High grade B-cell lymphoma with MYC, BCL2 and/or BCL6 rearrangements: unraveling the genetic landscape of a rare aggressive subtype of non-Hodgkin lymphoma

Abstract

High-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements (DH/TH-HGBL) still miss an in-depth genomic characterization. To identify accompanying genetic events, we performed a pilot study on 7 cases by applying DNA microarray and targeted NGS sequencing. Interestingly, the genetic background of DH/TH-HGBL is largely overlapping with that of other high-grade/poor prognosis lymphomas. Namely, copy number abnormalities were trisomy of chromosome 7 and chromosome 8q gain, encompassing MYC. Among gene variants, those affecting transcription factors (MYC, FOXO1), epigenetic modulators (KMT2D, EZH2 and CREEBP), and anti-apoptotic gene (BCL2), were recurrent. MYC and BCL2 were mutated in 3 and 5 cases, respectively. In addition, mutations of FOXO1, previously reported in Diffuse Large B-Cell Lymphomas, were also detected. Clarifying the genomic background of this subset of high-risk lymphomas will pave the way for the clinical use of new biomarkers to: (1) monitor treatment response and; (2) consider alternative targeted therapies.

Keywords:

• Double/Triple hit high-grade B-cell lymphoma
• BCL2
• MYC
• mutational screening

Introduction

The last updated World Health Organization (WHO) classification [1] introduced, for the first time, a new entity of High-grade B-cell lymphoma (HGBL), with distinguishing biological and clinical features, that includes two subgroups, i.e. HGBL NOS and double/triple-HGBL hit (DH/TH-HGBL) [2].

DH/TH-HGBL accounts for 3–10% of HGBL and is characterized by an increased risk of central nervous system involvement and chemo-refractoriness, for which intensified regimens of therapy are recommended [3–6]. Notwithstanding therapy is mainly based on high-dose chemotherapy, i.e. DA-EPOCH-R followed by consolidative and autologous stem cell transplantation [7,8], only half of cases can be cured [9].

Histology and immunohistochemistry of DH/TH-HGBL cases typically resemble either Diffuse Large B-cell Lymphoma (DLBCL), NOS or have intermediate features between DLBCL and Burkitt Lymphoma (BL). The diagnosis of DH/TH-HGBL cases, however, responds to strict criteria which are based on the presence of MYC and BCL2 and/or BCL6 translocations. In the routine diagnostic work-up, they are typically detected by fluorescence in situ hybridization (FISH) with the specific break-apart assay, identifying the rearrangement of these genes, regardless of the partner.

In 2019, however, a specific double hit signature has been identified by next-generation sequencing [10]. This gene expression profile identifies about 30% of DH/TH-HGBL which are missed by FISH due to the occurrence of cryptic translocations involving MYC, BCL2 and/or BCL6, focal MYC and MIR17H copy number gains, or focal deletion at the PVT1 promoter [10,11]. On the other hand, the presence of extra copies of MYC and/or BCL2 has been associated with the typical DH/TH-HGBL signature only in 3% of cases, indicating that Copy Number Abnormalities (CNA) alone are insufficient to produce the DH/TH-HGBL biology [11]. Taken together, these findings have warned on the limit of currently adopted molecular-cytogenetic diagnostic algorithms to correctly classify cases of DH/TH-HGBL and raised the problem of which test(s) to use in clinical practice.

Although, DH/TH-HGBL have been extensively studied to properly pick up all genomic markers, and assess the most reliable diagnostic criteria, only a little is currently known about additional concurrent abnormalities contributing to the genomic background, and their putative role as predictive and/or prognostic markers. A single Center study, on 22 DH/TH-HGBL found that this lymphoma subtype has a higher rate of mutations than DLBCL, most frequently affecting BCL2, CARD11, NOTCH2, MYC, TNFAIP3, TP53 and EZH2 [12]. Interestingly, they also observed a correlation between the type of gene mutations and the origin of DH/TH-HGBL, i.e. de novo or transformed from indolent forms. Namely, TNFAIP3 mutations were more frequently detected in the latter subgroup.

