The Composition of the B Cell Receptor Repertoire In 7428 Cases of Chronic Lymphocytic Leukemia: One Third Stereotyped, Two Thirds Heterogeneous - What Does This Mean?

Andreas Agathaggelidis, Nikos Darzentas, Anastasia Hadzidimitriou, Xavier Brochet, Fiona Murray, Xiao-Jie Yan, Zadie Davis, Ellen van Gastel, Cristina Tresoldi, Charles C Chu, Nicola Cahill, Veronique Giudicelli, Boris Tichy, Lone Bredo Pedersen, Letizia Foroni, Paola Francia di Celle, Agnieszka Janus, Karin Smedby, Achilles Anagnostopoulos, Helene Merle-Beral, Nikolaos Laoutaris, Gunnar Juliusson, Sarka Pospisilova, Jesper Jurlander, Athanasios Tsaftaris, Marie-Paule Lefranc, Anton W. Langerak, David Graham Oscier, Nicholas Chiorazzi, Chrysoula Belessi, Frederic Davi, Richard Rosenquist, Paolo Ghia and Kostas Stamatopoulos


Abstract 43

Chronic lymphocytic leukemia (CLL) is characterized by the existence of subsets (clusters) of cases with restricted, “stereotyped” immunoglobulin (IG) variable heavy complementarity-determining region 3 (VH CDR3) sequences within their B cell receptors (BcR), suggesting selection by common epitopes or classes of structurally similar epitopes. Emerging evidence indicates that the grouping of CLL cases into distinct clusters with “stereotyped” BcR is functionally and prognostically relevant. Further than that, several issues remain open: (i) the refinement of criteria for identification of BcR stereotypy and cluster assignment; (ii) the true frequency of BcR stereotypy; (iii) the total number of clusters and relative size of each; and, (iv) the identification of “CLL-biased” features in BcR stereotypes. To address these issues, we systematically examined VH CDR3 stereotypy in 7596 IGHV-D-J sequences from 7428 patients with CLL (168 cases, 2.2%, with two productive sequences), three times the size of the largest published series. Recent studies in both normal B cells and other (non-CLL) B cell malignancies along with accumulated experience in our group led to an advanced clustering bioinformatics algorithm applying more stringent criteria than before. A novel parameter was also included; the usage of IGHV genes, which takes into account the role of the germline-encoded specificities in (super)antigen recognition. The algorithm assigns sequences in a cluster only if exhibiting >50% amino acid identity and >70% amino acid VH CDR3 similarity and also carrying IGHV genes that share common ancestry and, thus, belong to the same IGHV phylogenetic clan. To increase the likelihood that cluster assignment reflects actual structural relatedness, we also required that each cluster consisted only of sequences with identical VH CDR3 length and identical offsets of common patterns. Following this new approach, 2308/7596 (30.4%) CLL sequences were assigned to 952 different ground-level clusters with shared patterns and unique characteristics, each containing 2 to 56 cases. Different types of VH CDR3 patterns were identified, independent of mutational status, as “mainly germline”, i.e. deriving from restricted associations of specific IGHD and IGHJ genes, and “junctional+germline”, i.e. extending over V-D and/or D-J junctions as well. In several clusters of mutated sequences, the cluster-defining features were ubiquitous junctional residues. Common sequences among ground-level clusters enabled grouping into clearly delineated, higher-order (HO) clusters that were considerably larger in size and displayed ‘CLL-biased’ features with regard to: IGHV gene usage, somatic hypermutation (for clusters with mutated sequences) and VH CDR3 pattern composition. As an example, the largest HO cluster, including 213 sequences (2.8% of the cohort), utilized the IGHV3-21 gene with an acidic residue at VH CDR3 position 107 (3 of 9), while the second-ranking HO cluster, including 184 sequences (2.1% of the cohort), utilized different IGHV genes of Clan I (e.g. IGHV1-2, 1–3, 1–8, 1–18, 5-a, 7-4-1) with a QWL motif at VH CDR3 positions 108–110 (4-6 of 13). Based on random set simulations (using the actual sequences) and starting from a critical mass of 2000 cases, each increase of the total set by a 1000 random cases resulted in an increase in the percentage of stereotypy to ∼2% (i.e. from 21% in 2000 cases to 25% in 3000 cases to 30% in 7000), though not proportional to the increase of the cohort. Perhaps most important, however, was the finding that the percentage of sequences in known major clusters was remarkably stable compared to previous studies on smaller series. These results strongly indicate that not all CLL belong to stereotyped subsets even if the cohort size is increased significantly, corroborating our previous hypothesis that CLL consists of two distinct categories, one with stereotyped and the other with heterogeneous BcR, likely of different ontogenetic origin. Furthermore, they demonstrate that the major clusters collectively represent a sizeable proportion of the cohort. Consequently, this deeper, more robust, compartmentalized examination of BcR structures in association with other biological and clinical information may eventually pave the way for the introduction of specialized treatment protocols applicable to a significant number of patients assigned to the same cluster.

Disclosures: No relevant conflicts of interest to declare.