Pemetrexed in the treatment of relapsed/refractory primary central nervous system lymphoma


  • Author responsibilities were as follows: M.S., drafting the article, acquisition and analysis of data; J.P.C. drafting the article, acquisition and analysis of data; C.C.G., drafting the article, acquisition and analysis of data; A.R., study design, analysis of data, statistics; C.T., drafting the article, acquisition and analysis of data; S.A.G., drafting the article, acquisition and analysis of data; J.J.R., drafting the article, study design, acquisition and analysis of data; A.M.E., drafting the article, analysis of data; L.R., drafting the article, acquisition and analysis of data.



Despite initial treatment with high-dose methotrexate-based regimens, many patients with primary central nervous system lymphoma (PCNSL) relapse and die from their disease. No standard of care exists at progression or relapse, but chemotherapy and in some cases radiation are usually used. Pemetrexed is a multitargeted antifolate, similar to methotrexate, but with a broader spectrum of activity. Because methotrexate is an integral part of PCSNL treatment, the authors assessed the antitumor activity and safety of pemetrexed in recurrent PCNSL.


Patients with relapsed/refractory PCNSL were enrolled in this trial. Treatment consisted of pemetrexed 900 mg/m2 given every 3 weeks with low-dose dexamethasone, folate, and B12 supplementation. Each cycle was 6 weeks, and follow-up imaging was done before each new cycle. Treatment was continued until complete remission, progression, or toxicity.


Eleven patients were treated, with a median age of 69.8 years and Karnofsky performance status of 70%; 10 of 11 patients had failed prior high-dose methotrexate. The median number of pemetrexed cycles given was 5, with an associated overall response rate of 55% and disease control rate of 91%. The 6-month progression-free survival (PFS) was 45%, median PFS was 5.7 months, and median overall survival was 10.1 months. Toxicities were primarily hematologic and infectious.


Pemetrexed has single-agent activity in relapsed/refractory PCNSL. Toxicities were seen likely because of the higher than standard dose used. Further investigation of this agent or other multitargeted antifolates in PCNSL is warranted to determine optimal dose and efficacy in a more homogeneous population. Cancer 2012. © 2011 American Cancer Society.


Primary central nervous system lymphoma (PCNSL) is an aggressive tumor limited to the central nervous system (CNS). Untreated, patients survive only a few months; however, in the last 2 decades, it has become apparent that high-dose (HD) methotrexate-based regimens can lead to median survivals of 30 to 50 months.1-6 Despite aggressive therapy, however, nearly 50% of patients will relapse within 24 months of diagnosis. Furthermore, the expectant survival after relapse of PCNSL is between 2 months and 14 months, with most series reporting about 4 months to 8 months, depending on whether further treatment is instituted.3, 7-9 There is no standard salvage treatment for relapsed or refractory patients, although numerous approaches have been tried using radiation therapy (RT) or chemotherapy with variable degrees of activity. Unfortunately, the majority of patients with relapsed PCNSL still die from their disease.10-14

Given the poor outcomes for these patients, new and novel agents should be evaluated. Pemetrexed (Alimta; Lilly Oncology, Indianapolis, Ind) has antifolate activity, similar to methotrexate, but with the advantage of targeting >1 site in folate metabolism. It interrupts purine synthesis via thymidylate synthase and dihydrofolate reductase inhibition, and pyrimidine synthesis via glycinamide ribonucleotide formyltransferase and aminoimidazole carboxamide formyltransferase inhibition.15 We performed a trial using pemetrexed in patients with relapsed/refractory PCNSL to determine single-agent activity; this was part of a larger single-institution trial assessing pemetrexed activity in recurrent gliomas, PCSNL, and brain metastases (NCT00276783). Our objective was to assess for single-agent activity based on response rates, progression-free survival (PFS) at 6 months (PFS-6), median PFS, overall survival (OS), and toxicity in 10 to 12 patients.


All patients signed an institutional review board-approved consent form before study enrollment and treatment. Patients were required to have a diagnosis of recurrent or refractory PCNSL, have a Karnofsky performance status (KPS) of ≥60, and be >18 years of age. Measurable disease was required to assess response. Patients had to have failed at least 1 prior chemotherapy regimen and were required to be >4 weeks from prior RT if this modality was used. Patients had to have blood tests performed within 14 days of registration and/or before starting treatment. The following criteria were used for inclusion: white blood cell count ≥3000/μL, absolute neutrophil count (ANC) ≥1500/mm3, platelet count ≥100,000/mm3, hemoglobin ≥10 mg/dL, aspartate aminotransferase/alanine aminotransferase <3.0× upper limit of normal (ULN), bilirubin <1.5× ULN, and creatinine <1.5 mg/dL. Patients were permitted on study if effusions or fluid collections were drained before treatment. Effective contraception was mandated while patients were on study, and pregnancy or breast-feeding was exclusionary. Patients could not be on nonsteroidal anti-inflammatory drugs and had to be able to take steroids, B12, and folic acid. Patients could not have significant medical illnesses or infection that could not be adequately controlled with appropriate therapy or would compromise the patient's ability to tolerate this treatment. Prior malignancies were okay if in complete remission and the patient had been off all therapy for a minimum of 3 years, except nonmelanoma skin cancer or carcinoma in situ of the cervix.

