Introduction

Diffuse large B cell lymphoma (DLBCL) is a clinically and genetically heterogeneous disease. MYC, BCL2, and BCL6 are the three most commonly altered oncogenes in DLBCL. BCL2 (chromosome 18q21) is altered in more than 30% of DLBCL and does not impact DLBCL outcomes when present as a sole abnormality [1]. BCL6 (chromosome 3q27) alteration is seen in about a third of patients and has no prognostic value when present alone [2]. However, MYC (chromosome 8q24) rearrangement is seen in up to 15% of cases of DLBCL and portends a worse prognosis following treatment with standard combination chemotherapy [3]. Double-hit lymphoma (DHL) is an aggressive form of DLBCL in which MYC rearrangement is coexistent with BCL2 or less commonly BCL6 rearrangement. When all three alterations are present, it is known as triple-hit lymphoma (THL) [1]. IHC can be used to identify MYC, BCL2, and BCL6 overexpression, but these are not diagnostic of DHL. DLBCL with overexpression of MYC and BCL2/BCL6 is referred to as dual- or double-expresser lymphoma (DEL) [4]. MYC and BCL2 rearrangements are more common than MYC and BCL6 in DHL [5]. DHL is typically of germinal center B cell origin (GCB, GC COO), while DEL is usually non-GCB. Both forms are usually aggressive, and standard R-CHOP therapy may not be adequate. In its most recent classification, WHO reclassified DHL from DLBCL NOS, or B cell unclassifiable between DLBCL and Burkitt lymphoma to high-grade B cell lymphoma, with MYC and BCL2 and/or BCL6 rearrangements [6]. In this review, we discuss the biology, testing for, and treatment of DHL in light of available literature.

Biology of DHL

The introduction of gene expression profiling (GEP) has helped in classifying DLBCL under two main groups, the germinal center B cell-like type (GCB) and the post germinal center B cell-like type (non-GCB), often called activated B cell type (ABC) [7]. GCB subtype has a better five-year survival with current therapy (CHOP or R-CHOP) as compared to non-GCB lymphomas [8]. Since GEP is not yet a commercially available tool, clinicians rely more on different algorithms using available IHC stains to differentiate the two types, all of whom show similar degree of correlation with GEP [9].

BCL2 rearrangement [t(14;18) (q32;q21)] is more common in GCB lymphomas vs non-GCB subtype (30 vs. < 5%). The expression of BCL2 protein does not always correlate with BCL2 rearrangement [t(14;18)]. In one study, BCL2 protein expression was observed in 44% of GCB and 62% of non-GCB DLBCL [10]. BCL2 overexpression was correlated with a poor outcome with R-CHOP therapy only in the GCB subtype. Same study showed that the non-GCB subtype of DLBCL rarely has the t(14;18), yet amplifications of 18q21 were seen in up to two thirds of cases, which could possibly provide mechanism for BCL2 overexpression in these tumors [10]. Similarly, MYC overexpression is found in much higher frequency in DLBCL as compared to MYC rearrangement [11], suggesting other mechanisms, such as copy number variations, mutations, and transcriptional upregulation (by B cell receptor and NF-κB signaling) may be responsible. MYC rearrangement in DLBCL is associated with increased risk of CNS relapse, poor progression-free survival (PFS), and overall survival (OS) with conventional therapy [12, 13]. These lymphomas also present with complex karyotype [14] in contrast with Burkitt lymphoma. This poor prognosis is largely seen when MYC rearrangement is present in conjunction with BCL2 and/or BCL6 rearrangement (DHL) [3, 15, 16]. In 90% of DHL, MYC, and BCL2, rearrangements are found [17, 18]. It has been shown that the combination of genetic abnormalities, MYC, and BCL2 rearrangements is predictive of refractory disease to conventional therapy (e.g., R-CHOP) [19]. Similarly, triple-hit lymphoma (THL) with rearrangement in MYC, BCL2, and BCL6 has also been defined; these are rare and are also associated with poor prognosis [20].

