Outcomes at a single institution with MTR induction and consolidation in the management of primary CNS lymphoma Open Access

Primary central nervous system (CNS) lymphoma (PCNSL) is a variant of diffuse large B-cell lymphoma (DLBCL) that involves the brain parenchyma, leptomeninges, orbits, or spinal cord without evidence of systemic disease.¹ PCNSL is a rare but aggressive form of non-Hodgkin lymphoma, comprising 4% of all primary brain tumors, 5% of extranodal lymphomas, and <1% of all new non-Hodgkin lymphoma diagnoses.² 

The incidence of PCNSL has risen fivefold in the last 50 years, particularly in the cohort aged >60 years, with a median age of diagnosis of 67 years, confirming PCNSL is a disease of older patients.^3,4 PCNSL can lead to rapid neurologic dysfunction and impairment in performance status (PS).⁵ Consequently, aggressive and effective induction therapy is necessary to optimize chances of neurologic recovery and prevent long-term debility. The treating clinician must remain conscious of treatment-related toxicities, particularly with older patients.

Given the paucity of phase 3 randomized clinical trials, there is no consensus on optimal induction and consolidation strategies in PCNSL. High-dose methotrexate (HD-MTX) remains the backbone in induction options. Studies have demonstrated that adding agents including cytarabine (Ara-C) to HD-MTX prolongs survival and increases response rates, suggesting polychemotherapy should be the preferred induction strategy, setting the stage for the multiagent chemotherapy regimens currently in use.⁶ 

Options that have shown efficacy include HD-MTX, Ara-C, rituximab and thiotepa (MATRix); HD-MTX, rituximab, and temozolomide (MTR); HD-MTX, rituximab, procarbazine, and vincristine; and HD-MTX, teniposide, carmustine, and prednisolone.^7-13 With the exception of MATRix, there are no head-to-head studies comparing these regimens to determine which is superior. Preferred regimens vary based on institutional preference, physician experience, and geographic location.

Whole brain radiation therapy (WBRT) with 36 to 45 Gy was historically the preferred consolidative strategy. However, given increased neurotoxicity, particularly in older patients, chemotherapy-only approaches including consolidative autologous stem cell transplantation (ASCT) have been increasingly used and have also shown an improved progression-free survival (PFS) over WBRT.¹⁴ Toxicity with ASCT is high in older patients because studies have shown a nonrelapse mortality of 21% in patients aged >60 years.¹⁵ The New Approaches to Brain Tumor Therapy (NABTT) 96-07 trial established the treatment paradigm of HD-MTX administered every 14 days for induction until complete response (CR) was achieved, with 2 additional cycles of consolidation followed by HD-MTX maintenance every 28 days for 11 cycles.¹⁶ Overall response rate (ORR) was 74% with a follow-up study showing median overall survival (OS) of 55 months.¹⁷ There were no deaths related to therapy. Other options for consolidation include Ara-C and infusional etoposide, rituximab, dexamethasone, etoposide, and carboplatin; and less toxic regimens including rituximab, HD-MTX, procarbazine, and lomustine, HD-MTX and temozolomide, temozolomide monotherapy, lenalidomide, and ibrutinib.^18-26 

Since 2008, our group has preferred MTR induction for both PCNSL and secondary CNSL based on CALGB 50202, with treatment administered every 2 weeks, then spaced to monthly at CR to complete between 6 to 8 doses of MTX in induction.¹¹ A prolonged course of MTR is administered monthly for consolidation to complete 1 year of treatment per NABTT 96-07.¹⁶ We have previously reported our institutional experience.²⁷ Here, we report an updated 15 year experience of our approach in PCNSL.

We conducted a retrospective cohort study of consecutive patients with PCNSL who were treated at the Hospital of the University of Pennsylvania. The study was approved by the institutional review board of the University of Pennsylvania.

