The impact of MYC and BCL2 structural variants in tumors of DLBCL morphology and mechanisms of false-negative MYC IHC

Diffuse large B-cell lymphoma (DLBCL) is a heterogeneous neoplasm that accounts for 30% to 40% of newly diagnosed non-Hodgkin lymphomas.¹  Approximately 50% to 60% of patients with DLBCL are cured by standard R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) immunochemotherapy.^2,3  However, patients with DLBCL that progresses or relapses after frontline immunochemotherapy have poor outcomes, making the identification of this subgroup of patients, at the time of initial diagnosis, a major priority to improve outcomes. Initial efforts to stratify DLBCL patients into clinically relevant subgroups used gene expression profiling (GEP) to identify 2 major molecular subtypes of DLBCL corresponding to different putative cells-of-origin (COO): germinal center-B cell–like (GCB-DLBCL) and activated B cell–like (ABC-DLBCL).^4,5  These molecular subtypes identify distinct biology, with activation of subtype-specific oncogenic programs as a result of different mutational landscapes.^6-9  However, recent studies have identified unique genetic subgroups that extend beyond the binary COO division.^10-13  Notably, these genetics-based classifications did not comprehensively incorporate structural variants (SVs) involving MYC, a key oncogene in aggressive B-cell lymphomas.

Tumors of DLBCL morphology with concurrent MYC and BCL2 rearrangements are generally associated with inferior outcomes.^14,15  This has contributed to the establishment of a new category in the 2017 revision of the World Health Organization (WHO) classification for lymphoid neoplasms: high-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements (HGBL-DH/TH), colloquially referred to as double- and triple-hit lymphoma.¹  The description of a gene expression signature (DHITsig) in HGBL-DH/TH tumors with BCL2 rearrangement (HGBL-DH/TH-BCL2) has supported the validity of this new WHO category by identifying a unified underlying biology among these tumors that extends beyond their often shared GCB origin.¹⁶  Interestingly, about half of DHITsig+ tumors do not display concurrent MYC and BCL2 rearrangements but share similar biology and outcomes. Although a portion of these tumors harbor rearrangements cryptic to fluorescent in situhybridization (FISH),¹⁷  other mechanisms leading to DHITsig+ require further exploration.

Copy number variations (CNVs) are another SV commonly associated with aberrant gene expression, offering a potential mechanism for DHITsig+ in the absence of gene rearrangements. However, the biological impact of these alterations remains unclear, as studies of MYC and BCL2 CNVs in DLBCL have primarily focused on outcome associations.^18-24  As such, although the association of MYC and BCL2 rearrangements with increased protein expression has been well described,^25-28  the impact of MYC and BCL2 CNVs on expression is less well known. Despite this uncertainty, the establishment of an “atypical double-hit” category has been proposed,²²  encompassing all tumors with concurrent MYC and BCL2 SVs other than cooccurring translocations. Because the diagnosis of HGBL-DH/TH requires FISH testing for MYC rearrangement in all tumors of DLCBL morphology, which is a method that can also detect CNVs, such a classification could be integrated into clinical practice. However, whether atypical double-hit tumors share biology similar to HGBL-DH/TH-BCL2 requires further investigation.

Furthermore, an atypical double-hit entity would substantially increase the number of tumors in which subsequent FISH for BCL2 SVs is required. Because routine FISH testing is already challenging where clinical pathology resources are constrained, the association of MYC rearrangement with increased protein expression has provided a potential avenue to screen tumors for subsequent FISH testing. However, ∼25% of tumors with MYC rearrangement are negative for MYC protein expression according to immunohistochemistry (IHC) (universally performed by using the Y69 antibody)²⁹  using a threshold of ≥40% tumor cell staining.²⁸  Because the study of MYC rearrangement with increased expression has largely been limited to IHC, it is unknown if the lack of detectable protein expression is due to low MYC expression or other mechanisms specific to IHC.

Here, we performed a comprehensive analysis of the association of MYC and BCL2 SVs (rearrangement and CNVs), as determined by FISH, with expression at the messenger RNA (mRNA) and protein level in a cohort of 802 de novo tumors with DLBCL morphology. Using this approach, we assessed the biological impact of MYC and BCL2 CNVs and elucidated mechanisms of IHC negativity in MYC rearranged tumors.

