Disease relapse involving the central nervous system (CNS) carries a dismal prognosis in patients with aggressive lymphoma. Risk factors for CNS-relapse have included involvement of lymphoma at specific anatomical sites, proliferation rate and high tumor burden. The risk and importance of CNS-relapse in peripheral T-cell lymphomas (PTCL) has been very little studied. So far, the only study focusing on CNS-relapse in PTCL found elevated LDH and paranasal sinus involvement to be independent risk factors and that CNS-relapse was associated with an inferior survival. Recently, several studies have highlighted the dismal prognosis of relapsed PTCL in general. In this study we aimed at analyzing risk factors for CNS involvement in first relapse and the outcome of these patients, in a large population-based cohort of PTCL patients.

Materials and methods

Through the Swedish Lymphoma Registry we identified all adult (>18 years) patients diagnosed with PTCL between January 1st 2000 and December 31st 2009. Primary cutaneous, primary leukemic types and T-cell lymphoblastic lymphoma were excluded. Information on first relapse was collected from patient files retrospectively after informed consent.


Out of 755 patients identified through the registry, 10 patients had CNS involvement at diagnosis and were excluded from further analyses. Data on first relapse was available for 618, and among these 369 patients experienced relapse or progression. Thirty-one patients (5.0 %) experienced progression or relapse involving the CNS after a median time of 4.3 months (range 1.1-45 months) from diagnosis. Thirteen patients experienced an isolated CNS relapse/progression and 18 as part of a systemic relapse. Sixteen patients relapsed after responding to primary therapy (10 CR, 6 PR), and 15 patients with refractory disease had CNS-progression (1 SD, 14 PD). Histologic subtypes among the 31 patients were ALK-positive anaplastic large cell lymphoma (ALCL) (n=3), ALK-negative ALCL (n=3), ALK-unknown ALCL (n=1), angioimmunoblastic T-cell lymphoma (n=3), peripheral T-cell lymphoma NOS (n=13), enteropathy-associated T-cell lymphoma (n=3), NK/T-cell lymphoma nasal type (n=1), hepatosplenic T-cell lymphoma (n=1) and unclassified T-cell lymphoma (n=1). All 31 patients had received chemotherapy with curative intent and 6 received intrathecal prophylaxis. In multivariable analysis, the presence of B-symptoms (HR 2.79 95% CI 1.14-6.82, p=0.024), >1 extranodal site of disease (HR 3.28[1.41-7.64], p=0.006) and skin involvement (HR 3.06 [1.10-8.49], p=0.032) were independent risk factors for CNS relapse.

Median survival for patients with CNS involvement at relapse or progression was 3.3 months (range 0-124) compared to 3.7 (range 0-129) for patients without CNS involvement and CNS involvement was not a risk factor for OS after relapse/progression (HR 1.08 [0.73-1.59], p=0.700).

For relapsing patients only age was predictive of inferior survival (1.014 [1.002-1.027], p=0.026) whereas in patients with primary refractory disease, male gender predicted inferior survival (2.93 [1.54-5.58], p=0.001) and angioimmunoblastic T-cell lymphoma was associated with better survival (0.41 [0.18-0.95], p=0.037).


CNS involvement, isolated or as part of a systemic first relapse, occurred in 5% of the patients in our series, which is similar to the incidence in larger series dominated by B-cell lymphomas. In our cohort, B-symptoms, >1 extranodal site of disease and skin involvement were independent predictors for CNS-relapse. The outcome of relapsed patients was very poor in general, and CNS-involvement did not convey worse outcome compared to patients with systemic relapse only. With current treatment strategies the uncommon CNS-relapses seem to be a problem of lower magnitude compared to the frequent systemic relapses. Future improvements in the treatment of PTCLs could possibly increase the importance of identifying patients at risk for CNS-relapse, and our study provides data on this poorly described complication.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.