Advanced OCT Analysis of Biopsy-proven Vitreoretinal Lymphoma

Abstract

PURPOSE:

Although diagnosing vitreoretinal lymphoma (VRL) can be challenging, early detection is critical for visual prognosis. We analyzed the spectrum of optical coherence tomography (OCT) findings in patients with biopsy-proven VRL and correlate these features with clinical parameters.

DESIGN:

This retrospective cross-sectional study was a multicenter chart review from 13 retina, uveitis, and ocular oncology clinics worldwide from 2008 to 2019. We included patients with a diagnosis of biopsy-proven VRL imaged with OCT at presentation. Ocular information, systemic information, and multimodal retinal imaging findings were collected and studied. The main outcome measure was the characteristics of VRL on OCT.

RESULTS:

A total of 182 eyes of 115 patients (63 women, mean age 65 years) were included in this study. The disease was bilateral in 81 patients (70%), and mean baseline visual acuity was 0.2 ± 0.89 logMAR (Snellen equivalent, 20/32). At baseline, 38 patients (33%) presented with isolated ocular involvement, 54 (45%) with associated central nervous system involvement, and 11 (10%) with other systemic lymphomatous involvement; an additional 12 patients (10%) presented with central nervous system and other systemic involvement. On OCT, tumor infiltration was identified in various retinal layers, including lesions in the subretinal pigment epithelium compartment (91% of eyes), the subretinal compartment (43% of eyes), and the intraretinal compartment (7% of eyes). OCT analysis of eyes with VRL identified 3 main regions of retinal infiltration. Subretinal pigment epithelium location, with or without subretinal infiltration, was the most common pattern of involvement and isolated intraretinal infiltration was the least.

Vitreoretinal Lymphoma (VRL) IS STRONGLY associated with primary central nervous system (CNS) lymphoma^1,2 and is the most prevalent and aggressive extranodal presentation of non-Hodgkin’s lymphoma. Histopathologically, the cells are of diffuse large B lymphocyte origin in 80% of cases.³ Ocular involvement in VRL can be primarily diagnosed in the eye, concurrent with or after CNS lymphoma diagnosis.⁴ Patients diagnosed with CNS lymphoma develop ocular involvement in 15% to 25% of cases.^5–9 In lymphoma primarily diagnosed in the eye, 16% to 34% of patients will have CNS involvement at ocular diagnosis^10,11 and 65% to 90% of the remainder will eventually develop CNS manifestations, usually within 2 to 3 years.^4,5 Overall, between 35% and 90% of patients with VRL show evidence of CNS lymphoma at some point during the course of their disease.

VRL typically occurs in older patients with a median age of 60 years.^12–14 The high mortality rate is due to CNS lymphoma, which has a mean survival of 32 months³ and a 5-year survival rate of 61%.¹⁵ Although early detection is critical, the diagnosis of VRL can be challenging.³ VRL represents less than 0.01% of all vitreoretinal diagnoses, and therefore may not be considered in the initial differential diagnosis.^2,16 The clinical findings are considered non-specific, but can lead to vision loss and even blindness. VRL can masquerade as a posterior or intermediate uveitis that may initially be responsive to local and systemic corticosteroids, but recurs with prolonged corticosteroid treatment.^17–19

Owing to its propensity to mimic other ocular diseases, VRL is considered a masquerade syndrome. At tertiary referral centers, VRL has been reported to comprise 33% of all masquerade syndromes, 81% of neoplastic masquerades, and approximately 2% of all uveitis referrals.^20,21 The classic ophthalmoscopic presentation includes vitreous cells with multiple placoid yellow–white subretinal pigment epithelium (RPE) infiltrates, and RPE alterations including a leopard spot pattern.^22,23 However, these fundus findings occur in only about 40% of patients.²⁴ Although macular edema is not a typical finding in VRL, intraretinal fluid has been identified in some studies.^24–28 Other reported clinical findings include cells in the aqueous humor²⁹ and sheets of vitreous cells associated with retinal phlebitis.³⁰

Noninvasive imaging modalities such as optical coherence tomography (OCT) have previously demonstrated the capability to display characteristic findings of VRL, although this evidence is largely limited to case reports or small case series^23,31–37 including cases following treatment where the diagnosis of VRL was already made.³⁸ Although such modalities have been proposed to enable earlier identification of VRL and may decrease dependence on invasive biopsy, there are few OCT studies investigating the morphologic features of VRL and the correlation with ocular and systemic findings. Furthermore, earlier diagnosis informed by an increased awareness of the OCT findings of this entity can enable earlier treatment, which, in turn, has the potential to decrease ocular morbidity and systemic mortality.

