Secondary central nervous system lymphoma (SCNSL) refers to lymphoma within the CNS that originated outside the CNS. The term SCNSL does not refer to a specific histologic subtype and does not represent a distinct biologic entity. DLBCL accounts for most cases and is the focus of this article.
Secondary CNS involvement always confers a poor prognosis with standard approaches, and most patients die from lymphoma. The prognosis for patients with CNS relapse after frontline therapy is especially poor, with a median overall survival of less than six months.
There is no consensus on the optimal approach to SCNSL and practice patterns vary. This article describes current practices for prevention and management of SCNSL, highlights fundamental challenges, and discusses emerging strategies.
Standard Management of SCNSL
Several factors make treating SCNSL especially challenging. One core issue is that combination chemotherapy regimens that cure DLBCL are composed of agents that do not reliably cross the blood-brain barrier and therefore do not result in durable remissions when lymphoma involves the CNS. Secondly, the current cornerstone of CNS lymphoma treatment, high-dose methotrexate (HD-MTX), has excellent CNS penetration but is not particularly effective for aggressive B-cell lymphomas.
To address the challenge of delivering highly active chemotherapy across the blood-brain barrier, researchers have pursued high dose-intensive chemotherapy regimens containing antimetabolites and CNS-penetrating alkylating agents followed by autologous hematopoietic stem cell transplant (AHSCT), a strategy that can only be applied to younger, fit patients.
One regimen, MATRix-R-ICE (HD-MTX 3.5 g/m2, intravenous rituximab, high-dose cytarabine, and thiotepa plus rituximab, ifosfamide, carboplatin, and etoposide), consists of three cycles of MATRix, three cycles R-ICE, intrathecal chemotherapy, and a carmustine/thiotepa-conditioned AHSCT in eligible patients.
The recent phase II MARIETTA trial evaluated MATRix-R-ICE in 75 patients with DLBCL involving the CNS. Among the patients, 49 (65%) responded to induction therapy, but only 37 were able to undergo AHSCT. Patients who underwent AHSCT had a two-year overall survival of 83%, highlighting a subset of patients with SCNSL who may achieve meaningful remissions with dose-intensive approaches.
Patients with disease that did not respond to MATRix derived no benefit from subsequent R-ICE and AHSCT, which suggests that while intensive chemotherapy may address the challenge of drug delivery imposed by the blood-brain barrier, it does not adequately overcome chemotherapy resistance.
Another limitation of a dose-intensive chemotherapy approach is that it excludes older and frail patients due to prohibitive hematologic toxicity.
Standard Approach to CNS Prophylaxis
Because CNS relapse is associated with dismal outcomes, the standard approach to SCNSL emphasizes prevention. This requires identifying patients at high risk of CNS relapse and delivering CNS prophylaxis along with frontline therapy.
Several multivariate risk models have been developed to integrate clinical and biologic factors and guide selection of patients for CNS risk stratification and/or prophylaxis. The most widely used among them is the CNS International Prognostic Index.
For patients selected to receive CNS prophylaxis, the standard options are intrathecal (IT) chemotherapy and/or intravenous HD-MTX. However, neither method has been studied in a randomized, prospective clinical trial, and all evidence is hampered by a small number of events and variable selection criteria, treatment protocols, and participation in these protocols; many patients who are eligible for prophylaxis do not receive it.
There are two central issues within the current paradigm for preventing CNS relapse: the lack of reliable models for predicting relapse and the available prophylactic therapies are not universally effective. Another issue may be limitations of the paradigm itself; in DLBCL, IT chemotherapy and HD-MTX are intended to prevent isolated CNS relapses by treating occult CNS disease at diagnosis but cannot be expected to prevent relapses caused by the spread of uncontrolled, chemotherapy-resistant systemic disease. Given that 50% of CNS relapses present with synchronous systemic relapse, the latter scenario may account for a significant portion of relapse events.
Novel Treatments and Future Directions
In activated B-cell (ABC) subtype DLBCL, aberrant B-cell receptor (BCR) signaling (termed “chronic active BCR signaling”) is highly dependent on the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway. A key enzyme in this pathway is Bruton’s tyrosine kinase (BTK), and ABC subtype DLBCL in both the periphery and the CNS is uniquely sensitive to the irreversible BTK inhibitor ibrutinib. However, aggressive lymphomas quickly develop resistance to ibrutinib monotherapy. To improve the rate of durable responses, ibrutinib-based combination regimens are being explored.
The immunomodulatory agents lenalidomide and pomalidomide also impair NF-κB pathway activation, have preferential activity in ABC subtype DLBCL, and demonstrate efficacy in CNS lymphomas. Future studies aim to combine targeted therapies.
Another promising treatment option for SCNSL is T-cell-directed therapy, including chimeric antigen receptor T cells targeting CD19. However, data are lacking since the pivotal trials of both axicabtagene and tisagenlecleucel in aggressive B-cell lymphomas excluded patients with CNS involvement.
In addition to novel treatments, the introduction of highly sensitive circulating tumor DNA into translational research and clinical practice has shown utility in primary CNSL and may improve future SCNSL management as well.
The prevention of SCNSL remains an unmet clinical need, and the standard approaches of delivering MTX either as IT therapy or as HD-MTX during frontline therapy are largely ineffective. Further, the management of patients who develop SCNSL at diagnosis or after frontline therapy often requires highly intensive chemotherapy, including AHSCT, which only applies to select patients.
Novel treatments including targeted pathway inhibitors and immunotherapy approaches are being studied, but patients with CNS involvement are often unnecessarily excluded from clinical trials testing novel approaches. Noninvasive biomarkers including cell-free DNA from cerebrospinal fluid or peripheral blood may prove to be a useful adjunct to current risk-stratification tools. Prospective studies testing novel approaches to SCNSL with strong translational emphasis on molecular correlates of response are urgently needed.