Patients were treated with pemetrexed at a dose of 900 mg/m2 intravenously over 10 minutes every 3 weeks; 1 cycle was 6 weeks. This dose was arbitrarily chosen by using the maximum tolerated dose from a single-agent phase 1 trial, which was 1200 g/m2, 16 and the theory that if pemetrexed was like methotrexate in its ability to penetrate the CNS, then higher than standard doses (500 mg/m2) may be needed. In addition, preclinical data suggested limited penetration into the CNS,17 so we wanted to try and optimize CNS penetration.

Patients were required to take 1 mg folic acid daily and vitamin B12 (1000 μg as an intramuscular injection every 9 weeks) starting 5 days before the first dose of pemetrexed and continuing until 21 days after the last dose. Patients not on dexamethasone received 4 mg orally twice a day the day before, the day of, and the day after pemetrexed treatment. The use of B12, folic acid, and dexamethasone was per package insert guidelines and to reduce toxicities related to pemetrexed. Patients remained on treatment as long as there was no evidence of tumor progression or intolerable side effects and a complete response (CR) was not achieved. Patients who had a CR could receive up to 2 more additional doses. The trial was modified to allow pemetrexed failures who initially responded to be re-treated.

Within 14 days before beginning treatment all patients were required to have a complete history and neurologic examination, electrocardiography, standard eligibility laboratory data and serum pregnancy test for women of childbearing potential, and contrast enhanced magnetic resonance imaging (MRI) of the brain and spinal cord and to be on a stable dose of steroids for 5 days if they were on steroids. While on therapy, a neurologic examination and contrast enhanced MRI of the affected area were performed approximately every 6 weeks. Blood counts were performed weekly, and electrolytes with blood urea nitrogen/creatinine were done within 48 hours of each pemetrexed dose.

Treatment could be delayed for up to 21 days for grade 3 or 4 toxicities, and if there was no recovery the patient was removed from the trial. If treatment was resumed, the dose was lowered by 25% for a platelet count of ≥50,000/mm3 but ANC <500/mm3 and by 50% if platelets were <50,000/mm3 with an ANC nadir of <100/mm3. For diarrhea requiring hospitalization (or of at least grade 3) and other grade ≥3 nonhematologic effects (except transaminase elevations, nausea, or vomiting), treatment was delayed until grade was ≤1. For grade 3 or 4 nonhematologic toxicities, the dose was reduced by 50%.

Statistical Methods

The activity of pemetrexed in the patients with recurrent PCSNL was done in a small cohort of patients (goal 10-12) to assess for antitumor activity. Response assessment was per the Macdonald Criteria18 as CR, partial response (PR), stable disease (SD), or progressive disease (PD). Because not all patients had brain, eye, and cerebrospinal fluid (CSF) evaluation at relapse, the International Primary CNS Lymphoma Collaborative Group response criteria were not used. PFS-6, median PFS, and OS were analyzed using the Kaplan-Meier product limit curve.


Eleven patients (8 men and 3 women) with a median age of 69.8 years (range, 40-82 years) were enrolled between April 13, 2006 and February 17, 2009; median KPS was 70% (range, 60%-100%) (Table 1). All patients had diffuse large B-cell lymphoma except 1 each who had low-grade histology and primary CNS T-cell lymphoma; all patients had only brain disease at recurrence with prior therapies listed in Table 1. All patients had complete neuraxis imaging; lumbar puncture; computed tomography of the chest, abdomen, and pelvis; slit lamp examination; bone marrow biopsy; and human immunodeficiency test at initial presentation confirming the diagnosis of PCSNL. At recurrence, 8 patients had brain only imaging, and 3 had brain, spine, and CSF evaluation. Number of prior treatments was 1 in 6 patients, 2 in 2 patients, 3 in 2 patients, and 5 in 1 patient; 2 patients had a prior stem cell transplant. Median time from initial PCNSL diagnosis to pemetrexed treatment was 31 months (range, 4-96 months). The median number of treatment cycles was 5.1-7 Median follow-up of the 11 patients was 10.1 months (range, 1.5-46.2 months). Before starting pemetrexed, 3 patients were on steroids, and 2 were could be weaned from them; the remainder of patients only received steroids as part of treatment. Response was as follows: 4 CR, 2 PR, 4 SD, and 1 PD for an overall response rate (CR + PR) of 55% (CR rate of 36%) and a disease control rate (CR + PR + SD) of 91%. One patient was re-treated twice for recurrences, attaining a CR and a PR before stopping treatment because of overall clinical decline.

