The outcome of patients with systemic diffuse large B-cell lymphoma (DLBCL) had improved over the past decade with the addition of monoclonal antibody therapy. Unfortunately, approximately 5% of these patients still developed a secondary central nervous system (CNS) recurrence followed invariably by rapid death. This rate is substantially increased in patients with certain high-risk features. Although prophylaxis against CNS recurrence with either intrathecal or intravenous methotrexate is commonly used for such patients, to the authors' knowledge, there is no standard of care. Retrospectively evaluated was the role of high-dose systemic methotrexate combined with standard cyclophosphamide, doxorubicin, vincristine, and prednisone with rituximab (R-CHOP) chemotherapy to decrease CNS recurrence in high-risk patients.
A total of 65 patients with DLBCL and CNS risk factors were identified at the study institution between 2000 and 2008 who received intravenous methotrexate as CNS prophylaxis concurrent with standard systemic therapy with curative intent. CNS recurrence rate, progression-free survival, and overall survival were calculated.
Patients received a median of 3 cycles of methotrexate at a dose of 3.5 gm/m2 with leucovorin rescue. The complete response rate was 86%, with 6% partial responses. At a median follow-up of 33 months, there were only 2 CNS recurrences (3%) in this high-risk population. The 3-year progression-free and overall survival rates were 76% and 78%, respectively. Complications associated with methotrexate therapy included transient renal dysfunction in 7 patients and a delay in systemic chemotherapy in 8 patients.
The rate of secondary involvement of the central nervous system (CNS) in lymphoma varies widely by histology but is nearly always devastating. CNS dissemination in Burkitt and lymphoblastic lymphomas approaches 30%,1 necessitating the routine incorporation of intravenous and intrathecal (IT) prophylaxis against secondary CNS disease in the treatment programs for these histologies. The risk of CNS involvement and subsequent need for prophylaxis are less well described for diffuse large B-cell lymphoma (DLBCL), the most common non-Hodgkin lymphoma, and there is no standard of care.
DLBCL carries a rate of secondary CNS involvement of approximately 3% to 5%, but the rate is considerably higher in patients with certain high-risk clinical features present at the time of diagnosis.1-3 These risk factors include the involvement of specific extranodal sites (bone marrow, testes, and paranasal sinuses, as well as perhaps the kidneys, adrenal glands, liver, and breast).1, 4-10 Multivariate risk models have further identified the combination of an elevated lactate dehydrogenase (LDH) and involvement of ≤2 extranodal sites as conferring a risk of CNS recurrence as high as 34%.1, 11 A large Norwegian retrospective analysis identified 5 independent risk factors for CNS recurrence: elevated LDH, age >60 years, involvement of >1 extranodal site, retroperitoneal lymph node involvement, and hypoalbuminemia.3 When >3 of these risk factors were present at diagnosis, the risk of CNS recurrence exceeded 25%. The clinical outcome of CNS recurrence in patients with DLBCL is poor, with rapid morbidity and death within 2 to 5 months.1, 3, 5
Identification of DLBCL patients at high risk for subsequent CNS recurrence provides the opportunity to offer prophylactic therapy at the time of diagnosis. High-dose methotrexate is uniformly included in regimens for the therapy for Burkitt lymphoma and lymphoblastic lymphoma and is the most active single agent in the therapy of primary CNS DLBCL. Despite these implications, to our knowledge there is no clear standard for CNS prophylaxis in high-risk DLBCL patients. IT methotrexate has been most commonly used historically,12 but evidence of benefit has been equivocal, with no protective benefit observed in 2 large randomized controlled treatment trials of DLBCL.2, 11, 13-15 By contrast, the combination of systemic and IT methotrexate has demonstrated effective reduction in CNS recurrence in DLBCL,16 suggesting that intravenous methotrexate may be primarily responsible for the risk reduction; however, the protective benefit of intravenous high-dose methotrexate without concurrent IT chemotherapy has never been formally evaluated. In the current study, we present what to the best of our knowledge is the first report of high-dose systemic methotrexate combined with standard chemoimmunotherapy to decrease CNS recurrence in high-risk DLBCL patients.
