IGH Repertoire Analysis In Multiple Myeloma (MM): Lack of Intra-Disease Homology and Occasional Clustering with Sequences of Other B-Cell Neoplasms Sharing Identical Geographical Origin

Abstract

AbstractAbstract 2951 Background:The identification of stereotyped immunoglobulin (IG) receptors has improved our knowledge on the pathogenesis of several B-cell malignancies, suggesting the role of antigen-driven stimulation in chronic lymphocitic leukemia (CLL), marginal-zone lymphoma (MZL) and mantle-cell lymphoma (MCL). Multiple myeloma (MM) is a terminally-differentiated neoplasm no longer expressing surface IG; however some reports suggest the existence of early B-lymphocyte precursors which could be susceptible to antigen-driven stimulation. IG heavy chain (IGH) repertoire has not been extensively investigated in MM, with the largest available reports containing less than 80 complete sequences. Aims:To address this issue we created a database of MM IGH sequences including our institutional records (mostly derived from minimal residual disease studies) and sequences available from the literature. We planned a two-step analysis: a) first we characterized the MM repertoire and performed intra-MM clustering analysis; b) then we compared our MM series to a large public database of IGH sequences from neoplastic and non-neoplastic B-cells in search of similarities between MM sequences and other normal or neoplastic IGH repertoires. Patients and methods:131 MM IGH genes were amplified and sequenced at our Institutions and belonged to Italian patients, while 214 MM IGH sequences from non-Italian patients were derived from published databases (NCBI-EMBL-IMGT/LIGM-DB) for a total of 345 fully interpretable MM sequences (out of 396). 28590 IGH sequences from other malignant and non-malignant B-cells were retrieved from the same public databases, including approximately 4500 CLL/Non-Hodgkin lymphoma (NHL) sequences and comprising 500 sequences from Italian patients. All sequences were analyzed using the IMGT database and tools (Lefranc et al., Nucleic Acid Res. 2005; http://imgt.cines.fr/) to identify IGHV-D-J gene usage, to assess the somatic hypermutation (SHM) rate and to identify HCDR3. HCDR3 aminoacidic sequences were aligned together using the ClustalX 2.0 software (Larkin et al., Bioinformatics, 2007; http://www.clustal.org/). Subsets of stereotyped IGH receptors were defined according to Stamatopoulos et al. (Blood, 2007). Result:IGHV analysis in MM was almost in keeping with the normal B-cell repertoire, showing a less remarkably biased IGH usage compared to CLL, MCL and MZL (with seven genes accounting for 40% of cases, compared to respectively five, three and two genes). However, a modest but significant over-representation of IGHV1-69, 2–5, 2–70, 3–21, 3–30-3, 3–43, 5–51 and 6-1 genes and under-representation of the IGHV1-18, 1–8, 3–30, 3–53 and 4–34 was noticed. The rate of somatic hypermutation in MM followed a Gaussian distribution with a median value of 7.8%. Intra-MM search for HCDR3 similarities never met minimal requirements for stereotyped receptors. When MM sequences were compared to non-MM database, only a minority of MM sequences (2.6%, n=9) clustered with sequences from lymphoid tumors and normal B-cells (figure 1A). In particular two non-Italian MM sequences clustered with previously characterized, uncommon CLL subsets (n.37 and n.71 according to Murray et al., Blood 2008). Moreover, novel provisional clusters were observed including three MM-CLL subsets, one MM-NHL subset, and three MM-normal B-cell subsets. While the MM-normal B-cell clusters involved non-Italian patients, we unexpectedly noticed that the four MM-CLL/MM-NHL clusters were composed exclusively of Italian patients, as shown in figure 1B, although Italian subjects represented less than 12% of the entire CLL-NHL database. Conclusion:The analysis of the largest currently available database of MM IGH sequences indicates the following: 1) MM IGH repertoire is closer to physiological distribution than that of CLL, MCL and MZL; 2) MM specific clusters do not occur to a frequency detectable with currently available databases; 3) 98% of MM sequences are not related to other “highly-clustered” lymphoproliferative disorders; 4) Uncommon clustering phenomena may follow a geographical rather than a disease-related pattern. [Display omitted] Disclosures:No relevant conflicts of interest to declare.