The study cohort consisted of 802 tumors with DLBCL morphology, as determined by central pathology review. The tumors represent all those with available GEP data in 3 cohorts drawn from a population-based registry of patients diagnosed and treated in British Columbia, Canada. Full inclusion and exclusion criteria are provided in the supplemental Methods (available on the Blood Web site). In brief, the study materials comprised diagnostic biopsy samples from adult patients with de novo DLBCL, excluding posttransplant lymphoproliferative disorders, primary mediastinal B-cell lymphoma, patients with central nervous system involvement, and those known to have a history of indolent lymphoma or HIV infection. This study was approved by the University of British Columbia/BC Cancer Research Ethics Board, in accordance with the Declaration of Helsinki.

FISH was performed on formalin-fixed paraffin-embedded (FFPE) biopsy specimens using commercially available break-apart probes (supplemental Table 1). Tumors displaying break-apart signals in ≥5% of cells were considered to harbor a rearrangement. Gain was defined as 3 to 4 fused signals in ≥25% of cells; high-level gain was defined as ≥5 signals in ≥10% of cells.²²  Genes that were rearranged with increased copy number were included only in the rearranged group. Tumors with amplification (defined by uncountable signals) were placed in their own category, irrespective of break-apart status. IHC staining was performed for MYC (Y69, Epitomics) and BCL2 (BCL2/124, Dako; and E17, Epitomics) on the BenchMark XT platform (Ventana). Thresholds of ≥40% and ≥50% of tumor cell staining were used to define positivity for MYC and BCL2, respectively.²⁶ 

Digital GEP was performed on total RNA extracted from FFPE material by using the DLBCL90 assay on the nCounter platform (NanoString) to determine MYC and BCL2 mRNA expression levels and assign COO and DHITsig status, as previously described.¹⁶  Tumors were classified into molecular subtypes, with DHITsig+ taking precedence over COO (supplemental Figure 1A). Tumors classified as DHITsig-indeterminate were included in the DHITsig+ subtype (supplemental Methods).

To determine potential mechanisms for unexplained MYC IHC negativity, MYC sequencing data from 446 tumors were analyzed. Mutational data for 325 tumors were generated based on deep targeted sequencing using the TruSeq Custom Amplicon assay.³⁰  The remainder were sequenced by using Agilent SureSelect custom hybrid capture assays.^31,32  Further details are given in the supplemental Material.

The N11S variant was incorporated into the MYC locus of DOHH-2 and Karpas-422 cell lines using CRISPR/Cas9. Incorporation of the N11S variant was validated by Sanger sequencing, and the allelic frequency was determined by using a droplet digital polymerase chain reaction mutation detection assay. Wild-type (WT) controls were expanded from single cells under the same experimental conditions.

Western blots were performed on cell lysates using the Y69 and 9E10 antibodies. For IHC, cell lines were formalin-fixed and paraffin-embedded, and IHC was performed by using the Y69 antibody. Further information, as well as details on further studies of the N11S variant in additional cell lines, is available in the supplemental Methods.

Comparisons between mRNA expression levels were performed by using 2-sided Student t tests with adjustment for multiple testing using the Holm-Bonferroni method. Odds ratios and associated significance of protein expression and DHITsig+ expression within subgroups were determined by univariable logistic regression. All statistical analyses were performed by using R 4.0.0 software (R Foundation for Statistical Computing).

Genetic and IHC analysis of MYC and BCL2 was performed on 802 de novo tumors of DLBCL morphology drawn from 3 population-based cohorts (supplemental Figure 2). Clinical characteristics are shown in Table 1.

Identification of MYC and BCL2 SVs by FISH was interpretable in 771 and 740 tumors, respectively. MYC gain (MYC^Gain; defined as 3-4 copies) was the most frequent MYC SV, detected in 20% (152 of 771) of tumors, whereas MYC rearrangement (MYC^Tr) and high-level MYC gain (MYC^Gain+; defined as ≥5 copies) were detected in 13% (104 of 771) and 4% (30 of 771) of tumors. In contrast, BCL2^Tr was the most frequent BCL2 SV, detected in 28% (206 of 740) of tumors; BCL2^Gain and BCL2^Gain+ were detected in 21% (155 of 740) and 8% (56 of 740) of tumors.