MATERIALS AND METHODS
We queried our Institutional Review Board-approved clinicopathologic database, derived from comprehensive tumor registry data at the Massachusetts General Hospital, for all patients aged ≥18 years who were diagnosed with DLBCL between 2000 and 2008. A total of 1283 patients with DLBCL were identified in the database, 65 of whom (5%) met inclusion criteria and are included in the analysis. Patients were selected for inclusion if they were treated with curative intent with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP), with or without rituximab (R-CHOP), plus at least 1 infusion of systemic intravenous methotrexate at a dose of 3.5 g/m2 for CNS prophylaxis. CHOP and R-CHOP were administered by standard protocol on a 21-day cycle.17 Patients were not included if they presented with primary DLBCL of the CNS, or concurrent CNS and systemic involvement by DLBCL at diagnosis. CNS involvement was determined by neuroimaging or cerebrospinal fluid (CSF) examination in symptomatic patients at the discretion of the treating physician. Patients were not included if their pathology was consistent with Burkitt lymphoma or high-grade B-cell lymphoma with intermediate features between DLBCL and Burkitt lymphoma. All patients were considered high risk for CNS recurrence based on published risk models, including involvement of >2 extranodal sites plus an elevated LDH1, 18; Hollender 5-point criteria3; or high-risk locations including bone marrow, paranasal sinuses, testes, epidural disease, liver, adrenal, renal, or orbit.1, 5, 6 Discordant bone marrow involvement by a low-grade lymphoma was not considered an indication for CNS prophylaxis. Progression-free and overall survival rates with 95% confidence intervals (95% CI) were calculated from date of diagnosis to disease progression or death, or death, respectively, and were plotted by the method of Kaplan and Meier.
Sixty-five patients met inclusion criteria and were included in the analysis. Patient characteristics are summarized in Table 1. The median age at diagnosis was 60 years (range, 25-79 years), and 63% of patients were male. The majority of patients had Ann Arbor stage IV disease (73%). Using the revised International Prognostic Index (IPI) score, 44 (68%) patients were high risk (score of 3-5), predicting for a 4-year overall survival rate of 55%.19 One patient had human immunodeficiency virus infection. LDH was elevated in 73% of patients, and 64 of 65 patients had extranodal involvement of their DLBCL. Lumbar punctures (LPs) were performed at diagnosis in 19 patients, all of which were negative by cell counts. Seventeen LPs underwent cytologic examination and 14 underwent flow cytometry, none of which were positive for malignant cells, although 2 flow cytometries were believed to be equivocal due to paucity of cells, and had negative accompanying cytology and cell count and are included in the analysis. These LPs were performed at the discretion of the treating physician. Cytogenetic information was available on 27 patients, 6 of whom had a BCL2 rearrangement, 3 patients had a c-MYC rearrangement, and 5 had complex karyotypes. Two patients had concurrent rearrangements of c-MYC and BCL2. Ki-67 staining was performed on 46 cases at diagnosis, among which the median Ki-67 proliferation index was 78% (range, 15-100%). Fourteen cases (30%) had a proliferation index of >90%.
Table 1. Patient Characteristics
DLBCL indicates diffuse large B-cell lymphoma; PMBCL, primary mediastinal B-cell lymphoma; LDH, lactate dehydrogenase; R-CHOP-M, cyclophosphamide, doxorubicin, vincristine, and prednisone with rituximab plus at least 1 infusion of systemic intravenous methotrexate; HD, high-dose; LP, lumbar puncture; IT, intrathecal; ECOG, Eastern Cooperative Oncology Group; IPI, International Prognostic Index.
