Primary central nervous system lymphoma: age and performance status are more important than treatment modality

To assess prognostic factors and treatment modalities of patients with primary central nervous system lymphoma (PCNSL) in terms of response rates, patterns of failure and overall survival.

Sixty-two patients presenting with PCNSL between 1982 and 1994 at Peter MacCallum Cancer Institute with no evidence of human immunodeficiency virus infection were included in the study. Their median age was 60 years; World Health Organisation (WHO) performance status was > or = 2 in 85%. All patients were planned to receive whole brain irradiation; 7 also received spinal irradiation. The median planned dose to the target volume was 50.4 Gy. Twenty patients were planned to receive chemotherapy as well. Patients were followed up to June 20, 1995, giving a median follow-up for 14 surviving patients of 5.4 years, range 0.3 to 10.2 years.

The clinical response rate to treatment was 77% [95% confidence interval (CI) 65 to 87%]. The estimated median overall survival was 20.6 months (CI 12.4 to 33.4 months). On univariate analysis male gender, age <60 years, WHO performance status < or = 1, treatment to the target volume > or = 45 Gy, and treatment with additional chemotherapy, were associated with a significantly better overall survival (p < 0.05). On multivariate analysis only age and performance status remained significant prognostic variables. Relapse involved the central nervous system or cerebrospinal fluid (CSF) in all patients with known sites of relapse except three who had ocular relapse only. There was a low incidence of relapse in the initial brain site (23% of known cases) and a high incidence (50%) of CSF/spinal cord relapse. Of 48 deaths, 15 were related to initial or subsequent treatment.

Patient outcome is strongly influenced by age and performance status. Studies suggesting better survival for patients treated with chemoradiation may reflect patient selection rather than treatment variables. Optimal management remains to be defined. The high CSF/spinal relapse rate deserves particular attention.