Background: Treatment of primary CNS lymphoma (PCNSL) is based on high-dose methotrexate (HD-MTX) containing chemotherapy (CT) followed by brain radiotherapy (RT). Initial CT allowed 30% to 63% of complete response (CR) in recent large series. After CT, consolidation RT can increase the CR rate up to 80%. Despite this high rate of response after initial treatment, the outcome of patients remained poor. The impact of the quality of response on outcome is not well known as well as the outcome of PR patients who converted to CR after RT. We assessed these questions in patients with newly diagnosed PCNSL treated with HD-MTX-containing CT followed by RT in the prospective LNHCP93 GELA study.
Methods: 99 patients were treated in this prospective phase II study between 1995 and 2002. Patients younger than 61 years received C5R protocol (Blay et al. Blood 1995), Patients aged 61-70 years received reduced doses of C5R protocol and patients older than 70 years received a specific schedule with MTX, vepeside and cyclophosphamide. After CT, brain RT was planned: 20 Gy whole brain and a 36 Gy boost to the tumor bed. Responses after CT and after RT were evaluated by MacDonald criteria. Evaluation of response was made at time of the beginning of RT, 21-35 days after the last course of CT, and one month after the end of RT.
Results: Median age of the 99 PCNSL patients was 63 years (range, 20-82), 51% were male, 51% had performance status >1, and 58% had involvement of deep structures of brain. Forty-five patients were younger than 61 years, 36 were aged 61-70 years and 18 older than 70 years. After a median follow-up 83 months, median overall survival (OS) and progression-free survival (PFS) were 33 and 20 months, respectively. Seventeen patients (17%) died of acute toxicity during CT; 3 patients (3%) did not receive RT; 8 patients (8%) progressed or had stable disease after CT and 3 patients (3%) had no available data. Thus, 68 patients were assessable for this exploratory study with thirty-six patients (36%) in PR and 32 patients (32%) in CR after CT. Sixteen of PR patients converted to CR after RT (44% of PR patients after CT). Median OS of patients in CR and PR after CT was 80 and 34 months with a 5-year OS probability of 65% and 29%, respectively (p=0.02). Median PFS of patients in CR and PR after CT was 60 and 21 months with a 5-year PFS probability of 56% and 17%, respectively (p=0.03). In univariate and multivariate analysis, age and response were the two prognostic factors for OS but not performance status, number of tumors at diagnosis, site of tumor (involvement of deep structures). Only response to CT was predictive of PFS in multivariate analysis but not age, performance status, number of tumors, site of tumor at diagnosis. 5-year OS was 65% for CR patients before RT compared to 31% and 28% for PR patients who converted to CR after RT and for patients not in CR after RT, respectively (p=0.06). The 5-year PFS probability was 56% for CR patients before RT compared to 13% and 20% for patients who converted to CR after RT and those not in CR after RT, respectively (p=0.09).
Conclusion: Despite the inherent bias of response analysis as a prognostic factor, this analysis of a prospective study of PCNSL patients showed that only patients achieving CR after CT may experience long term survival. This study also showed that PR patients who converted to CR after RT had a poor outcome, similar to patients that did not reach a CR after chemoradiotherapy.