Main

MYC is a proto-oncogene that encodes the Myc protein, which is critical for cell proliferation, growth, metabolism, differentiation, apoptosis, and immune responses. In mouse models, Myc inactivation induces sustained tumor regression via both tumor cell-intrinsic and host-dependent mechanisms.1 The selective small-molecule bromodomain inhibitor JQ1 has a potent antiproliferative effect that was associated with the effective downregulation of MYC and Myc target genes2 rendering this agent to have great therapeutic potential.3

The chromosomal rearrangement or translocation involving MYC and other genes (most commonly the immunoglobulin heavy-chain gene (IGH) locus) leads to Myc overexpression and occurs in ~10% of diffuse large B-cell lymphomas (DLBCLs).4, 5, 6, 7, 8 DLBCL is the most common type of non-Hodgkin's lymphoma, and among MYC-rearranged aggressive lymphomas, DLBCL is the entity that clinicians most commonly encounter.9, 10 Several studies have reported that MYC translocations independently predicted significantly poor survival in DLBCL patients.11, 12, 13, 14, 15, 16 However, other studies found inconsistent results17, 18, 19 or limitations of its prognostic significance.20, 21

The clinical significance of Myc overexpression in DLBCL has also been the source of much attention and controversy. Several groups including ours have found that DLBCL with high Myc protein expression detected by immunohistochemistry had inferior survival.5, 13, 22, 23, 24 Furthermore, the poor prognosis associated with Myc overexpression was contributed by cases with Myc/Bcl-2 coexpression—‘double-positive lymphoma’ (DPL)—which account for 18–44% of DLBCLs.13, 20, 23, 24, 25, 26 However, one study showed that the prognostic value of double-positive lymphoma was lost in younger DLBCL patients with poor prognosis.27 Inconsistent results have been reported regarding whether the prognostic significance of Myc or Myc/Bcl2 protein expression depends on MYC or MYC/BCL2 gene rearrangement status or not.8, 13, 20 Other issues include whether Myc/Bcl2 immunohistochemistry is robust and reproducible,28 and that immunohistochemistry cutoff values have varied among different study groups,13, 20, 23, 24, 25, 26 which may affect the specificity of this combined biomarker for poorer prognosis.

Possible molecular mechanisms underlying the inconsistent clinical results may include presence or absence of other genetic abnormalities and oncogenic pathways,29, 30, 31 as well as another aspect of Myc function: promoting apoptosis.32 Moreover, researchers have recently reported findings that Myc is a universal amplifier of 10–15% in human genome, suggesting that Myc function is nonspecific and that the consequences of MYC activities are affected by pre-existing molecular programs in the tumor cells.33, 34 Therefore, tumor biology associated with MYC/Myc (designated MYC herein) abnormalities may have important roles in the observed adverse prognosis.

Taken together, both MYC gene rearrangement and Myc protein overexpression have been correlated with significantly adverse prognosis in DLBCL. However, how much of these two biomarkers overlap and differ, how much their associated tumor biology affects the prognostic effects, and whether the MYC functional role is molecular context-dependent are not very clear. In this study, we compared the occurrence and clinicopathologic features of patients stratified by MYC rearrangement and Myc expression status, and analyzed the differential prognosis and gene expression profiling associated with these MYC abnormalities in a well-characterized DLBCL cohort to assess the utility of these two genetic and protein biomarkers and explore the prognostic determinants. This study is important for achieving the goal of precision medicine in DLBCL.

Patients and methods

Patients

The study cohort consisted of 539 R-CHOP-treated patients with de novo DLBCL from the International DLBCL Rituximab-CHOP Consortium Program, including 466 cases from the training set of a previous study,24 and additional 73 cases with either Myc immunohistochemistry or MYC gene rearrangement status determined. The diagnostic criteria, review process, and eligibility and exclusion criteria have been described previously.35, 36 The cell-of-origin classification as either the germinal center B-cell-like (GCB) or activated B-cell-like (ABC) subtype was determined using gene expression profiling and/or immunohistochemistry for CD10, BCL6, GCET-1, FOXP1, and MUM1 using the Visco-Young and/or Choi algorithms as described previously.24, 35, 36 Totally, 276 cases were classified as GCB, 259 cases were classified as ABC, and 4 cases were unclassifiable. All patients underwent standard R-CHOP or R-CHOP-like therapy, and the median follow-up time was 45 months (range, 30–176.1 months). This study was conducted in accordance with the Declaration of Helsinki and was approved as being of minimal to no risk or as exempt by the institutional review boards of all participating centers, including The University of Texas MD Anderson Cancer Center.

