Incidence and Risk Factors of Central Nervous System Relapse in Localized Extranodal NK/T-Cell Lymphoma, Nasal Type: An Analysis of Nkea Study

Background: More than 65% of patients (pts) with extranodal natural killer/T-cell lymphoma, nasal type (ENKL) have localized disease at diagnosis, and sites of ENKL involvement are confined to the nasal cavity and paranasal sinuses in most pts. Chemoradiotherapy with non-anthracycline-containing chemotherapy (eg, radiotherapy with dexamethasone, etoposide, ifosfamide, and carboplatin; RT-DeVIC) has improved the prognosis of pts with localized nasal ENKL and is regarded as the current standard approach. It is known that pts in the NK/T-cell lymphoma prognostic index (NK-PI) Group 3/4 is at a high risk of central nervous system (CNS) relapse in the CHOP era; however, the incidence and risk factors for CNS relapse with localized ENKL in the non-anthracycline era are not well known. In particular, it is unclear whether CNS relapse is more frequent in the pts with involvement adjacent to the skull base.

Methods: A dataset of our recent study (NKEA Part A, UMIN000015491) was used for analysis. We retrospectively analyzed data from pts with newly diagnosed localized nasal ENKL who were diagnosed at 31 institutes in Japan between 2000 and 2013. Pts with extranasal ENKL, those having distant lymph node involvement and/or CNS involvement at diagnosis, and those who received anthracycline chemotherapy were excluded. The incidence and risk factors for CNS relapse in the following two cohorts were analyzed: pts who received non-anthracycline therapy (n = 189) and pts treated with RT-(2/3)DeVIC (n = 165). The time to CNS relapse in this study was defined as the time from the diagnosis to disease progression with CNS involvement and treatment failure with CNS relapse after complete response. Univariate and multivariate analyses included the following parameters: male sex, age at diagnosis > 60 years, B symptoms, ECOG PS >1, high-risk NK-PI (Group 3/4), the high-risk group of the prognostic index of NK lymphoma (PINK), C-reactive protein > upper limit of normal (ULN), serum soluble IL-2 receptor (sIL-2R) > ULN, detectable EBV-DNA in peripheral blood, lymphocyte count < 1,000 /mL, hemoglobin < 11 g/dL, platelet count < 150 x 10³ /mL, lactate dehydrogenase > ULN, albumin > ULN, the presence of each site of lymphomatous involvement (paranasal sinuses, palate, orbit, gingiva, facial skin, Waldeyer's ring, and regional lymph nodes), CNS prophylaxis (intrathecal administration of methotrexate and/or cytarabine), and the use of RT-DeVIC. In 84 pts, pretreatment CT, MR, or PET-CT was centrally reviewed and evaluated the primary tumor invasion by two radiation oncologists. Subsequently, the relationship between the involvement of paranasal sinus and CNS relapse was analyzed.

Results: Of the 189 pts with localized nasal ENKL who received non-anthracycline therapy, the median age was 55 years (range: 16-83 years) and 34% were > 60 years old. The baseline clinical features were as follows: male, 71%; stage II, 28%; lymph node involvement, 21%; elevated LDH, 27%; ECOG PS > 1, 8%; B symptom present, 34%; elevated serum sIL-2R, 39%; and NK-PI high-risk, 22%. The most common site of involvement was the nasal cavity (93%) followed by the paranasal sinuses (27%), palate (6%), orbit (6%), and gingiva (2%). Fifteen pts received CNS prophylaxis. With a median follow-up of 5.8 years, the 5-year OS and PFS were 74% and 64%, respectively. The 2-year and 5-year CNS relapse rates were 5.1% and 7.2%, respectively. During the follow-up, 12 pts experienced CNS relapse. The NK-PI group 3/4 and elevated serum sIL-2R were not associated with CNS relapse in a univariate analysis. A multivariate analysis identified the involvement of the gingiva (HR, 16.74; 95% CI, 3.49-80.23) and that of the palate (HR, 6.27; 95% CI, 1.66-23.77) as independent risk factors for CNS relapse. In pts treated with RT-(2/3)DeVIC, multivariate analysis identified the involvement of the gingiva (HR, 40.28; 95% CI, 7.25-223.77) and the paranasal sinuses (HR, 5.00; 95% CI, 1.37-18.25) as independent risk factors. In the cohort of pts evaluated by central imaging review, involvement of the paranasal sinus was not significantly associated with CNS relapse.

Conclusion: The involvement to adjacent sites/organs was risk of CNS relapse in pts with localized nasal ENKL who received non-anthracycline therapy. Our findings suggest that the pts with the involvement of the gingiva and/or palate need more effective treatment approaches to reduce CNS relapse.