Adult patients (aged ≥18 years) were identified based on diagnosis code of CNS lymphoma in the electronic medical record between 1 April 2008 to 1 October 2024. To be included in the analysis, patients must have had documented B-cell lymphoma in their CNS (via biopsy or cerebrospinal fluid cytology) and no systemic lymphoma. Patients were excluded if they did not receive induction or received therapy alternative to MTR. In addition, patients with follow-up of <1 year after completing MTR chemotherapy or who were receiving active treatment at the time of data cutoff were also excluded from analysis (Figure 1).

Patients were treated with rituximab 375 mg/m² on day 1 in combination with HD-MTX 8 g/m² (days 1 and 15) and temozolomide 150 mg/m² (days 1-5) in 28-day cycles. HD-MTX was administered with leucovorin rescue. The HD-MTX dose was adjusted for creatinine clearance (supplemental Table 1) and other toxicities (ie, acute kidney injury [AKI], transaminitis, infection, and poor PS) and at the discretion of the treating physician. Once a CR was achieved, the therapy was given monthly without extra day 15 MTX. Induction was defined as the initial 6 to 8 doses of therapy. Consolidation options included WBRT, MTR monthly for 6 additional months, or rituximab plus etoposide followed by ASCT.

MTR consolidation was given monthly thereafter to complete 6 additional months of therapy. Based on the protocol originally described in NABTT 96-07, treatment was continued unless there was disease progression, significant adverse events (AEs), or patient preference to discontinue further consolidation.¹⁶ Growth factor support was not mandatory and used at the discretion of the treating physician. Standard antimicrobial prophylaxis with acyclovir was used in all patients.

Initial response assessment was performed after the first and second cycles of rituximab, temozolomide, methotrexate (RTM) with magnetic resonance imaging of the brain or cerebrospinal fluid cytology, as appropriate. Clinical response in the CNS was graded using the criteria from the International Workshop to Standardize Baseline Evaluation and Response Criteria for PCNSL.²⁸ Response assessments thereafter was performed at the discretion of the treating physician but typically every 2 to 4 months. Toxicities were recorded using the Common Terminology Criteria for AEs version 4, as previously reported.²⁹ 

For patients with vitreal involvement, response was based on clinical assessment or intravitreal sampling performed at the discretion of the treating ophthalmologist.

Descriptive and survival analyses using the Kaplan-Meier methodology were performed (Stata, version 18.5; Stata Corp, TX). A log-rank test was used to compare OS and relapse-free survival (RFS) between patient groups. For the primary outcome of OS, we performed a univariate Cox proportional hazard analysis to evaluate the association of each variable on OS. These variables were then analyzed in a multivariate Cox proportional hazard analysis to adjust for confounding variables. The proportional hazards assumption was tested and met based on Schoenfeld residuals. Logistic regression analysis was performed to determine association between characteristics. All tests were 2 sided and a P value <.05 was considered statistically significant. Secondary outcome included time to best response.

OS was defined as the time from initiation of therapy to last office visit or death from any cause. RFS was calculated from initiation of therapy to disease progression or last office visit. Event-free survival was calculated from initiation of therapy to disease progression, death, or last follow-up.

Steroid response was defined as improvement in neurologic symptoms and/or physical examination with the initiation of steroids before the initiation of systemic chemotherapy.

Univariate Cox proportional hazard analysis was also performed to evaluate factors that could predict for response to MTR lasting >10 years.

We identified 153 patients diagnosed and treated for PCNSL at our institution from 1 April 2008 through 1 October 2024 who met criteria for analysis. Of 153 patients, 83 (54%) were female; 137 (90%) were White, 8 (5%) were Asian, and 7 (4%) were Black. The median age at diagnosis was 63 years (range, 20-87). Of 153, 123 (80%) had an Eastern Cooperative Oncology Group (ECOG) PS score between 0 and 2 (median, 1; range, 0-4). Classification into the Memorial Sloan Kettering Cancer Center (MSKCC) prognostic score showed 23 of 153 patients (15%) had low-risk disease, 85 (56%) had intermediate-risk disease, and 45 (29%) had high-risk disease. Most had parenchymal disease, with 91 of 153 (59%) having multifocal lesions. Of 153, 59 (39%) had deep brain involvement, defined by involvement of the periventricles, cerebellum, basal ganglia, or brain stem; 18 patients (12%) presented with seizures; and 139 (91%) were cases of DLBCL.