The aim of this study was to analyze a large series of pretreatment OCT images from patients with biopsy-proven VRL and to describe the spectrum of OCT features in the early and late stages of the disease, as well as correlate these OCT findings with clinical data and outcomes.

METHODS

Institutional review board approval was determined by the respective coauthors according to the requirements established by their institutional centers. The study adhered to the principles set forth by the Declaration of Helsinki.

At each institution, patients with the diagnosis of VRL were identified through a retrospective review of medical records from 2008 to 2019. Inclusion criteria for the analysis included (1) VRL identified on the basis of either positive ocular tissue biopsy, including vitreous cytologic analysis or positive CNS tissue biopsy; (2) gradable baseline OCT raster scans through lymphomatous lesions, whether in the macula or extramacular; and (3) high-quality color fundus photographs. De-identified patient records and multimodal imaging data from the baseline and subsequent examinations were collected and reviewed.

Baseline and final follow-up Snellen visual acuity (VA) were recorded and converted to LogMAR values. Visual acuities of count fingers, hand motion, light perception, and no light perception vision were assigned LogMAR designations of 1.7, 2.0, 2.3, and 3.0, respectively. Biomicroscopic examination information were recorded as well as dilated fundus examination findings to determine the location of the lymphomatous lesions (macular, extramacular, or a combination of both). Additional information collected from the medical record included patient demographics including age, gender, and race; symptoms at baseline; time between presentation and definitive diagnosis of lymphoma; details regarding the investigational workup, including the surgical procedures required for diagnosis; methods of histopathologic and molecular diagnosis; time from diagnosis of VRL to CNS involvement; and the list of ocular and systemic treatments received. Medical records were reviewed to determine any evidence of CNS or non-CNS systemic involvement, and the time interval from presentation to CNS or non-CNS systemic progression was also evaluated. The term “VRL” was adopted for all modes of presentation where the main affected compartments did not affect the choroid on imaging with OCT. Clinical features at baseline ocular examination, including the pattern of retinal involvement (subretinal or intraretinal) were also recorded.

Baseline pretreatment OCT raster scans were independently assessed by two masked observers (F.P. and R.D.M.) for the presence of sub-RPE, subretinal, and intraretinal hyperreflective material, and for the presence of subretinal or intraretinal fluid. These OCT findings were reviewed with the senior author (D.S.) and correlated with the systemic and CNS findings.

Color fundus photographs and widefield multicolor fundus images were analyzed for the presence of vitreous haze, mass lesions, areas of atrophy, and RPE alterations. The optic nerve head was analyzed for the presence of peripapillary hemorrhages or blurred margins suggestive of lymphomatous involvement of the optic nerve. When available, fundus autofluorescence (FAF) images were evaluated for patterns of abnormal hyperautofluorescence or hypoautofluorescence, including a leopard spot pattern, defined as hyperautofluorescent spots ranging from 50 μm to 250 μm in diameter alternating with adjacent hypoautofluorescent spots in the posterior pole and/or periphery. Hypoautofluorescence owing to blockage was defined as RPE autofluorescence masked by a subretinal infiltrate. Hyperautofluorescence was defined as multiple hyperautofluorescent spots corresponding to areas of outer retinal disruption, more readily identifiable on FAF than on fundus photography.³⁹ When available, fluorescein angiography (FA) findings were evaluated for the presence of a leopard spot pattern showing granular areas of hyperfluorescence intermixed with hypofluorescence, areas of hyperfluorescent window defects owing to RPE loss, regions of RPE mottling with late staining, and hyperfluorescent foci or areas of dye leakage and/or dye pooling. Leakage of the optic nerve head was also assessed to identify optic nerve swelling and involvement.

