To the Editor:

Hepatitis C virus (HCV) infection has been found to be associated with extrahepatic immune manifestations such as mixed cryoglobulinemia (MC) and B-cell non-Hodgkin lymphoma (B-NHL), mostly marginal zone lymphoma (MZL), de novo or transformed diffuse large B-cell lymphoma (DLBCL) [1]. Two main pathways have been proposed to explain the role of the virus in lymphomagenesis: (1) chronic antigenic stimulation through the B cell receptor (BCR) leading from oligoclonal to monoclonal expansion and ultimately to overt lymphoma, mostly of marginal zone subtype or (2) transformation of B cells either directly by oncogenic viral proteins, or by a hit-and-run mechanism inducing a mutator phenotype, particularly in de novo large B-cell lymphoma [1]. In recent years, direct-acting antiviral agents (DAA) have demonstrated their efficacy to obtain sustained HCV eradication in chronic hepatitis C virus infection but also to induce remission in MC and indolent low-grade NHL. Thus, in this subset of patients, HCV treatment has been recommended as first-line therapy without chemotherapy [2]. Regarding high grade lymphoma, optimal management has yet to be determined and combination of DAA with chemotherapy is still under evaluation [3].

In this context, unraveling the molecular oncogenesis of HCV-related B-cell lymphoproliferations is essential to decipher their pathogenesis and to optimize therapeutic strategies. However, available molecular data of HCV-associated NHL are still limited so far and exclusively arise from targeted sequencing [4,5,6].

The present retrospective study aimed to explore the genetic landscape of a well-characterized series of 34 HCV-associated B-cell lymphomas from the observational multicentric ANRS HC-13 Lympho-C Study using tumor and matched germline whole exome sequencing (WES). This analysis first sought to detect recurrent somatic variants and copy number alterations (CNA), then identify somatic genetic abnormalities specific to HCV-induced B-cell lymphomagenesis.

Thirty-four patients were included in this study, with the following distribution of histological subtypes: MZL (n = 19) including nodal MZL (NMZL) (n = 7), extranodal MZL (EMZL) (n = 6) and splenic MZL (SMZL) (n = 6), DLBCL (n = 8) including transformed (n = 4) and de novo (n = 4) cases, FL (n = 3), mantle cell lymphoma (MCL) (n = 2) and chronic lymphocytic leukemia (CLL) (n = 2). Sample types were either peripheral blood mononuclear cells (PBMC) (n = 21), frozen biopsies (n = 8) or formalin-fixed, paraffin-embedded (FFPE) biopsies (n = 5). Main clinical characteristics of the cohort are detailed in Supplementary Table 1.

WES identified 36 recurrently mutated genes, of which 12 were affected in more than 10% of the cases (Fig. 1). Among them, some were already known to be recurrently mutated in DLBCL or MZL, such as KMT2D (29%), NOTCH2 (26%), KLF2 (18%), CARD11 (15%), MYD88 (12%), ARID1A (12%) and PIM1 (12%). Notably, KMT2D, PIM1 and H1-4 mutations were significantly enriched in HCV-associated DLBCL subgroup (p = 0.027, p = 0.004, and p = 0.017, respectively), as compared to HCV-associated MZL subgroup.

Fig. 1: Pattern of somatic mutations in HCV-associated B-NHL.
figure 1

Co-mutation plot showing the spectrum of somatic mutations in recurrently altered genes (n = 36) across 34 HCV-associated B-NHL. Each column represents a patient sample with histological lymphoma subtype and sample type indicated at the top. Each row represents a mutated gene with frequency of mutations indicated in the left. Only genes recurrently mutated in ≥3 cases are reported. NMZL Nodal Marginal Zone Lymphoma, SMZL Splenic Marginal Zone Lymphoma, EMZL Extranodal Marginal Zone Lymphoma, DLBCL_tr transformed Diffuse Large B-cell Lymphoma, DLBCL_novo de novo Diffuse Large B-cell Lymphoma, FL Follicular Lymphoma, MCL Mantle Cell Lymphoma, CLL Chronic Lymphocytic Leukemia.

