The double-hit signature identifies double-hit diffuse large B-cell lymphoma with genetic events cryptic to FISH

Diffuse large B-cell lymphomas (DLBCLs) are a genetically heterogeneous group of neoplasms with variable outcomes after standard chemoimmunotherapy (eg, R-CHOP [rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone]). There is an ongoing interest in delineating subgroups within DLBCL that share targetable biology, with the potential for precision medicine to overcome biology currently associated with treatment failure. Among the subgroups with the worst outcomes, high-grade B-cell lymphomas with MYC and BCL2 (and sometimes BCL6) rearrangements (HGBL-DH/TH-BCL2s) mostly belong to the germinal center B-cell (GCB) molecular subgroup (the subgroup generally considered to have favorable prognosis).^1,2  We recently defined a gene expression signature that identifies nearly one-third of GCB-DLBCL tumors as having a double-hit–like gene expression pattern (DHITsig), with only half harboring rearrangements of both MYC and BCL2 as determined by breakapart fluorescence in situ hybridization (FISH).³  In that cohort, the outcomes for the 20 non–HGBL-DH/TH-BCL2 patients were comparable to those of patients with HGBL-DH/TH-BCL2.

In contrast to Burkitt lymphoma (BL), where the partner of MYC is universally an immunoglobulin gene, in approximately half of HGBL-DH/THs with DLBCL morphology, the partner is nonimmunoglobulin.⁴  For this reason, a breakapart FISH strategy is favored for detecting MYC rearrangements in DLBCL rather than MYC/IGH dual fusion. Breakapart FISH purportedly fails to detect 2% to 4% of MYC rearrangements when the false-negative cases are identified using an MYC/IGH dual-fusion approach.^5-7  Recently, a BL tumor was described where the coding exons of MYC (8.6 kb) were inserted into the IGH locus, an event cryptic to both FISH approaches.⁸  This raises the possibility that a proportion of DLBCLs harbor MYC and/or BCL2 rearrangements that are cryptic to FISH.

We hypothesized that genetic dysregulations of MYC and BCL2 and their downstream signaling pathways are among the fundamental genetic events underlying the DHITsig in most, if not all, DHITsig⁺ DLBCL tumors. Using whole-genome sequencing (WGS) of 20 tumors lacking MYC and/or BCL2 rearrangements by FISH, we identified 6 cases with rearrangements cryptic to breakapart FISH, along with other genetic mechanisms of dysregulation of MYC and downstream pathways.

DNA was extracted from the 20 fresh-frozen GCB DHITsig⁺ tumors lacking MYC and/or BCL2 translocations (breakapart FISH) identified by Ennishi et al.³  Polymerase chain reaction–free libraries were constructed from sheared DNA that was end repaired and A-tailed using the NEBNext Ultra II DNA library prep kit for Illumina (New England Biolabs) before the ligation of dual-indexed TruSeq adapters and cleanup using 0.8X AMPure XP beads (Beckman-Coulter). Libraries were sequenced to an average depth of ×60 on an Illumina HiSeqX system. Reads were aligned using BWA-MEM (version 0.7.6a).⁹  Structural variants were identified using Manta.¹⁰  Copy-number variants (CNVs) were identified with Control-FREEC.¹¹  Copy-number data were obtained for the discovery cohort using Affymetrix SNP6.0 arrays as previously described.^1,12  All statistical tests were performed in R software (version 3.5.1). This study was reviewed and approved by the University of British Columbia–BC Cancer Research Ethics Board in accordance with the Declaration of Helsinki.

All FISH-detected BCL2 and MYC rearrangements were detected by WGS structural variant calling, confirming the sensitivity of WGS to these events (Figure 1A; supplemental Table 1, available on the Blood Web site). Additional BCL2 and MYC rearrangements were identified in the genomes of 3 cases each. Figure 1B describes 1 of 2 tumors in which the BCL2 gene was inserted into the IGH locus, placing BCL2 in close proximity to the Eµ enhancer. The breakpoints in IGH and near the 3′ untranslated region of BCL2 were consistent with typical BCL2-IGH translocations.¹³  Importantly, the removal of BCL2 from chromosome 18 left the regions targeted by breakapart FISH probes adjacent (supplemental Figure 1). In another case, 80 kb of the IGK locus, including the Eκ enhancer, was inserted telomeric to BCL2 (Figure 1C). Two tumors harbored cryptic enhancer insertions near MYC (eg, 170 kb of the ZCCHC7 locus including the PAX5 superenhancer inserted telomeric to MYC; Figure 1D). None of these enhancer insertions were large enough to separate the breakapart probes so as to be detectable by FISH (supplemental Figure 1).