Our study aimed to elucidate the genomic profile of DH/TH lymphomas by investigating copy number abnormalities and gene variants. The study was carried out on 7 cases, prospectively referred to our laboratory. The patient cohort included 6 DH/TH-HGBL, one of which arising from a previous indolent non-Hodgkin lymphoma, and 1 double-hit Follicular Lymphoma (DH-FL). Our comprehensive study detected recurrent copy number abnormalities, gene variants and the occurrence of chromothriptic events. We found that the genomic background of DH- and TH- HGBL, and low and high-grade histologic subtypes, are tightly related and that the affected genes are shared with other lymphoma subtypes. Many of the genes found mutated in our cases, are targets of inhibitors currently tested in Relapsed/Refractory (R/R) DLBCL, follicular lymphomas (FL), and/or other subtypes of High-Grade Lymphomas (e.g. CREBBP, EZH2, BCL2) (https://clinicaltrials.gov/). The routine use of genetic diagnosis, in this type of rare and aggressive pathology, could give us a tool to better evaluate the disease, suggesting alternative therapies and in some cases, explaining the resistance to treatments.

Materials and methods

Patients

From a cohort of 15 cases of non-Hodgkin lymphomas prospectively referred to our laboratory from 2015 to 2020, and classified according to fluorescence in situ hybridization patterns for MYC, BCL2 and BCL6, as DH/TH-HGBL (=13) or DH-FL (=2), 7 cases were investigated for copy number abnormalities and gene variants (Table 1). There were 3 TH- and 2 DH- HGBL, 1 DH-HGBL arising from a previous FL, and 1 DH-FL. DNA was extracted from Formalin-Fixed Paraffin-Embedded (FFPE) lymph nodes (=6) or bone marrow cells (=1; BM) by an automatic system (QIAamp DNA FFPE Tissue kit and Allprep DNA/RNA Mini Kit, Qiagen). Informed consent was obtained in accordance with the Declaration of Helsinki. The protocol was approved by CER (Comitato Etico Regionale, Umbria), November 28th, 2019 (number 3624/19).

Table 1. Patient characteristics.

Patient ID	Sex	Age	Clone %	Diagnosis	Ki-67	HITs	MYC IHC	BCL2 IHC/FISH	BCL6 IHC/FISH	FFPE/ BM	Microarry (CNA)	Lymphoma panel	SNV/Indel
1	M	71	90%	DLBCL	70%	Double	−	+/rearr	+/wt	BM	SNP array	Yes	7
2	M	57	70%	DLBCL	80%	Triple	n.a.	+/rearr	+/rearr	FFPE	ONCOSCAN	Yes	8
3	M	62	50%	DLBCL-GCB	80%	Triple	50%	+/rearr	+/rearr	FFPE	ONCOSCAN	Yes	8
4	F	61	60%	DLBCL-GCB	n.a.	Triple	15%	+/rearr	+/rearr	FFPE	n.a.	Yes	7
5	M	56	50%	FL	20%	Double	30%	+/rearr	+/wt	FFPE	ONCOSCAN	Yes	7
6	M	82	80%	DLBCL*	90%	Double	90%	+weak/rearr	+partial/wt	FFPE	ONCOSCAN	Yes	11
7	F	69	>90%	DLBCL-GCB	95%	Double	90%	+/rearr	+/wt	FFPE	ONCOSCAN	Yes	5

M: male; F: female; DLBCL: diffuse large B-cell lymphoma; DLBCL-GCB: diffuse large B-cell lymphoma Germinal Center; FL: Follicular lymphoma; B, n.a.: not available, +, positive, −, negative, rear: the presence of rearrangement; wt: absence of rearrangement; FFPE: formalin-fixed paraffin-embedded; BM: bone marrow; CNA: copy number abnormalities; SNV: Single nucleotide variant. *previous diagnosis of follicular lymphoma.

Copy number analysis

To study CNA, we used Oncoscan CNVplus (5 cases) or Cytoscan HD (1 case) microarray, according to manufacturer’s protocols (Affymetrix/Thermo Fisher Scientific). Experiments were done using 80 ng of DNA for Oncoscan CNVplus and 250 ng of high-quality genomic Cytoscan HD. The protocol was supported by Affymetrix GeneChip Command Console (AGCC) software. Results were analyzed using the Chromosome Analysis Suite 4.1 software (Affymetrix/Thermo Fisher Scientific) and annotated based on GRCh37 (hg19). Filter settings were 1.5 Mb for CNA and 10 Mb, with at least 50 markers, for cnLOH. Polymorphisms were excluded by using the publicly available CNV database of genomic variants (DGV; http://dgv.tcag.ca/dgv/app/home).