Full size image

Pathway analysis indicated that mutated genes belonged to the following categories: BCR/NF-KB signaling (21/34 patients, 62%), epigenetic modifiers and chromatin regulation (15/34, 44%), NOTCH signaling (14/34, 41%), MAPK-ERK signaling (9/34, 26%), JAK-STAT signaling (7/34, 21%), immune response and escape (6/34, 18%), B-cell development (5/34, 15%), cell cycle (4/34, 12%) and RHO GTPase signaling (3/34, 9%) (Supplementary Fig. 1).

To identify genetic events specific to HCV-induced lymphomagenesis, we then compared frequencies of the most mutated genes from our cohort with those from large published series of HCV-negative DLBCL [7, 8], NMZL [9] and SMZL [10]. Given their low representation in our cohort and weak or no association with HCV infection, other histological entities such as FL, MCL and CLL were excluded from this analysis. Because of the small sample size and large number of genes involved, we lacked power to properly assess the statistical significance of mutation frequency differences between the cohorts. Nevertheless, as compared to their HCV-negative counterparts, we observed a higher frequency of NOTCH2 mutations in HCV-positive NMZL (57% vs. 20%), of DTX1 (38% vs. 15%/12%), KMT2D (63% vs. 33%/25%) and ZFP36L1 (25% vs. 9%/8%) mutations in HCV-positive DLBCL and of DTX1 mutations in HCV-positive SMZL (17% vs. 2%). In contrast, KDR (25% vs. 1%/4%) and KRAS (25% vs. 4%/3%) mutations appeared to be very specific to HCV-positive DLBCL (Supplementary Table 2). As a result, DTX1, ZFP36L1, KDR and KRAS could represent novel candidate genes involved in HCV-induced lymphomagenesis (Supplementary Figs. 2 and 3).

In addition, WES allowed to perform CNA analysis. Not surprisingly, most of the CNA described here have been previously reported in HCV-negative MZL and DLBCL. Indeed, trisomy 3/3q gains were the highest recurrent CNA (9/34, 26%) followed by complete or focal 6q deletion (8/34, 24%), complete or partial 8q gain (5/34, 15%), trisomy 18 (3/34, 9%), trisomy 12 (3/34, 9%) and complete or focal 1q gain (3/34, 9%) (Fig. 2a and Supplementary Fig. 4).

Fig. 2: Landscape of somatic alterations in HCV-associated B-NHL.
figure 2

a Frequency of copy number gains and losses across 33 HCV-associated B-NHL. One sample was excluded from the analysis due to noisy coverage profile throughout the genome. Genomic positions are plotted along the x-axis. Y-axis indicates the percentage of cases with copy number gain (in red) or loss (in blue) of the corresponding chromosome region. Genes targeted by both inactivating mutations and focal deletions are shown. b Co-mutation plot showing the spectrum of somatic alterations in 6 recurrently altered genes across 34 HCV-associated B-NHL. Each column represents a patient sample with histological lymphoma subtype and sample type indicated at the top. Each row represents an altered gene with frequency of alterations indicated in the left. Only genes recurrently altered in ≥6 cases are reported. DEL focal deletion.

Full size image

Overall, by compiling copy number and mutational profiles, we identified 6 genes altered in at least 18% of the cases, including TNFAIP3 and ARID1A that were targeted by both somatically acquired inactivating mutations and deletions (Fig. 2b).

Finally, we sought to correlate mutational status with survival. HCV-positive MZL patients harboring mutations of NOTCH2 displayed better overall survival (p = 0.039) (Supplementary Fig. 5), as previously reported in HCV-negative MZL [11]. The effect on survival of HCV + DLBCL was not assessable in our cohort due to small numbers.

