Intraocular Lymphoma

The 2017 World Health Organization (WHO) Classification of Tumors and Hematopoietic and Lymphoid Tissues recognized five main categories of lymphoid neoplasms: 1) precursor lymphoid neoplasms; 2) mature B-cell neoplasms; 3) mature T- and NK-cell neoplasms; 4) Hodgkin lymphomas; and 5) immunodeficiency-associated lymphoproliferative disorders.^1–3 Intraocular lymphomas are generally uncommon, but when they do occur, are almost always non-Hodgkin, B-cell in type, and one of two broad categories. The first category is uveal lymphomas, which tend to be small-cell, low- to intermediate-grade aggressivity, relatively benign extranodal marginal zone or follicular in subtype, and rarely associated with central nervous system (CNS) involvement.⁴ The second category is vitreoretinal lymphomas (VRLs) – previously referred to simply as intraocular lymphomas – which tend to be diffuse large B-cell, high-grade aggressivity, and frequently associated with malignant primary CNS lymphoma (PCNSL).^5–8 Large B-cell lymphomas may involve the iris, ciliary body, and choroid, but these are rare, and when they occur tend to be more aggressive and with more frequent systemic, albeit non-CNS, involvement.⁹ Even more rarely, aggressive large B-cell intraocular lymphomas may be mostly intravascular, or angiotropic, in which case CNS involvement is common.^10,11

The diagnosis of both uveal and VRLs can be challenging, often leading to delays in diagnosis.^3–8 The reasons for this are complex, but include the fact that many patients present initially with fairly non-descript intraocular inflammation; that an early, albeit transient, therapeutic response to corticosteroids is not uncommon; and that most patients who present with intraocular lymphoma show no evidence of extraocular involvement, despite appropriate contrast-enhanced full-body computerized tomographic (CT) imaging in those with primary uveal involvement and contrast-enhanced magnetic resonance imaging (MRI) of the brain and spinal cord in those suspected of having VRL. Moreover, difficulties collecting and analyzing tissue, vitreous, and aqueous humor biopsy specimens outside of tertiary referral centers accustomed to running specialized cytologic, immunologic and genetic analyses are commonplace. Once diagnosed, therapeutic approaches to intraocular lymphoma vary by type, extent of involvement, treatment history, response to therapy, and treatment center.^3–8 In this issue of Ocular Immunology & Inflammation (OII), one expert consensus recommendation,¹² one invited review,¹³ seven original articles^14–20 and four letters^21–24 address aspects of the development, diagnosis, clinical and imaging characteristics, and management of intraocular lymphomas.

Carbonell et al.¹⁴ summarized the consensus results of an expert panel convened by the International Uveitis Study Group (IUSG) to improve the recognition and diagnosis of VRLs. Participants employed two rounds of the well-described Delphi method, which involved an online questionnaire completed by 28 respondents, followed by a live discussion involving 23 participants on November 26, 2018. Consensus statements included in the guidelines required 75% or greater agreement. While the results of the questionnaire emphasized the broad range in both demographics and clinical and imaging characteristics, consensus suggested the majority patients with VRLs are over 50 years of age and experience painless loss of vision and floaters without eye redness or photosensitivity. Most patients have bilateral involvement, often with keratic precipitates (KPs), mild anterior chamber inflammation, and vitritis that tends to be organized as sheets and/or vitreous veils. When active retinal lesions are present, they tend to be deep or outer retinal and creamy- or yellow-white in appearance. Retinal pigment epithelial (RPE) disruption producing a characteristic leopard-spot appearance is common, both clinically and on fluorescein angiography (FA) and fundus autofluorescence (FAF). Optical coherence tomography (OCT) is particularly useful for demonstrating the deep retinal or sub-RPE location of suspected lymphomatous infiltrates, which may resolve spontaneously and/or migrate with time.²⁵ For patients suspected of having VRL, contrast-enhanced brain MRI is recommended. If the brain MRI is negative, a spinal tap and/or a diagnostic vitrectomy for cytology, flow cytometry, and molecular analysis can be considered. Vitreous analysis that shows an elevated interleukin (IL)-10:IL-6 ratio, mutations in the myeloid differentiation primary response 88 (MYD88) gene—especially leucine to proline at codon 265 (L265P), or immunoglobulin heavy chain (IgH) and/or T-cell receptor (TCR) gene rearrangements are each strongly supportive of the diagnosis. Undiluted vitreous samples should be collected using a low cutting speed, preferably 2 weeks or more after discontinuing systemic corticosteroids. Uncommon features of VRL noted in the report included the presence of pseudo-hypopyon, anterior or posterior synechiae formation, the presence of pars plana exudates (snowbank), vitreous hemorrhage, retinal vasculitis, and macular edema – evident either clinically or on FA. While contrast-enhanced full-body CT with or without positron emission tomography (PET) can be used to search for involvement beyond the brain or eyes, and should be employed in subjects suspected of having uveal lymphoma, such systemic involvement is uncommon in VRLs. The authors concluded that the diagnosis of VRL requires a high index of suspicion and is often delayed by early false-negative findings.