Fluorescence In Situ Hybridization, Immunohistochemistry, and Gene Sequencing

MYC translocation was detected by fluorescence in situ hybridization using two probes (a locus-specific identifier IGH/MYC/CEP8 tri-color dual-fusion probe and a locus-specific identifier MYC dual-color break-apart probe) (n=344). Myc expression was assessed by immunohistochemistry using tissue microarray sections and a monoclonal anti-(c)MYC antibody, clone Y69 (Epitomics, Burlingame, CA, USA) (n=535). The experimental techniques and scoring processes have been described previously.15, 21, 24

Evaluation of other biomarker expression by immunohistochemistry was also performed on tissue microarray sections using corresponding antibodies: p53 (DO-7; Dako, Carpinteria, CA, USA), MDM2 (IF2; Calbiochem, Billerica, MA, USA), Bcl-2 (Clone-124; Dako), Ki-67 (MIB-1; Dako), pAKT (726E11; CST), Bcl-6 (PG-B6p; DAKO), FOXP1 (EPR4113; Abcam), MUM1/IRF4 (Dako), CD10 (56C6; Vantana), CD30 (clone BerH2; Dako), BLIMP-1 (EPR16655; Epitomics), NF-κB subunits (Dako), CXCR4 (Abcam, Cambridge, MA, USA), and survivin (EP2880Y; Epitomics). BCL6 and BCL2 translocations and TP53 mutations were detected as described previously.15, 21, 24, 35, 36, 37, 38, 39, 40, 41, 42

Gene Expression Profiling

Gene expression profiling for 457 patients was performed using the Affymetrix GeneChip Human Genome HG-U133 Plus 2.0 Array as described previously.21, 24, 35, 36, 37 The CEL files have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus repository (GSE31312). Microarray data were normalized for further supervised clustering analysis. Multiple t-tests were used to identify differentially expressed genes and the P-values obtained were corrected for the false discovery rate using the β-uniform mixture method.

Statistical Analysis

The clinical and pathobiological features of DLBCL patients at the time of presentation were compared using the χ2 test. The mean expression levels of biomarkers between DLBCL groups with or without MYC abnormalities were also compared by unpaired t-test. Overall survival (OS) was calculated from the time of diagnosis to last follow-up or death from any cause. Progression-free survival (PFS) was calculated from the time of diagnosis to disease progression, disease relapse, or death from any cause. Patients who were alive and disease progression-free at last follow-up were censored. Survival analysis was performed using the Kaplan–Meier method with the Prism 5 program (GraphPad Software, San Diego, CA, USA), and differences in survival were compared using the log-rank (Mantel–Cox) test. Multivariate survival analysis was performed using a Cox proportional hazards regression model with the SPSS software program (version 19.0; IBM Corporation, Armonk, NY, USA). All differences with P≤0.05 were considered statistically significant.

Results

Myc Expression in Germinal Center B-Cell-Like and Activated-B-Cell-Like Subtypes of DLBCL

We observed variable levels of nuclear Myc expression in DLBCL (n=535) (Figures 1a–c). The mean expression level of Myc in ABC-DLBCL was significantly higher compared with that in GCB-DLBCL (Figure 1d). Among the cases successfully tested for MYC rearrangement status (n=344), 27 (16.3%) of 166 GCB-DLBCL cases and 13 (7.3%) of 177 ABC-DLBCL cases had MYC rearrangement/translocation. Most but not all MYC-rearranged cases (MYC-R+) had high MYC-mRNA and Myc expression, with an average percentage of Mychigh nuclei in the tumor samples of 70.5% (Supplementary Figure S1A). We thus set the cutoff for Myc overexpression (Mychigh) at ≥70%, so that Myc expression levels in MYC-non-rearranged (MYC-R−) Mychigh patients comparably ‘matched’ those in MYC-R+/Mychigh cases (Supplementary Figure S1C) in comparisons of clinical and biological significance of Myc overexpression vs MYC translocation.

Figure 1
figure 1

(a–c) Histograms of Myc expression according to immunohistochemistry staining in all diffuse large B-cell lymphoma (DLBCL) patients, germinal center B-cell-like (GCB) DLBCL patients, and activated B-cell-like (ABC) DLBCL patients. (d) The mean Myc protein levels of ABC-DLBCL were significantly higher compared with those of GCB-DLBCL. (e and f) MYC rearrangement-positive (MYC-R+) DLBCL without high Myc immunohistochemistry scores (either GCB or ABC subtype) had significantly lower MYC-mRNA levels. (g) MYC-R+ DLBCL patients without high Myc immunohistochemistry scores had significantly better overall survival compared with MYC-R+/Mychigh DLBCL. Mychigh, high Myc protein expression; Myclow, low Myc protein expression.

Full size image

Using this cutoff, we found that 175 (32.7%) of 535 DLBCL patients, including 76 (27.9%) of 272 GCB-DLBCL patients and 98 (37.8%) of 259 ABC-DLBCL patients, were Mychigh (Figures 2a and b). Nineteen (73%) of the 26 MYC-R+ GCB patients (one MYC-R+ GCB case had no expression data available) and 7 (54%) of the 13 MYC-R+ ABC patients were Mychigh, who had significantly higher MYC transcripts and significantly worse survival compared with the MYC-R+/Myclow cases (Figures 1e–g and Supplementary Figures S1D). Compared with MYC-R− patients, MYC-R+/Mychigh patients had significantly worse survival in GCB-DLBCL (P=0.0001 for OS and P<0.0001 for PFS) but not in ABC-DLBCL (P=0.56 for OS and P=0.14 for PFS) (Supplementary Figures S1E and F). Comparison of MYC expression levels between the MYC-R+ GCB, MYC-R− GCB, MYC-R+ ABC, and MYC-R− ABC Mychigh groups revealed that MYC-R+/Mychigh GCB-DLBCL had significantly higher Myc protein expression levels compared with all other subsets, and that among both the GCB and ABC subtypes, MYC-R+/Mychigh cases had a significantly higher level of MYC-mRNA levels compared with MYC-R−/Mychigh cases (Figures 2c and d).