Total no. of patients
Median age (range), y
Extranodal sites >1
Any extranodal disease
Median no. of extranodal sites
2 (range 0-5)
Bulky disease (>7cm)
Median no. of CHOP cycles (range)
Median no. of HD methotrexate cycles (range)
Diagnostic LP (all negative)
IT at diagnostic LP
ECOG Performance status 0-1
IPI 0, 1
IPI 3, 4, 5
Sixty-three patients received R-CHOP, and 2 patients received CHOP without rituximab. CHOP/R-CHOP was given for a median of 6 cycles, with 44 patients receiving 6 cycles, 12 receiving 7 to 8 cycles, and 9 receiving 3 to 4 cycles. Methotrexate was administered to all patients at a dose of 3.5 g/m2 intravenously for a median of 3 cycles (range, 1-8 cycles). Methotrexate was most commonly administered on Day 15 of alternating cycles of R-CHOP (ie, with Cycles 2, 4, and 6), as an inpatient with leucovorin rescue. Four patients received a single dose of IT methotrexate at the time of diagnostic LP. Sixteen patients received consolidative involved field radiotherapy.
The indications for CNS prophylactic therapy are summarized in Table 2. Most prophylactic therapy was given to patients with multiple extranodal sites plus an elevated LDH (46%), followed by involvement of bone marrow (22%); epidural disease (22%); a Hollender score of 4 or 5 (17%); and involvement of the kidney or adrenal gland (14%), orbit (14%), liver (12%), paranasal sinuses (9%), testes (8%), and breast (2%). Patients frequently had >1 of the above risk factors present at diagnosis.
Table 2. Summary of CNS Risk Factors
CNS Risk Factor
CNS indicates central nervous system; LDH, lactate dehydrogenase.
>1 extranodal site
>1 extranodal site and elevated LDH
Hollender score of 4-5
The median follow-up is 33 months. Outcome is summarized in Table 3. The overall systemic response rate was 92%, with 56 patients (86%) achieving a complete response, 4 patients (6%) achieving a partial response, and 5 (7%) developing disease progression during induction therapy. There were 11 systemic recurrences outside of the CNS at a median of 14 months (range, 3-51 months) from diagnosis. The 3-year progression-free survival rate was 76% (95% CI, 62-86%), and the overall survival rate was 78% (95% CI, 64-88%) (Fig. 1). Among all patients with recurrent disease, the median Ki-67 proliferation index of diagnostic biopsy specimens was 70% (range, 50-90%). Among 11 patients with recurrent systemic disease, the median follow-up is 2 years with 4 patients remaining alive at last follow-up.
CNS recurrence occurred in only 2 patients (at 4 months and 9 months, respectively from diagnosis), yielding a CNS recurrence rate of 3% in this high-risk population. Both patients died of their CNS recurrence at 32 days and 127 days from recurrence, respectively. One patient had received IT methotrexate at the time of initial diagnostic LP, which was equivocal by flow cytometry and negative by cytology and cell count. The other patient did not have an LP at diagnosis. Both patients received R-CHOP. One patient developed disease recurrence during treatment exclusively in the leptomeninges after receiving 3 cycles of R-CHOP and 1 infusion of systemic methotrexate. The other patient completed 6 cycles of R-CHOP and 5 infusions of systemic methotrexate, and presented 48 days after completing chemotherapy with CNS recurrence in the brain parenchyma and leptomeninges, as well as systemic disease recurrence. The indication for CNS prophylaxis was bone marrow involvement by DLBCL in both patients. LDH was markedly elevated in both patients at diagnosis (2764 U/L and 798 U/L; normal range for our laboratory, 110-210 U/L). Ki-67 staining at initial diagnosis was available in only one of these patients and was 70%. Neither patient had a detectable c-MYC rearrangement.
The median creatinine at baseline was 0.9 mg/dl (range, 0.5-1.7 mg/dl). The median maximal creatinine within 2 weeks of the first methotrexate infusion was 1.1 mg/dl (range, 0.6-2.8 mg/dl). Twenty-six patients had renal toxicity, defined as an increase in serum creatinine above the upper limit of normal; 10 of these patients (38%) were aged ≤60 years and 16 patients (62%) were aged >60 years. After the first methotrexate infusion, 7 patients had an elevation in their creatinine to ≥2.0 mg/dl, 3 of whom did not receive further methotrexate infusions but subsequently had complete recovery of renal function. One of these cases of renal failure occurred in the setting of recent intravenous contrast administration.