Steffen et al.¹³ reviewed VRL in human immunodeficiency virus (HIV)-infected patients, who are more likely to be male and younger than non-immunocompromised patients with the same malignancy. The authors noted that PCNSL, and by extension VRL, is 15 to 20 times more common in HIV-positive patients, typically as CD4 + T-cell counts drop below 50 cells/microliter. For this reason, the differential diagnosis in severely immunosuppressed patients suspected of having VRL should include infections known to occur in those with acquired immune deficiency syndrome (AIDS) – most notably cytomegalovirus (CMV), syphilis, and toxoplasmosis.^26,27 Many HIV-infected patients with PCNSL, and again by extension VRL, have active Epstein-Bar virus (EBV) infection – presumably related to relative loss of EBV suppressing CD4 + T-cells. In addition to the diagnostic studies mentioned above by Carbonell et al,¹² Steffen et al¹³ noted that testing of tissue or fluid samples for the presence of EBV DNA may have value.²⁸ Those HIV-infected patients found to have VRL without PCNSL should undergo anti-retroviral therapy (ART) to strengthen their immune system along with intravitreal chemotherapy with methotrexate (MTX), rituximab, and/or local external beam radiation, whereas those with PCNSL involvement should receive ART with systemic chemotherapy and/or full-brain radiation. The use of EBV therapy with foscarnet or ganciclovir has been suggested by some, but is not universally employed.²⁸ The review reminds us to consider VRL in HIV-positive patients with intermediate or posterior uveitis and, conversely, to consider HIV infection is those found to have VRL with or without PCNSL.

Giuffrè et al¹⁴ retrospectively reviewed the five-year experience with 59 eyes of 31 patients with diffuse large B-cell VRLs seen at their tertiary referral center in Milan, Italy, from 2014 to 2019. All diagnoses were made histologically, and CD19, CD20 and/or CD22 immunopositivity with restricted expression of either κ or λ light chains confirmed the diagnosis. Molecular analysis performed on selected patients identified the missense mutation L265P in the MYD88 gene. Patients with VRL without CNS involvement were treated with alternating intravitreal injections of MTX and rituximab. Those with PCNSL received varying combinations of systemic and intrathecal chemotherapy, full-brain radiation, and intravitreal MTX and rituximab. Mean time from first ocular symptoms to diagnosis of VRL was 7.4 months (range = 1 to 24 months), but tended to be shorter in those subjects with preceding PCNSL (mean = 4 months) verses those without PCNSL (mean = 9.9 months; nominal p-value = 0.02). The cohort had a mean age of 69 years (median = 70 years; range = 51 to 87 years), and contained slightly more women (17; 54.8%). Involvement was bilateral in 28 subjects (90.3%). At baseline, PCNSL was identified in 11 subjects (35.5%), but increased to 25 patients (80.6%) by final visit, which ranged from two to 15 months (mean = 15 months). Mean vision at presenation was 0.43 LogMAR (~ Snellen 20/54) and at last visit was 0.52 (~ Snellen 20/66). Thirty eyes (50.8%) had both anterior and intermediate inflammation, whereas 29 (49.2%) had no anterior inflammation. Thirty six eyes (61.0%) had retinal findings, most commonly RPE disruption producing a leopard-spot appearance (28 eyes; 47.5%), which was typically visible on both FA and FAF. Among the 33 subjects who had vitreous collected for cytology, 12 (36.4%) were positive. The MYD88 L265P mutation was identified in 19 or 23 (82.6%) vitreous samples and 31 or 41 (75.6%) of aquesous samples tested. Among the 23 eyes that had MYD88 L265P testing on both the aqueous and vitreous collected on the same day, four were (17.4%) negative, three (13.0%) were positive from the vitreous only, and 16 (69.6%) were positive from both the aqueous and vitreous. Treatments varied considerably, but included: systemic plus intravitreal chemotherapy (16 patients; 51/6%); systemic chemotherapy with full-brain radiation (12 patients; 38.7%); systemic chemotherapy alone (8 patients; 25.8%); autologous hematopoietic stem cell transplantation (8 patients; 25.8%); intravitreal chemotherapy alone (3 patients; 9.7%); and systemic plus intrathecal chemotherapy (2 patients; 6.5%). Among patients with PCNSL, five-year mortality was 32.3%, with a mean time from diagnosis to death of 11.3 months (range = 3 to 26 months). While the authors acknowledged the diagnostic predictive power of any single clinical or multimodal imaging feature, they emphasized the importance of the combined presence of multiple suggestive features, such as abnormal sheets or veils or vitreous cells, subretinal infiltration, and a leopard spot pattern of RPE disruption – particularly in older, at-risk patients. Their work confirmed the utility of MYD88 mutation testing,^29–37 including the use of minimally invasive paracentesis – as reported recently by Hiemcke-Jiwa et al³⁸ and Miserocchi et al.³⁹

