No standard chemotherapy regimen exists for primary CNS lymphoma, reflecting an absence of randomised studies. We prospectively tested two promising methotrexate-based regimens, one more intensive and a milder regimen, for primary CNS lymphoma in the elderly population, who account for most patients.

In this open-label, randomised phase 2 trial, done in 13 French institutions, we enrolled immunocompetent patients who had neuroimaging and histologically confirmed newly diagnosed primary CNS lymphoma, were aged 60 years and older, and had a Karnofsky performance scale score of 40 or more. Participants were stratified by Karnofsky performance scale score (<60 vs ≥60) and treating institution and randomly assigned (1:1) to receive methotrexate (3·5 g/m(2)) with temozolomide (150 mg/m(2)) or methotrexate (3·5 g/m(2)), procarbazine (100 mg/m(2)), vincristine (1·4 mg/m(2)), and cytarabine (3 mg/m(2)). Neither regimen included radiotherapy; both included prophylactic G-CSF and corticosteroids. The primary endpoint was 1-year progression-free survival. Analysis was intent to treat, in a non-comparative phase 2 trial design. This study is registered with ClinicalTrials.gov, number NCT00503594.

Between July 16, 2007, and March 25, 2010, 98 patients were enrolled, of whom 95 were randomly assigned and analysed; 48 to methotrexate with temozolomide and 47 to methotrexate, procarbazine, vincristine, and cytarabine. 1-year progression-free survival was 36% (95% CI 22-50) in the methotrexate, procarbazine, vincristine, and cytarabine group and 36% (22-50) in the methotrexate with temozolomide group; median progression-free survival was 9·5 months (95% CI 5·3-13·8) versus 6·1 months (3·8-11·9), respectively. Objective responses were noted in 82% (95% CI 68-92) of patients in the methotrexate, procarbazine, vincristine, and cytarabine group versus 71% (55-84) of patients in the methotrexate with temozolomide group. Median overall survival was 31 months (95% CI 12·2-35·8) in the methotrexate, procarbazine, vincristine, and cytarabine group and 14 months (8·1-28·4) in the methotrexate with temozolomide group. No differences were noted in toxic effects between the two groups. The most common grades 3 and 4 toxicities in both groups were liver dysfunction (21 [4%] in the the methotrexate and temozolomide group and 18 [38%] in the methotrexate, procarbazine, vincristine, and cytarabine group), lymphopenia (14 [29%] and 14 [30%]), and infection (six [13%] and seven [15%]). To date, 33 (69%) patients in the methotrexate and temozolomide group have died, versus 31 (55%) in the methotrexate, procarbazine, vincristine and cytarabine group. Quality-of-life evaluation (QLQ-C30 and BN20) showed improvements in most domains (p=0·01-0·0001) compared with baseline in both groups. Prospective neuropsychological testing showed no evidence of late neurotoxicity.

In this study of two different methotrexate-based combination regimens in elderly patients, the efficacy endpoints tended to favour the methotrexate, procarbazine, vincristine, and cytarabine group. Both regimens were associated with similar, moderate toxicity, but quality of life improved with time, suggesting pursuing treatment in these poor prognosis patients is worthwhile. New alternatives are needed to improve response duration in this population.

Schering-Plough/Merck and French Government.