Diffuse large B-cell lymphoma is the most common adult non-Hodgkin lymphoma, accounting for about a third of all cases. The World Health Organization classification recognizes over 15 subtypes of DLBCL, based on the primary tumor site, specific genetic alterations, or association with specific viruses.
Avyakta Kallam, MD
Diffuse large B-cell lymphoma (DLBCL) is the most common adult non-Hodgkin lymphoma (NHL), accounting for about a third of all cases.1The World Health Organization classification recognizes over 15 subtypes of DLBCL, based on the primary tumor site, specific genetic alterations, or association with specific viruses.2
DLBCL is considered potentially curable, and although central nervous system (CNS) involvement is infrequently discovered at diagnosis, it often occurs within months of initial disease identification. This suggests it may have remained undetected at presentation.3,4CNS relapse is seen in about 5% of patients, ranging from less than 1% in young patients with low-risk disease to up to 35% in patients with a combination of risk factors.5 When it occurs, CNS relapse has an extremely poor prognosis, with a median survival time of a few months.4,5
Standard-of-care chemoimmunotherapy with rituximab (Rituxan [R]) added to cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) has been responsible for high complete response rates in DLBCL.3Nevertheless, the addition of rituximab to CHOP has been a controversial prophylactic measure for CNS disease.5
Therefore, there is great need to develop accurate risk assessment scores and prophylactic strategies for CNS relapse in patients with DLBCL to provide interventions for patients at high risk for this fatal complication and to spare low-risk patients treatment that has little benefit and potential risks.
Patient- and disease-related factors that contribute to an increased risk of CNS relapse in DLBCL and the guidelines that have been developed using these factors to identify high-risk patients are detailed below. Additional prophylactic strategies for reducing the risk of CNS relapse, which to date have not been standardized, are still under investigation.
Scoring systems have been developed to assess the risk of CNS relapse in DLBCL. Effective risk assessment allows for patients at high risk to receive prophylaxis, while sparing low-risk patients exposure to the toxicities of additional therapy.6 Risk assessments have been based on clinical factors, presence of disease in particularly high-risk extranodal sites, and lymphoma molecular biology.7
The National Comprehensive Cancer Network (NCCN) lists the following factors as being associated with an increased risk of developing CNS relapse in patients with DLBCL: >60 years, elevated lactate dehydrogenase (LDH) levels, 2 or more extranodal sites, and involvement of specific extranodal sites, such as kidney or adrenal glands.8Other risk factors may include testicular lymphoma, including stage I disease at diagnosis8; stage IE primary DLBCL of the breast; and HIV-associated lymphoma.8
Six of these clinical factors have been incorporated into a prognostic model, the CNS-International Prognostic Index (IPI), to predict the risk of CNS relapse in DLBCL in patients treated with R-CHOP5,8:
These risk factors can be used to assign patients to 1 of 3 risk groups5,8:
Other negative risk factors for CNS relapse include high- grade B-cell lymphoma with or without translocations ofMYC,BCL2, and/orBCL6.8DLBCL with translocations in MYCplus BCL2and/or BCL6occurs in 5% to 10% of cases and is referred to as double- or triple-hit lymphoma.9Double-hit lymphomas have translocations in both MYCand eitherBCL2or BCL6whereas triple-hit lymphomas have translocations of all 3 genes. The risk for CNS relapse in patients with these characteristics may be as high as 50%.7
Similarly, 1 study found that patients with dual expression ofMYCand BCL2 specifically, also known as dual expressers, by immunohistochemistry had an increased risk of CNS relapse compared with non dual expressers. The 2-year risk for dual expressers was 9.7% versus 2.2% in non– dual expressers (P= .001).10
A multivariate analysis showed that dual-expresser status and CNS-IPI score were the only factors associated with CNS relapse. However, some studies have shown the molecular subtypes of activated B-cell and nongerminal center B-cell–type DLBCL have been associated with an increased risk of CNS relapse.10
Recently, the contribution of the cell of origin (COO) to the risk of CNS relapse has been investigated using gene expression profiling of tissue samples from patients with previously untreated CD20-positive DLBCL who enrolled in the GOYA study, a randomized phase III trial (NCT01287741) comparing the safety and efficacy of CHOP in combination with either rituximab or obinutuzumab (Gazyva).11In this analysis, dual expression of MYC and BCL2 was not associated with an increased risk of CNS relapse. The analysis did show that activated B-celllike (ABC) (HR, 5.2) or unclassifiable COO subtypes (HR, 4.2) were independently associated with an increased risk of CNS relapse and that these factors may confirm the predictive value of the CNS-IPI.7,11The 2-year CNS relapse rate was 15.2% in the highest risk group defined as presence of high CNS-IPI score and ABC or unclassified COO (FIGURE).11
The goal of prophylaxis of CNS relapse is to identify and treat patients with high-risk disease and thereby reduce the incidence of CNS relapse while avoiding the toxicity of treatment in patients at low risk.7
With few exceptions, patients in the low- and intermediate-risk CNS-IPI groups may be spared additional CNS-directed diagnostic and prophylactic procedures. Exceptions to this rule are patients with testicular involvement in the low-risk group and those with kidney or adrenal gland involvement in the intermediate-risk group.5