To determine the significance of MYC and BCL2 SVs, the association of these alterations with mRNA and protein expression was examined compared with expression levels in tumors lacking SVs (Figure 1). MYC and BCL2 mRNA expression was assessed according to GEP in all tumors, whereas dichotomous levels of MYC and BCL2 protein expression were determined by IHC in 709 and 769 tumors, using thresholds of ≥40% and ≥50% cell staining to define positivity, respectively. Both MYC^Tr and BCL2^Tr were strongly associated with increased mRNA and protein expression (P < .0001). In contrast, MYC and BCL2 CNVs were observed to have different effects. Both BCL2^Gain and BCL2^Gain+ were strongly associated with increased mRNA and protein expression (P < .0001), with BCL2^Gain+ conferring similar mRNA levels as BCL2^Tr. In contrast, MYC^Gain and MYC^Gain+ were not associated with increased mRNA (P = .95 and P = .78) or protein (P = .21 and P = .18) expression. Notably, no increase in mRNA expression was seen even in tumors with >7 copies of MYC (P = .79) (Figure 2). Taken together, although rearrangements of both MYC and BCL2 are associated with deregulated expression, only CNVs of BCL2 produced aberrant expression.

Of note, amplification of MYC (MYC^Amp; defined by uncountable signals) was observed in 4 tumors and was associated with elevated MYC mRNA expression levels (P = .04), with all tumors displaying levels of expression comparable to MYC^Tr (Figure 2). Although only 1 tumor displayed amplification without a break-apart pattern (supplemental Figure 3), due to the rarity and limited understanding of MYC^Amp, these tumors were excluded from other FISH subgroups.³³ 

The increase in expression associated with MYC and BCL2 rearrangements and BCL2 CNVs was observed across all molecular subtypes (supplemental Figure 4). However, some alterations failed to reach significance due to low incidence within specific subtypes. To determine the role of SVs within each molecular subtype, the subtype-specific frequency of MYC and BCL2 SVs was explored.

In total, 155 tumors were classified as DHITsig+, including 88% (46 of 52) of HGBL-DH/TH-BCL2 tumors. Highest expression levels of both MYC and BCL2 were observed in DHITsig+ tumors (supplemental Figure 5). MYC^Tr and BCL2^Tr were the only SVs frequently observed in the DHITsig+ subtype, detected in 49% (74 of 151) and 67% (97 of 144) of tumors, and were both associated with increased levels of expression (Figure 3; supplemental Figure 4A-B).

HGBL-DH/TH-BCL2 tumors are primarily of GCB COO.²⁸  In agreement, 90% (139 of 155) of DHITsig+ tumors had a GCB COO (supplemental Figure 1B). After the exclusion of DHITsig+ tumors from the GCB subgroup (the following text refers to GCB-DLBCL as DHITsig– by definition), MYC^Tr was observed in only 4% (11 of 274) of GCB-DLBCLs and was not significantly associated with increased expression (Figure 3A; supplemental Figure 4A). Furthermore, MYC mRNA expression levels in MYC^Tr GCB-DLBCLs were significantly lower than those observed in MYC^Tr DHITsig+ tumors (P = .02) (supplemental Figure 6). In general, MYC expression levels were lowest in GCB-DLBCL (supplemental Figure 5), with 77% (210 of 274) of GCB-DLBCLs expressing MYC at levels below the cohort mean (Figure 3A). In contrast, BCL2^Tr was observed in 30% (81 of 266) of GCB-DLBCLs and was strongly associated with increased expression (Figure 3B; supplemental Figure 4B). Thus, although MYC overexpression was not a common feature of GCB-DLBCL, inappropriate BCL2 expression, achieved through translocation, is a feature of a subset of these tumors.

