Article Text

Download PDFPDF

Intravitreal methotrexate for treating vitreoretinal lymphoma: 10 years of experience
  1. S Frenkel1,
  2. K Hendler1,
  3. T Siegal2,
  4. E Shalom2,
  5. J Pe’er1
  1. 1
    Department of Ophthalmology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel
  2. 2
    Neuro-oncology Center, Hadassah-Hebrew University Medical Center, Jerusalem, Israel
  1. J Pe’er, Department of Ophthalmology, Hadassah University Hospital, PO Box 12000, Jerusalem 91120, Israel; peer{at}


Aim: To describe our experience in treating vitreoretinal involvement of primary central nervous system lymphoma, by intravitreal injections of methotrexate (MTX).

Methods: Patients with suspected intraocular lymphoma underwent a diagnostic vitrectomy. Samples were sent for cytology, genetic evaluation and for interleukin level measurements. Treatment protocol included injection of 400 μg/0.1 ml MTX intravitreally twice weekly for 4 weeks, once weekly for 8 weeks, and then once monthly for 9 months, for a total of 25 injections. Data were collected from the patients’ records and included, inter alia, response to intravitreal MTX measured by time to disappearance of vitreal cells and retinal infiltrates, changes in visual acuity, and clinical recurrence rate.

Results: In the past 10 years we have treated 44 eyes of 26 patients; seven patients had monocular involvement, and 19 binocular. Six patients were initially diagnosed as having a non-responsive uveitis, and 16 with either CNS or systemic lymphoma with later involvement of the eye. Four patients had systemic lymphoma; one of them was found to have CNS lymphoma after the ocular involvement. Three patients had T cell lymphoma, and the rest had B cell lymphoma. Clinical remission was reached after 6.4 (3.4) (2–16) injections of MTX (mean (SD) (range)), with 95% of the eyes needing 13 injections or less to be cleared of malignant cells. None of the patients had an intraocular recurrence. Among the side effects, the most common was corneal epitheliopathy, which usually appeared after the third injection and began to subside when the intervals between injections increased.

Conclusions: Vitreoretinal involvement of lymphoma can be controlled effectively and without serious adverse reactions by intravitreal MTX injections. The treatment protocol described herein has resulted in no intraocular recurrence so far and has had bearable side effects. The accumulating clinical results bring us to propose the consideration of this protocol as a good first-line treatment option for intraocular lymphoma.

Statistics from

Primary central nervous system lymphoma (PCNSL) is usually a B cell lymphoma that arises within and is generally confined to the central nervous system. The incidence of PCNSL is 0.27 per 100 000 person-years in the immunocompetent population and 4.7 per 1000 person-years among patients with AIDS.14 Vitreoretinal involvement, generally referred to as intraocular lymphoma (IOL), occurs in approximately one-quarter of patients with PCNSL.5 The hallmark of intraocular involvement of PCNSL is the presence of vitreal cells, mainly in clumps, which is often misdiagnosed as a non-responsive uveitis, and subretinal infiltrates.6

The diagnosis of IOL is based on integrating results of several tests. Clinically, vitreal cells mainly in clumps, and yellow-white subretinal infiltrates can be seen and may reduce the visual acuity. Based on the clinical examination, patients who are suspected of IOL undergo a diagnostic vitrectomy or vitreous tap. Vitreal samples can be analysed for the presence of malignant cells by cytopathology with possible immunohistochemical staining. Other diagnostic methods are flow cytometry, molecular analysis for immunoglobulin heavy chain (IgH) rearrangement and monoclonality by polymerase chain reaction (PCR), and enzyme-linked immunosorbent assay (ELISA) test for ratios between interleukins (IL10/IL6).6 7 In the lack of diagnostic vitreal cells, diagnosis can be made with a chorioretinal biopsy of subretinal infiltrates.8