Targeted next-generation sequencing

A next-generation sequencing (NGS) capture-based target panel (Lymphoma Solution; SOPHiA GENETICS, CH) was applied to all cases. The panel targets a wide selection of the 54 most mutated lymphomas genes (full coding sequences of 32 genes and specific exons of 22 additional genes; Table S1) starting from 60/90 ng FFPE DNA and from 50 ng for BM DNA. Libraries preparation and pair-end sequencing were performed according to manufacturer´s instructions on a MiSeq with 151 × 2 cycles using Reagent Kit V3 600 cycles cartridge (Illumina, San Diego, CA, USA). The percentage of target regions with coverage of >500 and >1000 reads was 99.5% and 98.5% respectively. Alignment, base calling and variant annotation were performed with SOPHiA DDM^TM software. As recommended by the producer we retained variants with a variant allele frequency (VAF) >4%. To software default settings for variant retention we added further filters: a) we kept variants that had a frequency population less than 0.01, into at least one public databases of human polymorphisms used in the software [esp5400 (https://evs.gs.washington.edu/EVS/); ExAC (http://exac.broadinstitute.org/) and GnomAD (https://gnomad.broadinstitute.org/)]; b) we discarded synonymously, 3′/5′ untranslated regions, intronic and intergenic variants; and c) we held back variants annotated by ClinVar as pathogenic or likely pathogenic and we also rejected benign/likely benign variants (https://www.ncbi.nlm.nih.gov/clinvar/; Gene Cards: The Human Gene database).

Results

Copy number abnormalities

Overall 48 CNA (range: 4–14) were detected. There were 20 gains (range: 1–6), 22 losses (range: 2–6), and 6 cnLOH (range: 0–4), with similar distribution in DH- and TH- HGBL. Patient n. 1, investigated by Cytoscan, harbored 3 CNA (gain 1q25.2-q44, loss 12p13.31p13.1, and loss 13q32.2q34), and 2 LOH (1p36.33-p31.1 and 16p13.3p12.1). In cases undergoing Oncoscan analysis, recurrent CNAs were trisomy of chromosome 7 (cases #3 and 7) and a gain at chromosome 8q, with a share duplicated region from position 127208252 to 128751563, that included MYC at its telomeric side (cases #2 and 7). In case n. 7, we also found a gain at chromosome 3 starting from the 3p telomere (position 63411) to the RFTN1 gene (16569746), a known translocation partner of MYC in high-grade lymphomas. Genomic losses included an interstitial deletion at 2q21.1 extending for 1.8 Mb (cases #2 and 5), and monosomy 13 (in case #5) or deletions at 13q (cases # 1 and 2). A chromothripsis event, at the long arm of chromosome 3, was found in case n. 6.

Mutational screening of newly diagnosed DH/TH-HGBL patients by deep targeted sequencing

Panel coverage was high in both FFPE and BM samples, with a median of >1000x of 98.5% (range 91.32–100%; Table S2). Considering retained variants with coverage up to 1000 reads, we obtained 7764 median reads (range 1434- 16801; Table S3). A total of 53 filtered variants were detected in 20 genes (range 5–11; median: 8) (Table S3). Eight mutations (15.1%) were INDELs and 45 (84.1%) were single nucleotide variants (SNVs); the latter included 4 non-sense point mutations. All variants are detailed in Supplemental Table S3, and their distribution per gene is shown in Figures 1A,B.

Figure 1. High grade B-cell lymphoma mutational landscape. (A) Gene mutation number (N) observed sequencing 54 genes in 7 patients with high-grade B-cell lymphoma. (B) Frequency of detected gene mutations in two different high-grade B-cell lymphoma subtypes: Double Hit (blue) and Triple Hit (grey). (C) Heat-map of 7 patients detected mutations across Double Hit (DH) and Triple Hit (TH) subtypes.

