ICML 2017 | Aggressive B-Cell Lymphomas: molecular genetics, classification, and treatment – Educational Session 

At the 14^th International Conference on Malignant Lymphoma (ICML) held in Lugano, Switzerland, an Educational Session on aggressive Lymphoma took place, chaired by Elias Campo, MD, PhD, from Hospital Clinic, University of Barcelona, August Pi i Sunyer Biomedical Research Institute, Barcelona, Spain.

Molecular genetics of aggressive B-Cell Lymphomas

The first talk during the session was given by Riccardo Dalla-Favera, MD, from Columbia University, New York, NY, USA. He began with an overview focusing on recent insights into the mechanisms that regulate germinal center development and those relevant to human B-cell lymphomagenesis (Basso & Dalla-Favera. 2015).

Dalla-Favera then gave an overview of DLBCL, stating it is the most common and aggressive B-cell NHL subtype in adults, making up approximately 40% of all diagnoses. DLBCL responds poorly to therapy, nearly one-third of patients are not cured with the therapeutic strategies currently available to us. DLBCL displays marked heterogeneity in morphologic, molecular, and clinical features, and Gene Expression Profiling (GEP) studies have identified two distinct molecular subgroups: Germinal Center B-Cell-Like (GCB) and Activated B-Cell-Like (ABC). 

The speaker also discussed a paper by Pasqualucci et al. (Nat Gen. 2008) who combined Next Generation Sequencing (NGS) with copy number analysis and showed that, per case, the Diffuse Large B-Cell Lymphoma (DLBCL) coding genome contains an average of 30+ clonally represented gene alterations. Previously unreported mutations in genes not thought to contribute to the pathogenesis of DLBCL were also discovered, such as those which regulate chromatin methylation (MLL; 24% of samples) and immune recognition by T-cells.

Next, the talk focused on Activation-Induced Cytidine Deaminase (AID), which is a key player in Class Switch Recombination (CSR) and Somatic Hypermutation (SHM) of the Ig gene. It is expressed in various types of human B-cell Lymphomas and Leukemias. AID is an effective mutator due to its involvement in DNA breakage of Ig but also other genes, including proto-oncogenes. Investigation into this indicates that AID is needed for chromosomal translocations involving cMYC or BCL-6 and Ig loci, contributing to DLBCL or Burkitt Lymphoma (BL) lymphomagenesis. AID presents as a novel target for future therapeutic strategies.

Riccardo Dalla-Favera emphasized that in future we should reduce the focus on individual genomic alteration data and broaden the findings to pathway alteration data – simplifying understanding of lymphomagenesis and identifying novel targets. GCB- and ABC-DLBCL can be distinguished by a number of distinct pathways, but also share some common pathways:

Dalla-Favera also discussed aberrations in epigenetic regulation in Lymphoma:

• Loss of EP300 and/or CREBBP (a tumor suppressor gene in lymphomagenesis) occurs in 50–70% of Follicular Lymphomas (FLs) and 25% of DLBCLs
• Loss of MLL2 occurs in 80% of FLs and 30% of DLBCLs
• Overexpression of SUZ12/EZH2 occurs in 10–22% of FLs and 20% of GCB-DLBCLs

It has been reported that mutations of MLL2 and CREBBP result in truncated proteins which lack their C-terminal catalytic domains (Pasqualucci et al. Nat. 2011). Furthermore, various studies have found that MLL2 loss is an event that takes place early during clonal expansion and co-operates with BCL-2 de-regulation in Lymphoma.

BCL-6 as an oncogene was discussed in detail. Numerous studies have implicated BCL-6 in contributing to genetic instability and aberration of a number of different pathways including DNA damage response, cell cycle arrest, apoptosis, signal transduction, cytokine response, and differentiation. De-regulated BCL-6 expression can result from chromosomal translocations (leading to promoter substitution) or mutations (causing impaired auto-regulation and IRF4-mediated repression). In Iμ-HA-BCL6 knock-in mice, increased GC formation was observed, as well as perturbed post-GC differentiation and an increased incidence of B-Cell Lymphomas (Cattoretti et al. Cancer Cell. 2009). Inhibiting BCL-6 is an important therapeutic strategy.

Dalla-Favera moved on from BCL-6 to discuss expression of cell-surface molecules by DLBCL cells. It has been found that DLBCLs often lack the cell-surface molecules required for immune-effector cells to recognize the tumor cells. Mutations and deletions which inactivate the β2-Microglobulin (β2M) gene have been reported in 29% of cases; this prevents HLA class-I (HLA-I) expression on the cell-surface, and so the tumor cell cannot be recognized by CD8+ cytotoxic T-cells. Moreover, analogous lesions involving the CD58 gene (encodes a molecule involved in T- and NK-cell-mediated responses) were identified in 21% of cases. Furthermore, as well as gene inactivation, other mechanisms result in aberrant expression of HLA-I and CD58 in nearly two-thirds of DLBCLs. Therefore, evasion of immune-surveillance plays a crucial role in the pathogenesis of DLBCL (Challa-Malladi et al. Cancer Cell. 2011).