Table 1. Patient Clinical and Treatment Data
Age, y/SexPrior TreatmentDexamethasone Dose Before PemetrexedKPSPathologyCycles, No.ResponseReason Off StudyTreatment PostpemetrexedCause of Death
  1. Abbreviations: Ara-C, cytarabine; CR, complete remission; DLBCL, diffuse large B-cell lymphoma; F, female; HD, high-dose; I/E, ifosfamide/etoposide; IT, intrathecal; IV, intravenous; KPS, Karnofsky performance status; M, male; MPV, methotrexate, procarbazine, vincristine; MTX, methotrexate; R, rituximab; PD, progressive disease; PR, partial remission; RS, radiosurgery; RT, radiation therapy; Auto SCT, autologous stem cell transplantation; SD, stable disease; T, thiotepa; TMZ, temozolomide; V, vincristine; WBRT, whole brain radiotherapy.

53/F1) MPV × 5, WBRT + Eyes, Ara-C; 2) R; 3) Auto SCT6 mg twice daily60DLBCL3PRPDNoneDisease
70/M1) MPV, Ara-C4 mg twice daily80DLBCL6 for 2nd relapse; 6 for 3rd relapse; 2 for 4th relapseCRCompleted treatmentNoneDisease
80/M1) MPV ×1, Ara-CNone70DLBCL2SDToxicity/PDNonePneumonia
73/M1) MPV, Ara-CNone70DLBCL7PRDeathNoneUnknown
57/F1) MTX IV/IT, P, V followed by RT and Ara-C; 2) MTXNone60DLBCL7CRCompleted treatmentLost to Follow-upUnknown
70/M1) MPV ×5, Ara-CNone100DLBCL3SDDeath from infectionNonePneumonia
49/M1) R-MPV ×5, Ara-c ×2None100Low-grade B-cell lymphoma6SDPatient stopped treatment1) Ara-C; 2) RAlive
40/M1) MPV ×5, Ara-c ×2None90DLBCL7CRCompleted treatment1) HD-MTX/R; 2) thiotepaDisease and PE/infection
76/F1) MPV/RT; 2) intraocular MTX and V4 mg daily100DLBCL5CRToxicity: rashTMZ/RDisease
50/M1) RT; 2) MPV, Ara-C; 3) I/E; 4) T/TMZ; 5) MTX; 6) Auto SCTNone60T-cell lymphoma3SDWithdrawal/ clinical declineRSDisease
82/M1) MPV ×2, Ara-c ×7; 2) R-MPV; 3) TMZ + R ×4None60DLBCL1PDPDNoneDisease

The PFS-6 and PFS-12 were 45% (95% confidence interval [CI], 16%-74%) and 27% (95% CI, 20%-70%), respectively, with a median PFS of 5.7 months (95% CI, 1.6-21.4 months; Fig. 1). The median OS was 10.1 months (95% CI, 2.4-33.3 months), with an OS-12 of 45% (95% CI, 16%-74%; Fig. 2). Reason for trial discontinuation was completion of treatment in 3 patients (1 treated with 6 and 2 with 7 cycles); PD in 2 patients; toxicity in 4 patients—2 with pneumonia from pneumocystis jirovecii pneumonia only 1 of whom was on steroids (neither was on prophylaxis; both had PD on imaging), 1 with cellulitis, and 1 with clinical decline in the setting of grade 2 thrombocytopenia; and death in 1 of unknown cause. One patient with SD withdrew from the trial. Only the patient with a low-grade central nervous system lymphoma remains alive; the other died of PCNSL (5 patients), pneumonia and associated pulmonary complications (2 patients), infection/gastrointestinal perforation (1 patient), and unknown cause (2 patients lost to follow-up).

Figure 1.

Progression-free survival of patients treated with pemetrexed is shown.

Figure 2.

Overall survival of patients treated with pemetrexed is shown.

At progression, 6 patients had no further treatment, 1 patient had radiosurgery to a cerebellar lesion, 3 patients received further salvage chemotherapy, and 1 patient was lost to follow-up.

Toxicities seen while on trial are listed in Table 2. The most common adverse events were hematologic, and the most severe where pneumonia, combined in 1 patient with neutropenia. Only 1 patient required a dose reduction for hematologic toxicity, and 1 patient was unable to restart because of a grade 3 thrombocytopenia that did not resolve before the patient was removed from the trial because of clinical deterioration.

Table 2. Toxicities
  1. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase.

