Auto-Immune Origin of B Cells from HCV-Associated Lymphoma

Hepatitis C virus (HCV) infection is a well-established risk factor for the development of B cell non-Hodgkin lymphoma (B-NHL). Two main pathways have been proposed to explain the role of the virus in lymphomagenesis: 1) transformation of B cells either directly by oncogenic viral proteins, or by a hit-and-run mechanism inducing a mutator phenotype, and 2) chronic antigenic stimulation through the B cell receptor (BcR) in conjunction to the binding of the viral E2 protein to its receptor CD81. There are however conflicting data regarding whether the BcR of the lymphomatous cells are capable of directly recognizing viral proteins or, rather, they possess a rheumatoid factor (RF) activity, binding instead polyclonal immunoglobulin (IG) within immune complexes trapping the virus.

To further elucidate the role of antigenic stimulation in the natural history of HCV-associated B-NHL, we analyzed the IG gene repertoire of both heavy and light chains expressed by neoplastic cells in 41 cases of HCV-associated NHL, including: 29 marginal zone lymphoma (MZL); 7 diffuse large B-cell lymphoma (DLBCL) of which 3 originated from transformed MZL ; and 5 other low-grade B-cell NHL. Tumor cells were obtained from blood in 39 cases, bone marrow and a lymph node biopsy in 1 case each.

Forty-three productive IGHV-IGHD-IGHJ gene rearrangements were obtained, which displayed a clear biased gene composition as 3 IGHV genes contributed to almost half of the repertoire: IGHV1-69 (11/43, 25,6%), IGHV3-7 (5/43, 11.6%), IGHV3-21 (4/43, 9.3%). This was also true for both IGHD genes (IGHD3-22: 12/43, 27.9%), and IGHJ genes (IGHJ4: 17/43, 39.5%; IGHJ3: 17/43, 25.6%). All but 3 sequences carried somatic hypermutations (SHM) in their IGHV genes with a median identity to the germline of 97.6% (range 86.5-100%).

Thirty-eight productive IG light chain rearrangement sequences were obtained from 36 cases, of which 30 (78.9%) were IGK. Similarly, they exhibited strong gene usage bias with an over-representation of IGKV3-20 (9/30, 30%), as well as IGKJ1 (11/30, 33.3%) and IGKJ2 (6/30, 20%). IG light chain sequences were found to harbor similar SHM load with a 97% median identity to germline (range 91-100%). As previously described, we observed preferential pairing of heavy and light chain genes, with 6 of the 9 IGHV1-69 cases (with available light chain sequence data) being associated with IGKV3-20.

Using established criteria, we found 5 cases (11.6%) carrying stereotyped BcR i.e. IGHV-IGHD-IGHJ gene rearrangements with quasi-identical amino-acid (AA) sequences, including the highly variable complementary determining region 3 (CDR3). Three cases concerned IGHV3-7/IGHD3-22/IGHJ3 rearrangements with an 18 AA-long CDR3, and 2 concerned IGHV1-69/IGHD3-22/IGHJ4 rearrangements with a 13 AA-long CDR3. Identity within the CDR3 extended to AA encoded by randomly inserted N-region nucleotides.

We failed to establish correlations between histological categories and IG repertoire, probably due to the uneven distribution of lymphoma subtypes within our cohort. In contrast, most if not all sequences with biased IG gene usage, including the 5 stereotyped BcR ones, were found amongst the 28/41 cases (68.3%) with mixed cryoglobulinemia type II and/or positive for RF (MC/RF+). These two categories of patients differed also regarding the SHM load of their IGHV genes since MC/RF+ cases were signlificantly less mutated than MC/RF- cases (median identity to germline: 97.9% vs 95.9%, p= 0.048).

We then searched for similar sequences in public databases and collaborative studies. Stereotyped BcR sequences similar to those of our cases were detected in HCV-associated lymphoma, but also in other HCV-negative B-cell maligancies e.g. MALT lymphoma (some associated with RF) and chronic lymphocytic leukemia, non malignant B cells with RF activity, and non malignant marginal zone splenic B cells. Sequence similarity extended to some shared AA replacements, e.g. identical AA introduced by HSM at the same positions.

In conclusion we confirm the highly biased IG repertoire of HCV-associated lymphoma. However this feature seems to be linked essentially to the presence of a MC and/or RF. As quasi-identical sequences are found in HCV-negative malignant and normal B cells, our data support the hypothesis that HCV-associated lymphomatous cells originate from precursors endowed with auto-immune properties rather than B cells expressing an anti-virus BcR.