Although ABC-DLBCLs expressed higher levels of MYC compared with GCB-DLBCL, no MYC SVs associated with increased expression were observed at an appreciable frequency in ABC tumors (Figure 3A), suggesting that alternative mechanisms drive MYC expression in ABC-DLBCL. In agreement, MYC expression in the absence of SVs was highest in ABC-DLBCL (supplemental Figure 7A). Although BCL2 expression in the absence of SVs was similarly highest in ABC-DLBCL (supplemental Figure 7B), BCL2 CNVs were frequently observed in ABC tumors and were associated with a further increase in expression (Figure 3B; supplemental Figure 4B). This increase was apparent for both BCL2^Gain and BCL2^Gain+, which were detected in 38% (96 of 253) and 13% (34 of 253) of ABC-DLBCLs, respectively. Therefore, although both MYC and BCL2 seem to be elevated in ABC tumors by alternative mechanisms, BCL2 CNVs further contribute to deregulated expression, whereas MYC CNVs do not.

A strong association between DHITsig score and HGBL-DH/TH-BCL2 was observed, with an area under the curve of the receiver-operator characteristic curve of 0.93 (supplemental Figure 8). Therefore, DHITsig+ was used as a surrogate for HGBL-DH/TH-BCL2 biology to assess whether atypical double-hit tumors share biology similar to that of HGBL-DH/TH-BCL2.

Concurrent MYC and BCL2 CNVs (MYC^CNV/BCL2^CNV) were observed in 12% (88 of 733) of tumors, whereas MYC^CNV/BCL2^Tr and BCL2^CNV/MYC^Tr were observed in 8% (55 of 733) and 2% (14 of 733) of tumors, respectively. Only 3% (3 of 88) of MYC^CNV/BCL2^CNV tumors were DHITsig+ (Figure 4A), suggesting that CNVs alone are insufficient to produce HGBL-DH/TH-BCL2 biology. Although CNVs in the context of a coexisting translocation were more likely to belong to the DHITsig+ subtype, the rarity of these events prevented discernment of whether this observation could be attributed to the rearranged locus alone. Although 67% (6 of 9) of MYC^Tr/BCL2^Gain tumors and 80% (4 of 5) of MYC^Tr/BCL2^Gain+ tumors were DHITsig+, these tumors were not significantly more likely to belong to the DHITsig+ subtype than tumors harboring MYC^Tr alone (MYC^Tr/BCL2^Normal), wherein 45% (15 of 33) were DHITsig+ (P = .27 and P = .18) (Figure 4B). Although this analysis is limited by the rarity of MYC^Tr/BCL2^CNV, these tumors were also distinct from other DHITsig+ tumors in that they tended to display a higher COO linear predictor score (P < .0001) (supplemental Figure 9), consistent with the observed association of BCL2 CNVs with ABC-DLBCL.

In agreement with the lack of an mRNA correlate for MYC^Gain, BCL2^Tr/MYC^Gain was not associated with DHITsig+. Although 37% (34 of 93) of tumors with BCL2^Tr alone (BCL2^Tr/MYC^Normal) were DHITsig+, only 20% (9 of 45) of BCL2^Tr/MYC^Gain tumors belonged to the DHITsig+ subtype. Despite BCL2^Tr/MYC^Gain+ tumors more frequently being DHITsig+ (50% [5 of 10]), they were similarly not significantly more likely to be classified as DHITsig+ than tumors harboring BCL2^Tr alone (P = .41) (Figure 4C). Of note, all tumors with amplification of MYC were DHITsig+.

Similar to previous reports,²⁸  30% (29 of 98) of MYC^Tr tumors were negative for MYC protein according to IHC (<40% cell staining). IHC positivity was strongly associated with increased levels of MYC mRNA expression (P < .0001) (supplemental Figure 10). Notably, of the 11 MYC^Tr tumors with mRNA expression levels below the cohort mean, 82% (9 of 11) were negative for MYC IHC, providing one mechanism of MYC IHC negativity in MYC^Tr tumors.

To determine a potential mechanism for IHC negativity in MYC^Tr tumors with high mRNA expression levels, MYC sequencing data for 446 tumors were analyzed. Four of the five highest mRNA-expressing MYC^Tr tumors that were negative for MYC IHC harbored mutations in exon 2 of MYC. Interestingly, 3 of these tumors carried a single nucleotide polymorphism resulting in an asparagine-to-serine substitution at the 11th amino acid residue of MYC (MYC-N11S), whereas the fourth tumor had mutations affecting nearby residues (T8I and L14V). In total, the MYC-N11S variant was observed in five MYC^Tr tumors, with 4 of 5 negative for MYC protein expression (Figure 5A).