Sixty-one patients received a second infusion of high-dose methotrexate, at which time the baseline creatinine was again 0.9 mg/dl (range, 0.5-1.7 mg/dl). Only 2 patients experienced elevations in their creatinine ≥2 mg/dl, both of whom had also had developed creatinine >2 mg/dl after the first methotrexate infusion. Among patients without renal dysfunction after Cycle 1, there was no renal dysfunction with Cycle 2.
Fifty-eight patients received a third infusion of high-dose methotrexate, with a median baseline creatinine of 0.9 mg/dl (range, 0.6-1.3 mg/dl). Two patients experienced creatinine elevations ≥2 mg/dl, and received no further methotrexate. One patient developed a creatinine of 6 and required temporary hemodialysis with subsequent recovery to dialysis independence. No other patient in this series required renal replacement therapy. Nineteen patients received a fourth infusion of high-dose methotrexate, 12 a fifth cycle, 6 a sixth cycle, and 1 a seventh and eighth cycle. There were no episodes of acute renal failure during methotrexate cycles 4 to 8. For the entire cohort, the median end of treatment creatinine was 1.1 mg/dl (range, 0.6-4.1 mg/dl), with only 2 patients having creatinine ≥2 mg/dl.
Nine patients (14%) had planned additional methotrexate treatments aborted due to nephrotoxicity. Eight patients (12%) experienced a delay in R-CHOP because of methotrexate toxicity. Six patients had delay of 1 cycle, and 1 patient had 2 cycles delayed. Of the 8 delayed R-CHOP cycles, 3 were by 1 week, 3 were by 2 weeks, and 2 were by 3 weeks. Reason for dose delay was renal toxicity in 4 cycles, mucositis in 2 cycles, cytopenias in 2 cycles, and both cytopenias and renal dysfunction in 1 cycle. One patient died during treatment of chemotherapy-related complications, including nephrotoxicity, mucositis, pneumonia, and pancytopenia.
The results of the current study demonstrate that incorporation of intravenous systemic methotrexate at a dose of 3.5g/m2 into the standard R-CHOP treatment regimen is associated with decreased CNS recurrence risk compared with the published outcome in high-risk patients who did not receive CNS prophylaxis. Our 3-year progression-free survival rate approaching 80% is particularly encouraging in a cohort of patients with a preponderance of high-risk features for systemic disease recurrence, including high-risk IPI scores and high Ki-67 proliferation fractions.
Previous series demonstrate a CNS recurrence risk of 2% to 8% in patients with DLBCL in general,1-3, 5, 11 but none of these series were limited to high-risk patients, among whom the incidence of CNS recurrence is known to be significantly higher. Specific extranodal sites are highly associated with an elevated risk of CNS recurrence, perhaps due to differential expression of integrin or chemokine receptor profiles that allow colonization of these sites.20 Involvement of the bone marrow, testes, and paranasal sinuses as well as the kidneys, adrenal glands, liver, bladder, breasts, mediastinum, bone, epidural space, and peripheral blood have all been associated with an increased risk of CNS recurrence.1, 4-10, 21 In addition, univariate risk factor analysis of patients who develop CNS recurrence also has identified LDH elevation, the presence of extranodal sites of disease, age, and B symptoms as conferring an increased risk of CNS recurrence.1, 3, 18 On multivariate analysis, the presence of an elevated LDH in concert with the involvement of multiple extranodal sites by DLBCL appears to be the most strongly predictive clinical algorithm, predicting a CNS recurrence risk of 17% to 34%.1, 5, 11 In our series, 30 patients met these criteria, none of whom experienced a CNS recurrence. Both CNS recurrences in this series occurred in patients with bone marrow involvement by their DLBCL, which yields a 9% rate of CNS recurrence among the 22 total patients with bone marrow involvement at baseline. Involvement of the bone marrow has been associated with a nearly 20% risk of CNS recurrence,1 although this has not been uniformly observed across published series. No CNS recurrences were observed in patients with risk due to other identified high-risk locations including testes, paranasal sinuses, kidneys, orbit, breast, or epidural space, but the small subset of patients in each of these groups prevents drawing conclusions regarding the efficacy of systemic methotrexate in preventing recurrence due to a specific site of high-risk disease.