To our knowledge, we report herein the first WES-based genetic analysis of a series of 34 HCV-positive B-cell lymphomas. As compared to targeted sequencing, this non-biased technology allowed to confirm the somatic origin of the mutations by comparison with germline DNA and also to generate copy number data. With this approach and by considering copy number alterations, we finally found TNFAIP3 to be as frequently altered as KMT2D (29% of cases), both genes being regularly affected in DLBCL and MZL [7, 9].

NOTCH signaling pathway, especially NOTCH2, NOTCH1 and PTEN members, was already known to be involved in HCV-positive DLBCL [4] as well as in MZL, regardless of the HCV status [9, 11]. In addition to frequent NOTCH2 mutations (26%), one major finding of our study was the identification of mutations in other genes involved in NOTCH signaling pathway or regulation. We indeed detected recurrent mutations of DTX1, a negative regulator of the NOTCH signaling pathway, in 18% of the cases. Variants were mostly missense mutations affecting the WWE domain of the protein, which mediates physical interactions between DTX1 and NOTCH, and nonsense mutations, including one already described in CLL (Supplementary Fig. 2a). This suggests that genetic alterations of DTX1 result in protein loss-of-function and hence, NOTCH activation. We also identified ZFP36L1 mutations in HCV-positive DLBCL and MZL. All mutations, i.e. one missense and two affecting splice site, were targeting the same amino acid located in a domain involved in mRNA decay activation (Supplementary Fig. 2b). Interestingly, ZFP36L1 which encodes a RNA-binding protein reported to negatively regulate NOTCH1 expression in T cells [12], has been proposed as a tumor suppressor gene in germinal center–derived B-cell lymphomas [13]. So far, mutations of ZFP36L1 have been only described in DLBCL at a low frequency [7, 14].

Moreover, WES analysis revealed KDR and KRAS as novel candidate genes of HCV-associated lymphomagenesis, with a more pronounced specificity for the DLBCL subtype. Three KDR mutations, including two affecting the kinase domain, were indeed observed in 3 different patients (Supplementary Fig. 2c). KDR, also called vascular endothelial growth factor receptor 2, is a tyrosine kinase receptor for VEGF. Notably, HCV-associated upregulation of VEGF has been reported to promote viral transmission in liver cells and use of inhibitors targeting VEGF-R kinases significantly inhibited HCV entry [15]. In addition, KRAS mutations, including two located at the G12 and G13 hotspots (Supplementary Fig. 2d), were detected in 12% of the cases (2 DLBCL, 1 EMZL, and 1 CLL), which was higher than expected frequency from the literature (2% in DLBCL, not reported in MZL) [7].

Recently, Defrancesco et al. [5] used targeted sequencing on a cohort of 22 patients with HCV-associated indolent B-NHL and 7 patients with type II MC. The most frequently mutated genes in their study (TNFAIP3 (28%), KMT2C (20%), FAT4 (20%), TBL1XR1 (20%)) differ from the ones we reported herein. This is probably due to sampling differences, their cohort consisting in pre-lymphomatous and indolent cases, probably responding to antiviral therapy alone, mostly of unspecified type, whereas ours was composed of a majority of MZL cases as well as DLBCL. Notably, they detected only 1 DTX1 mutation and no KRAS mutation whereas ZFP36L1 and KDR were not targeted by their gene panel.

Our study still suffers from some limitations. The heterogeneity of B-NHL among our cases, resulting in small numbers in each subtype, prevented from identifying genomic alterations related to a specific mode of transformation. In addition, due to partial data availability regarding IGHV gene usage, we were not able to find any correlation between this parameter and molecular profile, as previously described [5] (data not shown).

Nevertheless, based on a well-characterized cohort, this study reports the first series of HCV-positive lymphoma patients analyzed by WES. Our results provide an overview of the genetic landscape associated to HCV-related B-cell lymphomas and highlight some novel putative candidate genes. Further investigations are thus warranted to validate these findings in large-scale prospective cohorts and to determine mechanisms of resistance in patients treated by DAA.