Among all sequenced tumors, 7.6% (34 of 446) harbored the N11S variant, and 91% (30 of 33) were negative for MYC protein expression, as detected by IHC using the Y69 antibody (Figure 5A). Analysis of MYC mRNA expression revealed no significant difference between N11S-positive and N11S-negative tumors (P = .33) (Figure 5B; supplemental Figure 11). In contrast, tumors positive for the N11S variant showed significantly reduced MYC IHC staining (P < .0001) (Figure 5C). Interestingly, the Y69 antibody is known to bind within the N terminus of MYC,³⁴  suggesting that the association of the N11S variant with reduced protein detection may be due to disrupted binding of the Y69 antibody.

To determine if the N11S variant disrupts binding of the Y69 antibody, MYC-N11S was modeled in the MYC rearranged DLBCL cell line DOHH-2 using CRISPR/Cas9. Both heterozygous and homozygous mutants were produced in replicate (supplemental Figure 12A). Importantly, all cell lines expressed comparable levels of MYC mRNA (supplemental Figure 12C), and there was no difference in proliferation across N11S and WT cell lines (P = .59) (supplemental Figure 13), a process regulated by MYC.³⁵  Western blots using the Y69 antibody detected significantly reduced levels of MYC in N11S cell lines compared with WT (Figure 6A-B). In contrast, nearly equivalent levels of MYC were detected between N11S and WT cell lines using the 9E10 antibody, which binds to a different region of MYC.³⁶  This finding was verified in HEK-293T cells overexpressing WT or N11S MYC (supplemental Figure 14). Taken together, these results suggest that the N11S variant disrupts Y69 binding, leading to a reduction in detectable protein.

To assess the impact of MYC-N11S in a routine clinical setting, MYC IHC was performed on FFPE sections of N11S cell lines. A significant reduction in the proportion of stained cells was observed in N11S cell lines (30%-50% staining) compared with WT cell lines (70%-90% staining), with heterozygous and homozygous cell lines showing comparable levels of staining (Figure 6C). The lack of a difference in staining between heterozygous and homozygous cell lines was explained by the predominant expression of the N11S variant in heterozygous cells (supplemental Figure 12B), suggesting incorporation of the N11S variant on the translocated allele. Interestingly, MYC-N11S was also observed in the MYC^Tr DLBCL cell lines WSU-DLCL2 and SU-DHL-10. Expression of the N11S variant was only seen in WSU-DLCL2, which was correspondingly MYC IHC negative, whereas SU-DHL-10 expressed the WT allele and was positive for MYC IHC (supplemental Figure 15). The presence of the N11S variant on the nontranslocated allele of SU-DHL-10 was confirmed by linked-read sequencing (data not shown).

Because most N11S positive tumors do not harbor MYC^Tr, the N11S variant was modeled in Karpas-422 cells, which lack MYC^Tr. In contrast to DOHH-2 cell lines, no apparent difference in MYC staining was observed between WT (70%-90% staining) and heterozygous (70%-80% staining) cell lines (supplemental Figure 16). Importantly, equal expression was observed from the WT and N11S allele (supplemental Figure 12B). Because Karpas-422 WT cells are strongly positive for MYC IHC staining, it is likely that very high levels of biallelic MYC expression can overcome the attenuating effect of the N11S variant.

We provide a comprehensive analysis of the impact of MYC and BCL2 SVs, as detected by FISH, on expression at the mRNA and protein level in 802 de novo tumors of DLBCL morphology. Consistent with previous reports,^25-28 MYC^Tr and BCL2^Tr were both strongly associated with increased expression. In contrast, although BCL2 CNVs were strongly associated with increased expression, MYC CNVs were not.

