Secondary Central Nervous System Lymphoma: A Single Center Experience

Secondary CNS non-Hodgkin lymphoma (SCNSL) is estimated to occur in ∼5% of patients with non-Hodgkin lymphoma, and it carries a poor prognosis. Early CNS relapse (i.e. during treatment or within the first 6 months following chemotherapy) may represent subclinical CNS lymphoma present at the time of diagnosis. DLBCL is the most common histology seen in SCNSL and primary CNS (PCNSL) lymphoma. PCNSL DLBCL is usually characterized by an activated B-cell like (ABC) phenotype; it is unknown whether SCNSL follows the same pattern or if it includes germinal center B-cell (GCB) and non-ABC/non-GCB phenotypes.

We queried our IRB approved clinicopathologic database of hematologic malignancies for patients with lymphoma diagnosed between 1999 and July 2011 who had brain or spine MRI, head CT, lumbar puncture or intrathecal chemotherapy. Clinical data and patient characteristics were extracted. We also obtained pathologic features of tumors including immunohistochemistry and cytogenetics, when available. Descriptive statistics were used to characterize patients at baseline. Overall survival (OS) is defined as the time from time from systemic lymphoma diagnosis (i.e. disease outside of the CNS) to death (event), or censored at last known date of survival. Overall survival curves were obtained using the Kaplan-Meier method with 95% confidence intervals calculated using Greenwood's formula. We did an analysis of CNS lymphoma diagnosis within the first 6 months from systemic lymphoma diagnosis (Early CNS involvement) versus greater than 6 months from systemic lymphoma diagnosis (Late CNS relapse), based on Multivariable logistic regression. The associations between potential risk factors and OS after CNS relapse have been explored using the stepwise Cox regression models.

One-hundred and twelve patients met the criteria of systemic lymphoma with CNS relapse/involvement with 68% DLBCL and the remainder including Mantle Cell lymphoma (5%), BL (5%), CLL/SLL (4%) and others. Of the DLBCL patients, 25/76 (33%) were GCB type, 20/76 (26%) were non-GCB, and the remaining unclassifiable with the data available. The median OS for all patients was 23 mo (95% CI=11–41), and 6.7 mo (95% CI=4.6–9.9) after CNS lymphoma diagnosis. Sixty-one of 112 patients (54%) were found to have CNS lymphoma within 6 mo of systemic lymphoma diagnosis including 40 patients (36%) where CNS involvement was present at the time of, or within the first month of, systemic diagnosis, suggesting that these are concurrent presentations. Median OS for patients with early (i.e. concurrent systemic and CNS lymphoma and early CNS relapse patients) v. late CNS relapse was 8.5 mo (95% CI=5.5–10.4) compared to 57.8 mo (95% CI=33.2–94.7), (p < 0.003). The OS after CNS diagnosis was similar between the two groups at 5.5 mo (95% CI=2.85–11) and 8.2 mo (95% CI=3.9–9.7), respectively (p=0.5). Regression analysis of clinical and pathological features (e.g. age, sex, stage, IPI score, LDH, histology, cytogenetics, immunohistochemistry, initial treatment and others) was performed to determine if there were factors associated with early versus late CNS lymphoma involvement and none were significant or clinically relevant. Hazard ratios (HR) for OS after CNS relapse favored patients who obtained a CNS complete response (CR1, HR=0.18, p=<0.01, 95% CI=0.08–0.40) or partial response (PR1, HR=0.31, p=0.004, 95% CI=0.14–0.68). Fifteen patients underwent autologous stem cell transplantation (ASCT) for CNS lymphoma therapy; the median OS after CNS relapse was over 1000 days (HR=0.19, p=0.002, 95% CI=0.07–0.56), suggesting a benefit for intensive therapy in selected patients. HR's for OS after CNS relapse were worse for DLBCL patients with non-GCB phenotype (HR=5.61, 95% CI=2.85–11) and those with an elevated LDH and >1 site of EN disease (HR=2.50, p=0.004, 95% CI=1.34–4.68) at the time of systemic diagnosis, perhaps reflecting more aggressive disease at initial presentation.

The prognosis of secondary CNS lymphoma remains poor, though selected patients who undergo high-dose chemotherapy with ASCT may enjoy prolonged remissions. In contrast to PCNSL, both GCB and non-GCB DLBCL relapse in the CNS, with the non-GCB subtype associated with an inferior prognosis. For patients at high-risk of CNS involvement, efforts must be directed at prophylactic strategies and novel therapeutic approaches.

No relevant conflicts of interest to declare.