Of patients for whom cell of origin (COO) was determined, 82 of 105 (78%) had activated B-cell (ABC) subtype. None had double-hit lymphoma, however, 39 of 88 (44%) had double-expressor lymphoma with at least 40% MYC expression and 50% B-cell lymphoma 2 expression on immunohistochemistry. Overall, 10 of 107 (9%) were Epstein-Barr virus (EBV) positive, detected via Epstein-Barr encoding region in situ hybridization. All but 1 patient with EBV⁺ tumors had a history of immunosuppression due to HIV, renal transplant, myasthenia gravis, dermatomyositis or Crohn disease; 7 of 153 patients (5%) were HIV⁺ (Table 1).

The median length of follow-up was 63 months (range, 1-198). The median OS for the entire cohort was 65 months (95% confidence interval [CI], 33.0-102.9). The median RFS was 36 months (95% CI, 16.7-97.9; Figure 2). The 4-year OS rate was 49%. Four-year event-free survival rate was 35%. The mean number of doses of MTR administered was 7 (range, 1-15). Of 153 patients, 72 (47%) remained disease free after MTR. Overall, 87 of 153 patients (57%) had died at the time of analysis; 73 of 87 patients (84%) died from complications related to their lymphoma; 2 patients (1%) died from an unrelated solid tumor malignancy (lung or breast cancer); and the remainder of the patients died from causes unrelated to their lymphoma.

MTR induction was administered every 2 weeks until CR, then spaced to monthly to complete between 6 and 8 treatments over a span of ∼6 months although adjustments may have been made at the discretion of the treating physician.

ORR was 84%; 91 of 150 (61%) had a CR and 34 of 150 (23%) had a partial response as best response to therapy. The median time to best response was 1.93 months (95% CI, 1.67-2.53) or after ∼2 cycles of therapy.

Of 153 patients, 92 (60%) completed induction with MTR. Of the 61 patients who did not complete induction, 39 (64%) stopped because of progressive disease, 11 opted for hospice, 4 had toxicity limiting additional MTX administration, 3 died from another cause, 2 preferred no additional MTX, and 2 were lost to follow-up. For the 39 patients who progressed during induction, the median time to progression was 1.6 months (between 0.7 and 6 months); 36 of 39 patients (92%) were receiving MTR every 2 weeks at the time of progression.

Patients who did not complete the full induction course of MTR had a median OS of 7 months (5.49-11.05; P < .001).

Univariate Cox proportional analysis demonstrated that sex, race, focality (unifocal vs multifocal), COO, seizure at presentation, and deep brain involvement were not associated with OS. However, ECOG PS score of 0 to 2, age of <60 years at diagnosis, HD-MTX–sensitive disease (defined by no progressive disease within 6 months after starting HD-MTX–based therapy), low or intermediate disease by MSKCC score, steroid-responsive disease, and CR/partial response after 2 cycles of therapy were all associated with improved OS (Table 2).

Nevertheless, multivariate Cox proportional hazards model showed that only age and steroid responsiveness were significant predictors when adjusted for ECOG PS, MTX resistance, and disease response after 2 cycles of treatment (Table 3).

Of 92 patients who completed induction, 57 (62%) received MTX-based consolidation. The remaining patients received other therapy for a variety of reasons including patient preference for no consolidation (9/35 [26%]), Ara-C–based consolidation with or without ASCT (7/35 [20%]), progressive disease (7/35 [20%]), temozolomide-based consolidation (5/35 [14%]), radiation, Bruton tyrosine kinase inhibitor, lenalidomide, or died from another cause. There was no significant difference in OS and RFS among patients who received MTR or Ara-C–based consolidation.