A study looking at mutational profiling showed DHL to fall intermediate between Burkitt lymphoma (BL) and DLBCL, hence the previous WHO category of B cell lymphoma NOS [21]. In this study, DLBCL pattern showed mutations in BCL2, EZH2, CREBBP, EP300, MEF2B, and SGK1, while BL pattern showed mutations in BL-associated genes including ID3/TCF3, CCND3, and MYC. DHL and DLBCL with MYC rearrangement only showed pattern which was combined between DLBCL and BL. MYC/BCL2 rearranged DHL shows increased incidence of TP53 mutation when compared with MYC/BCL6 [22].

In DHL, BCL2 translocation is thought to be the first hit followed by MYC which could be an early or a late event in germinal centers leading to DHL [23, 24]. DHL often has alterations in genes in addition to MYC, BCL2, and BCL6 like TP53 [17]. In some studies, high expression of BCL2 and MYC has been associated with worse risk of relapse and poor OS [25, 26], while smaller studies show that DHL (about 15–20%) with low or no expression of MYC and BCL2/BCL6 may have relatively better prognosis despite presence of characteristic translocations [3, 27,28,29].

Clinical Presentation and Diagnosis of DHL

There is lack of consensus regarding which patients need to be tested for DHL status. Both NCCN and ESMO do not require, rather suggest evaluation in their most recent guidelines [30, 31]. A majority of DHL present as high-grade lymphomas with features suggesting poor prognosis, often with CNS disease and high stage [3, 32•]. In a cohort of Japanese patients, the presence of B symptoms, extra nodal sites, advanced stage, high serum lactate dehydrogenase level, and bone marrow involvement was more prevalent in patients with DHL [33]. Landsburg et al. published a series of 53 patients, in which the double gene rearrangements were detected in 32% of patients; however, no baseline factors, including age International Prognostic Index (IPI) score or histology, were statistically significantly associated with DHL status [34]. In another study of 152 patients with high-grade B cell non-Hodgkin lymphoma (NHL), 21 patients displayed MYC rearrangement, 9 of whom had a concurrent BCL2 rearrangement with one patient with triple hit. The overall frequency of DHL was 6%. All patients had GCB/GC COO by Hans criteria and suggested referral of all patients with DLBCL for MYC rearrangement testing at the time of initial diagnosis [35]. Fluorescence in situ hybridization (FISH) testing in all patients with large B cell lymphoma have identified DHL with low-risk features, with somewhat better outcomes [36]. While these patients are usually treated with aggressive regimens as DHL, it is unclear if this is helpful in this subset of patients. One strategy would be to test all patients presenting with high-grade large B cell lymphomas = with IHC for MYC, BCL2, and BCL6, which is rather inexpensive, and if positive, further testing with FISH should be undertaken. As previously discussed, high expression of these proteins in addition to FISH testing portends worse prognosis compared to patients positive for FISH abnormality for DHL but no overexpression. The downside to this strategy is the inter-observer variability in reporting expression of these proteins. There is no consensus what is considered high expression. A good reference point would be 40% for MYC and 50% for BCL2, as this has been shown to have prognostic significance [28]. Another starting point could be to evaluate patients presenting with GCB DLBCL, as 95% of DHL are GCB subtype which can decrease the number of patients to be tested by half, keeping in mind that only a small fraction of GCB DLBCL will be DHL [17, 28].

DHL Treatment

Initial Treatment of DHL

Due to low incidence and resultant paucity of prospective trials in DHL, there is no consensus about the standard treatment for DHL. DHL treatment studies available are listed in Table 1.