ALT/AST 21  
Cellulitis 1   
Infection 12 2
Leukopenia 231 
Neuropathy 1   
Skin rash 1   
Thrombocytopenia 232 
Weight loss 1   


This is the first report of using pemetrexed in PCNSL. Our objective was to assess the efficacy of pemetrexed in relapsed/refractory PCNSL patients, as there remains a significant need for effective salvage therapies in this patient population. Pemetrexed's similarity to methotrexate but ability to target >1 aspect of folate metabolism led us to hypothesize that this agent might also be active in PCNSL. Clinical activity was seen with an overall response rate of 55% and a disease control rate of 91%. The interval of disease controlled was variable but durable in some cases. Importantly, responses occurred in patients who had failed prior HD methotrexate, suggesting that methotrexate exposure does not preclude a response from pemetrexed. In addition, responses were seen in patients who failed multiple prior treatments, including prior autologous stem cell transplantation. All patients received dexamethasone during their treatment, which we acknowledge may have contributed to the response rates, but this was needed to minimize systemic toxicities.

Our PFS-6 was 45%, median PFS was 5.7 months, and median OS was 10.1 months; this is comparable to other chemotherapeutic agents used in the relapsed/refractory PCNSL setting.11-14, 19-21 Plotkin et al14 reported the outcomes of 22 patients rechallenged with HD methotrexate. The response rate was 91%, with a median time to second relapse of 25.8 months and median survival of 61.9 months; many went on to receive RT at progression. A trial using Procarbazine, CCNU, Vincristine (PCV) in 7 patients at relapse reported an 86% response rate and a median OS of 16 months.22 Two studies have used topotecan to treat recurrent PCNSL.11, 19 A total of 42 patients were treated on both trials, with response rates of 33% and 40%. The median PFS was 2 months in each trial, with a PFS-6 of 52% in the Fischer trial.19 The median OS was 8.4 months and approximately 33 months, with most patients salvaged with WBRT in the Voloschin trial.11 Reni at al13 treated 36 patients with temozolomide; a 31% response rate was seen, with a median PFS of 2.8 months, median OS of 3.9 months, and 1-year OS of 31%. Wong et al21 treated 7 patients with recurrent PCNSL with temozolomide and rituximab; they had a 100% response rate, with median response duration of 6 months and OS of 8 months. Enting et al12 treated 15 patients with refractory PCNSL with rituximab and temozolomide; their response rate was 53%. The median PFS for all patients was 2.2 months but 7.7 months for complete responders; median OS was 14 months. Batchelor et al23 recently published data in 12 patients treated with rituximab after failing HD methotrexate. They reported a 36% response rate, with a median PFS of almost 2 months and OS of 20.9 months.

Two additional studies used intensive multiagent chemotherapy, which had higher morbidity than single-agent trials, as well as several treatment-related deaths. Arellano-Rodrigo et al20 treated 16 patients with recurrent PCNSL with etoposide, ifosfamide, and cytarabine. The response rate was 37% (6 CRs), with a PFS of 4 months, PFS-12 of 22%, and OS-12 of 41%. Soussain et al24 treated 22 patients with refractory or relapsed PCNSL with HD cytarabine and etoposide and then transplanted with thiotepa, busulfan, and cyclophosphamide. After this regimen, 16 patients entered CR, and 2 remained in PR; the 3-year probability of survival and event-free survival was 60% and 53%.

The toxicities seen in our patients were higher than expected and in some cases severe, which may be a result of using a higher than standard dose (900 mg/m2 vs 500 mg/m2) and their degree of prior therapy. The dose chosen was empiric but increased over the standard dose to increase CNS penetration. The increased toxicity may have been less if the standard dose was used.

Our results with single-agent pemetrexed in relapsed/refractory PCNSL are not dissimilar from other salvage regimens, given the differences in patient populations. Despite minimal CSF penetration in an animal model, antitumor activity occurs in patients. In a separate trial, we measured CSF and plasma in patients treated with pemetrexed and found that the levels were 1% to 3% of plasma (manuscript in preparation); tumor penetration likely occurs from a dysfunctional blood-brain barrier leading to tumoricidal levels of drug. Furthermore, this agent is relatively easy to administer in the outpatient setting, unlike methotrexate, which requires hospitalization for continuous hydration. Continued investigation of this agent, or other multitargeted antifolates, in a more homogeneous population, either alone or in combination, for PCNSL is warranted to determine an optimal dose with less toxicity and also true efficacy.


This study was in part supported by Lilly Oncology.


We thank Katie McCarthy and Amy Youngkin for their assistance in data collection and management.


J.J.R. has received advisory board and speaking honoraria from Genentech and Merck, and has served as a paid consultant for Geron. S.A.G. has received a speaking honorarium from Genentech.