The median number of doses of prolonged MTR administered was 10 (range, 8-15). Of 57 patients, 12 (21%) received a shorter course of consolidation, between 1 and 3 cycles; 2 discontinued MTX because of toxicity and 10 discontinued because of patient preference. Overall, 7 patients did not relapse with shorter consolidation. There was no significant difference in OS between shorter or longer (4-6 cycles) consolidation.

The median OS in patients who completed induction and received prolonged MTR was 143 months (95% CI, 95.3 to not reached). Median RFS was 122 months (95% CI, 52-174.2; Figure 3). In this cohort, the 4-year OS rate was 85% and 4-year RFS was 72%. Of 57 patients, 34 (60%) remained in remission after completing prolonged MTR. Of the patients who relapsed, the median time to relapse was 30 months (range, 14-174). Of 23 patients, 5 (22%) relapsed more than a decade after completing treatment; 4 patients were retreated with MTR and 1 was treated with rituximab and zanubrutinib.

Overall, 100 of 129 (78%) patients had improvement in neurologic symptoms with the initiation of dexamethasone before initiating chemotherapy. Dosing and duration were determined at the discretion of the treating physician. There was a significant difference in OS between those with steroid-responsive and -unresponsive disease, with median OS of 90.1 months (95% CI, 54.8-163.1) vs 10.6 months (95% CI, 3.5-17.0; P = .002). The hazard ratio for death in the steroid nonresponsive groups was 2.64 (95% CI, 1.57-4.44; P < .001; Figure 4). Logistic regression showed that factors including age, ECOG PS, MSKCC prognostic score, histologic factors, and focality and location of the tumor did not predict for steroid-responsive disease. However, patients who were HIV⁺ were significantly more likely to respond to steroids with an odds ratio of 16.7 (95% CI, 1.69-100.0; P = .02) and had a nonstatistically significant better OS than those who are HIV^– (103.1 vs 59.0 months; P = .89).

Overall, 11 of 57 patients (19%) who were treated with prolonged MTR remained event free for ≥10 years after completing therapy. The median number of doses of MTR these patients received was 12 (range, 8-15). Logistic regression found that focality (unifocal vs multifocal), ECOG PS, seizure at presentation, elevated lactate dehydrogenase, deep-brain involvement, COO, double-expressor lymphoma status, length of consolidation (long vs short), HIV status, EBV status, and MSKCC prognostic index did not predict for long-term response to MTR. Age of <60 years did however predict for long-term response to MTX (odds ratio, 0.10; 95% CI, 0.20-0.57; P = .01). The median age of patients with long-term response was 54 years (range, 37-72) vs 62 years (range, 42-80). However, it is possible that this could be confounded by older patients being more likely to die from other causes than young patients. In addition, 3 patients developed difficulty with memory; however, 1 patient was 87 years old when this developed, and thus, it was considered age related.

Next-generation sequencing (NGS) data were available for 26 patients; 18 of 26 patients (69%) had mutations in MYD88. The most common mutation site was L265P (15/18). The remaining patients had mutations in MYD88^L273P. Mutations in MYD88 were associated with worse OS although logistic regression showed that these patients were more likely to have higher risk disease per MSKCC prognostic scoring, which could explain the difference in OS (Figure 5). In patients with MYD88 mutations, 8 had high-risk MSKCC disease and 10 had intermediate-risk disease. In patients without MYD88 mutations, 1 had high-risk, 4 had intermediate-risk, and 3 had low-risk disease. Of 15 patients with MYD88^L265P, 9 (60%) had comutations in CD79b, indicating MYD88/CD79b (MCD) cluster of DLBCL. There was no significant difference in OS between those with and without these comutations. Unlike in systemic DLBCL, MCD cluster did not favor ABC COO. Three of 26 patients (12%) had mutations in TP53, which did not predict for worse survival.