Statistical analysis was completed using Stata 13.1 statistical software (StataCorp). Two-group comparison of continuous variables was performed using the t test for comparison of mean values. Visual outcome across multiple groups was analyzed by comparing mean values of baseline and change in logMAR VA using 10-way analysis of variance with the Bonferroni correction for multiple comparisons. Categorical variables were analyzed by comparison of proportions using Fisher’s exact test. Hypothesis testing used two-tailed statistics and significance was defined by a P value of less than or equal to.05.

RESULTS

In this analysis, there were 152 patients with a diagnosis of VRL affecting 256 eyes. There were 23 patients (15%) excluded owing to a lack of tissue diagnosis for VRL. Of the remaining 129 patients (235 eyes), the diagnosis of VRL was based on vitreous cytologic findings in 94 patients (73%), chorioretinal biopsy in 11 patients (8%), a combination of vitreous and chorioretinal biopsy in 14 patients (11%), and brain biopsy and/or cerebrospinal fluid cytologic findings in 10 patients (8%), of which only 3 eyes (1%) also had a vitreous biopsy. Aqueous humor analysis of IL–10 and IL-6 was performed in 34 patients (26%) and the IL–10/IL–6 concentration ratio was greater than 1 in 100% of cases; however, this factor was not considered an inclusion criterion for diagnosis of VRL.

Of the 235 eyes from 129 patients with biopsy-proven VRL, high-quality, baseline OCT raster scans through lymphomatous lesions were available for 182 eyes (77%) of 115 patients. Of these 115 patients, 63 (55%) were women and 52 (45%) were men. The mean age at baseline presentation was 64.73 ± 10.86 years (range, 25–91 years) and 67 of the 115 patients (58%) exhibited bilateral disease at baseline (Table 1). The mean baseline VA of this final cohort of 182 eyes was 0.2 ± 0.89 logMAR (Snellen equivalent, 20/30).

TABLE 1.

Demographic, Ocular, and Tomographic Features of 182 Eyes of 115 Patients with Biopsy-proven Vitreoretinal Lymphoma

Features	No. ( %)
Demographics (n = 115 patients)
Sex
Male	52 (45)
Female	63 (55)
Race
Caucasian	97 (84)
African American	7 (6)
Asian	7 (6)
Indian	4 (3)
Age at presentation (y)
Mean (range)	65 (25–91)
Systemic lymphoma
Central nervous system lymphoma	54 (47)
Noncentral nervous system lymphoma	11 (10)
Both	12 (10)
Ocular features (n = 182 eyes)
Laterality (n = patients)
Unilateral	48 (26)
Bilateral	134 (74)
Visual acuity (Snellen), mean	20/30
Visual acuity (LogMAR), mean	0.2
SD-OCT features (n = 182 eyes)
Intraretinal deposits	1 (0.5)
Subretinal deposits	16 (9)
Sub-RPE deposits	101 (55)
Thickened RPE	89 (49)
Shallow PED	39 (21)
Large PED	37 (20)
Intraretinal + subretinal deposits	0 (0)
Intraretinal + sub-RPE deposits	1 (0.5)
Subretinal + sub-RPE deposits	53 (29)
Intraretinal + subretinal + sub-RPE deposits	10 (5)

PED = pigment epithelial detachment; RPE = retinal pigment epithelium; SD-OCT = spectral domain optical coherence to-mography.

Of the 115 patients who met both the tissue diagnostic and OCT inclusion criteria, 38 (33%) presented with VRL, of whom 21 had bilateral disease. Fifty-four patients (47%) (33 with bilateral ocular disease) presented with a concomitant diagnosis of CNS lymphoma. Of these, 11 (10%) were diagnosed with CNS lymphoma before ocular involvement. Eleven patients (10%) (6 bilateral) presented with a concomitant history of non-CNS systemic lymphoma and in 4 of these 11 patients (3%) the diagnosis preceded the eye involvement. An additional 12 patients (10%; 7 patients with bilateral involvement) presented with both CNS and non-CNS systemic involvement.

Among the 182 eyes of 115 patients included in the analysis, the earliest available (baseline) OCT volume scan was analyzed, before tissue diagnosis, and before any lymphoma-specific treatments. Of these 182 eyes, 124 (68%) showed a combination of both macular and extramacular manifestations, 45 (25%) displayed macular changes alone, and 13 (7%) exhibited extramacular findings alone. From the 13 patients with only extramacular manifestations, peripheral OCT scans were available in all cases. Hyperreflective lesions were observed across multiple retinal layers as summarized in Figure 1 and Table 1.