However, accurate identification of patients who will experience CNS relapse has not been consistently reliable, particularly on an individual patient basis, nor have optimal prophylactic regimens been established. In addition, the use of some risk scoring schema and prophylactic regimens is controversial.6,8There have been no randomized, controlled trials specifically designed to test the efficacy of CNS prophylactic regimens.7
The National Comprehensive Cancer Network (NCCN) recommends CNS prophylaxis for patients at risk for CNS relapse and recommends methotrexate after completion of chemoimmunotherapy. Intrathecal methotrexate is now being replaced by intravenously administered therapy; the latter may be more effective for preventing CNS relapses. The NCCN currently recommends CNS prophylaxis with 4 to 8 doses of intrathecal methotrexate and/or cytarabine or systemic methotrexate at a dose of 3 to 3.5 g/m2.8
According to an article from David A. Qualls, MD, clinical fellow in medicine, and Jeremy S. Abramson, MD, director of the Jon and JoAnn Center for Lymphoma, both at Massachusetts General Hospital Cancer Center in Boston, CNS prophylaxis should be considered for most patients with double- or triple-hit lymphoma, those with a high-risk CNS-IPI score (≥4), those with an interme diate-risk score who have an ABC subtype with dual MYC and BCL2 expression, and those with high-risk extranodal disease, including primary testicular, orbital, or directly infiltrating spinal neuroforaminal. Noting the lack of benefit of intrathecal methotrexate in trials in high-risk patients receiving rituximab-containing regimens, they generally did not recommend this treatment. Instead, they suggested high- dose, intravenous methotrexate on day 15 of alternating 21-day cycles of R-CHOP, acknowledging that toxicity may result in delay of R-CHOP, even discontinuation in some cases.7
The Spanish Lymphoma Group has developed an algorithm for identifying high-risk patients that includes CNS-IPI and genetic risk factors and adds testing for the presence of CNS disease at diagnosis.12The investigators acknowledge wavering reports on the efficacy of intrathecal methotrexate prophylaxis, the standard recommended prophylaxis, and conclude that high-dose intravenous methotrexate alone or in addition to intrathecal methotrexate may be more effective in preventing CNS relapse. There is also some evidence for the efficacy of high-dose intravenous cytarabine. The high-dose methotrexate administration regimen is similar to that of the previous report from Qualls and Abramson, with ≥3 g/m2 given in tandem with primary immunochemotherapy. Alternative intrathecal agents include liposomal cytarabine, which is not licensed for CNS prophylaxis and is supported by limited data; rituximab, which is efficacious at the treatment level but has no supportive data for prophylaxis; or methotrexate, cytarabine, and hydrocortisone triplet therapy, which is a frequently used regimen in Spain despite a lack of studies comparing it with intrathecal methotrexate. A regimen of intrathecal methotrexate or triplet therapy may be considered for selected patients.12
Although not designed as a CNS prophylactic study, the addition of ibrutinib (Imbruvica) to R-CHOP reduced the incidence of CNS relapse to 2.4% compared with 3.8% with placebo plus R-CHOP in a trial of 838 patients with untreated nongerminal center DLBCL. Ibrutinib plus R-CHOP was associated with better outcomes and a similar safety profile in patients <60 years, but with increased toxicity, poorer outcomes, and frequent discontinuation in older patients, the population at highest risk for CNS relapse. The addition of ibrutinib did not improve event-free survival in the overall patient population (HR, 0.934).13
In a chart review study of patients in 2 trials of DLBCL treated with lenalidomide (Revlimid) plus R-CHOP, only 1 of 136 patients developed CNS relapse, yielding an estimated 2-year CNS relapse rate of 0.9% for the entire cohort. Patients were classified into low- (10.3%), intermediate- (71.3%), and high-risk (18.4%) groups using CNS-IPI. In the trial conducted in the United States (n = 87), 1.1% of patients received CNS prophylaxis; in the other trial conducted in Italy (n = 49), 38.8% of patients received CNS prophylaxis. There were more high-risk patients in the Italian group (24.5% vs 14.9%) and no low-risk patients (0% vs 16.1%). Prophylaxis was with intrathecal methotrexate; no systemic prophylaxis was allowed. Lenalidomide was added to R-CHOP because it is known to cross the bloodbrain barrier and has been well tolerated, with clinical activity in aggressive, relapsed/refractory NHL. The estimated 2-year CNS relapse rates for these risk groups in the entire cohort were 0%, 0%, and 5%, respectively. The single patient with CNS relapse had a high CNS-IPI score of 4. The authors suggest further testing of the role of lenalidomide in reducing CNS relapse in patients with DLBCL, as well as the effect of other small molecule agents, as they consider intravenous methotrexate to be expensive, inconvenient, and toxic.14
Other agents that may warrant testing include those targeting PD-1, such as nivolumab (Opdivo), which has shown activity in patients with primary relapsed CNS lymphoma or testicular DLBCL.15
Avyakta Kallam, MD, assistant professor of internal medicine in the Division of Oncology & Hematology at the University of Nebraska Medical Center in Omaha, said that her institution uses the CNS-IPI score to define which groups of patients have high versus low risk for CNS relapse of DLBCL and that patients who score ≥4 high-risk factors are automatically candidates for CNS prophylaxis.