It is not routine practice at our institution to perform LPs at the time of DLBCL diagnosis in the absence of neurologic signs or symptoms; however, 19 of the 65 patients had an LP performed before the initiation of systemic therapy at the discretion of the treating physician. Sixteen (84%) of these LPs were sent for flow cytometry, which has been shown to be the more sensitive than cell counts or cytology in detecting occult CNS involvement of the CSF.22 Two of the 16 samples sent for flow cytometry were equivocal for the presence of clonal B cells due to a paucity of cells for analysis; in both of these cases, the cytology was negative. One of the patients with equivocal flow cytometry results at diagnosis subsequently developed a CNS recurrence.
The majority of CNS recurrences in patients with DLBCL occur during or shortly after the completion of induction chemotherapy.3, 5 The timing of these recurrences suggests that occult microscopic disease may already be present at the time of diagnosis. Before the addition of rituximab to chemotherapy regimens, these recurrences preferentially involved the leptomeninges in approximately 70% of cases.1, 3, 23 However, DLBCL of the testes has consistently demonstrated a unique natural history with late recurrences and a predilection for the brain parenchyma.2, 7, 8 The prominent leptomeningeal pattern of DLBCL recurrence led to the adoption of IT prophylactic therapy with methotrexate, with some retrospective analyses suggesting benefit,13, 15, 18 and others no benefit.2, 11, 14
Two prospective controlled trials have randomized DLBCL patients to therapeutic regimens either containing or not containing prophylactic CNS therapy with IT methotrexate. Southwest Oncology Group 8516 (SWOG) was a 4-arm randomized trial of front-line therapy in DLBCL that included 2 arms with CNS prophylaxis for patients with bone marrow involvement (prednisone, procarbazine, doxorubicin, cyclophosphamide, etoposide, cytarabine, bleomycin, vincristine, and methotrexate [ProMACE-CytaBOM] and methotrexate, doxorubicin, cyclophosphamide, and vincristine [MACOP-B]), and 2 arms without CNS prophylaxis (CHOP and methotrexate, bleomycin, doxorubicin, cyclophosphamide, vincristine, and dexamethasone [m-BACOD]). Patients treated with ProMACE-CytaBOM received consolidative whole-brain radiotherapy to 24 gray (Gy) after chemotherapy if bone marrow was involved at diagnosis. Patients on the MACOP-B arm with a positive bone marrow at diagnosis received prophylaxis with IT methotrexate and cytarabine twice weekly for 6 doses. Both arms also included intravenous methotrexate, but at significantly lower doses than are required to achieve significant CNS penetration. The overall rate of CNS recurrence in this trial was 2.8%, with no benefit noted in patients treated with prophylactic IT therapy or whole-brain irradiation.2 The RICOVER-60 trial randomized patients to R-CHOP at 14-day or 21-day intervals and included IT prophylaxis with methotrexate for all high-risk patients (22%) defined by disease involvement of the bone marrow, paranasal sinuses, orbits, oral cavity, tongue, or salivary glands.11 Of the total 1217 patients, 58 (4.8%) experienced a CNS recurrence; however, this rate rose to 34% in those patients with an elevated LDH and at least 2 extranodal sites of involvement. The incidence of CNS recurrence was not affected by the incorporation of IT therapy.
