Site of central nervous system (CNS) relapse in patients with diffuse large B‐cell lymphoma (DLBCL) by the CNS‐IPI risk model

With the addition of rituximab to cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP), 50–60% of patients with diffuse large B-cell lymphoma (DLBCL) are cured (Sehn & Gascoyne, 2015). For those who relapse, the outcome is poor, particularly for patients with involvement of the central nervous system (CNS) at relapse where the median overall survival (OS) is typically only 3–6 months (Boehme et al, 2009; Yamamoto et al, 2010; Chihara et al, 2011; Schmitz et al, 2012; Tomita et al, 2012). Thus, identification of patients at risk for CNS relapse is critical both to guide diagnostic work-up and develop CNS prophylaxis strategies. The German High-Grade Non-Hodgkin Lymphoma Study Group (DSHNHL) proposed a CNS prognostic model which incorporates the five International Prognostic Index (IPI) factors (age >60 years, elevated lactate dehydrogenase [LDH], stage 3 or 4 disease, performance status [PS] ≥2 and number of extranodal sites [EN] ≥1) as well as kidney/adrenal involvement (CNS-IPI) that stratifies DLBCL patients into three risk groups: low risk (LR), 0–1 factors; intermediate risk (IR), 2–3 factors; and high risk (HR), 4–6 factors, with a 2-year risk of CNS relapse of ≤1%, ~5% and >10%, respectively (Schmitz et al, 2016). This model was successfully validated using a large cohort of DLBCL patients treated with R-CHOP chemotherapy from the British Columbia Cancer Agency (BCCA) (Schmitz et al, 2016). Although this risk model effectively stratifies patients into risk groups, a detailed evaluation of the affected CNS compartment(s) and whether the CNS relapse occurs in isolation or with systemic disease has not been reported. Herein, we evaluated the site of CNS relapse according to the CNS-IPI risk model with a focus on the site of relapse and whether it occurs with systemic disease involvement. The BCCA Lymphoid Cancer Database was used to identify all patients ≥16 years of age diagnosed with de novo DLBCL, without CNS involvement, treated with at least one cycle of curative intent R-CHOP. All diagnostic biopsies were centrally reviewed by BCCA haematopathologists. CNS relapse was confirmed utilizing stereotactic brain biopsy and/or lumbar puncture for cerebrospinal fluid (CSF) analysis and/or new radiological findings, with unequivocal clinical signs or symptoms, consistent with CNS disease. Data were analysed using the Statistical Package for the Social Sciences (SPSS version 14·0 for Windows; SPSS Inc, Chicago, IL). The University of British Columbia/BCCA Research Ethics Board approved the study. A total of 1732 patients with de novo DLBCL treated with at least one cycle of curative intent R-CHOP chemotherapy were included in the final analysis, of which, 1597 were included in the validation study (Schmitz, 2016). The median age was 66 years (range 16–94) with 65% of patients >60 years. Forty-six per cent had an IPI ≥3, 71 (4·1%) patients had kidney/adrenal involvement and 52 (3%) patients had testicular involvement. Bone marrow involvement with DLBCL was detected in 8·4% and low-grade lymphoma in 5%. Median number of R-CHOP cycles received was 6 (range 1–8). The distribution of patients in the LR, IR and HR CNS-IPI risk groups was 519 (30%), 797 (46%) and 416 (24%), respectively. In addition to identified high risk factors constituting the CNS-IPI, patients in the HR CNS-IPI group were more likely to have bone marrow involvement. Testicular involvement was more common in the LR CNS-IPI group as most patients present with limited stage disease. In total, only 48 patients (2·8%) received CNS prophylaxis. With a median follow-up among living patients of 4·3 years (range 0·2–13·6), 84 (4·8%) patients experienced CNS relapse. The 2-year risk of CNS relapse was 1·3%, 4·6% and 12·6% for the LR, IR and HR groups, respectively (P < 0·01) (Table 1). Of those who relapsed in the CNS, median time to CNS relapse was 7·2 months (range 1–96); this was significantly longer for the LR patients compared to IR and HR group (22·7 [range 2·2–96] vs. 7·2 [range 2·5–54] and 6·4 months [range 0·7–22·7], respectively; P < 0·01). Early CNS relapse (defined as recurrence less than 1 year from the date of initial DLBCL diagnosis) was noted more frequently in HR patients (88%) compared to the IR (69%) and LR (33%) groups (P < 0·01). Overall, 47 patients (56%) had an isolated CNS relapse (parenchymal N = 31, parenchymal and leptomeningeal N = 9, leptomeningeal N = 7) and 37 (44%) had concurrent systemic disease involvement. LR patients were more likely to have an isolated CNS relapse (66·7%). Of the 47 patients with an isolated CNS recurrence, only 3 (6·4%) experienced a subsequent systemic relapse. Of those with CNS relapse, 54 (64%) patients were primary refractory: 28 (52%) had synchronous systemic disease and 26 (48%) had isolated CNS recurrence. Focusing on the overall site of CNS relapse, with or without systemic disease, 51 of all 84 CNS relapses (61%), had CNS parenchyma disease (in isolation) and an additional 10 patients had concurrent parenchymal and leptomeningeal involvement: thus, 73% had involvement of the brain parenchyma as part of their CNS relapse. Any leptomeningeal involvement occurred more frequently in HR patients (47·5%, P = 0·06) in comparison to LR and IR groups. With a median follow-up among living patients of 3·9 years (1·2–9·4 years) from the time of CNS recurrence, the median OS from the time of CNS recurrence was only 2·5 months with no difference across the risk groups. Patients with any leptomeningeal disease had a trend towards an inferior 2-year OS compared to those with isolated parenchymal involvement (6% vs. 21%, P = 0·06). The 2-year OS was better in those with an isolated CNS recurrence compared to those with concurrent systemic disease (OS 23% vs. 5%, P = 0·01). We demonstrate that, in DLBCL patients treated with R-CHOP who develop CNS relapse, approximately three-quarters involve the brain parenchyma. In the rituximab treatment era, most CNS relapses occur as an isolated event in the absence of systemic disease. In the HR group, leptomeningeal disease is also frequent and relapses occur early (~ 6 months), warranting CSF analysis, including flow cytometry, at diagnosis. These data support the investigation of CNS prophylaxis with high-dose methotrexate and other emerging novel agents that can penetrate all CNS compartments, especially in the HR group. RK and KJS designed and performed the research; RK and KS analysed the data. All authors interpreted the data, provided patients and research material and wrote the paper.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.