Table 1 Currently published studies on DHL
Full size table

One study evaluated the outcomes of patients with MYC+ and DHL and demonstrated that only age and achievement of complete response (CR) were correlated with better outcomes. After median follow up of 28.5 months, the median PFS for patients with documented double-hit NHL who achieved a CR had not yet been reached (95% CI, NR to NR) vs 3.9 months (95% CI, 1.8–8.0 months) for those with no CR (p < .0001). The median OS for those patients who did not achieve a physician-assessed CR was 7.0 months (95% CI, 2.0–12.5 months) compared with a median OS not reached for those who did achieve a CR (95% CI, NR to NR; p < .00001) [44].

Petrich et al. [37] published a multicenter retrospective analysis looking at DHL patients, comparing the outcomes of those that received standard R-CHOP vs more intensive regimens (R-Hyper CVAD, DA R-EPOCH, or R-CODOX-M/IVAC). Patients who had CR and underwent autologous stem cell transplant (ASCT) were compared with patients who had CR and did not get ASCT. After a median follow up of 23 months, the median PFS and OS for all patients were 10.9 and 21.9 months, respectively, with no difference in OS for those that received intensive regimens vs R-CHOP. However, median PFS was significantly better for intensive regimen patients over R-CHOP patients 26.6 vs 7.8 months (p = 0.0463 for DA R-EPOCH group, p = 0.001 for R-HyperCVAD group and p = 0.036 for R-CODOX/M IVAC group). Of note, there was no difference between the three intensive treatment regimens. In multivariate analysis, advanced stage, central nervous system involvement, leukocytosis, and LDH, > 3 times the upper limit of normal were associated with higher risk of death. When corrected for these variables, intensive induction was associated with improved OS. Based on these findings, a novel risk score for DHL was developed based on PS ≥ 2 and bone marrow involvement with 0–2 scoring, which divides patients into high-risk (2), intermediate-risk (1), and low-risk (0) groups. In another study, Oki et al. [39•] analyzed the outcome of 129 cases of DHL; DHL was defined as B cell lymphoma with translocations and/or extra signals involving MYC plus BCL2 and/or BCL6. The 2-year event-free survival (EFS) rate in all patients was 33%; however, when analyzed by individual regimen, those who received R-CHOP, R-EPOCH, and R-HyperCVAD/MA had 2-year EFS of 25, 67, and 32%, respectively.

A multicenter phase II study using R-DA-EPOCH for MYC rearranged aggressive B cell lymphoma demonstrated, though with a short-term median follow-up time of 14 months, PFS, time to progression (TTP), and OS were 79, 86, and 77%, respectively [43]. A phase I study incorporating lenalidomide (LEN) into dose-adjusted EPOCH plus rituximab (DA-EPOCH-R) in patients with double-hit (DHL) or double-expressing (DEL) lymphomas was recently presented and concluded that LEN can safely be added to DA-EPOCH-R in DHL and DEL patients and preliminary safety/efficacy data appear promising. With a median follow-up of 10.7 months (range 1.3–18.6 months), for 15 patients enrolled in the study, OS was 93%. A phase II study in this patient population with LEN + DA-EPOCH-R is underway [45].

First-line consolidation with autologous stem cell transplant (ASCT) has been evaluated and the results are mixed. In one study with 311 patients, there was a trend seen toward improved survival with first-line consolidation with ASCT in patients who achieved CR with first-line induction chemotherapy [32•]. SWOG9704 also showed a trend in improved PFS with ASCT after first-line RCHOP, but study included only one DHL patient who had ASCT(n = 1) [46]. In the study by Oki et al., ASCT after CR did not improve OS in patients achieving complete response with initial therapy (n = 71); 2-year EFS rates in patients who did (n = 23) or did not (n = 48) receive frontline stem cell transplantation were 68 and 53%, respectively (p = 0·155). A prospective study recently published looked at patients with DHL receiving ASCT after achieving first CR [47•]. A total of 159 patients who achieved CR after RCHOP or intensive therapy were included. No difference in 3-year relapse-free survival (RFS) or OS was seen in patients who achieved ASCT vs no ASCT. Patients showed inferior 3-year RFS with RCHOP-based induction vs intensive therapy (56 vs. 88%, respectively).