The most compelling evidence for effective prophylaxis against CNS recurrence comes from the GELA LNH 93-5 trial, which randomized patients with aggressive lymphomas and an age-adjusted IPI score of ≥1 to either CHOP-21 or the intensive ACVBP regimen.16 The majority of patients (80%) had doxorubicin, cyclophosphamide, vindesine, bleomycin, and prednisone (DLBCL). CHOP patients received no CNS prophylaxis, whereas all ACVBP patients received 4 doses of IT methotrexate (15 mg), and 2 infusions of intravenous methotrexate at a dose of 3000 mg/m2 with leucovorin rescue. The primary outcome of the trial demonstrated an improved overall survival for ACVBP-treated patients complicated by a higher treatment-related death rate. Notably, ACVBP-treated patients experienced significantly fewer CNS recurrences than CHOP patients (2.7% vs 8%). Although patients in this trial received both IT and intravenous prophylaxis, the preponderance of evidence, in conjunction with the SWOG and RICOVER-60 data, suggests that the intravenous methotrexate provided the improvement in efficacy. Our data validate the hypothesis that effective CNS prophylaxis can be provided with intravenous methotrexate alone in a high-risk cohort of DLBCL patients. High-dose methotrexate overcomes multiple potential liabilities of IT therapy, including uneven distribution within the neuroaxis, and failure to significantly penetrate the brain parenchyma. Lumbar administration of IT methotrexate results in marked differences in peak levels throughout the subarachnoid space, and subtherapeutic levels are common, resulting largely from differences in bulk CSF movement, choroidal uptake, and drug clearance.24 Intraventricular administration by means of an implantable Ommaya reservoir improves drug distribution around the brain, but lumbar concentrations remain lower.24, 25 In contrast, intravenous administration of methotrexate produces therapeutic drug levels that are evenly distributed throughout the entire subarachnoid space.24
All but 2 patients in our analysis received rituximab-containing therapy. The addition of rituximab to standard CHOP-21 appeared to have no influence on CNS recurrences in the landmark GELA trial of R-CHOP versus CHOP, with an overall 5% risk of CNS recurrence reported among elderly patients with DLBCL, regardless of treatment arm.23 However, when the RICOVER-60 trial compared R-CHOP with CHOP in a dose-dense manner, there were modestly fewer recurrences in the rituximab-treated patients noted on univariate analysis (6.9% vs 4.1%), although this was not significant on multivariate analysis. A notable observation in both the GELA and RICOVER-60 trials was that CNS recurrences in R-CHOP-treated patients were predominantly parenchymal, not leptomeningeal, as had been reported before the introduction of rituximab. This intriguing observation raises the question of whether small amounts of rituximab are able to penetrate and protect the leptomeningeal compartment and not the parenchyma. The increase in parenchymal recurrences in modern series further supports the role of intravenous prophylaxis against CNS recurrence in combination with rituximab-containing combination chemotherapy regimens.
We also demonstrate that high-dose methotrexate with concurrent R-CHOP can be administered safely, even to elderly patients. Although renal dysfunction may occur, it is uncommon and almost always self-limited. Patients with renal impairment at baseline are not candidates for this approach, nor are patients with significant effusions or ascites, which serve as a reservoir for methotrexate, resulting in extended toxicity. Careful attention to adequate hydration, alkalinization of the urine, and leucovorin rescue is critical to safe administration of intravenous methotrexate, and methotrexate levels and renal function must be closely followed after drug administration.
This study is limited by its retrospective nature, but to the best of our knowledge there are no prospective studies specifically evaluating the optimal method of preventing CNS recurrence in high-risk DLBCL patients. Patients were selected for inclusion only if they had no evidence of CNS involvement at the time of DLBCL diagnosis. The majority of patients did not have diagnostic LPs, which is consistent with standard of care when LPs are not part of routine staging in DLBCL. Patients with diagnostic LPs demonstrating CNS involvement were excluded from our analysis to focus purely on the question of prophylaxis. It is possible that our CNS recurrence rate was low, in part, due to a better job of excluding patients with occult CNS disease than previously published series due to exclusion of patients with positive flow cytometry of diagnostic CSF, thus making our data appear more favorable. It is likewise possible that some patients with occult involvement of their CSF were included in the analysis because most patients did not undergo diagnostic LP in this series.
Placing our results in the context of the existing data, we suggest that systemic intravenous methotrexate at a dose of 3500 mg/m2 followed by leucovorin rescue be adopted for CNS prophylaxis in high-risk patients, and that IT therapy be considered only for those patients who cannot tolerate systemic therapy. Further investigation is necessary to better understand the optimal dose, quantity, and timing of high-dose systemic methotrexate required to achieve effective prophylaxis against recurrent CNS disease while minimizing potential toxicities associated with this therapy and will optimally be investigated in the context of prospective clinical trials.
CONFLICT OF INTEREST DISCLOSURES
Dr. Hochberg has received consulting and speaking fees from Enzon pharmaceuticals.