Orbital irradiation, sometimes in combination with systemic chemotherapy, has been used in many centres for treatment of IOL and induced an ocular remission.9 10 However, the intraocular lymphoma recurred in 5/9 patients reported by Hoffman9 and in 1/15 patients reported by Isobe.10 Another study showed no relapse of IOL after treatment with radiotherapy.11 Systemic intravenous high-dose methotrexate (MTX) may lead to relapse after an initial response.12 Radiation to the eye can have a significant ocular toxicity including cataract formation, radiation retinopathy, optic neuropathy, dry eyes and prolonged corneal epithelial defects.9 In recent years, a protocol for treating vitreoretinal involvement of PCNSL with intravitreal injections of MTX was developed at the Oregon Health & Science University and modified at the Hadassah-Hebrew University Medical Center.13 14 This treatment was found to be effective in inducing clinical remission of vitreoretinal involvement in PCNSL with acceptable morbidity.14 15 We describe here our 10 years of experience using this treatment modality.


All patients with biopsy-proven vitreoretinal lymphoma, who had been treated by intravitreal injections of MTX between March 1997 and March 2007 at the Ocular Oncology clinic of the Hadassah-Hebrew University Medical Center, were included in the study. The intravitreal injections were approved for individual patients by an institutional review board. The protocol for intravitreal chemotherapy was described previously.14 A dose of 400 μg/0.1 ml of MTX was injected at the level of the pars plana using a 30-gauge needle, under topical anaesthesia with topical benoxinate HCl 0.4% (Localin, Fischer Laboratories, Israel) drops, and recently with Amethocaine HCl 0.5% drops. Eight twice-weekly intravitreal methotrexate injections constituted the induction phase, followed by eight weekly consolidation injections. Subsequently, a maintenance phase involved nine monthly methotrexate injections to complete 1 year of treatment (25 injections). Intravitreal chemotherapy was stopped if patients were systemically unwell due to the extraocular disease or suffered from severe ocular complications.

At each visit, patient examinations included measurement of visual acuity (VA) and intraocular pressure (IOP), slit-lamp biomicroscopy to view individual cells within the vitreous cavity, and dilated funduscopy to see malignant cellular infiltrates involving the retina and/or optic nerve head. Lately, a gonioscopic examination was added to the monitoring protocol to identify iris and anterior chamber angle neovascularisation. Increases in IOP immediately after the injection (up to 70 mm Hg in some cases) were treated by digital massage. With this treatment, IOP returned to normal values within 30 min after the injection in all cases.

We collected the following information from the patients’ medical records: demographic details; duration of symptoms prior to diagnosis; diagnosis by biopsy; initial site of tumour; status of the extraocular disease; laterality of ocular involvement; total number of intravitreal injections; clinical response to intravitreal chemotherapy, including indicators of remission (absence of cells from the vitreous and absence of retinal and optic nerve head infiltrates); number of injections to clinical remission; complications during the study period and their outcome; length of follow-up; other interventions, including vitrectomy, systemic and intrathecal chemotherapy, and radiotherapy; length of survival from diagnosis; and cause of death if applicable. Visual acuity before treatment, at the end of the treatment course, if applicable, and at the last known clinic visit were also recorded, and the reason for significant loss of visual acuity (decrease in Snellen visual acuity by two or more lines) was determined. Early Treatment Diabetic Retinopathy Study (ETDRS) VA charts were not available to measure the VA of the patients in this study. Because of the non-linear change in the Snellen VA chart, average change in visual acuity cannot be calculated using Snellen VA measurements. The Snellen VA measurement is merely a representation of the visual angle. This can be presented in different ways. Thus, the decimal Snellen VA measurements were translated into the log of the minimum angle of resolution (LogMAR) units. Once converted to LogMAR, calculations become statistically valid. Changes in the IOP during the treatment period were noted.

Statistical analysis was done with JMP statistical software (SAS Institute, Cary, NC).


Forty-four eyes of 26 consecutive patients with vitreoretinal lymphoma were treated by intravitreal injections of MTX at the Ocular Oncology clinic of the Hadassah-Hebrew University Medical Center from March 1997 to March 2007. Eight of the patients were male. The average age at diagnosis was 56.9 (17.4) (mean (SD)). Seven patients had a monocular disease. Of the 19 patients who had binocular disease, three presented with monocular disease with involvement of the other eye weeks to months later.