FIGURE 1.

Venn diagram illustrating baseline distribution of hyperreflective lymphomatous lesions across the various optical coherence tomography (OCT) retinal levels.

In the 182 eyes with biopsy-proven vitreoretinal lymphoma studied with OCT analysis, subretinal pigment epithelium lymphomatous lesions were the most common finding (91%) followed by subretinal lesions (43%), and 12 eyes (7%) exhibited intraretinal involvement. Isolated sub-RPE infiltrates (55%) or combined sub-RPE and subretinal infiltrates (29%) represented the most common baseline presentations. Ten eyes (5%) illustrated lesions at all 3 levels at baseline. Isolated intraretinal lesions were extremely rare.

Subretinal pigment epithelium (sub-RPE) infiltrates associated with thickening or detachment of the RPE were observed in 165 of 182 eyes (64%) (Figure 2, A through C). Sub-RPE material was further classified into 3 categories (Table 1): (1) a thickened RPE, defined as the presence of thickening of the RPE–Bruch’s membrane complex without a distinct separation between RPE and Bruch’s membrane (Figure 2, A). (2) A shallow RPE detachment or pigment epithelial detachment (PED), defined as a separation between RPE and Bruch’s membrane owing to hyperreflective material, with a maximal PED height of less than 250 μm (Figure 2, B). (3) A large PED, defined as a dome-shaped PED with a maximal height of more than 250 μm showing anterior hyperreflective material posterior hyporeflective shadowing closer to Bruch’s membrane (Figure 2, C). The most common sub-RPE feature was a thickened RPE (89 eyes [49%]), whereas shallow and large PEDs were less common (39 eyes [21%] and 37 eyes [20%], respectively).

FIGURE 2.

Optical coherence tomography illustrating hyperreflective lymphomatous lesions in all 3 retinal spaces.

In this study, hyperreflective lesions in patients with vitreoretinal lymphoma were most commonly identified in the subretinal pigment epithelium (RPE) space (A, B, C), but were also present in the subretinal space (D, E, F) and in some cases were identified in the inner and/or outer retina (G, H, I). Sub-RPE lesions displayed 3 morphological subtypes with OCT. The most common, a “thickened RPE pattern” (A, yellow arrow) was usually associated with concomitant lesions in the subretinal (E, F, yellow arrows) and intraretinal (G, H, I, yellow arrows) compartments. The “shallow retinal pigment epithelium detachment” subtype (B, yellow arrow) and the “large retinal pigment epithelium detachment” subtype (C, yellow arrow) were more rarely associated with hyperreflective lesions in the subretinal space. Subretinal lesions either extended horizontally in a band-like pattern under the retina (D, F, red arrows) or displayed a focal round hyperreflective pattern (E, red arrow) and were more commonly seen in association with intraretinal lesions (H, I, red arrows). Intraretinal extension of the hyperreflective lymphomatous lesions (G, H, I, green arrows) were very rarely isolated and were almost always identified in association with subretinal pigment epithelium and subretinal lesions.

Subretinal hyperreflective material (Figure 2, D through F) was the second most common finding with OCT and was detected in 79 of 182 eyes (43%) (Table 1). These hyperreflective subretinal infiltrates often appeared as broad or band-like shallow lesions (Figure 2, D), but others were more focal and more round in morphology (Figure 2, E). Cloudy vitelliform lesions,⁴⁰ defined as a macular detachment owing to thick hyperreflective subretinal material resembling an acquired vitelliform lesion above an irregularly thickened or rippled RPE–Bruch’s membrane complex, was detected in 16 of 182 eyes (9%) (Figure 3, A and C).

FIGURE 3.

Multimodal imaging of a case of cloudy vitelliform maculopathy associated with vitreoretinal lymphoma (VRL).

Color fundus photography of the right eye (A) illustrates cloudy creamy yellow submacular material with extensions along the superior arcade and multiple, small, punctate yellow-white spots scattered superior to the macula. Fundus autofluorescence shows low-intensity hyperautofluorescence of the cloudy vitelliform submacular lesion with speckled hyperautofluorescence in the superior macula. Spectral-domain optical coherence tomography through the fovea illustrates the dense hyperreflective subretinal material above the RPE band which can be highly suggestive of VRL, but was present in only 9% of our cohort.