Wang et al.¹⁵ evaluated six distinct fixatives for processing VRL fluid specimens with regard to preservation of cellular morphology in histology/immunohistochemistry and DNA/RNA integrity. All tests were run on B-cell-like diffuse large B-cell lymphoma Pfeiffer cells. The six fixatives included PreserveCyt Solution, Shandon^TM Cytospon^TM Collection Fluid, HOPE® I solution, PAXgene Tissue FIX solution, 10% buffered formalin, and RNAlater^TM. Several clinical samples were tested with PreserveCyt Solution. Investigators then examined cellular morphology on histology/immunohistochemistry, nucleic acid quantity and integrity, flow cytometry, MYD88 mutation detection, and IgH gene clonality. While the different fixatives had relative strengths and weaknesses across the assays performed, the authors concluded that PreservCyt provided optimal preservation of cytologic and nucleic acid integrity prior to VRL molecular analysis.

Shi et al.¹⁶ used droplet digital polymerase-chain reaction (ddPCR) to identify a L265P missense mutation in MYD88 from vitreous samples from 29 eyes of 20 subjects with intraocular lymphomas seen at a tertiary referral center in Shanghai, China. The cohort included 18 diffuse large B-cell lymphoma (DLBCL), one natural killer T-cell lymphoma (NKTL), and one lymphoplasmacytoid lymphoma (LPL). All subjects also had PCNSL. The L265P mutation was identified in 20 of 29 vitreous samples (69.0%). The authors acknowledged comparably high rates of this specific MYD88 mutation in other diverse cohorts,^29–37 and concluded that ddPCR may offer a rapid, sensitive, and specific approach to diagnosing VRL.

Mahendradas et al.¹⁷ used in vivo confocal microscopy (IVCM) to describe the morphological patterns of KPs observed in six eyes of three patients with biopsy-proven VRL seen at a tertiary referral center in India. All six eyes had an anterior chamber reaction with fine to medium-sized KPs, which were pigmented in three eyes and white in three eyes. One eye has a pseudo-hypopyon. Confocal microscopy showed the active KPs to have a central cellular conglomerate with a surrounding partial or complete petaloid appearance. The authors proposed that this unique floral or petaloid pattern to KPs on IVCM might be used to support the diagnosis of VRL.

Koç et al.¹⁸ described a 59-year-old man who presented to their referral center in Ankara, Turkey with treatment refractory anterior uveitis and a hemorrhagic hypopyon who was found to have a primary ciliary body marginal zone lymphoma. The authors cited three previously reported cases of lymphoma localized to the ciliary body^3,40,41 and reports of iris lymphomas, some of which also involved the ciliary body.⁴² Many were indolent and responded to treatment with external beam radiotherapy, and several displayed pseudo-hypopyon. The authors emphasized the treatment refractory nature of the inflammation, including pseudo-hyopyon formation, as signs that might warrant analysis of aqueous humor cells and/or excisional biopsy, as was performed in their case.

Zhou et al.¹⁹ presented two patients, a 78-year-old woman and a 59-year-old man, who presented with a unilateral pseudo-hypopyon and in whom diagnostic cytologic, immunohistologic, and molecular analyses were performed on aqueous humor aspirates. The affected eye of 78-year-old patient presented with ocular hypertesion and hyphema, whereas a hyphema developed in the affected eye of the 59-year-old patient within a few months of presentation. In the female patient, histology was consistent with large B-cell lymphoma, immunohistochemical staining showed CD20+ lymphocytes, flow cytometry revealed CD19 and CD20 immunophenotypes, and molecular analyses showed IgH gene rearrangement. In the male patient, analysis revealed large, malignant lymphoma cells, an elevated IL-10:IL-6 ratio, and IgH gene rearrangement. The authors reviewed 20 previously reported cases of intraocular lymphoma with pseudo-hypopyon formation and made the observations that all but three of the 22 total cases (86.4%) occurred in the setting of systemic lymphoma; that inflammation was bilateral in 10 cases (45.5%); and that six eyes each had hyphema or increased intraocular pressure (IOP), including the two eyes of their two patients, as described above. In most of the 20 previously reported cases the diagnosis was based on histology and/or immunostaining of cells obtained from the aqueous humor.