Twenty-four patients were diagnosed by a diagnostic vitrectomy; 23 patients had positive cytology and/or IgH PCR results, and one patient had suspicious-looking cells, but PCR was unavailable at the time. One patient underwent a vitreal tap on both of her eyes, and suspected lymphoma cells were found. One patient underwent an iris biopsy and was diagnosed as having iris T cell lymphoma. In addition, he also had dense vitreal cells. This case was previously published separately as a case report.16 Recently we also analysed the vitrectomy samples for IL-6 and IL-10 levels, but data are available for too few patients to draw conclusions at this stage.

Tumour type

Twenty-three patients had intraocular lymphoma of the B cell type, and three patients had T cells. In six patients, the ocular involvement either preceded the CNS or was a presenting symptom. The mean interval between ocular and CNS involvement was 14 months (range 1–54 months). In 13 patients, the CNS involvement preceded the ocular involvement, with a mean interval of 36 months (range 2.5–132 months). Four patients had ocular involvement 8 months to 14 years after remission from systemic lymphoma. One of them developed CNSL 2 months after initiation of treatment of the ocular involvement. Two of these patients died of recurrence of their systemic disease without developing a CNS disease, and the fourth is still under follow-up 19 months since her ocular diagnosis without showing CNS involvement. One patient is being treated for ocular involvement for 9 months without any evidence of CNS involvement.

Response to treatment

Seventeen (39%) eyes completed the treatment protocol of 25 injections and were cleared clinically of malignant cells after a median of five methotrexate injections (range 2–11). Twenty-three additional eyes (52%) did not complete the entire treatment protocol but were cleared clinically of malignant cells after a maximum of 16 methotrexate injections (range 3–16). Data regarding the completion of treatment of the remaining four eyes (two patients) are missing (patients who disappeared from follow-up and/or died). Patients did not complete the treatment protocol either because they were too sick or dead due to their CNS involvement, or the systemic involvement (1), or because they developed neovascular glaucoma (two patients, see below). Remission was reached after 6.4 (3.4) (2–16) injections (mean (SD) (range)), with 95% of the eyes needing 13 injections or less to be cleared of malignant cells and retinal infiltrates. A cumulative incidence graph, showing the number of injections to clinical remission is presented in fig 1. A representative response to treatment is shown in figs 2 and 3. Patients who had IOL in both of their eyes showed no difference in the response to the intravitreal MTX injections between the eye that underwent a diagnostic vitrectomy and the other eye which did not undergo surgical intervention (matched pairs t test showed a correlation of 0.95 and p<t of 0.7653). The length of follow-up from commencement of the methotrexate injections was 3–120 months (median, 21 months). Time from remission to last known follow-up or death was 16.2 (6.9) (0–92) months (mean (SE) (range)). Fourteen patients died of their CNS or systemic lymphoma within a median of 17 months (range 3–84 months) from the time of their ocular diagnosis. Twelve patients have survived for median 24 months (range 4–120 months) from the time of their ocular diagnosis. Eight patients have survived for over 3 years after the last injection with a median of 63 months (range 41–107 months). None of them has had an ocular recurrence.

Figure 1 Rate of injections needed for remission presented in a cumulative incidence graph, showing the percentage of eyes in remission after each injection in relation to the number of injections given until clinical remission was achieved.
Figure 2 Fundus colour images of the RE of a patient diagnosed as having intraocular lymphoma 8 months after the diagnosis of systemic lymphoma and 4 months after complete remission from the systemic lymphoma. (A) Image taken before initiation of treatment with methotrexate (MTX). Note the haziness caused by the dense vitreal cells and the thick retinal infiltrates. (B) Image taken after the 10th MTX injection (in the 6th week of treatment). Note the marked clearing of the vitreal cavity and of the retinal infiltrate.
Figure 3 Fundus colour images of the RE of a patient diagnosed as having intraocular lymphoma 8 months after the diagnosis of primary CNS lymphoma (PCNSL). (A) Image taken before initiation of treatment with methotrexate (MTX). Note the massive subretinal lymphoma infiltrates. (B) Image taken at the end of treatment (after the 25th MTX injection). Note the complete clearing of the lymphoma.