The most recurrently mutated genes encode for transcription factors, epigenetic modulators and proteins involved in apoptosis (Figure 1C). Namely, 9 mutations were found on both KMT2D (# 1, 4–7) and BCL2 (# 3–7). BCL2, involved in 2 DH-, 2 TH- HGBL, and in the DH-FL, showed mutations in four main domains: the folded loop domain (FLD; n = 4), the apoptosis regulator domain (n = 3), and the Bcl-2 homology 2 and 4 domains (BH2 and BH4; n = 1) (Figure 2A). BCL2 variants were all missense mutations; in 44% of cases, the amino-acidic substitution involved an Alanine (Table S3,S4). In 3 cases, two-three mutations of BCL2 co-occurred.

Figure 2. Genetic profiling of BCL2, MYC and FOXO1 genes. (A) BCL2 protein diagram, domain annotations and the position of the nine detected missense mutations; 85.7% of patients carried BCL2 mutations. Bcl-2 homology 1, 2, 3 and 4 domains (BH1, BH2, BH3 and BH4). (B) MYC protein diagram, domain annotations and site of the three found somatic variants (42.9% of patients). (C) FOXO1 protein diagram, domain annotations and site of the four found somatic variants.

KMT2D mutations were frameshift (=6), non-sense (=2) and missense (=1) variants (Table S4). They were detected in 3 DH-, 1 TH- HGBL, and in the DH-FL (Figure 1B). Interestingly, DH-HGBL and DH-FL cases harbored 2–3 KMT2D variants (#1, 5 and 7). Besides KMT2D, other epigenetic modulators, recurrently mutated in this lymphoma subtype, were EZH2 in 4 cases (# 3, 4, 6 and 7) and CREEBP in 3 cases (Table S3).

Transcription factors/regulators were mutated in all but one case of DH-HGBL (#1). Among them, the transcription repressor FOXO1 and the MYC oncogene were involved in 3 patients each (Figures 2B,C). All MYC mutations (2 in DH/TH-HGBL and 1 in the DH-FL) affected the N-terminal region (NTR) of the gene (see Table S3) (Figure 2B). Cell cycle, signaling and immunity-related genes GNA13 and CIITA are altered in two cases; TP53 in one.

Discussion

The genomic background of DH/TH-HGBL is still largely uncharacterized. Whether this lymphoma subtype has a distinctive bio-molecular profile or if it resembles that found in other high-grade lymphomas, is currently unknown. Herein, we report for the first time results of comprehensive CNA and mutational analyses in cases of primary and transformed DH/TH-HGBL, and of DH-FL. To diagnose DH/TH-HGBL we used FISH assays for MYC, BCL2, and BCL6 which, however, did not pick up cases with the recently described cryptic abnormalities, identifiable only by applying next-generation sequencing approaches [9]. Although the study was carried out in a small cohort of cases, it provided new insights into the genomic landscape of this rare subtype of non-Hodgkin lymphomas and paved the way for future confirmatory studies.

Overall, we found that the genomic background of our DH/TH-HGBL and DH-FL largely resembles that of other B-cell non-Hodgkin lymphoma subtypes, for both recurrent CNA and gene variants [13–29]. As far as to regard CNA, trisomy of chromosome 7, which we detected in 1 DH- and 1 TH- HGBL, it is a recurrent numerical chromosome abnormality in both DLBCL and FL [13,14], where it appeared to be associated with a more aggressive outcome [15,16]. Instead, the 8q gain encompassing MYC in case n. 7 was probably the result of unbalanced t(3;8) translocation juxtaposing MYC to the non-IG partner RFTN1. In fact, this case also harbors a gain of chromosome 3, starting from the 3p telomere and ending within the RFTN1 gene, at the 3p24.3 band. RFTN1, which encodes for a protein that is involved in protein trafficking via association with clathrin and AP2 complex, is a recurrent non-IG rearrangement partner of MYC [17,18]. It is worth mentioning, that 65% of MYC rearrangements in DH/TH-HGBL, have non-IG partners [18].

Interestingly, MYC and BCL2 were not only the two most frequently rearranged genes in DH/TH-HGBL and DH-FL, but also among the most frequently mutated. The involvement in rearrangements, as well as mutations, suggests a strong impact on disease pathogenesis/progression. Mutational screening of these genes could clarify, in future studies, whether the presence of mutations, in addition to translocations, cause any differences in terms of gene expression, outcome and response to therapies.