Figure 2
figure 2

Occurrence of MYC rearrangement (MYC-R) and overexpression (Mychigh) and MYC expression levels in diffuse large B-cell lymphoma (DLBCL). (a and b) Schematic diagrams showing the frequencies of MYC translocation (MYC-R+) and Myc overexpression (Mychigh) and their overlaps in germinal center B-cell-like (GCB) and activated B-cell-like (ABC) DLBCL in the current study cohort. (c and d) Comparisons of Myc protein and MYC-mRNA expression levels in DLBCL patients with or without MYC dysregulation. Among Mychigh groups, MYC-R+ GCB-DLBCL had highest Myc protein levels; MYC-R+ DLBCL (both GCB and ABC subtypes) had significantly higher MYC-mRNA levels compared with MYC-R− DLBCL. Note: Each dot represents one patient in the study cohort. MYC-R+, MYC rearrangement-positive; MYC-R−, MYC rearrangement-negative; Mychigh, high Myc protein expression/Myc overexpressing; Myclow, low Myc protein expression; MYC-RN/A, MYC rearrangement status not available.

Full size image

Clinical Parameters Associated with Myc Overexpression and MYC Rearrangement

We analyzed the clinical parameters associated with Myc overexpression (Table 1) and MYC rearrangement (Supplementary Table 1). We further compared MYC-R+/Mychigh and MYC-R−/Mychigh cases (Supplementary Table S2). Results of significant characteristics are summarized in Supplementary Table S3. Largely, both Myc overexpression and MYC rearrangement could identify a subgroup of patients with adverse clinical features in GCB-DLBCL, whereas patients with these abnormalities did not show such distinct characteristics among ABC-DLBCLs.

Table 1 Clinicopathologic features of DLBCLa
Full size table

For example, in GCB-DLBCL, Mychigh patients compared with Myclow patients more often had stage III/IV disease, ≥2 extranodal sites, ECOG performance status ≥2, tumor size ≥5 cm, International Prognostic Index >2, bone marrow involvement at clinical presentation, and less likely to have complete response. In ABC-DLBCL, Mychigh patients compared with Myclow patients had a higher proportion of women, and did not show significant association with other clinical parameters. The GCB and ABC subtypes of Mychigh patients differed significantly only in their frequencies of patients with age ≥60 years (Table 1). Similarly, in GCB-DLBCL, MYC-R+ patients were enriched in patients with adverse clinical features, whereas in ABC-DLBCL, MYC rearrangement was only associated with primary extranodal origin. The GCB subtype compared with the ABC subtype of MYC-R+ patients showed trends toward having a higher proportion of patients with bone marrow involvement and non-complete treatment response (Supplementary Table S1).

In both the GCB- and ABC-DLBCL groups, MYC-R+/Mychigh patients appeared to have similar clinical features with MYC-R−/Mychigh patients (except complete response rate), yet may have elevated serum lactate dehydrogenase levels and higher frequency of extranodal origin (Supplementary Table S2).

Molecular Biomarkers Associated with Myc Overexpression and MYC Rearrangement

Mychigh DLBCL compared with Myclow DLBCL, in addition to having higher Myc levels and significantly higher frequencies of MYC translocation and ABC subtype (Table 1), had higher expression levels of p53, MDM2, Bcl-2 (Mychigh ABC-DLBCL only), Bcl-6, FOXP1, IRF4/MUM1 (Mychigh ABC-DLBCL only), Ki-67, pAKT, CXCR4, and CD10 (Mychigh GCB-DLBCL only), but lower expression levels of BLIMP-1 (in ABC-DLBCL only) and nuclear expression of the NF-κB subunits c-Rel and RelB (in ABC-DLBCL only, P=0.019, figure not shown) (Figure 3). At the mRNA level, only MDM2 (in GCB-DLBCL only), FOXP1 (in ABC-DLBCL only), IRF4 (in ABC-DLBCL only), and MME/CD10 transcript levels showed corresponding correlations consistent with those at the protein level (Figures 4a–d). PRDM1 mRNA levels were significantly lower in Mychigh GCB-DLBCL compared with that in Myclow GCB-DLBCL (P=0.0005). The GCB and ABC subtypes of Mychigh patients differed in frequencies of MYC translocation, BCL2 translocation, TP53 mutation (all higher in the GCB subtype), and expression levels of Bcl-2, p50, c-Rel (all higher in the ABC subtype), pAKT (higher in the GCB subtype), and cell-of-origin-related biomarkers (Bcl-6, BLIMP-1, GCET-1, CD10, FOXP1, MUM1) (Table 1, Figure 3, and Supplementary Table S3).

Figure 3
figure 3

Expression of pathobiological markers in germinal center B-cell-like (GCB) and activated B-cell-like (ABC) diffuse large B-cell lymphoma with or without Myc overexpression. In both GCB- and ABC-DLBCL, Myc overexpression was associated with significantly higher levels of p53 (a), MDM2 (b), Ki-67 (d), Bcl-6 (f), and FOXP1 (g), and a significantly lower level of c-Rel (k). In GCB-DLBCL only, Myc overexpression was associated with significantly higher levels of CD10 (e), pAKT (j), and CXCR4 (l). In ABC-DLBCL only, Myc overexpression was associated with significantly higher levels of Bcl-2 (c) and MUM1 (h) levels and a significantly lower level of BLIMP-1 (i). Mychigh, high Myc protein expression; Myclow, low Myc protein expression.