Effect of treatment on visual acuity

The entire dataset of visual acuity measurements is presented in table 1. Eleven eyes of nine patients improved their initial VA from mean initial VA LogMAR 0.98 (equivalent to 0.1) to mean final VA LogMAR 0.39 (equivalent to 0.4) (fig 4). Eyes with poor initial VA (⩽CF 1.5 m) did not change much, or improved from a mean initial VA LogMAR 2.81 (equivalent to CF 15 cm) to a mean final VA LogMAR 2.47 (equivalent to CF 40 cm). On the other hand, eyes with a good initial VA (0.8–1.0) did not change much, or deteriorated slightly from the mean initial VA LogMAR 0.06 (equivalent to 0.9–) to a mean final VA LogMAR 0.13 (equivalent to 0.7+) (table 2). One eye had no light perception (NLP), due to optic nerve and brain infarcts as part of the CNS disease.

Figure 4 Visual acuity (LogMAR) of 11 eyes of nine patients improved between the first and the last follow-up visit. The decimal equivalent mean VA of these eyes improved from 0.1 to 0.4. LogMAR, log of the minimum angle of resolution.
Table 1 Initial and last known visual acuity measurements
Table 2 Change in mean visual acuity (VA) from the time of initiation of treatment to the last follow-up visit available

Some of the eyes had an initial improvement of visual acuity as seen in the example of the patient in fig 2 with the clearing of vitreal haze and the thick peri-papillary retinal infiltrate. A later deterioration in the VA due to moderate keratopathy usually improved with the clearing of the keratopathy.

Timing is of the essence in this treatment as demonstrated by the patient presented in fig 3, who developed “uveitis” in her RE 8 months after diagnosis of CNS lymphoma. Three months later, her VA deteriorated to LogMAR 3.2 (equivalent to hand motion), after which she was diagnosed as having IOL, and treatment was started. Four weeks later, vitreoretinal involvement appeared in her LE with a reduction in VA from LogMAR 0.15 to 0.7 (equivalent to a reduction from 0.7 to 0.2), and treatment was initiated immediately. On her last visit, for the last MTX injection, the VA in her RE remains LogMAR 3.2 (equivalent to hand motion) despite the complete clearing of the lymphoma from this eye as presented in fig 3, whereas the VA in her LE improved to LogMAR 0.1 (equivalent to 0.8).


All of our patients developed conjunctival hyperaemia and some form of keratopathy. Keratopathy, which ranged from diffuse punctate keratopathy to severe epitheliopathy, appeared after the third injection, and usually subsided completely when the patients were treated with injections once a month. One patient developed band keratopathy, which required surgical intervention. Another complication that occurred in some patients was acceleration of existing cataract. Two of our patients developed iris and anterior chamber angle neovascularisation (NVI), with subsequent neovascular glaucoma (NVG) that needed filtration valve surgery, as reported recently.15 Both patients had a history of diabetes, but neither had retinal neovascularisation. NVI and a rise in IOP were noted at the 17th injection in the first patient, and after the 18th injection in the second patient. Intraocular injections of MTX were discontinued. No recurrence of the lymphoma was noted in the follow-up of these two patients (3 years for the first patient, and 8 months for the second). As mentioned above, one patient developed optic atrophy due to multiple brain infarcts—a complication of his CNS disease. It does not seem to be a true complication of the vitreal MTX injections. Severe inflammatory reaction that responded rapidly to topical steroids appeared in two of our patients. In one of them, the reaction appeared after eight injections, and was described as sterile endophthalmitis. The injections were postponed and continued after the resolution of the inflammatory reaction. The other patient was seen in another clinic after his 20th injection with what was described as an anterior chamber reaction with a chamber filled with fibrin, resembling toxic anterior segment syndrome (TASS). The patient refused further injections and died 5 months later of his systemic disease, still free of vitreal cells.


In this study, we describe the clinical characteristics of patients with vitreoretinal involvement of primary central nervous system lymphoma, and the long-term follow-up of their response to a treatment protocol of intraocular injections of methotrexate. This includes the numbers of injections required for remission, effects on visual acuity and resultant complications.

