Patterns of relapse after successful completion of initial therapy in primary central nervous system lymphoma: a case series

Primary central nervous system lymphoma (PCNSL) is a subtype of non-Hodgkin’s lymphoma that involves the brain, spinal cord, or leptomeninges, without evidence of systemic disease. This rare disease accounts for ~ 3% of all primary central nervous system (CNS) tumors. Methotrexate-based regimens are the standard of care for this disease with overall survival rates ranging from 14 to 55 months. Relapse after apparent complete remission can occur. We sought to understand the outcomes of patients who relapsed. This is an IRB-approved investigation of patients treated at our institution between 12/31/2004 and 10/12/2016. We retrospectively identified all cases of PCNSL as part of a database registry and evaluated these cases for demographic information, absence or presence of relapse, location of relapse, treatment regimens, and median relapse-free survival. This analysis identified 44 patients with a pathologically confirmed diagnosis of PCNSL. Mean age at diagnosis was 63.1 years (range 20–86, SD = 13.2 years). Of the 44 patients, 28 patients successfully completed an initial treatment regimen without recurrence or toxicity that required a change in therapy. Relapse occurred in 11 patients with the location of relapse being in the CNS only (n = 5), vitreous fluid only (n = 1), outside CNS only (n = 3), or a combination of CNS and outside of the CNS (n = 2). Sites of relapse outside of the CNS included testes (n = 1), lung (n = 1), adrenal gland (n = 1), kidney/adrenal gland (n = 1), and retroperitoneum (n = 1). Median relapse-free survival after successful completion of therapy was 6.7 years (95% CI 1.1, 12.6). After successful initial treatment, PCNSL has a propensity to relapse, and this relapse can occur both inside and outside of the CNS. Vigilant monitoring of off-treatment patients with a history of PCNSL is necessary to guide early diagnosis of relapse and to initiate aggressive treatment.