Twelve eyes (7%) exhibited intraretinal hyperreflective lesions extending from the outer nuclear layer as far as the nerve fiber layer (Figure 2, G through I).

The number of eyes showing various combinations of sub-RPE, subretinal, and intraretinal involvement is displayed in Figure 1. Sub-RPE lesions were present in isolation or in combination with subretinal and/or intraretinal lesions in 91% of eyes (165/182) in this study. Isolated sub-RPE (101/182 [55%]) and combined sub-RPE and subretinal (53/182 [29%]) involvement were the most common baseline OCT features identified. Of note, 5% of eyes (10/182) displayed hyperreflective material in all 3 locations (sub-RPE, subretinal, and intraretinal) and only 1 eye (0.5%) showed isolated intraretinal hyperreflective lesions (Figure 1). Of the 3 subcategories of RPE involvement, no statistical difference was noted between eyes with a thickened RPE and a shallow RPE detachment regarding the presence of associated subretinal material (25% and 18%; P = .1). By contrast, few eyes with large RPE detachments showed hyperreflective material in the subretinal space (3%) (Figure 2, C).

An analysis of OCT features based on age did not show any statistical difference in age of patients with sub-RPE, subretinal, and intraretinal lesions (mean ages of 65, 58, and 61 years, respectively; P = .3). However, a subgroup analysis within the sub-RPE group showed that patients with a large RPE detachment on OCT were significantly older compared with the other groups (78 years; P = .05).

Intraretinal fluid was observed in 23 of 182 eyes (13%) (Figure 4, A and C) and subretinal fluid (SRF) in 20 of 182 eyes (11%) (Figure 4, B and D). Ten eyes (5%) displayed a combination of both intraretinal fluid and SRF (Figure 4, C). The presence of SRF was more commonly associated with sub-RPE material (18/20 eyes [90%]) (Figure 4B through D) and in particular with large RPE detachments (16/18 eyes [89%]; P > .001) (Figure 4, C and D). Conversely, intraretinal fluid was more commonly associated with subretinal material (14/23 eyes [61%]), but this difference was not statistically significant (P = .1).

FIGURE 4.

Optical coherence tomography (OCT) showing intraretinal and subretinal fluid in cases of vitreoretinal lymphoma.

Note the presence of intraretinal fluid (A), observed in 13% of eyes with vitreoretinal lymphoma, and subretinal fluid (B, C, D), identified in 11% of eyes in this study. Subretinal fluid was most commonly associated with subretinal pigment epithelium lesions (B, C, D), especially with large RPE detachments (B, D), that may be secondary to impairment of the retinal pigment epithelium pumping capacity especially when the RPE is further displaced from the underlying choroid.

Mean baseline VA for the 182 eyes was 0.2 ± 0.89 logMAR (Snellen equivalent, 20/30). An analysis of baseline VA according to OCT features showed that patients with a combination of subretinal and intraretinal material at baseline presented with the worst VA (1.30 ± 1.41 logMAR, 20/400 Snellen), followed by patients with combined sub-RPE, subretinal, and intraretinal material (0.96 ± 0.88 logMAR, 20/200 Snellen) and patients with subretinal material alone (0.62 ± 0.62 logMAR, 20/80 Snellen) or in combination with sub-RPE material (0.64 ± 0.50 logMAR, 20/80 Snellen). The absence of hyperreflective material in the macula or the presence of only sub-RPE lesions at baseline was associated with relatively preserved VA (0.33 ± 0.56 logMAR [20/40 Snellen] and 0.24 ± 0.37 logMAR [20/30 Snellen], respectively).

Baseline color and/or multicolor fundus photographs (Figure 5, A, E, and I) were available for all 182 eyes. No-table fundus photographic findings included vitritis in 101 eyes (55%), subretinal lesions in 148 eyes (81%) (Figure 5, E and I), retinal hemorrhage in 4 eyes (2%), leopard spotting in 17 eyes (9%; Figure 7, E) and optic nerve head edema in 6 eyes (3%).

FIGURE 5.

Multimodal imaging of 3 patients with vitreoretinal lymphoma.