Twenty-six patients with 44 study eyes is the largest study group so far in the published literature. The rarity of the entity described and the lethal nature of the CNS involvement hinder the ability to follow a large group of patients in one centre and to draw general conclusions. However, we hope that the long-term follow-up of the surviving patients may be helpful for other clinicians in choosing a treatment protocol.

In this study, patients developed vitreoretinal involvement of primary central nervous system lymphoma both after CNSL and after systemic lymphoma. Thus, it seems prudent to suspect this diagnosis in any patient with a history of either CNSL or systemic lymphoma with vitreal cells or retinal infiltrates, disregarding the time from the diagnosis of the extraocular disease to the appearance of the ocular signs. Clinical examination is not sensitive enough to distinguish whether the vitreal cells are malignant cells, benign lymphocytes or resident hyalocytes. In addition, individual cells cannot be detected clinically within the retina. Therefore, an early diagnostic vitrectomy with immediate cytological evaluation before the cells disintegrate can ascertain the clinical diagnosis.17 18 Later, PCR for IgH rearrangement and monoclonality can finalise the diagnosis.19

One must bear in mind though, that PCR’s sensitivity in the diagnosis of IOL is only 64% (although it has 100% specificity).20 Thus, the use of cytokines in distinguishing IOL from uveitis has been investigated by several groups. Interleukin 10 (IL-10) is produced by malignant B lymphocytes in intraocular and CNS lymphoma, while IL-6 is produced in high levels by inflammatory cells in uveitis.21 Whitcup et al found an IL-10 to IL-6 ratio of greater than 1.0 in vitreal samples from patients with PCNSL with IOL.22 We recently started to collect data in order to evaluate whether this ratio can improve the diagnostic efficacy. The preliminary results are still inconclusive.

While it is known that the majority of the PCNSLs and the IOLs are of B cell origin, the biopsy in three of our patients yielded T cells. Thus, caution should be taken in analysing the vitreal material, in order not to miss the diagnosis.

The appearance of vitreoretinal lymphoma can be a presenting sign or may precede, even by years, the CNS disease (Invest Ophthalmol Vis Sci 2006;47:E-Abstract 2830). Moreover, one of our patients who was in remission from systemic lymphoma developed CNS lesions after developing vitreoretinal lymphoma. We therefore consider the vitreoretinal lymphoma as part of the CNS disease, even in patients who were previously diagnosed as having systemic lymphoma. The appearance of the disease in one of these sites after the other should be considered not as a recurrence but as part of the same multifocal disease. In addition, the treatment in one site cannot affect the other. Even the newer form of chemotherapy for PCNSL, intrathecal MTX alone, was unable to reach detectable levels in the eye,23 let alone expecting the low intravitreal doses to be therapeutic in CNS.

Four of our patients developed vitreoretinal lymphoma after systemic lymphoma. One of them developed CNS lesions 2 months after the ocular diagnosis. Three of them did not develop CNS involvement, and two of them have died of a flare-up of their systemic disease. We cannot know if the two patients who died would have developed CNS lesions, given enough time. We are still following the fourth patient.

There is a clear difference in the final visual acuity of the patient presented in fig 3 between the RE in which treatment was initiated 3 months after the initial signs and the LE which received treatment immediately after the initial involvement. Since cytology can be evaluated within a short time of the diagnostic procedure, it seems beneficial for the patient’s final visual acuity to initiate treatment based on cytology results when possible, with later additional diagnosis by PCR and interleukin levels.

All eyes for which we have complete information (91%) were clinically clear of lymphomatous involvement. Except for one patient who needed 13 injections and one who needed 16 injections to be clear of vitreal cells, all patients were clear at the end of the second month of the treatment. Of note is the fact that in patients who had IOL in both of their eyes, there was no difference in the response to the intravitreal MTX injections between the eye that underwent a diagnostic vitrectomy and the other eye which did not. One may thus seek to reduce the number of injections and finish the treatment protocol after only 2 months. However, it is not certain that there will not be a recurrence if the treatment is not maintained after the consolidation period. Answering this question may require a large international collaboration to reach a reasonable number of patients.