Treatment of Relapsed DHL

Standard treatment for relapsed DHL has not been established yet. Most recently, in a phase II study, the efficacy of autologous T cells genetically modified to express a chimeric antigen receptor consisting of an external anti-CD19 single-chain murine antibody domain with CD3ζ and 4-1BB signaling domains (CTL019 cells) was tested in patients with relapsed/refractory GCB and non-GCB DLBCL, DHL, and transformed follicular lymphoma (tFL). Thirteen patients (7 pts. GCB, 5 pts. NGC, 1 undetermined) were enrolled. The median number of prior therapies were 5 (range 2–8) and number of pts. with prior transplant 7 (54%). Lymphodepleting chemotherapy regimens were given prior to giving CTL019 cells. At 3 months’ post CTL019, overall response rate (ORR) was 52% (7/13 pts); ORR for GCB 71% (5/7 pts) and non-GCB was 40% (2/5 pts). Best response for all patients was CR in 46%, CR for GCB 57%, and non-GCB 40%. Three of 7 pts. with GC DLBCL had tFL, and all 3 achieved CR; two of seven patients with GCB DLBCL had DHL and both achieved CR. At the time of this data presentation, no patient achieving CR relapsed. Median PFS was 5.8 months for all patients, 3.0 months for non-GCB patients, and not reached for GCB patients (57.1% [95%CI: 17.2–83.7%] progression-free at median follow-up 21.9 months). At median follow-up 23.3 months for responding patients, 85.7% [95%CI: 33.7–97.9%] maintain response [48]. KTE-C19, which is autologous CD3ζ/CD28 chimeric antigen receptor (CAR) modified T cells, has shown promising results. ZUMA 1 study of KTE-C19 included two cohorts, one with relapsed/refractory DLBCL (N = 72) and the other with transformed follicular lymphoma and primary mediastinal B cell lymphoma. In the intention to treat analysis, an ORR of 82% was seen in DLBCL patients with 49% CR rate. Median duration of response was 8.2 months with majority of patients having durable responses if they achieved CR [49•].

ASCT for patients in second line has been evaluated as well in a retrospective series of 117 patients having DEL and DHL [50]. The study showed inferior outcomes for patients who underwent ASCT.

Summary

Considering the available data, it can be concluded that DHL has an inferior outcome when treated with standard R-CHOP therapy. Further, the data imply that this inferior outcome may be overcome by using more intense regimens such as DA R-EPOCH, R-Hyper CVAD, or R-CODOX/M IVAC. Based on the MD Anderson Experience, DA R-EPOCH may be best tolerated while maintaining the largest improvement in PFS. Until a prospective trial demonstrates improved survival, these suggestions will remain somewhat speculative. Several prospective trials are ongoing that will hopefully answer some of these questions. Several of these trials are looking at multiple targets involved in the pathogenesis of lymphoma at the molecular and genetic level.

Future Directions

Multiple pathways have been studied in the treatment of double-hit lymphoma. Preclinical data has shown that a second-generation proteasome inhibitor, MLN9708/ixazomib, degrades MYC and induces cell death at nanomolar concentrations in lymphoma models and resulted in significant tumor/growth inhibition (p < 0.001) and improvement in survival (p < 0.001) compared with controls in Jurkat and L540-derived SCID xenograft models [51]; however, its single-agent activity was modest in relapsed/refractory lymphoma in another study [52]. Of 22 response-evaluable patients, one achieved CR, three achieved PR, and four patients had stable disease (SD).