Full size image
Figure 4
figure 4

(a–d) Myc overexpression was correlated with significantly higher levels of MDM2, IRF4, and MME/CD10 mRNA expression in germinal center B-cell-like (GCB) diffuse large B-cell lymphoma and significantly higher levels of FOXP1 and IRF4 mRNA expression in activated B-cell-like (ABC) diffuse large B-cell lymphoma (DLBCL). (e) The MYC-R+/Mychigh compared with MYC-R−/Mychigh group had significantly lower levels of MUM1 expression. (f–j) The MDM2, CD30, CD10, and FOXP1 levels in Mychigh GCB-DLBCL patients with MYC translocation were significantly different from those of Mychigh GCB-DLBCL patients without MYC translocation. (k and l) The Bcl-6 and pAKT levels of Mychigh ABC-DLBCL patients with MYC translocation were significantly lower compared with those of Mychigh ABC-DLBCL patients without MYC translocation. Mychigh, high Myc protein expression; Myclow, low Myc protein expression; MYC-R+, MYC rearrangement-positive; MYC-R−, MYC rearrangement-negative; MYC-RN/A, MYC rearrangement status not available.

Full size image

Some of the pathobiological associations with Mychigh DLBCL were shared by MYC-R+ GCB-DLBCL patients (compared with MYC-R− GCB-DLBCL; Supplementary Tables S1 and S3). On the other hand, in addition to the difference in Myc activation mechanisms (MYC translocation or not), MYC-R+/Mychigh compared with MYC-R−/Mychigh patients had higher frequencies of GCB subtype and BCL2 translocation, and lower frequencies of BCL6 translocation, MUM1, CD30, and p52 expression (Supplementary Table S2 and S3). Among GCB-DLBCL patients, MYC-R+/Mychigh compared with MYC-R−/Mychigh patients had significantly higher expression levels of MDM2, CD10, and FOXP1 but a lower expression level of CD30; among ABC-DLBCL patients, MYC-R+/Mychigh compared with MYC-R−/Mychigh was associated with lower expression levels of Bcl-6 and pAKT (Figures 4f–l).

Prognostic Effect of Myc Overexpression with or without MYC Translocations and Associated Gene Expression Profiling Signatures

We combined the survival analysis with gene expression profiling comparisons to assess the role of MYC abnormalities in DLBCL and to study potential mechanisms. As in earlier studies,13, 22, 23, 24 in the present study, high expression levels of Myc was associated with significantly worse OS and PFS in both GCB- and ABC-DLBCL (Supplementary Figure S2). Similar prognostic impact was also shown by high MYC-mRNA levels in GCB- and ABC-DLBCL (Supplementary Figures S3A–D). Gene expression profiling analysis showed that Mychigh GCB- and ABC-DLBCL had different gene expression profiling signatures (Table 2, Supplementary Figures S2D and F, and Supplementary Table S4A). The significantly differentially expressed genes between these two groups included CDCA7L, IGF2BP3, and RUVBL2, which are known to have roles in the oncogenic transformation by MYC or to interact with MYC.

Table 2 Gene expression profile signatures of Myc protein overexpression in DLBCL, GCB-DLBCL, and ABC-DLBCL
Full size table

MYC rearrangement did not show prognostic significance in the Myclow cases [for OS: P=0.25 (OS of MYC-R+/Myclow patients was slightly better than MYC-R−/Myclow DLBCL); for PFS: P=0.71]. Dividing the Mychigh cases with unfavorable prognosis into MYC-R+/Mychigh and MYC-R−/Mychigh two types and comparing their prognosis and gene expression profiling features, we further found that MYC-R+/Mychigh GCB-DLBCL and MYC-R−/Mychigh ABC-DLBCL were the main contributors to the overall worse prognosis and distinct gene expression profiling signatures of the Mychigh GCB-DLBCL and Mychigh ABC-DLBCL groups, respectively.

In the GCB-DLBCL group, MYC-R+/Mychigh patients had significantly worse survival than both Myclow and MYC-R−/Mychigh patients did. Although MYC-R−/Mychigh GCB-DLBCL showed trends towards worse survival compared with the overall Myclow GCB-DLBCL (P=0.40 for OS, Figure 5a; P=0.48 for PFS; Supplementary Figure S4A), and the MYC-R−/Myclow GCB-DLBCL (for OS: P=0.32; Supplementary Figure S4C; for PFS: P=0.21), the differences were not significant. Biologically, only MYC-R+/Mychigh (but not MYC-R−/Mychigh) GCB-DLBCL compared with Myclow GCB-DLBCL showed a distinct gene expression profiling signature (false discovery rate <0.01) (Figure 5b,Supplementary Table S4B, and Table 3), and this signature mostly overlapped the Mychigh gene expression profiling signature identified in the overall GCB-DLBCL group (Table 2 and Supplementary Figure S2D) involving cell proliferation, gene expression, metabolism, apoptosis, microenvironment and immune response, and microRNA genes. In contrast, the MYC-R−/Mychigh GCB-DLBCL patients compared with Myclow GCB-DLBCL (Supplementary Figures S4D, Table 3, and Supplementary Table S4B) or MYC-R+/Mychigh GCB-DLBCL (Figure 5c, Table 3, and Supplementary Table S4B) only showed a few differentially expressed genes below false discovery rate threshold of 0.30 (not including MYC), most of which are involved in cell proliferation, gene expression, ribosome biogenesis, and metabolism, suggesting the presence of heterogeneity and post-transcriptional regulation of MYC as a cause of Myc overexpression within the MYC-R−/Mychigh GCB-DLBCL group. The comparison between MYC-R+ GCB-DLBCL and MYC-R− GCB-DLBCL overall is shown in Supplementary Figure S4E.