BCL2 displayed 9 variants in 5 patients. All BCL2 mutations have been already reported even though the functional consequences have not been established yet (https://varsome.com/). More than half of BCL2 variants (5/9; 55.6%) fell in the BH4 and FLD domains. The H20Q mutation, reported in DLBCL and in Marginal zone lymphoma, affected the inhibitory action of the inositol triphosphate receptor (IP₃R) [19–22] (https://cancer.sanger.ac.uk/cosmic/mutation/). It appeared to confer resistance against the inhibitor, by enhancing the IP₃R binding ability of the BH4 domain, implicating gain-of-function of this mutation in the progression of lymphoma [30]. In the FLD domain, instead, we identify 4 different point mutations, i.e. P53T, A60D, A61T and S62P, all altering the negative regulatory region at amino acids 32–68. This specific region interacts with p53 preventing the link between Bax and BCL2 [31]. Lastly, we found the N192H mutation, in the highly conserved BH2 domain, which was predicted to be probably damaging (score= 0.99; PolyPhen-2). The BH2 domain is a dimerization motif, required for the BCL2/BAX heterodimerization and consequently, for mediating the pro-apoptotic activity of the protein [32].

Concurrent BCL2 translocations and mutations have been already reported in DLBCL, more frequently in the germinal center B-cell-like (GCB) subtype, and in FL where if they are present at diagnosis, appeared to be associated with increased risk of transformation and shortened overall survival [23].

Although MYC variants did not occur in a specific hot spot, they all fell at the N-terminal region of the gene and were predicted to be probably damaging (P72S score= 0.99; S218N score = 0.95; E268Q score = 1; PolyPhen-2), and pathogenic/likely pathogenic (https://varsome.com/). The MYC variants P72S and/or S218N have been already described in Burkitt lymphoma, in DLBCL, and in post-transplant lymphoproliferative disorders, while the E268Q substitution has not been previously reported (https://cancer.sanger.ac.uk/cosmic/mutation/). Interestingly in Burkitt lymphoma, mutations in this region may have opposite effects, i.e. activating or repressing, on cell cycle, apoptosis, and transformation [24].

Besides MYC, another transcription factor recurrently involved in DH/TH-HGBL was FOXO1, harboring 4 mutations in 3 cases. All mutations affected the conserved NTR cluster region of the gene (#2: S22P and H119Y; #4: T24I and #7: L77V) and were predicted to be probably damaging. Alterations at this domain result in diminished T24 phosphorylation, loss of interaction with 14-3-3, and nuclear retention, with a consequent, increased in cell proliferation and survival [25,26]. Variants of FOXO1, indifferently involving the first exon or the terminal region, have been already reported in 8.6% of DLBCL, where they behave as independent poor prognostic markers [26]. A possible prognostic impact of FOXO1 mutations in DH/TH-HGBL should be investigated.

Other genes typically altered in DH/TH-HGBL were those coding for epigenetic modulators. Indeed, mutations affecting these genes were found in all cases, with a range of 1–5; genes related to histone methylation (KMT2D, EZH2) and histone acetylation (CREBBP) were mainly involved. All these three genes are typically mutated in DLBCL and FL. In particular, KMT2D loss-of-function mutations have been described in 89% of FL and 32% of DLBCL [19], EZH2 activating mutations, in 21.7% of GCB-DLBCL and 20% of FL cases [27,28], and CREBBP in 50% of FL and 16.9% of DLBCL [19]. It is worth noting that epigenetic alterations play an important role in the tumor progression of DLBCL [29].

The affordability of this NGS panel offers an exceptional opportunity to intensify the search for new biomarkers and elucidate the genomic background of DH/TH-HGBL cases. Recurrently involved genes and CNA, are largely overlapping with those found in other lymphoma subtypes. Among them, putative prognostic markers have been identified. Thus, the introduction of NGS in clinical diagnosis would likely improve treatment choice and would be useful to identify actionable targets to design a personalized therapy for high-risk patients. Due to the cost of running a large gene sequencing panel, it would be useful to select a small number of HGBL-specific genes. The integration of FISH with a targeted mini-NGS (Table S5) at diagnosis or, after first-line therapy, in unresponsive cases, will be undoubtedly useful for identifying prognostic and actionable targets to be used as selection criteria for prospctive clinical trials aimed to assess the efficacy of tailored treatments (https://clinicaltrials.gov/). Lastly, gene mutations are also suitable markers for noninvasive monitoring of peripheral blood cell-free DNA.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

GILEAD Fellowship program 2018, 2019, 2021.