Kallam also stated that because different physicians had their own practice patterns, her institution looked at differences in the guidelines and developed a consensus on which patients should be given prophylaxis. “Regardless of the score, if we have what we call a double-hit lymphoma or a triple-hit lymphoma, we do CNS prophylaxis,” she said.
Her institution also administers CNS prophylaxis to patients who have testicular, epidural, adrenal, or renal involvement. “We consider CNS prophylaxis in patients with HIV-[associated] lymphoma, because these patients have a higher risk for CNS disease,” she said. “We also consider CNS prophylaxis in patients with breast or extensive skin involvement.”
She also pointed out that prophylaxis for CNS disease should not interfere with initial therapy for lymphoma. “Once we identify a patient who is at high risk for CNS disease, we perform a diagnostic lumbar puncture to determine CNS involvement by lymphoma. We also administer 1 dose of intrathecal methotrexate,” she said.
After patients receive that single dose of methotrexate, they are evaluated for additional treatment with high-dose methotrexate, taking into account their renal function, comorbidities, and fitness. “We know that high-dose methotrexate is more superior in preventing CNS relapse than intrathecal methotrexate, so our goal is that we try to give high-dose methotrexate if possible,” Kallam said. Patients who are eligible for high-dose methotrexate receive 6 cycles of standard-of-care chemotherapy such as R-CHOP or rituximab plus etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin—first, followed by 2 cycles of high-dose methotrexate. Although this differs from regimens reported in some studies, Kallam said that this schedule used by her institution minimizes interruption or delay of initial chemotherapy due to methotrexate- induced cytopenias.
Although recent data suggest that high-dose methotrexate is superior to intrathecal methotrexate for prophylaxis, how best to incorporate it into the patient’s current chemotherapy regimen is something that remains to be determined with additional clinical trials.
An open-label, phase III trial currently recruiting patients (NCT02777736), and is randomly assigning participants at risk of CNS relapse to either 2 cycles of prophylactic intravenous methotrexate with doses administered after the third and the sixth doses of systemic chemotherapy or to 1 dose of prophylactic intrathecal methotrexate per each cycle of systemic chemotherapy for 6 cycles. Patients with low risk will not receive CNS prophylaxis; all patients will be observed for relapse for 1 year following the end of first-line therapy.
Kallam said that the CNS-IPI risk model needs improvement and that better criteria are needed. “There has been an increasing interest in circulating tumor DNA [ctDNA] in lymphomas in general. Analyzing circulating tumor DNA might be the next step in determining which patients are at high risk of relapse. By doing that, we might be coming up with more personalized therapy for these patients,” she said.
An ongoing, single-institution, phase II trial is measuring levels of ctDNA in patients with early DLBCL (NCT03758989) to identify prospective markers of treatment failure in patients receiving R-CHOP and as a potential tool for response-adapted therapy. However, this trial does not appear to be assessing the association of ctDNA with CNS relapse.
Another ongoing, phase I trial is using ctDNA to detect early relapse of DLBCL after chemotherapy and the ability of nivolumab to abrogate relapse in patients who test positive for ctDNA (NCT03311958). Again, CNS relapse is not specifically indicated as an outcome measure, although it might be possible to assess this from trial data.
Another potential molecular diagnostic could be based on levels of 2 differentially expressed microRNAs that have been shown to be associated with risk of CNS relapse. Additional testing is required.16
For now, the accurate prediction and prevention of CNS relapse in DLBCL remains an unmet clinical need.