Treatment of Primary CNS Lymphoma in the Elderly with High-Dose Methotrexate Containing Chemotherapy Followed Radiotherapy or Chemotherapy Alone Plus Deferred Radiotherapy: Evaluation of Modification of Treatment Modalities in Leon Berard Cancer Center.

Treatment of Primary CNS lymphoma (PCNSL) is based on high-dose methotrexate (HD-MTX) and cytarabine containing chemotherapy (CT) followed by brain radiotherapy (RT). However, the risk of leukoencephalopathy (LE) is important after chemoradiotherapy, especially for patients older than 60 years old. To reduce neurotoxicity in clinical practice, RT is restricted at relapse or for not responding patients after CT, but no formal randomized trial had evaluated this two modalities of treatment. The objective of this study is to evaluate this modification of clinical practice in our institution over time.

74 patients older than 60, median age 68 years old (range, 60–81) were treated in our institution between 1986 and 2008. Before 2003, 51 patients were treated with HD-MTX CT followed by RT. Most patients (59%) were treated with HD-MTX + CHOP-like with intermediate dose of MTX (1.5g/m²) and only 25% received 3g/m² of MTX per course. Evaluation of this schedule in the prospective phase II GELA trial LNHCP93 showed excessive rate of toxic death during CT. Since 2004, 23 patients were treated with expected less toxic protocol and increased dose of MTX (3g/m², 83% of patients), and RT was omitted for patients in complete response (CR) in order to limit risk of LE. Acute toxicity, initial response, risk of relapse, overall survival (OS) and risk of neurotoxicity were compared between these two periods. As acute treatment toxicity had a powerful impact for these categories of patients especially for patients treated in first period, we access if hazard ratio (HR) was constant over time by determining Schoenfeld residuals. A piecewise Cox proportional model was used in defining time periods after visual inspection of Schoenfeld residuals. Strategy (CT+RT vs.CT only) was tested during each time period in order to detect a difference on survival.

Comparison of clinical characteristics showed that patients in second period were slightly older (71 vs. 67, p=0.008) but Performance Status, LDH level, rate of involvement of deep structures of brain, CSF protein level, leptomeningeal involvement were not different between the 2 cohorts. Toxic death rate during treatment was significantly more important in first period (31% vs. 9% p=0.04). CR after CT was significantly better for patients treated after 2003 (33% vs. 48% p=0.01), but was similar at the end of treatment, CT + RT (49%) vs. CT strategy (48%). The follow-up of cohort 1 and 2 was 95 and 27 months, respectively. Median OS was 16.1 and 18.1 months (p=0.8) and progression free survival (PFS) was 9.7 and 5.1 months (p=0.24) between the 2 cohorts, respectively. With CT + RT strategy, 10 patients did not relapse after first line therapy and eleven were alive at last follow-up. With CT strategy four patients were in persistent CR at 13.5, 14.9, 20.3, 21.5 months, respectively. Among patients in partial response (1 pt), progressive disease (9 pts) or in relapse after CR (7 pts) after CT strategy, 4 died after supportive care, 3 patients were treated with CT alone but died rapidly, and 10 had RT alone (4 pts) or after CT (6 pts). Among patients who received deferred RT, 4 were alive at last follow-up. Regarding neurotoxicity, in first cohort, 7 patients (14%) presented treatment-related neurologic toxicity with 4 deaths of LE. Among patients treated with CT strategy, 2 patients died of LE but received deferred RT at relapse and 2 patients presented severe LE after CT alone and deferred RT, respectively (17% of treatment-related neurologic toxicity). The piecewise Cox proportional model could define two time periods after visual inspection of Schoenfeld residuals, before and after 3 months. We observed that CT + RT strategy presented after the first 3 months a better PFS than CT alone strategy (HR 2.71 [1.29–5.73] p=0.009) but OS was not significant with this method. The cumulative risk of LE in the 2 cohorts was not significantly different (p=0.95)

This institutional study of modification of treatment practice for elderly patients with PCNSL showed that i) MTX + CHOP like CT had an excessive acute toxicity without any benefit in term of CR rates over others schedules with 3g/m² of MTX per course ii) OS of patients was in accordance to recent series with the similar median age iii) 59% of patients treated with CT strategy received finally deferred RT iv) a time varying effect was observed for PFS with after treatment period a benefit for CT + RT. An effort to conduct a randomized trial testing CT alone + deferred RT versus CT + RT for patients with PCNSL remained necessary.

No relevant conflicts of interest to declare.