Stereotyped Patterns of HCDR3 Sequences in Splenic Marginal Zone B-Cell Lymphoma (SMZL): SMZL-Biased Subsets Are Associated with a Worse Outcome.

Abstract 760

Splenic marginal zone B-cell lymphoma (SMZL) is frequently associated with HCV infection and with autoimmune disorders. Subsets of “stereotyped” Complementarity-Determining Region 3 (HCDR3) sequences are a frequent feature of CLL patients. The aim of this study was to perform clustering analysis of HCDR3 aminoacid (AA) sequences in SMZL. We characterized IGHVDJ rearrangements from 133 SMZL patients diagnosed according to Matutes criteria (Leukemia, 2008) and performed clustering analysis of SMZL HCDR3 AA sequences. Sources for analysis were bone marrow and peripheral blood in 106 cases and spleen tissue + lymph node in 27 cases. SMZL HCDR3 sequences were also compared to 27,413 productive, non redundant HCDR3 AA sequences retrieved from public databases (EMBL, NCBI, IMGT/LIGM-DB; 25,655 sequences) and from our unpublished laboratory databases (1,758 sequences), including sequences from malignant B-cell clones (3,197 CLL, 86 SMZL, 1,131 other B-cell malignancies) and from non malignant B-cell repertoire (22,999 sequences). Alignments were performed using the multiple sequence alignment software ClustalX (2.0). Criteria for assigning HCDR3 AA sequences to stereotyped subsets included AA identity ≥60%, sharing of junction residues and/or usage of the same IGHD-J genes (Stamatopoulos et al, 2007). Male/female ratio of the 133 SMZL patients was 0.7:1; median age at diagnosis was 67 years (40-82 years). HCV serology was positive in 25/125 cases (20%). Median follow-up was 2.7 yrs (range 1-20 yrs); 5-yrs OS for the entire series was 80% and 5-yrs PFS was 39%. The most frequent IGHV families were IGHV3 (n=66, 50%), IGHV1 (n=40, 30%) and IGHV4 (n=23, 17%); the most frequent IGHV genes were IGHV1-2 (n=26, 20%), IGHV 3-23 (n=24, 18%), and IGHV 4-34 (n=10, 8%). Using the 98% identity cut-off value, 40/133 (30%) sequences were “unmutated” (15 with 100% identity, 10 with 99-99.9%, 15 with 98-98.9%). By means of clustering analysis, 16/133 SMZL cases (12%) met the minimal criteria to be included in subsets with “stereotyped” HCDR3. Additional 7/86 (8%) SMZL public sequences were included in subsets. We identified 3 major groups of HCDR3 subsets: 1) “SMZL-biased subsets”: 5 novel subsets were composed only by HCV-negative SMZL cases: 9 from this study cohort and 3 from public databases. Two subsets included both SMZL cases from this study series and SMZL cases from public databases, while 3 subsets were composed by 3 pairs of SMZL from the present study series; 2) “CLL subsets”: 4 SMZL cases (2 from our series) were assigned to 4 subsets previously reported in CLL (subsets 1, 6, 9, 25). In addition, 2 SMZL cases (1 from our series) formed 2 novel subsets with CLL cases; 3) “Non-Hodgkin's lymphoma subsets”: in 5 new subsets HCDR3 of 5 SMZL cases (4 from present series) were homologous to HCDR3 of non-splenic MZL and extranodal indolent lymphomas; in particular 2 HCV+ SMZL clustered with 1 HCV-related lymphoma and 1 HCV+ low grade B-cell lymphoma with splenomegaly, respectively. Distribution of mutated and unmutated cases did not differ between SMZL cases belonging or not to HCDR3 subsets. PFS was statistically shorter for cases belonging to “SMZL-biased subsets” (5-yrs PFS 14%) compared to the remaining cases of the series (5-yrs PFS 41%) (p=0.04). Mutational status was not predictive of OS and PFS. Multivariate analysis selected “SMZL-biased” clustering [hazard ratio (HR), 2.4; 95% CI: 1.003-5.684; p = 0.04], but not mutational status, as predictor of PFS. Sequencing of light chains and 3D modeling of cases belonging to subsets are ongoing. In conclusion, 1) “SMZL-biased” stereotyped HCDR3 sequences appear to be unique to HCV-negative SMZL and suggest a potential role for antigen(s) different from HCV in SMZL lymphomagenesis, and may determine a worse outcome in a subset of SZML patients; 2) only a small number of SMZL cases is related to previously reported CLL subsets; 3) SMZL also share stereotyped HCDR3 sequences with non-splenic MZL and extranodal indolent lymphomas, suggesting a role for a common antigen stimulation in the pathogenesis of these B-cell malignancies.