Figure 5
figure 5

Combined prognostic and biologic analysis of germinal-center-B-cell-like (GCB) and activated-B-cell-like (ABC) diffuse large B-cell lymphoma (DLBCL) with or without MYC translocation (MYC-R) and/or Myc overexpression (Mychigh) in the current study cohort. (a) GCB-DLBCL patients with both MYC translocation and Myc overexpression (MYC-R+/Mychigh) had significantly worse overall survival compared with GCB-DLBCL patients with low Myc expression (Myclow) and Mychigh patients without MYC translocation (MYC-R−/Mychigh). The MYC-R−/Mychigh group did not have significant poorer survival compared with the Myclow group. (b) Genes significantly differentially expressed between MYC-R+/Mychigh GCB-DLBCL patients and Myclow GCB-DLBCL patients (false discovery rate <0.01, fold change >2.38). (c) Genes significantly differentially expressed between MYC-R+/Mychigh and MYC-R−/Mychigh GCB-DLBCL patients (false discovery rate <0.30). (d) Only MYC-R−/Mychigh but not MYC-R+/Mychigh ABC-DLBCL patients had significantly OS compared with Myclow ABC-DLBCL patients. (e) Genes significantly differentially expressed between MYC-R−/Mychigh ABC-DLBCL and Myclow ABC-DLBCL (false discovery rate <0.01, fold change >1.57). (f) Genes significantly differentially expressed between ABC and GCB subtypes of MYC-R+ DLBCL. (g) MYC-R+/Mychigh GCB-DLBCL showed trend towards worse progression-free survival compared with MYC-R−/Mychigh ABC-DLBCL patients with a borderline P-value. (h) Genes significantly differentially expressed between MYC-R+ and MYC-R−DLBCL (false discovery rate <0.01, fold change >1.66). (i) Genes significantly differentially expressed between MYC-R+/Mychigh and MYC-R−/Mychigh DLBCL (false discovery rate <0.01, fold change >2.05). ABC, activated-B-cell-like; GCB, germinal-center-B-cell-like; Mychigh, high Myc protein expression; Myclow, low Myc protein expression; MYC-R+: MYC rearrangement-positive; MYC-R−, MYC rearrangement-negative.

Full size image
Table 3 Gene expression profile signatures of Myc protein overexpression in GCB-DLBCL
Full size table

In contrast, in ABC-DLBCL, the prognostic significance (Figure 5d) and gene expression profiling features of Myc overexpression did not depend on MYC translocation. MYC-R−/Mychigh ABC-DLBCL had significantly poorer survival compared with the overall Myclow or MYC-R−/Myclow ABC-DLBCL (Figure 5d and Supplementary Figures S4G–I) and distinct gene expression profiling signatures (false discovery rate <0.01; Figure 5e and Supplementary Table S4C). The genes in the MYC-R−/Mychigh ABC-DLBCL gene expression profiling signature, overlapped with differentially expressed genes between the overall Mychigh vs Myclow ABC-DLBCL patients, included typical Myc targets mainly related to cell proliferation, the cell cycle, gene expression, ribosome biogenesis, metabolism (Table 4), and cooperating oncogenes such as RUVBL2, as well as IGF2BP3 involved in post-transcriptional regulation of MYC, and HINT1 modulating p53 levels and the p53 pathway. Different from the MYC-R−/Mychigh ABC-DLBCL cases, MYC-R+/Mychigh ABC-DLBCL compared with overall Myclow ABC-DLBCL only showed nonsignificant trends towards worse OS and PFS (Figure 5d and Supplementary Figures S4G–I) and only a few differentially expressed genes (false discovery rate <0.30, Supplementary Figure S4F). No genes were found differentially expressed between MYC-R+/Mychigh ABC-DLBCL and MYC-R−/Mychigh ABC-DLBCL, or between MYC-R+/Mychigh ABC-DLBCL and the overall MYC-R− ABC-DLBCL group. This is in contrast with the distinct gene expression profiling feature of MYC-R+/Mychigh GCB-DLBCL shown in Figure 5b and Supplementary Figures S4E. Comparison between overall MYC-R+/Mychigh ABC-DLBCL and MYC-R+/Mychigh GCB-DLBCL indicated their different and potentially heterogeneous tumor biology (Figure 5f) (between MYC-R+/Mychigh and MYC-R+/Myclow GCB-DLBCL, or between MYC-R+/Mychigh and MYC-R+/Myclow ABC-DLBCL, we did not find significant differentially expressed genes below false discovery rate thresholds of 0.05–0.50). MYC-R+ ABC-DLBCL appears to have decreased B-cell receptor signaling compared with MYC-R+ GCB-DLBCL (false discovery rate <0.30; Table 4 and Supplementary Table S4C). However, the ‘loss’ of the gene expression profiling signature may also be due to the small case number.