Of 153 patients treated with MTR, 78 (51%) had no major AEs, 50 (33%) developed AKI, with 2 patients (1%) requiring hemodialysis, and 11 (7%) developed cytopenias and infectious complications. Although 4 patients had pancytopenia, 3 had thrombocytopenia only, 2 had neutropenia only, and 2 had anemia only. One patient required platelet transfusions, and 1 required blood transfusions. Overall, 7% (10/153) had transaminitis. Less common AE included leukoencephalopathy, pneumonitis, and mucositis. Of 153 patients, 33 (22%) required dose reduction because of toxicity, which was most commonly AKI (24/33), cytopenias (5/33), transaminitis (2/33), and leukoencephalopathy (2/33). Of 153 patients, 20 (13%) required discontinuation of MTX because of AKI (15/20), leukoencephalopathy (4/20), and pancytopenia (1/20). In addition, 9 patients (6%) required glucarpidase rescue, with a median creatinine of 3.77 mg/dL (range, 1.40-6.11). Six patients were successfully rechallenged with MTR, whereas 3 patients subsequently received alternative therapy. No patients died from complications related to treatment.

We evaluated 153 patients who were treated with a uniform frontline treatment regimen at our institution from April 2008 to October 2024 and present our outcomes. ORR to MTR induction was 84%, with 61% of patients achieving CR as best response. The median OS and RFS were 65 and 36 months, respectively. In patients who were able to receive consolidation with MTR, median OS and RFS were 143 and 122 months, respectively. The 4-year OS rate was 85% and 4-year RFS was 72% in this group. Although 49% of patients did experience toxicity from treatment, there was no treatment-related mortalities. These results indicate that prolonged MTR is safe and can be an alternative treatment option in the management of PCNSL.

The International Extranodal Lymphoma Study group-32 (IELSG32) trial found that induction with MATRix followed by ASCT or WBRT yielded an ORR of 87% with 7-year OS of 70%. However, grade 4 toxicity occurred in 70% of patients and treatment-related mortality was high at 7%.³⁰ Additionally the median age of patients treated in IELSG32 was 57 years. As the incidence of PCNSL rises in patients aged >60 years, it is imperative to establish treatment regimens that are not only effective but also well tolerated.

Although more aggressive consolidation to HD-MTX–based induction may improve outcomes, more recent trials have reduced the number of HD-MTX cycles proposed by NABTT 96-07 in exchange for aggressive consolidation. Whether the potential benefits of WBRT, nonmyeloablative chemotherapy, or ASCT outweigh the risks of toxicity compared with prolonged HD-MTX outlined by NABTT 96-07, are the subject of ongoing study.

A recent retrospective study from The Netherlands with 346 patients treated with MTX-based induction followed by ASCT or WBRT consolidation reported a median OS of 49 months and median PFS of 31 months.³¹ A retrospective Israeli study with 222 patients treated with MTX-based induction and most commonly ASCT consolidation reported a median OS of 44 months and a median PFS of 22 months.³² Another study from Ohio with 80 patients with most treated with MTX-based induction followed by Ara-C consolidation found the median OS was 32 months and median PFS was 16 months.³³ Our results are comparable with other institutional experiences with less overall toxicity.

As patients with PCNSL and secondary CNSL live longer, exposure to high-doses of chemotherapy raises the concern for the later development of a secondary malignancy. It is unclear what the true incidence is in this population given short follow-up in trials, however, at 88 months, IELSG32 found 4% of patients developed a second cancer in contrast to 1% in our cohort.³⁰ 

Notable factors that predicted for prolonged OS include age and steroid responsiveness. It has been well established that older patients are more likely to have worse outcomes as per the MSKCC and IELSG prognostic scores.^34,35 This is thought to be because of poor PS and increased disease-related debility leading to poor tolerance of therapy along with medical comorbidities that may limit doses of MTX. Corticosteroids directly lead to apoptosis of lymphoma cells and improvement in cerebral edema so it is not surprising that they may lead to a short-lived improvement in symptoms. We found that those with improvement had prolonged OS. Although not well studied, this could be because of a unique tumor biology in these patients, increased disruption of the blood-brain barrier, or amenable location of their tumor leading to increased exposure to steroid and subsequent MTX from their vasculature.