Corresponding color fundus photograph, fundus autofluorescence and spectral domain OCT are illustrated for all 3 cases (A–D, E–H, and I–M). The level of the corresponding OCT is indicated by the colored line in the fundus photograph. Fundus photographs in patients with vitreoretinal lymphoma illustrate extra-macular (A, E) and macular (I) lesions either in the form of mild retinal pigment epithelium mottling (A) or as discrete subretinal pigment epithelium lesions (E, I). The most common fundus autofluorescence finding of vitreoretinal lymphoma was hyperautofluorescence in the form of a “leopard spot” pattern (B, F) which strongly correlated with hyperreflective “thickened retinal pigment epithelium” lesions on optical coherence tomography (C, D). The “large retinal pigment epithelium detachment” pattern noted with optical coherence tomography (H) typically illustrated a diffuse and homogeneous hyperautofluorescence (F) with fundus autofluorescence, while “shallow RPE detachments” of lymphomatous material (K–M) were associated with faint hypoautofluorescence (J).

FAF images were available for 162 of 182 eyes (89%). Abnormal autofluorescence on FAF was identified in 89 eyes (55% of the 162 eyes). A leopard spot hyperautofluorescence (hyperAF) pattern was present in 61 eyes (33%) and was observed in various locations in the retina (Figure 5, F) and was not limited to the location of the large subretinal infiltrates. A correlation of hyperAF with OCT patterns showed a higher association of the leopard spotting with the thickened RPE pattern (49/61 eyes; P = .05) and with large RPE detachments (12/61 eyes) (Figure 5, B through D). Nine eyes (5%) showed a pattern of hyperAF associated in 100% of cases with a focal disruption of the ellipsoid zone on OCT. Sixteen eyes (9%) showed diffuse subretinal hyperAF of mild intensity in the macular area corresponding to a cloudy vitelliform OCT pattern (Figure 3, C) and in 4 eyes (2%) the areas of diffuse hyperAF corresponded with large RPE detachments on OCT. HypoAF blockage by lymphomatous mass lesions was noted in 5 eyes (3%) with shallow RPE detachments (Figure 5, J), and hypoAF from RPE atrophy was noted in 7 eyes (4%).

FA was available for 60 eyes (33%). Twenty-five eyes (15%) displayed a leopard spot pattern with a granular appearance. Seventeen eyes (10%) showed late hyperfluorescent staining of the lymphomatous lesions, 1 eye (1%) illustrated hyperfluorescence with well-defined borders suggestive of pooling, and 2 eyes (1%) displayed early blockage with late staining of mass lesions. In 15 eyes (9%) of patients with vitritis alone, ultra-widefield FA demonstrated vascular leakage along medium and small retinal vessels in all 15 eyes. Petaloid macular leakage, that is, cystoid macular edema, was not identified in any eyes of our series.

Despite the cross-sectional nature of this study, we briefly report the follow-up OCT analysis that was available for 42 of the enrolled 182 eyes (23.08%) and illustrated persistent retinal lesions in 7 eyes (17% or 4% of the 182 eyes), resolution of all retinal lesions in 17 eyes (40% or 9% of the 182 eyes), and resolution of all retinal lesions associated with atrophy of the outer retina in 18 eyes (43% or 10% of the total).

DISCUSSION

The diagnosis of VRL is a notoriously challenging task, because the baseline presentation can masquerade as an inflammatory or infectious posterior uveitis. Given the challenges of clinical diagnosis, a number of immunologic analyses have been proposed using either vitreous or aqueous fluid, with diagnosis supported by the presence of B-lymphocyte biomarkers such as an elevated IL-10:IL-6 ratio versus the usual T-lymphocyte–derived mediators. Such analyses are only possible with highly differentiated malignancies and may be limited by the development of a T-lymphocyte–driven response to tumor antigens. A definitive diagnosis of VRL is made by vitreous and/or chorioretinal biopsy, often in conjunction with flow cytometry and cytology analysis. However, cytology specimens obtained through vitreoretinal biopsy have been shown to generate false-negative results in approximately 30% to 45% of analyses.^1,31 Vitreous samples are often insufficient, with a low cellular yield, and the results can be challenging to interpret unless the pathologist has prior experience in the diagnosis of VRL. Furthermore, cytology samples degrade rapidly unless suitably fixed, rendering results inaccurate unless analyzed promptly.¹³ Molecular testing for the presence of MYD88 mutations may enhance the diagnosis of VRL.^32,33 A definitive diagnosis requires a tissue biopsy. However, advancements in multimodal imaging can now provide an avenue to enhance the identification of features of this elusive disease and characteristic patterns, suggesting that VRL can be identified with OCT, FAF, and FA. Early suspicion and detection of the ocular disease may be aided by a greater appreciation of the spectrum of clinical findings on OCT.