Table 4 Gene expression profile signatures of Myc overexpression in ABC-DLBCL
Full size table

Comparing between GCB and ABC subtypes of Mychigh patients, the overall GCB and ABC subtypes of Mychigh patients showed no significant difference in survival (ABC subtype showed nonsignificant trends towards poorer survival; Supplementary Figures S3E and F). However, MYC-R+/Mychigh GCB-DLBCL showed unfavorable trends compared with Mychigh ABC-DLBCL (either MYC-R− or MYC-R+). The P-value for the difference in PFS between the MYC-R+/Mychigh GCB-DLBCL and MYC-R−/Mychigh ABC-DLBCL patients was 0.058 (Figure 5g). MYC-R+/Mychigh GCB-DLBCL compared with MYC-R−/Mychigh ABC-DLBCL or Myclow ABC-DLBCL showed distinct gene expression profiling signatures overlapping with the one comparing with MYC-R− GCB-DLBCL. Comparisons between MYC-R+ DLBCL and MYC-R− DLBCL overall (regardless of Mychigh or Myclow, GCB or ABC), and between MYC-R+/Mychigh DLBCL overall and MYC-R−/Mychigh DLBCL overall (regardless of GCB or ABC) are shown in Figures 5h and i (Table 5 and Supplementary Table S4D). These analyses suggest the distinctive biology of MYC-R+/Mychigh GCB-DLBCL and cell of origin during lymphomagenesis may have a role in defining its biological feature.

Table 5 Gene expression profile signatures of MYC-R+ GCB-DLBCL and MYC-R+ DLBCL overall
Full size table

Concurrent Evaluation of Myc/Bcl-2 Overexpression and MYC Translocations

We examined whether the prognostic value of Myc overexpression is contributed by or depends on the molecular marker associations with Mychigh DLBCL as shown in Figure 3. As shown previously, the prognostic significance of Mychigh and Bcl-2high in DLBCL significantly depend on each other.13, 20, 23, 24, 25, 26 In addition, to a certain extent, the prognostic significance of Mychigh showed dependence on CXCR4,40 FOXP1, and MUM1 overexpression, which are also associated with Bcl-2 overexpression (the association of MUM1 was only in ABC but not in GCB), high Ki-67 (for OS but not for PFS), and low BLIMP-1 expression (Supplementary Figure S5).

The lack of prognostic significance of Myc overexpression without MYC translocation in GCB-DLBCL (Figure 5a) could be attributable to the favorable prognosis of Mychigh/Bcl-2low GCB-DLBCL patients (Figures 6a and b and Supplementary Figures S6A and B). In contrast, in the ABC-DLBCL group, most MYC-R−/Mychigh patients (64%) also had Bcl-2 overexpression, which contributed to this group’s worse survival. Concurrent Bcl-2 overexpression also had significant prognostic impact in MYC-R+/Mychigh GCB-DLBCL (Figures 6c and d), but not in overall MYC-R+ ABC-DLBCL, MYC-R+/Mychigh ABC-DLBCL, or MYC-R+/Myclow cases (Supplementary Figures S6C and D); however, we could not distinguish whether the synergy in GCB is with Bcl-2 protein or with BCL2 translocation. Compared with Myclow, MYC-R−/Mychigh GCB-DLBCL, or overall MYC-R− GCB-DLBCL, MYC-R+/Mychigh GCB-DLBCL showed nonsignificant trends toward higher Bcl-2 levels (P=0.34, 0.27, and P=0.17, respectively; figures not shown; Supplementary Tables S1 and 2). The gene expression profiling analysis revealed no significant differentially expressed genes between MYC-R+ GCB-DLBCL with Bcl-2 expression and MYC-R+ GCB-DLBCL without Bcl-2 expression. Correspondingly, patients with MYC/BCL2 double-hit lymphoma and those with single MYC rearrangements did not have significant differentially expressed genes, which is consistent with an earlier study.16 The survival of MYC-R+ Mychigh/Bcl-2high GCB-DLBCL patients was markedly poorer compared with that of MYC-R− Mychigh/Bcl-2high GCB-DLBCL patients but this difference was not significant (Figures 6e and f). However, this difference was significant when we used cutoffs of ≥40% or >50% for Myc overexpression.

Figure 6
figure 6

Prognostic analysis in Mychigh germinal-center-B-cell-like (GCB) diffuse large B-cell lymphoma (DLBCL) and in MychighBcl-2high DLBCL patients. (a and b) Among MYC-non-rearranged (MYC-R−) Mychigh GCB-DLBCL patients, those with Bcl-2 overexpression had significantly worse overall survival and progression-free survival compared with those who did not have Bcl-2 overexpression. (c and d) Among patients with MYC-R+/Mychigh GCB-DLBCL, those with Bcl-2 overexpression had significantly poorer progression-free survival compared with those who did not have Bcl-2 overexpression. The P-value for overall survival was not significant. (e and f) Among patients with MychighBcl-2high GCB-DLBCL, those with MYC rearrangement had poorer overall and progression-free survival compared with those without MYC rearrangement did, but these differences were not statistically significant. (g and h) Among MychighBcl-2high DLBCL patients, GCB-DLBCL patients with MYC rearrangement had significantly poorer overall and progression-free survival than MychighBcl-2high patients without MYC rearrangement did. ABC, activated-B-cell-like; Bcl-2high, high Bcl-2 protein expression; Bcl-2low, low Bcl-2 protein expression; GCB, germinal-center-B-cell-like; Mychigh, high Myc protein expression; MYC-R+, MYC rearrangement-positive; MYC-R−, MYC rearrangement-negative.

