from the ELM, resulting in a lack of visible myoid. The degenerating ORT cones contain small and spherical mitochondria and, sometimes, lipofuscin.22
Before the advent of SD-OCT, time-domain OCT was not capable of differentiating ORT from IRF. Thus, ORT was possibly misdiagnosed and treated as IRF with its prevalence underestimated.5,8,14 In the present review, the prevalence of ORTs in nAMD varied greatly among the studies, ranging from 8.2% to 97%.11,13 The reason for this variability might be due to the fact that the majority of studies were retrospective and different methodologies were used.
ORT incidence tends to increase with the duration of the disease, ranging from 2.5% at the diagnosis of AMD to 41.6% at 4 years of follow-up.10 Prospective studies confirmed that ORT is also associated with worse VA.5,10–12,14 This fact is compatible with the presence of ORT in advanced macular diseases.
ORT has a predilection for areas with large CNV, sub-retinal hyperreflective material, and areas with previous IRF.2,14 The majority of patients with ORT had classic or minimally classic (or mixed) CNV due to its proximity to the IS/OS junction and ELM.6,11,13 ORT originates from RPE and/or photoreceptors dysfunction.2 There are four histologic phases of photoreceptor degeneration present in the lumenal walls of ORT. The end-stage has no recognizable photoreceptors, only ELM that is considered a requirement for ORT formation.16
It is not clear if anti-VEGF therapy has any effect on ORTs or if the dynamic changes seen during treatment are just their natural course. This fluctuation could be explained by a vascular component in the ORT structure or by a communication between the ORT and the CNV through a subretinal space.2,9 According to Massamba et al., switching from ranibizumab to aflibercept reduced the prevalence of ORTs. One hypothesis for this response is that aflibercept could induce a restoration of the outer retinal layers, particularly the ELM.13,23
The association between GA enlargement and the presence of ORT remains controversial due to the lack of published articles on this matter with different results.
A significant limitation of reviewing data regarding ORT is that most of the available studies are retrospective or small prospective reports. A long-term, prospective study with a larger number of participants is needed to further investigate the pathogenesis and ORT response to anti-VEGF therapies.
In conclusion, ORT is a structure present in advanced stages of macular diseases with degeneration of photoreceptors and RPE dysfunctions. It is associated with poor VA and does not require treatment with anti-VEGF, unless fluid is present.