Marchese et al.²⁰ retrospectively reviewed the clinical and OCT features of six eyes from six subjects with VRL who had retinitis-like lesions. Clinical and OCT features identified in these lesions were compared to the same numbers of lesions obverved in patients from their clinics with viral retinitis or toxoplasmic retinochoroiditis. Eyes were examined at baseline and following treatment. On clinical examination at presentation, the VRL eyes were less likely to have anterior chamber inflammation, KPs, elevated IOP, or feathery edges at the retinal infiltrates than eyes with either viral retinitis or toxoplasmic retinochoroiditis. Intraocular lymphomatous infiltrates were also more likely to be multifocal and to lack an adjacent or nearby retinochoroidal scar compared to eyes with toxoplasmic retinochoroiditis. Analysis using OCT showed that retinitis-like lesions in VRL tended to be thicker (> 600 vs ~ 240 microns) with undulations at the vitreoretinal interface that the authors described as having a ‘rounded roof’ appearance, and to have associated sub-retinal or sub-RPE infiltrates. Following treatment, eyes with viral retinitis or toxoplasmic retinochoroiditis showed retina thinning or atrophy with loss of retinal architecture due to necrosis. In contrast, those eyes with retinitis-like lesions due to VRL tended to show varying degrees of restoration of the normal retinal architecture with resolution of the subretinal/sub-RPE infiltrates. The authors acknowledged the limited sample size of their study, but suggested that identification of distinctive clinical and OCT features may be useful in the diagnosis of VRL.

Marchese et al.²¹ described a 44-year-old woman with chronic low-grade fever who presented to their tertiary referral clinic in Milan, Italy with bilateral granulomatous anterior uveitis, low IOP, and hypotony maculopathy. The inflammation improved in response to systemic and topical corticosteroids, whereas the IOP slowly normalized with the addition of topical 2% ibopamine administered six times daily. Diagnostic biopsy of an enlarged cervical lymph node lead to the diagnosis of Hodgkin lymphoma. The authors noted how uncommon it is for Hodgkin lymphoma to metastasize to the eye and reminded us of the utility of topical ibopamine for the treatment of ocular hypotony.

Sone et al²² described an 81-year-old Japanese woman seen in their tertiary referral clinic in Tokyo, Japan and who was taking long-term MTX for rheumatoid arthritis, and who developed decreased vision associated with anterior and intermediate uveitis in each eye. Vitreous biopsy lead to the diagnosis of large B-cell lymphoma based on characteristic histology, elevated IL-10:IL-6 levels, and IgH gene rearrangement. The authors noted that while intraocular lymphoma had not been described following long-term MTX use, systemic lymphoproliferative disorders have occurred in this setting, which lead them to consider the diagnosis in their patient. The patient’s intraocular lymphoma regressed following diagnostic vitrectomy and discontinuation of the MTX.

Choi et al.²³ described a 49-year-old woman seen in their tertiary referral clinic in South Korea who had bilateral VRL with immunoglobulin gene rearrangement, an elevated IL-10:IL-6 ratio, and MYD88 L265P gene mutation identified from vitreous biopsy specimens. The study used serial PCR-based L265P mutation analysis of aqueous humor to demonstration loss of detection of the mutation following two to four intravitreal MTX injections, which was accompanied by clinical regression of the inflammation. The authors suggested that serial aqueous analysis for L2656P mutation may serve as a surrogate marker for disease activity in VRL patients.

Mohd Fauzi Yap et al.²⁴ described a 45-year-old man with systemic diffuse large B-cell lymphoma who developed panuveitis with RPE disruption bilaterally and optic disc infiltration with venous congestion and tortuosity and hemorrhages in all four quadrants in the right eye. Diagnoses of both CMV retinitis and tuberculous uveitis with optic nerve involvement were considered given the patient’s immunosuppressed status and positive tuberculin skin test history, but antimicrobial treatment for these presumed infections failed to produce improvement. The authors noted that 5% or fewer systemic lymphoma cases show metastasis to the eye,⁴³ and cited rare reports of secondary lymphomatous optic disc infiltration.⁴⁴

Together, these studies highlight continued challenges associated with the diagnosis and management of intraocular lymphoma. While clinical and ophthalmic imaging findings often suggest the diagnosis, additional testing typically involves contrast-enhanced CNS or full-body CT or CT/PET imaging to evaluate for extraocular involvement in patients suspected of having VRL or uveal lymphoma, respectively. Diagnosis frequently involves tissue, vitreous, or aqueous humor biopsy followed by specialized cytologic, immunologic, and/or genetic analyses. Treatment is most often multidisciplinary and patient-specific, with options including both local and systemic chemotherapy, with or without regional radiotherapy. Testing for HIV infection should be performed in patients found to have VRL, and those who are HIV-positive should be considered for ART.

Declaration of interest

ETC, EM, JRS, and JAG have no relevant financial conflicts.  MZ is a consultant to AbbVie, Alimera, Santen, and Gilead. The Authors alone are responsible for the content and writing of the paper.

Additional information

Funding

The authors received no financial support for this editorial.