Full size image

Although concurrent evaluation of Myc and Bcl-2 expression improves the specificity of Myc biomarker in DLBCL, MYC rearrangement continues to demonstrate clinical value. In our cohort, there was no significant difference in survival between patients with the GCB or ABC subtypes of MYC-R− Mychigh/Bcl-2high (ie, double-positive lymphoma). Among overall double-positive lymphoma patients, patients with MYC-R+ GCB double-positive lymphoma patients had significantly poorer survival compared with those double-positive lymphoma patients without MYC rearrangement (Figures 6g and h). The survival of MYC-R+ double-positive lymphoma patients with the GCB or ABC subtypes appeared to be different, although P-values were not significant and the case numbers were small.

Discussion

Previously, we reported MYC translocation and Myc overexpression as adverse prognostic biomarkers individually.15, 24 In this study, we analyzed the occurrence of MYC translocation and Myc overexpression in GCB- and ABC-DLBCL (Figures 2a and b), and compared with the clinical features and tumor biology associated with these two overlapping biomarkers which have not been done by previous studies, and examined the dependence/independence between their indicated prognoses. To reduce the difference in Myc expression levels as a causing factor for the differential prognostic effect and tumor biology between MYC translocation and Myc overexpression activated by other mechanisms, the cutoff for Mychigh was set at ≥70% in this study, which is the optimal cutoff for predicting MYC translocation according to previous studies.6, 7, 31 Using this cutoff, the frequency of Myc overexpression was 32.7% in overall DLBCL patients (close to the frequencies by other independent studies13, 22, 23), 73% in MYC-R+ GCB-DLBCL, and 54% in MYC-R+ ABC-DLBCL (lower than the 93% in Green et al6 and 85% in Horn et al10). Tables 6–7 summarize the results of published MYC studies including ours. The current study shows that MYC translocation and Myc overexpression in DLBCL only partially overlaps and evaluation of both is critical for stratifying patients and predicting treatment outcomes. MYC-rearranged DLBCL without Myc protein overexpression did not show significantly worse survival (Figure 1g). However, evaluation of Myc overexpression alone is also insufficient for predicting poorer prognosis in MYC-R− cases, especially in MYC-R− GCB-DLBCL (Figure 5a). This low specificity for Mychigh as an adverse prognostic factor can be improved by concurrent evaluation of Bcl-2 expression (Supplementary Figures S5 and S6), which is overexpressed mainly in ABC-DLBCL and associated with Myc overexpression. However, the survival of MYC-rearranged MychighBcl-2high GCB double-positive lymphoma patients remains significantly worse than other double-positive lymphoma patients (Figures 6g and h).

Table 6 Summary of frequencies and prognostic significance of MYC-R in DLBCL in the literature and reported by the current study
Full size table
Table 7 Summary of frequencies and prognostic significance of Myc overexpression (Mychigh) and MYC mRNA levels in DLBCL in the literature and reported by the current study
Full size table

The biological investigation (in this regard, ≥70% is a better cutoff compared with ≥40% for Mychigh in our cohort) revealed that MYC activation was associated with significantly increased or decreased expression of genes and proteins involved in cell proliferation (e.g., pAKT, Ki-67), apoptosis (p53, Bcl-2, FAS, BCL2A1, PEG10, HINT1, TRAF1), differentiation (PRDM1, BLIMP-1, BACH2 (which represses PRDM1)43), noncoding RNAs (eg, LINC00152, GAS5, SNHG1, NAPSB) and microRNAs, microenvironment, and immune responses, as well as cell-of-origin markers (Figure 3 and Tables 3, 4, 5). Corresponding to the differences in prognostic effect between various Mychigh subtypes, only MYC-R+/Mychigh GCB-DLBCL and MYC-R−/Mychigh ABC-DLBCL, but not MYC-R−/Mychigh GCB-DLBCL or MYC-R+/Mychigh ABC-DLBCL, demonstrated distinct gene expression profiling feature compared with the Myclow subgroup (Figures 5b and e and Supplementary Figures S4D and F). Remarkably, MYC-R+/Mychigh GCB-DLBCL had a characteristic gene expression profiling in DLBCL. Myc activation was associated with gene expression profiling signatures suggesting decreased immune responses and a number of microRNAs overlapped with the molecular Burkitt lymphoma signature,29, 30 including MIR17HG, which was markedly upregulated, and MIR21 (ref. 44) and MIR155HG, which were significantly downregulated. Other studies have shown these microRNAs to be regulated by the Myc, Bcl-6, STAT3, and NF-κB pathways, and the MIR17HG locus was frequently amplified in Burkitt lymphoma.45, 46, 47, 48 Downregulation of MIR155HG expression may contribute to the pathogenesis of MYC translocation, as miR-155 suppresses activation-induced cytidine deaminase, which mediates MYC/IGH translocation.49 These microRNA signatures may also be implicated in defining the gene expression profiling features of MYC-R+/Mychigh GCB-DLBCL. In lymphoma cells, ectopic expression of miR-155 is associated with downregulation of IGJ, FAS, SMAD3/5, and BACH1, as well as HLA genes.45 IGJ, FAS, and genes involved in BMP/SMAD pathways (such as SMAD1, BMP3, and BMP7), as well as SMARCA4 (miR-21 target gene50), were upregulated in MYC-R+ GCB-DLBCL (Tables 3 and 5). miR-155 target gene BCL2 showed increased Bcl-2 protein expression in ABC-DLBCL (Figure 3c). Phosphorylated SMAD proteins have roles in BMP-induced cell growth inhibition (this inhibition can be overcame via MYC translocations),51, 52 immunoregulation,53 and ionizing radiation-induced double-strand break signaling.54 In addition, in MYC-R+/Mychigh GCB-DLBCL, antiapoptotic PEG10 and ZNF385B, which modulates p53 activity resulting in cell-cycle arrest over apoptosis, were significantly upregulated.

