Excellent outcomes in children and adolescents with CNS+ Burkitt lymphoma or other mature B‐NHL using only intrathecal and systemic chemoimmunotherapy: results from FAB/LMB96 and COG ANHL01P1

Abstract

Burkitt lymphoma (BL) is the most common Non-Hodgkin lymphoma (NHL) in children, representing 40–50% of paediatric NHL (Cairo et al, 2012). BL is frequently advanced, involving the bone marrow (BM), central nervous system (CNS) or both, which requires aggressive therapy. BL is also the most frequently CNS-positive (CNS) paediatric NHL (9–13% of cases) (Cairo et al, 2007, 2012; Patte et al, 2007; Gerrard et al, 2008). Trials have strived to improve outcomes in CNS BL and other mature B cell NHL (B-NHL), where prognoses are inferior (Salzburg et al, 2007). Specifically, CNS patients who were also BM-positive (BM) had the worst outcomes on the French-American-British (FAB)/Lymphome Malins B (LMB)96 trial (Cairo et al, 2012). Yet most patients do achieve long-term survival, so studies also aim to reduce therapy-induced sequelae. The international FAB/ LMB96 cooperative group trial previously demonstrated that CNS B-NHL patients treated with intensified CNS-directed systemic and intrathecal (IT) therapies had similar event-free and overall survival (EFS, OS) to prior CNS radiotherapycontaining regimens (Cairo et al, 2007). Besides altering CNS-directed treatment, FAB/LMB96 also randomized standardversus reduced-intensity arms: reduced-intensity was proven inferior (72% vs. 84% EFS) (Cairo et al, 2007). Here, we focus CNS patients treated on the standard FAB/LMB96 C1-arm. Rituximab has proven beneficial in adult B-NHL, so shortly after FAB/LMB96, we investigated adding rituximab to the C1-regimen in the Children’s Oncology Group (COG)ANHL01P1 trial (Goldman et al, 2014). Here, we present high-risk CNS patient data from both the FAB/LMB96 C1arm and COG-ANHL01P1 and compare them to prior LMB89 results. Overall, excellent EFS and OS were achieved, with further improvement with rituximab. Clinical trial information and approvals, CNS-defining criteria, chemotherapy details and statistical methods are listed in Data S1. Forty-four CNS mature B-NHL patients were treated on FAB/LMB96-C1 (Cairo et al, 2007), with 84% having BL or BL-like histology and 52% dually CNS/BM. Fifteen COGANHL01P1 patients were CNS, 100% with BL and 53% CNS/BM. Thus, these results pertain to BL, because 88% of patients shared this pathological diagnosis (P = 0 45). No significant differences in age, gender, or BM/CNS status were seen between the patient groups (Table I and data not shown), with both cohorts mostly male (83%) and aged ≤14 years (88%). The types of CNS patients on both trials were also similar, with 55% (FAB/LMB96-C1) vs. 53% (COG-ANHL01P1) cerebrospinal fluid-positive (CSF), 36% vs. 33% isolated CSF, and 36% vs. 33% dually BM/CSF. Each group also contained ~13% of patients with isolated cranial nerve palsies. After correcting for the inferior arm of the FAB/LMB96 CNS patients, estimated 4-year EFS was 75 4 5% (FAB/ LMB96-C1; Table II); versus 93 3% on COG-ANHL01P1, with EFS in BM/CNS and BM/CSF subgroups likewise similar (data not shown). Overall, both trials illustrate that children, adolescents and young adults with high-risk CNS mature B-NHL have outstanding outcomes when CNS radiation is replaced with CNS-directed systemic and IT chemotherapy (Cairo et al, 2007; Goldman et al, 2014). Prior studies like LMB89 used similar regimens, but with cranial radiation in CNS patients. These results in CNS patients show that CNS-directed systemic (high dose methotrexate, high dose cytarabine) and IT therapy (13 doses) can limit neurotoxicity by eliminating irradiation, yet still offer superior EFS and OS. The COG-ANHL01P1 study added rituximab to therapy that was almost identical to FAB/LMB96-C1, but reduced the doxorubicin infusion time. Despite an estimated 20% improvement in 4-year EFS in CNS patients on COGANHL01P1 versus FAB/LMB96, this was not statistically significant due to the small sample size of COG-ANHL01P1. However, directly comparing these trials is complex, because these studies had key differences: (i) FAB/LMB96 randomized patients after the cytoreduction and induction cycles, thus excluding refractory/progressing patients and those who died of toxicity prior to randomization. This over-estimates FAB/LMB96 outcomes. (ii) Conversely, the reduced therapy arm of FAB/LMB96 demonstrated inferior outcome. Therefore, it would be inappropriate to compare the addition of rituximab to standardversus reduced-intensity chemotherapy of FAB/LMB96. The actual 4-year EFS of all CNS patients enrolled (not just randomized) on FAB/LMB96 was 70 4 3%. Statistical analysis of FAB/LMB96, once corrected for the inferior arm of therapy, estimated the 4-year