The mammalian target of rapamycin (mTOR) complex 1-dependent evasion of senescence is essential for cellular transformation and lymphoma maintenance by MYC in B-lymphocytes. In one study, everolimus selectively cleared premalignant B cells from the bone marrow and spleen, restored a normal pattern of B cell differentiation, and strongly protected against lymphoma development. Established Eμ-Myc lymphoma also regressed after everolimus therapy [53]. In a phase II study, temsirolimus (an mTOR inhibitor) demonstrated a single-agent activity in DLBCL with an overall and complete response rate of 28.1 and 12.5%, respectively, and median PFS of 2.6 months and median OS of 7.2 months. [54]. The platelet-sparing BCL2 inhibitor, ABT-199, has single-agent activity shown in a preliminary report in three of the eight patients (38%) with relapsed/refractory DLBCL treated in the higher-dose cohorts (at doses ≥ 600 mg) [55]. Another study showed ABT-199 enhancement of the antitumor activity of chemotherapy agents including doxorubicin, cytarabine, and bortezomib in DHL cell lines [56].

Early preclinical data demonstrated that an aurora A kinase inhibitor (Aki) in combination with a histone deacetylase inhibitor enhanced lymphoma cell death through repression of MYC and MYC-responsive microRNAs. In one study, either of the AKi alone at 100 to 500 nmol/L resulted in approximately 50% reduced cell growth and 10 to 40% apoptosis. Addition of vorinostat reactivated proapoptotic genes and enhanced lymphoma cell death. [57]; however, in a small clinical trial of this combination, the three patients with DHL developed progressive disease [58]. In another clinical trial of a selective aurora A kinase inhibitor in patients with relapsed or refractory DLBCL, only three out of 21 patients had clinical response [59].

Inhibition of the BET bromodomain diminishes the effect of MYC overexpression by preventing signal transduction, essential in regulating MYC transcriptional initiation and elongation [60]. JQ1, bromodomain BRD4 inhibitor, showed preclinical activity, including specifically in DHL cell lines [56].

Clinical trials of the Bruton’s tyrosine kinase inhibitor ibrutinib [61], and phosphatidylinositol-3-kinase delta isoform inhibitor idelalisib, showed activity for inhibitors of signaling downstream of the B cell receptor in DHL [62]. In a phase III, multicenter, randomized, double-blind, placebo-controlled study, the efficacy and safety of idelalisib were assessed in combination with rituximab vs rituximab plus placebo. Two hundred and twenty patients were assigned to receive rituximab and either idelalisib (at a dose of 150 mg) or placebo twice daily. The median PFS was 5.5 months in the placebo group and was not reached in the idelalisib group (hazard ratio for progression or death in the idelalisib group, 0.15; p < 0.001). Patients receiving idelalisib vs those receiving placebo had improved rates of overall response (81 vs. 13%; odds ratio, 29.92; p < 0.001) and OS at 12 months (92 vs. 80%; hazard ratio for death, 0.28; p = 0.02) [63].

Other promising therapies for double-hit lymphoma include modified autologous T cells engineered to recognize other B cell surface targets such as CD22, CD20, CD30, and CD79a. which are in various stages of development [62, 64]. Also, small-molecule inhibitors of BCL6 could have potential therapeutic activity as well [65].

In previous translational studies, CUDC-907 treatment has demonstrated dose-dependent decreases in MYC protein expression and a more potent inhibition of MYC expression than various HDAC and PI3K inhibitors, either alone or in combination [66]. In an ongoing study by Landsburg et al., patients with RR DLBCL, including those who have MYC-altered disease per central testing, are being enrolled to receive either CUDC-907 alone or in combination with rituximab (NCT02674750). Multiple ongoing studies are evaluating different pathways in order to improve the outcomes (Table 2).

Table 2 Ongoing studies in treatment of DHL
Full size table

Conclusion

DHL is an aggressive disease with no consensus on what would be the standard treatment for it. The most consistent result of these retrospective analyses indicates that DHL has an inferior outcome when treated with standard R-CHOP therapy. However, those studies indicate that this inferior outcome may be overcome by using more intense regimens such as R-DA-EPOCH, R-HYPERCVAD, or R CODOX/M IVAC.

Until a prospective trial demonstrates improved survival, these suggestions will remain somewhat speculative.