Integrated Analysis of IGHV Gene Status, Cell-of-Origin Signature and Genomic Features in Diffuse Large B-Cell Lymphoma

The molecular heterogeneity of diffuse large B-cell lymphoma (DLBCL) has been highlighted by gene expression profiling (GEP), dividing DLBCL into the following three main molecular subtypes with different clinical outcomes and responses to immunochemotherapy: the germinal center B-cell-like (GCB) subtype, the activated B-cell-like (ABC) subtype and the primary mediastinal B-cell lymphoma (PMBL) subtype. Despite frequent translocations involving the IGH (heavy chain) locus, B-cell receptor (BCR) expression is retained in almost all DLBCL, indicating that BCR signaling is crucial for the survival of malignant cells. Furthermore, DLBCL are characterized by recurrent somatic mutations that are supposed to be oncogenic drivers. Here, with the aim to more accurately define the DLBCL pathogenic subgroups, we report a detailed immunogenetic analysis of IGH rearrangements in a well-defined GEP DLBCL cohort and correlate these features with the somatic mutation status of recurrently mutated genes involved in lymphomagenesis. Methods: 204 DLBCL enrolled in the prospective LYSA LNH03 trial program were classified into molecular subtypes by GEP using Affymetrix arrays [82 ABC (40.2%), 77 GCB (37.7%), 29 (14.2%) unclassified and 16 PMBL (7.8%)]. The amplification of complete VDJ rearrangements was realized using BIOMED-2 protocols. Somatic hypermutation (SHM) characteristics were studied (mutation rate, glycosylation N-X-S/T sites, RGYW hotspots, composition of CDR3) and correlated with the molecular subtypes. To obtain a more comprehensive molecular portrait of the DLBCL cases, GEP and IGH VDJ analyses were correlated with the mutational status of a panel of 34 recurrently mutated genes in DLBCL such as MYD88, EZH2, CD79B, TNFAIP3, CARD11 and GNA13 (Dubois et al. Clinical Cancer Research 2016). Results: A total of 153/204 (75%) clonal VDJ rearrangements were successfully determined. Failures to detect clonotypic sequences were predominantly seen in PMBL (56%) and unclassified (31%) cases. The ABC subtype display a significantly lower SHM rate than the 3 other subtypes, especially in contrast to the PMBL subgroup that was characterized by a higher SHM rate (p<0.0001 for each). While IGHV gene usage generally reflected that of normal B cells, a marked increase in IGHV4-34 in ABC cases (30%) was confirmed and those cases tended to have less SHM. Acquisition of new N-glycosylation (N-Gly) sites, a hallmark of follicular lymphoma, was more prevalent in the GCB cases (48%) than in ABC (22%) (p=0.012) but was also observed in 5/7 (71%) PMBL cases with functional IGH rearrangement. This acquisition is correlated with a more frequent t(14;18)(q24;q32) in GCB subtype. As reported in FL, this suggests that lectin binding signals via surface Ig that contain N-Gly sites may provide an extrinsic antigen-independent signal in some DLBCL, and more specifically in t(14;18) DLBCL (Linley et al. Blood 2015 ). Finally, we correlated these immunogenetic features with the mutational status of some drivers genes frequently mutated in DLBCL. Within ABC subtype patients, MYD88 or TNFAIP3 mutated cases displayed a higher SHM rate (13.4% vs 8.4%, p <0.0001; 15.44% vs 11.13%, p=0.03), without any IGH family distribution bias.9/56 (16%) GCB DLBCL harbored an EZH2 mutation. All expressed an IGHV3 rearrangement (100% vs 52% for the GCB-EZH2 wt, p=0.014). By contrast to MYD88 status, a similar IG SHM rate was observed in mutated and wt patients regarding EZH2, CARD11 or GNA13 genes. Conclusion: The IG characterization and the driver gene mutation analysis according to the cell of origin of the tumor allowed us to refine DLBCL subgroups. Our data suggest that there is a strong interaction between extrinsic mechanisms, including lectin-mediated antigen-independent activation or, in the IGHV4-34 cases, via N-acetyl lactosamine ligands, and intrinsic mechanisms to activate the BCR/NFKB pathway. These data could help us classify the different subgroups of DLBCL more accurately and, in particular, to identify new molecular features for BCR target therapy.