Initial Treatment Response Is Prognostic and Predictive of Relapse Patterns: Outcomes of >500 Patients Treated for Primary CNS Lymphoma

Introduction: Given concerns for neurotoxicity of standard dose (SD) whole brain radiotherapy (WBRT) after high-dose methotrexate (MTX), consolidation for primary CNS lymphoma (PCNSL) has evolved to either autologous hematopoietic cell transplantation (AHCT) or reduced dose (RD; ≤24Gy) WBRT with cytarabine. We sought to evaluate factors influencing consolidation approach and survival outcomes by consolidation strategy. For a disease that has historically been considered multifocal, we analyzed patterns of relapse in a large cohort of contemporarily treated patients.

Methods: PCNSL patients treated at Memorial Sloan Kettering from 1983-2020 were analyzed. Initial site was determined by T1 post-contrast-enhancing MRI. Patients were categorized by response (complete/unconfirmed [CR/CRu], partial [PR], stable, progression [POD]) to induction and consolidation (AHCT, RD vs SD WBRT, cytarabine, other). Factors impacting consolidation strategy were assessed by multinomial logistic regression models. Site of initial progression was categorized as local (involving/adjacent to initial site) vs distant intraparenchymal, leptomeningeal (CSF and/or MRI), or other. Cumulative incidence of local failure was assessed with death as a competing risk. Progression-free survival (PFS) was assessed from consolidation start with Kaplan Meier and proportional hazards.

Results: Of 645 patients, 559 were eligible for inclusion (diffuse large B-cell, sufficient data, ≥1 MRI response assessment). Median follow-up was 7.4 years. From diagnosis, median PFS was 2.2 years and median overall survival (OS) was 5.7 years. Median age was 63 years (IQR 54-72), and most (57%) were recursive partitioning analysis (RPA) class 2 (age≥50, KPS≥70). Most had supratentorial (420, 81%), multifocal (274, 53%), bilateral (224, 43%), and deep structure involvement (314, 56%). Nearly all received MTX-based induction (532, 95%); common regimens were rituximab, MTX, procarbazine, vincristine (R-MVP; 257, 46%) and MVP (144, 26%). Most (403, 72%) responded to induction (305 [76%] CR/CRu, 98 PR [25%]).

385 patients received consolidation: 116 (21%) WBRT, of which 55 (47%) had RD; 80 (15%) AHCT; 170 (31%) cytarabine; 19 (3%) other. Consolidation approaches evolved over time (SD WBRT 65% in 1980s-1990s vs 3% in 2000s-2020s, RD WBRT 0% in 1980s-1990s vs 18% in 2000s-2020s, AHCT 4% in 1980s-1990s vs 25% in 2000s-2020s; Fisher's exact test, p-value <0.001). RPA and induction response were significantly associated with consolidation approach, while baseline tumor size and leptomeningeal involvement were not. Patients with higher RPA were less likely to undergo AHCT than RD WBRT (RPA 3 vs 1, HR 0.15 [0.04, 0.53], p=0.003). Compared to patients with CR, those with PR were more likely to receive SD WBRT (HR 4.56 [1.70, 12.3] p=0.003) and less likely to receive AHCT (HR 0.34 [0.13, 0.93] p=0.036) than RD WBRT.

Of the 385 patients consolidated, 352 (91%) achieved CR/CRu, of whom 145 (41%) later relapsed. Most relapses after CR to consolidation were multifocal, non-local (1- and 5-year cumulative incidences: 43% and 86%, respectively) as opposed to unifocal local (8% and 16% at 1- and 5-years) or unifocal non-local (31% and 70% at 1- and 5-years). Upon controlling for diagnosis year, induction regimen, RPA, consolidation approach, and the interaction between consolidation and RPA, there was no significant difference in PFS or OS by consolidation strategy (AHCT vs RD WBRT: PFS HR 1.25 [0.31, 5.05], p=0.755; OS HR 0.88 [0.36, 2.19], p=0.788; Figure).

Among those who did not achieve CR to consolidation (POD n=16, PR n=14, stable n=4), patients who progressed had the worst prognosis (median OS: POD 1.04 years, PR 2.7 years, Stable 6.1 years; p=0.005) with 6 (38%) experiencing local only progression. Of the 81 patients with POD on induction, 46 (57%) had local only progression.

Conclusion: In a large PCNSL cohort, modern consolidation (AHCT, RD WBRT) achieved similar survival outcomes upon controlling for clinical variables in patients who achieved CR to consolidation. Performance status, age, and response to induction influenced consolidation strategy and outcomes, suggesting a role for further personalization. While delayed relapses after achieving CR to consolidation were often multifocal and non-local, early progression patterns in refractory patients (with many remaining localized) may suggest a role for focal RT for residual disease.

Scordo:Omeros Corporation: Consultancy, Research Funding; Medscape, LCC (CME): Honoraria; McKinsey & Company: Consultancy; Kite - A Gilead Company: Other: Ad-hoc advisory board (past); i3Health (CME): Honoraria; Angiocrine Bioscience, Inc.: Consultancy, Research Funding; Amgen, Inc.: Research Funding. Schaff:BTG International: Other: Intellectual property rights; Debiopharm: Other: provision of services. Grommes:Scripps Conference Services & CME: Other: provision of services; Ono Pharma: Other: provision of services; Kite Pharmaceuticals: Other: provision of services; Ampressa Therapeutics, Inc: Other: provision of services; BTG International: Other: provision of services. Imber:GT Medical Technologies: Honoraria.