Treatment patterns and outcomes for large B-cell lymphoma patients with central nervous system relapse: A multicenter retrospective study (CoNSensus Study)

Abstract Background: Secondary central nervous system lymphoma (SCNSL) occurs in patients (pts) with large B-cell lymphoma (LBCL) and is associated with poor prognosis. Based on onset timing, it is classified as either de novo (at initial LBCL diagnosis) or relapsed (after treatment of LBCL) SCNSL. A standardized treatment protocol has yet to be established, possibly owing to heterogeneity in CNS involvement patterns, concurrent extra-CNS disease, and variation in patient fitness. The efficacy of high-dose (HD) chemotherapy with autologous stem cell transplantation (HDT/ASCT) on SCNSL outcomes remains unclear. We performed a multicenter retrospective study of pts with de novo or relapsed SCNSL to gather clinical data supporting development of an optimal treatment strategy. Here, we report our analyses of relapsed SCNSL (rSCNSL) pts data. Methods: Records of pts diagnosed with LBCL between 2012 and 2022 who developed rSCNSL after initial treatment for LBCL were obtained from 38 institutions in Japan. Treatment patterns for rSCNSL were categorized according to the first-line salvage treatment received: “intensive treatment”, “radiation therapy (RT)-only”, or “best supportive care (BSC)”. BSC was defined as supportive care ± single agent chemotherapy with palliative intent. Intensive treatment was defined as systemic chemotherapy regimens that were more intensive than those categorized as BSC. Survival outcomes were defined from the time of rSCNSL diagnosis: to death for overall survival (OS), and to relapse or death for progression-free survival (PFS). Results: A total of 455 rSCNSL pts were enrolled in this study. The median age at rSCNSL diagnosis was 70 years (range, 18-91). Most of these pts (89.9%) had DLBCL-NOS. Isolated CNS relapse was observed in 349 pts (76.7%). The remaining pts had concurrent extra-CNS disease. CNS lesions were located in brain parenchyma-only in 67.7%, leptomeninges-only in 13.6%, both in 15.6%, and ocular-only in 3.1% of pts. With a median follow-up of 28.3 months (IQR: 7.6–54.5), median OS of pts who received intensive treatment (330 pts), RT-only (50 pts), and BSC (75 pts) were 20.2, 4.9, and 1.4 months, respectively (p < 0.001). Given the remarkably poor survival observed in pts who received RT-only or BSC, further prognostic analyses were performed in pts who received intensive treatment. In the intensive treatment cohort, 229 pts (69.4%) received rituximab (R) containing therapy. Most pts (272 pts, 82.4%) received HD-MTX based regimens as salvage therapy at first cycle, including HD-MTX ± R (52.2%) and MPV (HD-MTX, procarbazine and vincristine) ± R (46.7%). The remaining 58 pts (17.6%) received non-HD-MTX-based regimens, including HD-cytarabine-based (60.4%), CHOP-like (8.6%) and platinum-based regimens (12.1%). Consolidative HDT/ASCT and CAR T-cell therapy were administrated to 15.5% and 1.8% of pts, respectively. With a median follow-up for survival of 31.0 months (IQR: 9.9–55.7) in the intensive treatment cohort, median OS was 24.9 (95%CI: 15.5 - 32.6) vs. 7.8 (95%CI: 6.4 - 21.7) months in pts who received HD-MTX–based regimens at first cycle vs. those who did not (p = 0.099), and median PFS was 10.6 (95%CI: 7.2 - 12.4) vs. 4.8 (95%CI: 4.2 – 6.8) months (p = 0.037), respectively. In multivariable analysis, non-use of R, older age, elevated serum LDH levels, sites of CNS involvement, refractoriness to prior therapy, and ≥2 prior lines of treatment were identified as adverse prognostic factors, and HD-MTX-based regimens at first cycle were not identified as a favorable prognostic factor for PFS and OS. Pts who underwent HDT/ASCT had significantly better PFS and OS than those who did not (3-year OS: 75.9% vs. 33.7%, p < 0.001). These survival benefits were still observed after restricting the analyses to pts who achieved a partial response or better following first-line salvage therapy. Pts who received busulfan and thiotepa (BuTT) conditioning (n = 23) showed better outcomes than those who received non-BuTT conditioning (n = 28) (3-year OS: 87.5% vs. 66.0%, p = 0.033). Conclusion: Although HD-MTX-based regimens were most frequent, this study revealed heterogeneity in first-line salvage therapies for rSCNSL, highlighting the need for optimal tretament strategies. Addition of R and consolidative HDT/ASCT (particularly BuTT) might be associated with improved OS and PFS.