A subset of long-term responders to prolonged MTR remained event free for >10 years after treatment completion. These patients tended to be younger, with a median age of 54 years. Otherwise, there were no factors that predicted for these patients, which would be important because they could be a cohort that justifies less intensive therapy to reduce treatment-related mortality. Additional studies are necessary to determine whether any molecular or genetic features could correlate with long-term response to MTX.

We also evaluated the prognostic value of NGS. Most patients had activating MYD88^L265P leading to increased NF-κB signaling promoting tumoral growth.³⁶ In systemic DLBCL, co-occurrences of mutations in MYD88^L265P and CD79b are classified into the MCD genetic subtype, although more common in ABC tumors, compromise only 8% of systemic DLBCL.³⁷ MCD patients with systemic DLBCL have a poor prognosis, however, a phase 1/2 clinical trial found that these patients have increased response to ibrutinib, indicating dependence of B-cell receptor signaling.^38,39 In our cohort, although there was no significant difference in OS by MCD cluster, there was a trend toward worse survival in those with MYD88 mutations. Several studies have shown response rates of 80% with ibrutinib-based regimens in the relapsed setting, even in MCD subtype tumors.^40,41 

In conclusion, here we show that prolonged MTR is a well-tolerated and alternative induction and consolidative treatment strategy for patients with PCNSL.

This study had several limitations including that it is a retrospective, single-institution analysis and data were abstracted via chart review. This analysis is not a prospective clinical trial and, as such, there were variations in dosing and scheduling of therapy. Complete toxicity data may not have been available or missing, and some toxicities were not graded.

Contribution: N.K.D. designed research, performed research, analyzed data, and wrote the manuscript; V.C. designed research; A.C.W. and W.-T.H. analyzed the data; J.S.C., M.C., and C.T. reviewed the manuscript; D.E.T. and J.N.G. designed research; S.J.S., S.K.B., J.S., E.A.C., and D.J.L. designed research and reviewed the manuscript; and S.D.N. designed research, performed research, analyzed data, and wrote and reviewed the manuscript.

Conflict-of-interest disclosure: E.A.C. reports research funding from AbbVie, AstraZeneca, CARGO Therapeutics Inc, Genentech, Genmab, and Nurix; reports consulting role with Genmab/AbbVie, AstraZeneca, and BeiGene; and reports honoraria from IDEOlogy Health, and Intellisphere. D.J.L. reports research funding from ADC Therapeutics. C.T. serves on an advisory board for Acrotech; consulted for Versed; and reports honoraria from Reckner and Olson Research. S.D.N. reports research support from Takeda/Millenium, Genetech/Roche, Loxo/Lilly, Atara, Astex, Ono, and Caribou; and received honoraria/reports consulting role with Pierre Fabre, Genmab, ADC Therapeutics, Ono, Genentech, and served on the data safety monitoring board for Merck. J.S. reports consultancy with Pfizer, Incyte, Genmab, Bristol Myers Squibb (BMS), Atara, AstraZeneca, Adaptive, and Crisper; and reports research funding from Pfizer, Pharmacyclics, Merck, Incyte, BMS, AstraZeneca, Adaptive, and Kite. The remaining authors declare no competing financial interests.

Correspondence: Nikita K. Dave, Hematology/Oncology, Hospital of the University of Pennsylvania, 3400 Spruce St, Philadelphia, PA 19104; email: daven@pennmedicine.upenn.edu.