Treatment with intravitreal MTX injections resulted in few complications. Some form of keratopathy developed in all of our patients. All but one improved as the frequency of the injections decreased to monthly injections. At that time point, they had mild dryness signs that resolved within a few days with lubricants. Therefore, keratopathy can be considered a relatively mild and temporary side effect. It might be appropriate to try topical folinic acid 0.003% in the infrequent cases of severe keratopathy, but this treatment was not available for our patients. However, it is our impression that vitamin A-containing gel can accelerate the healing of the keratopathy.

Cataract formation or acceleration is attributed to MTX toxicity, and no special type of cataract was noted in our patients. Neovascularisation of the iris and anterior chamber angle developed in two patients. Although we cannot be certain that the neovascularisation is related to the intraocular MTX injections, we now follow all patients with gonioscopy and recommend cessation of the MTX injections should neovascularisation appear. The patient who developed optic atrophy had multiple brain infarcts, and it seems that the optic nerve lesion was part of the CNS disease. Sterile endophthalmitis appeared in one patient, and severe anterior segment sterile inflammation resembling TASS appeared in one patient; both responded to corticosteroid treatment. TASS is associated with cataract or other anterior segment surgery and typically appears within 12–48 h after the procedure. The reaction in the anterior segment of our patient resembled the reaction in TASS, despite the lack of the defining criteria of anterior segment intervention. It could be attributed to a toxic effect of the intravitreal MTX, but should not be a cause to stop treatment as the reaction responds to topical steroids within days. Maculopathy has been previously described as one of the side effects found in patients treated with intravitreal MTX for IOL.14 However, we have recently found maculopathy to be associated with the blood–brain barrier disruption (BBBD) treatment received by patients with CNS involvement of PCNSL, either with or without IOL. Fifteen of 23 (65.2%) patients who received BBBD developed maculopathy, but none of the patients who did not receive BBBD, whether they received intravitreal MTX or not, developed maculopathy.24 Similar findings were reported by Galor and colleagues.25 According to our findings, this type of pigmentary maculopathy does not affect the visual acuity permanently. Overall, there is a low rate of severe complications.

Berenbom and colleagues recently reported on their experience in treating seven IOL patients with radiotherapy.11 Their patients received a combination of radiotherapy and chemotherapy, chemotherapy alone or radiotherapy alone. None of their patients received intravitreal MTX injections per our protocol or any alternative protocol. Berenbom proposed that the lower radiation doses (35–40 Gy given in 15 fractions to both globes), the improved dosimetric techniques and the customised blocks made following computed tomographic simulation minimised normal tissue toxicity and yielded a lower complication rate than previously reported by several groups. Still, two of these seven patients (28.6%) suffered from irradiation retinopathy, and four (57.1%) developed cataract. Long-term studies using newer radiotherapy techniques are needed to prove no recurrences and few side effects. On the other hand, there is a markedly expanding experience in repetitive intraocular injections in treating age-related macula degeneration (ARMD) with minimal side effects related to the injection procedure itself. This body of evidence along with the reasonable complications reported here should alleviate the concerns raised regarding intravitreal chemotherapy with MTX.

Rituximab, a monoclonal antibody against the CD20 antigen of B cells’ cell membrane, has been studied as an adjunctive therapy for IOL.26 27 Only initial data are available, and only in conjunction with methotrexate. Hence, it might be used in the future to either reduce the MTX concentration or reduce the number of injections needed to prevent recurrence.

Treatment of vitreoretinal lymphoma by intravitreal injections of methotrexate is now used by a growing number of groups with encouraging results.14 28 29 Long-term studies of the newer radiotherapy treatments are not yet available, and alternative treatments such as Rituximab are not yet mature enough to be proposed as a first line of treatment. However, the protocol presented herein provides complete remission of vitreoretinal lymphoma with, so far, no recurrences and with bearable side effects. Thus, the accumulating clinical data bring us to propose the consideration of this protocol as a treatment option for vitreoretinal lymphoma.



  • Presented in part at the ARVO annual meeting, May 2007.

  • Competing interests: None.

  • Patient consent: Informed consent was obtained for every patient we treated.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • At a glance
    Harminder S Dua Arun D Singh, Editors-in-Chief