Thus, 6 apparently non–HGBL-DH/TH tumors were confirmed to be HGBL-DH/TH through the use of WGS. All of the cases with cryptic BCL2 or MYC rearrangements were positive for BCL2 or MYC protein by immunohistochemistry, respectively (Figure 1A). Cryptic MYC translocations exclusively involved a nonimmunoglobulin partner, whereas BCL2 rearrangements all involved immunoglobulin partner loci. In 2 tumors, the MYC locus was translocated adjacent to the BCL6 superenhancer (supplemental Table 1), an event cryptic to both MYC and BCL6 FISH. No other cryptic BCL6 rearrangements were detected in this patient group. Considering these findings, breakapart FISH failed to detect a clinically significant proportion of HGBL-DH/THs, with cryptic HGBL-DH/TH-BCL2 tumors representing 19% (6 of 31) of all HGBL-DH/TH-BCL2 tumors in the original DHITsig discovery cohort.

Although focal MYC gains were not exclusive to DHITsig⁺ tumors, several of the genomes had notable CNVs affecting MYC and/or pathways downstream of MYC, including 2 focal MYC gains and 1 case of double minute (Figure 2A). MIR17HG encodes the miR-17∼92 microRNA cluster, which is a direct target of MYC.^14,15  Copy gains affecting MIR17HG were identified in 6 of 20 DHITsig⁺ cases (Figure 2B) and may represent an avenue for dysregulation of MYC and its downstream targets in these cases.

A comparison of SNP array copy-number profiles from 179 GCB-DLBCLs, including the 20 cases described here, revealed that amplifications (copy number >3) of MIR17HG and FCGR2B and deletions of CDKN2A and 22q11.22 (IGL) were significantly enriched among DHITsig⁺ tumors (supplemental Figure 2). FCGR2B amplification is associated with rituximab resistance and may therefore contribute to treatment resistance among DHITsig⁺ tumors.^12,16  Deletions of CDKN2A lead to decreased TP53 protein¹⁷  and represent an additional mechanism of TP53 inactivation in DHITsig⁺ tumors, along with the frequent TP53 mutations described previously.³  The 22q11.22 deletion breakpoints observed by WGS were consistent with IGL VJ recombination.

A single genome harbored the 11q aberration pattern, which has been described among MYC⁻ BLs.¹⁸  The SNP array copy-number data identified 6 cases with the 11q aberration: 1 each in GCB DHITsig⁺ cases with or without genetic double hit, and 4 in GCB DHITsig⁻ tumors lacking MYC or BCL2 rearrangements. Four 11q aberrations were confirmed by FISH (supplemental Figure 3). These results suggest that the 11q aberration is not associated with the DHITsig in DLBCL.

Among the genomes, there were 3 instances of focal promoter deletions of PVT1, including 1 in the double-minute chromosome (Figure 2C). The SNP array copy-number data did not identify any additional cases with this deletion. The PVT1 long noncoding RNA gene contains intragenic enhancers that drive expression of PVT1, but in the absence of an active PVT1 promoter they instead drive expression of MYC.¹⁹  Although the number of occurrences limits our power to determine whether PVT1 deletions are associated with DHITsig positivity, these results suggest that this is a rare, albeit recurrent, event that could promote MYC expression in DHITsig⁺ tumors.

In summary, genetic events affecting both MYC and BCL2 were identified in 13 of 20 DHITsig⁺ tumors that were deemed non–HGBL-DH/THs by breakapart FISH. The identification of FISH-cryptic MYC and BCL2 rearrangements in 6 cases highlights the limitation of FISH in identifying HGBL-DH/TH. The DHITsig phenotype can thus be attributed to a range of genetic events affecting MYC, BCL2, and downstream pathways, some of which are not revealed even by WGS, demonstrating that gene expression profiling is a more appropriate method to identify this biological group that constitutes high-risk GCB-DLBCL.

This research was supported by grants from the Terry Fox Research Institute (1061) and Canadian Institute for Health Research (R.D.M.) and MITACS Accelerate (R.D.M. and L.K.H.), a Scholar Award from the ASH Foundation (R.D.M.), and the BC Cancer Foundation. R.D.M. is a Michael Smith Foundation for Health Research Scholar.

Contribution: L.K.H., B.M.G., J.T., M.A., C.K.R., A.J., and R.D.M. analyzed and interpreted the data; L.K.H. generated all figures; M.B., B.M., and S.B.-N. processed the tissues and performed FISH assays; and L.K.H., D.W.S., and R.D.M. conceived the study and wrote the manuscript.

Conflict-of-interest disclosure: A.J., D.W.S. and R.D.M. are named inventors on a patent application pertaining to the DHITsig. The remaining authors declare no competing financial interests.

Correspondence: Ryan D. Morin, Department of Molecular Biology and Biochemistry, Simon Fraser University, 8888 University Dr, Burnaby, BC, Canada; e-mail: rdmorin@sfu.ca; and David W. Scott, Department of Lymphoid Cancer Research, BC Cancer, 675 West 10th Ave, Vancouver, BC, Canada; e-mail: dscott8@bccancer.bc.ca.