OCT provides the capability to easily and rapidly acquire images that are reproducible and readily interpreted. In our series of 182 cases of biopsy-proven VRL, SD-OCT showed a preponderance of isolated hyperreflective deposits in the sub-RPE space (55%), followed by a combination of the same in the sub-RPE and subretinal space (29%). Isolated subretinal deposits were less commonly identified (9%) and intraretinal deposits extremely rare (0.5%).

Despite the power to detect a spectrum of significant pathological findings, there can be significant overlap of morphological features with other diseases, including posterior uveitis, age-related macular degeneration, and even diabetic retinopathy. Isolated extra-macular lesions were present in less than 10% of eyes on dilated fundus in this study, indicating the importance of macular OCT as a modality to enhance the early detection and diagnosis of VRL. Further OCT was successfully performed in 100% of the extramacular lesions.

Our study included the largest number of biopsy proven VRL cases captured and analyzed with OCT with relatively early findings of VRL before definitive tissue diagnosis and ocular and systemic therapy. In a recent publication, Barry and associates³⁹ described the OCT findings of VRL in a cohort of 22 patients and noted hyperreflective nodules in the sub-RPE and in the subretinal space in 16% and 53%, respectively. Casady and associates⁴¹ noted a higher incidence of sub-RPE deposits (43%) in 18 eyes with biopsy-confirmed VRL. In a recent OCT study⁴² of 55 eyes with VRL, RPE abnormalities (64%) and sub-RPE deposits (64%) were the most common findings and were isolated in 11% of eyes. Intraretinal and subretinal deposits were never detected as an isolated OCT finding, but always in combination with other OCT features (in 15% and 36% of eyes, respectively).

In our series of 115 patients (182 eyes), sub-RPE lymphomatous infiltrates were the most common OCT finding (91%). Autopsy eyes have demonstrated that lymphoma cells in VRL typically reside between the RPE and Bruch’s membrane.¹² Isolated sub-RPE lesions or in combination with subretinal lesions represented the most common baseline presentation (85%) in our study. By contrast, a baseline presentation that included infiltrates in sub-RPE, subretinal, and intraretinal locations was identified in only 5% of eyes. Intraretinal lesions were encountered rarely and almost never occurred without associated sub-RPE infiltration.

There remains uncertainty about the pathophysiology of VRL. Our understanding of VRL is increasing, but because of the rarity of the disease and the difficulty in obtaining diagnostic tissue specimens, most of our knowledge is derived from studies in patients with primary CNS lymphoma. The origin of the abnormal lymphocytes and the mechanisms that guide the targeting of specific tissues have not been elucidated.^2,4,11 Whether the lymphoma cells originate from the retinal circulation or from the choroid through Bruch’s membrane and the RPE remains controversial.¹ Only 8 cases of VRL with immunohistopathologic analysis are reported in the literature.^43–47 Three of these cases illustrated localization of diffuse B-cell lymphoma cells only within the subretinal space.^44,45 Two cases showed that the majority of the cells were located within the subretinal space with a few B cells identified within the choroid.⁴⁶ Finally, 3 cases demonstrated clear evidence of lymphocytes within the choroid.^43,44

Deák and associates⁴⁸ were the first to describe with OCT vertical hyperreflective lesions extending from the inner layers of the retina (ganglion cell layer and retinal nerve fiber layer) to the RPE in 5 of 7 patients with VRL and concomitant non-CNS or CNS systemic lymphoma and, therefore, proposed a retinal vascular origin for VRL, which is a pathway supported by many experts in the uveitis field. Our OCT findings suggest a possible progression of lymphoma cells from the choroidal vasculature across the RPE and into the sub-RPE and subretinal space, with final extension into the retina and vitreous. However, this hypothesis needs to be validated with a prospective longitudinal OCT analysis. The main support for a choroidal origin is the presence of B-cell chemokines present on RPE cells.⁴⁹ In a seminal study, Chan and associates⁴⁹ demonstrated that RPE cells expressed chemokines (BLC, SDF-1) in high levels. These chemokines bind to receptors highly expressed in B-cell lymphomas (CXCR4 and CXCR5) and are not found anywhere else in the retina. In these 3 autopsy cases of primary VRL, lymphoma cells were located between the RPE and the choroid; therefore, the authors postulated that B-cell chemokines secreted by the RPE may attract lymphoma cells from the choroidal circulation.