MYC expression in normal B-cell differentiation is tightly regulated, with transient expression observed immediately before germinal center (GC) formation following interaction with cognate T cells and in a small subset of centrocytes in the GC light zone undergoing cyclic reentry into the dark zone.^37,38  Most DLBCL tumors display a GEP corresponding to a light zone COO, a cell not expected to express MYC.³⁹  Tumors with MYC^Tr circumvent this regulation by placing MYC under the control of regulatory elements of genes that are transcriptionally active in the GC, such as immunoglobulin loci.^40-42  In contrast, although focal MYC CNVs have been reported, most affect a larger region, likely retaining the native MYC regulatory elements.^17,43,44  We therefore postulate that transcriptional regulation of MYC is maintained in the majority of DLBCL tumors harboring MYC CNVs, explaining the lack of impact on expression. Of note, although this study was restricted to FISH due to its use in clinical practice, this observation highlights the need for further studies using methodologies with greater resolution to fully elucidate the spectrum of MYC CNVs and their associated significance. Such methods could additionally account for the confounding effects of cryptic translocations.^17,45,46 

Distinct features of MYC expression were observed across molecular subtypes. MYC^Tr was the only SV associated with increased expression and was almost exclusively observed in DHITsig+ tumors. With removal of DHITsig+ tumors from the GCB subtype, deregulation of MYC was absent from the majority of remaining GCB tumors, suggesting that MYC deregulation is not a fundamental feature of GCB-DLBCL biology. In contrast, MYC expression was slightly elevated in ABC-DLBCL in the absence of SVs, suggesting that alternative mechanisms may contribute to MYC deregulation, such as NF-κB activation.^47,48 

Distinct mechanisms of BCL2 deregulation were also observed across molecular subtypes. Rearrangement of BCL2 was predominantly a feature of DHITsig+ tumors, although also observed in a portion of GCB-DLBCLs. In contrast, deregulation of BCL2 expression by CNVs was predominantly a feature of ABC-DLBCL. Of note, BCL2 expression in ABC-DLBCL was also elevated in the absence of BCL2 SVs, likely also due to NF-κB activation.^25,49 

When the WHO defined HGBL-DH/TH, the relevant SVs were limited to rearrangements, with ongoing uncertainty about the biological and clinical impact of CNVs.¹  Studies examining the clinical impact of CNVs have raised the possibility of an “atypical double-hit” category. Although outcome associations are essential for identifying patients with an unmet clinical need, these associations are intrinsically linked to current treatment strategies. To accommodate the changing landscape and evolution of DLBCL treatment strategies, a classification entity should encompass tumors that share a similar underlying biology. As such, although HGBL-DH/TH encompasses tumors harboring concurrent MYC and BCL6 rearrangement, analysis of BCL6 was not included in this study because the biology of these tumors is distinct from that of HGBL-DH/TH-BCL2.¹⁶  In contrast, the shared biology among HGBL-DH/TH-BCL2 tumors has been shown by a common gene expression pattern (DHITsig+) shared among these tumors. Of interest, although the DHITsig was highly sensitive in identifying HGBL-DH/TH-BCL2 tumors, here DHITsig+ identified an additional 108 tumors sharing a similar biology that routine clinical FISH failed to capture.

Notably, MYC and BCL2 CNVs, in the context of atypical double-hit, did not confer a similar GEP as HGBL-DH/TH-BCL2 tumors. Although a trend toward DHITsig+ was observed for MYC^Tr/BCL2^CNV and BCL2^Tr/MYC^Gain+ tumors, this association was not significantly stronger than the association of translocation alone nor was it as definitive as the association of HGBL-DH/TH-BCL2 tumors with DHITsig+. Furthermore, tumors with MYC^TR/BCL2^CNV typically displayed an ABC COO. Therefore, irrespective of prognostic significance, these results strongly support the current WHO classification, whereby only rearrangements are considered in defining HGBL-DH/TH.

Although MYC IHC has been proposed as a screening method for subsequent FISH,⁵⁰  MYC IHC negativity was observed in 30% of MYC^Tr tumors, consistent with previous reports.²⁸  Two major mechanisms were seen that uncoupled MYC^Tr from MYC IHC positivity, namely low resultant MYC mRNA levels and the presence of the MYC-N11S variant. The N11S variant was detected in 7.6% of all sequenced tumors, comparable with the global minor allele frequency of 3.4% (rs4645959).⁵¹  Modeling the N11S variant in a cell line harboring MYC^Tr led to a decrease in detectable protein. Interestingly, although this effect only seems to be active when the N11S variant is present on the translocated allele, 4 of 5 patient tumors harboring the N11S variant with concomitant MYC^Tr were negative for MYC IHC.