One possible reason for the differential prognoses and gene expression profiling signatures among GCB or ABC subtype of Mychigh and MYC-R+ cases is the difference in Myc protein levels. For example, MYC-R+/Mychigh GCB-DLBCL had highest Myc expression levels (Figure 2c) and significant or nonsignificant trends towards worse prognosis compared with all other three Mychigh groups. MYC-R+ cases with low Myc protein expression had good prognosis even though having an MYC rearrangement. Different MYC translocation partners, and/or breakpoints outside of MYC gene leaving MYC repressor element intact during rearrangement,4, 17 may cause the low MYC-mRNAs in these MYC-R+/Myclow DLBCL cases (Figures 1e–g). Compared with MYC-R+/Mychigh GCB-DLBCL, ABC subtype of MYC-R+/Mychigh DLBCL had similar MYC-mRNA but significantly lower Myc protein levels (Figure 2d) and trends of better survival. MYC-R−/Mychigh GCB-DLBCL compared with Myclow GCB-DLBCL had significantly higher levels of MYC-mRNA and Myc protein, but the false discovery rate for MYC upregulation in gene expression profiling analysis was high (>0.45), which may suggest either Mychigh or Myclow group are heterogeneous and the molecular mechanisms inducing Myc in this Mychigh group include post-transcriptional regulation. A previous Myc study in immature and mature GCB cells during germinal cell formation also demonstrated the lack of correlation between Myc protein and mRNA levels.55 The lack of distinct gene expression profiling signatures and better survival of MYC-R−/Mychigh GCB-DLBCL and MYC-R+/Mychigh ABC-DLBCL compared with MYC-R+/Mychigh GCB-DLBCL may indicate lower Myc activities corresponding to intracellular Myc protein levels; however, small case numbers and/or heterogeneity among these two Mychigh DLBCL groups could also be possible causes.

However, the Myc immunohistochemistry levels were similar between MYC-R−/Mychigh GCB-DLBCL and MYC-R−/Mychigh ABC-DLBCL (although GCB type had slightly lower level of MYC-mRNA) but their prognosis showed differences. MYC-R+/Mychigh ABC-DLBCL had similar Myc protein level to that of MYC-R−/Mychigh ABC-DLBCL but only the latter showed typical Myc gene expression profiling signatures (Figure 2c). Therefore, Myc-associated molecular mechanisms in GCB or ABC subtype of MYC-R+/− Mychigh cases impacted the prognostic and biological effect of Myc. GCB and ABC subtypes of Mychigh DLBCL had difference in frequencies of TP53 mutation, MYC translocation, BCL2 translocation, and Bcl-2 expression, as well as cell-of-origin biomarkers (Table 1 and Figure 3). MYC-R+/Mychigh and MYC-R−/Mychigh DLBCL had significantly different levels of MDM2, the cell-of-origin markers, and CD30 (in GCB-DLBCL) and pAKT (in ABC-DCLBL) (Figure 4), as well as significantly different gene expression profiling signatures at the mRNA level (Figure 5i). Recently, studies have posited a ‘c-Myc function rule,’ in which c-Myc is a ‘universal amplifier’ of active (expressed) genes in lymphocytes.33, 34, 57

In summary, both MYC rearrangement and Myc overexpression have advantage and limitations as a single biomarker in DLBCL, and their prognostic importance is significantly different in GCB- vs ABC-DLBCL. GCB subtype (opposite to the general association of ABC-DLBCL with poorer survival) of MYC-R+ DLBCL with Myc overexpression was associated with significantly poorer survival, likely contributed by significantly higher Myc protein levels as well as associated tumor biology. In ABC-DLBCL, Myc overexpression associated with Bcl-2 overexpression was a significantly adverse biomarker independent of MYC translocation. Our results suggest that fluorescence in situ hybridization analysis for MYC rearrangements and immunohistochemistry evaluation for Myc and Bcl-2 expression are both needed to determine the prognosis in subsets of patients.57 Insights gained into the tumor biology associated with MYC abnormalities are important for understanding the functional role of MYC in lymphomagenesis and chemoresistance, and help identify oncogenic targets for therapeutic intervention (Table 7).