The possibility that lymphoma cells started in the retinal vasculature and then migrated to and proliferated in the sub-RPE space is still plausible. However, this hypothesis would require that the endothelium of the retinal vasculature, the external limiting membrane, and the RPE tight junctions are leaky enough to allow a lymphoma cell to reach the sub-RPE space. In contrast, the choroidal origin of the B cells requires only Bruch’s membrane tight junction to be disrupted.

There were additional OCT findings of interest in this study. Various forms of RPE involvement were identified including RPE thickening versus detachment. Large RPE detachments of hyperreflective lymphomatous material were associated with an older age (78.2 years; P = .05) and were frequently isolated lesions. Only rarely were large RPE detachments associated with hyperreflective material in the subretinal space (3% of total eyes in the study). Gass⁴⁸ has hypothesized that the RPE is more adherent to the underlying Bruch’s membrane in younger patients compared with older patients, based on studies with macular neovascularization. This finding may explain the propensity for large PEDs to develop in the older patients in our study, in which the lymphomatous infiltrates can more easily establish a plane for growth and extension as opposed to younger eyes, in which the lesions are more likely to break through the RPE into the subretinal space because of greater adhesion of the RPE-BM complex. Furthermore, large RPE detachments strongly correlated (P > .001) with OCT detection of pockets of SRF (Figure 4, B and D) that may be secondary to impairment of the RPE pumping capacity when the RPE is further displaced from the underlying choroid, which is the primary oxygen provider of the RPE.^50,51

The cloudy vitelliform appearance of lymphomatous infiltration described by Pang and associates⁴⁰ in 3 patients and reported by Barry and colleagues³⁹ in 31.3% of their cases is now widely recognized as highly suggestive of VRL. We identified this characteristic central finding in only 9% of our patients. However, more subtle patterns of round or band-like hyperreflectivity in the subretinal space was a common finding and was noted in 43% of eyes in this study.

This series provides considerable support for the use of OCT as an aid to identify the characteristic features of VRL, with the potential to positively influence patient outcomes. However, the conclusions we have drawn may be limited by the retrospective analysis and uncontrolled methodology of the study. In addition, ascertainment bias associated with tertiary care referrals may have confounded our analysis. It is possible that our cohort of VRL cases may represent a more advanced or severe process that is not representative of the general population of VRL cases in the community. Furthermore, owing to the retrospective and multicenter nature of the study, there was a lack of uniformity in imaging modalities used to analyze the lymphomatous lesions. For example, the OCT analyses were largely limited to the macula; thus, some extramacular lesions may have been missed.

We cannot say that our findings improve patient outcomes or increase the likelihood of pretest or prebiopsy probability. However, we believe that this article will help practitioners to better identify VRL and will enhance diagnosis and may ultimately increase the yield of tissue diagnosis; however, this awaits further validation correlating OCT with tissue diagnosis yield.

The strengths of the study included the relatively large dataset of VRL cases that were all biopsy confirmed and all imaged with high-resolution spectral domain OCT before VRL treatment.

In summary, the present study analyzed 115 patients and 182 eyes with biopsy-confirmed VRL and illustrated 3 main lesion types with OCT: sub-RPE, subretinal, and intraretinal infiltrates. Isolated sub-RPE infiltration or a combination of sub-RPE and subretinal infiltration were the most common patterns of baseline presentation. Intraretinal infiltration was uncommon and almost never occurred without associated sub-RPE infiltration. Although a diagnosis of VRL cannot prescind from tissue analysis, these findings may point clinicians toward early detection. Spectral domain OCT, a commonly used office tool with rapid acquisition and readily interpretable image analysis, may enhance the identification of lymphomatous infiltrates.