It was previously reported that the MYC-N11S variant results in lower protein levels.⁵²  Here, we showed that the introduction of the variant did not alter mRNA or protein levels when an antibody (9E10) to the C terminus of the protein was used. The discrepancy between the levels of protein detected by the 9E10 and Y69 antibodies in manipulated cell lines suggests that, rather than a true decrease in MYC expression, an apparent decrease of MYC protein is observed due to N11S-mediated disruption of the Y69 epitope. Although use of an IHC-suitable antibody that binds to the C terminus would be ideal for accurate detection of MYC protein, development of such an antibody has proven challenging.³⁴ 

In summary, this study strongly supports the current exclusion of CNVs of MYC and BCL2 from the definition of HGBL-DH/TH, acknowledging that the analysis was limited to DLBCL morphology and excluded transformation from indolent lymphoma. It also highlights mechanisms that uncouple MYC^Tr from MYC protein expression, including those reflecting true low mRNA and protein levels, possibly related to the MYC partner gene,²⁷  and others that are spurious, related to disruption of an antibody epitope. Building on the finding of rearrangements of MYC and BCL2 that are not detected by FISH, our study supports moving from a category defined simply by the presence of gene rearrangements to one defined by the strong molecular biology that these rearrangements typically produce.

This study was supported by the Canadian Cancer Society Research Institute (704848 and 705288), Genome Canada (4108), Genome British Columbia (171LYM), the Canadian Institutes of Health Research (GPH-129347 and 300738), the Terry Fox Research Institute (1061 and 1043), and the BC Cancer Foundation. B.C. is supported by a Canada Graduate Scholarship–Masters Award. C.S., R.D.M., and D.W.S. are supported by a Michael Smith Foundation for Health Research Career Investigator Award (5120), Scholar Award (16805), and Health Professional Investigator award (18646), respectively. D.W.S. is supported by the BC Cancer Foundation.

Contribution: B.C. and D.W.S. conceived and designed this study, analyzed and interpreted data, and wrote the manuscript; B.C., S.B.-N., M.B., T.M.-T., P.F., J.W.C., G.W.S., D.E., A.M., R.D.G., and D.W.S. performed research; B.C. and A.J. performed the statistical analysis; and all authors participated in the collection and assembly of data, and all authors read and edited the manuscript.

Conflict-of-interest disclosure: G.W.S. reports consulting and honoraria from Seattle Genetics. D.V. reports consulting and honoraria from Roche, Celgene, Janssen, Teva, Lundbeck, Seattle Genetics, AstraZeneca, Gilead, Kite, Novartis, and NanoString. C.F. reports honoraria from Seattle Genetics, Janssen, Amgen, Celgene, and AbbVie; and research funding from Roche and Teva. A.S.G. reports consulting and honoraria from Janssen, AbbVie, and Gilead. K.J.S. reports consulting and honoraria from Bristol Myers Squibb, Merck, Seattle Genetics, AstraZeneca, Gilead, and AbbVie; consulting for Servier; and steering committee membership for BeiGene. L.H.S. reports consulting and honoraria from Roche/Genentech, AbbVie, Amgen, Apobiologix, AstraZeneca, Acerta, Celgene, Gilead, Janssen, Kite, Karyopharm, Lundbeck, Merck, MorphoSys, Seattle Genetics, Teva, Takeda, TG Therapeutics, and Verastem. C.S. reports consulting and advisory board member for Seattle Genetics, Bayer, Curis, and Roche; and research funding from Bristol Myers Squibb and Trillium Therapeutics. D.W.S. reports consulting for AbbVie, AstraZeneca, Celgene, and Janssen; research funding from Janssen and NanoString Technologies; and is a named inventor on patents, including one licensed to NanoString Technologies. A.J., D.E., R.D.M., and D.W.S. are inventors on the DLBCL90 patent. The remaining authors declare no competing financial interests.

Correspondence: David W. Scott, British Columbia Cancer Research Centre, 675 West 10th Ave, Vancouver, BC V5Z 1L3, Canada; e-mail: dscott8@bccancer.bc.ca.