This portion of the educational session was concluded by stating that ABC-DLBCL is extensively dependent on NFkB, and so a number of studies are exploring the potential of targeting upstream components of the NFkB signaling pathway.

Pathology and classification of aggressive mature B-Cell Lymphomas

The session chair, Elias Campo, MD, PhD, gave the second talk and began by stating that aggressive mature B-Cell Lymphoma is a heterogeneous category. Campo also briefly outlined the 2016 revision of the Large B-Cell Lymphoma category in the WHO classification, in particular the DLBCL Not Otherwise Specified (NOS) group (Swerdlow et al. Blood. 2016). DLBCL-NOS appearance is thought to take place in the immunoblast to centroblast phase, and its morphological presentation is dependent on the molecular characteristics of its Cell of Origin (COO). Therefore, this group contains both GCB and ABC subtypes. This presents a challenge: how do we distinguish the two subtypes of DLBCL? Due to limitations with immunohistochemistry (low prognostic value in R-CHOP treated patients and low concordance in patient classification across different algorithms) and RNA expression studies, Campo recommended that GEP should be implemented where possible. 

Campo then suggested a new technique for the molecular classification of DLBCL: the Lymphoma/Leukemia Molecular Profiling Project's Lymph2Cx assay is a digital gene expression (NanoString)-based test for COO assignment in formalin-fixed paraffin-embedded tissue (FFPET). The 20-gene assay was trained using 51 FFPET biopsies and validated using an independent cohort of 68 FFPET biopsies. This assay was compared with COO assignments made using the original model that utilizes matched frozen tissue. In the validation cohort, only one case was found to have COO incorrectly assigned, so was accurate, as well as robust (>95% concordance of COO assignment between 2 independent laboratories). This new assay also has a very quick turnaround time, making it an attractive option to implement in clinical trials and, eventually, the clinic (Scott et al. Blood. 2014). Further testing of the Lymph2Cx assay in FFPET biopsies from 344 de novo DLBCL patients treated with R-CHOP found that it provided a concordance of 96% of 49 repeatedly sampled tumor biopsies and in 100% of 83 FFPET biopsies tested across reagent lots. Importantly, no misclassification (ABC to GCB or vice versa) were reported. ABC patients were found to have substantially poorer outcomes than those with GCB-DLBCL (P < 0.001 for Time to Progression [TTP], PFS, disease-specific survival, and OS). In multivariate analysis, COO was associated with outcomes independent of IPI score and MYC/BCL2 immunohistochemistry. The prognostic significance of COO was especially noted in patients with intermediate IPI scores and the non–MYC-positive/BCL2-positive subgroup (P < 0.001 for TTP; Scott et al. JCO. 2015).

The talk quickly mentioned Primary Mediastinal B-Cell Lymphoma (PMBCL), which carries a special gene expression signature different from GCB- and ABC-DLBCL. This is important as not all tumors in the mediastinum originated there; it is key to know the COO in order to select the appropriate therapeutic strategy.

Furthermore, EBV-positive DLCBL of the elderly is now becoming classified at EBV-positive DLBCL-NOS; a broader category that no includes pediatric cases. Cases are usually more pleomorphic and RS-like cells are present, and confer a better outcome. NOS was added to emphasize that the diagnosis of these cases do not meet the criteria for a more specific type of EBV-positive DLBCL. In addition to this, HHV8-positive DLBCL-NOS has been added as a conditional entity.

Campo then went on to list the aggressive Lymphoma types which harbor MYC genetic and protein alterations:

• ALK-positive Large Cell Lymphoma (ALCL)
• Plasmablastic Lymphoma
• ABC-DLBCL
• GCB-DLBCL
• Burkitt Lymphoma (BL)
• Double Hit Lymphoma (DHL)

Alterations of MYC are identified in routine practice commonly by FISH. Campo also spent some time focusing on BL. Aberrations of MYC are observed in around 70% of cases and TP53 is mutated in 50–60% of cases.

The WHO 2008 classification contains a category titled “B-Cell Lymphoma, Unclassifiable, with Features Intermediate Between DLBCL and BL (BCLU).” This group includes: morphologically intermediate between DLBCL and BL, BL with atypical features (BCL2-positive), DHL (inconsistency with DLBCL or blastoid morphology). Transformed FL or MCL with MYC rearrangements are excluded from this group. The development of this category led to the acceptance of DHLs and Triple Hit Lymphomas, read more about their diagnosis and treatment in our coverage of the 2017 ASCO Annual Meeting here.

MYC rearrangement and protein expression specifically in DLBCL was covered. Similar distribution of MYC aberrations are seen between GCB and non-GCB subtypes. Other genetic alterations are often observed with MYC rearrangements included BCL2-R (60–70%), BCL6-R (6%), triple hit (15–20%), and complex karyotypes.

Campo then discussed BCL2 protein overexpression and genetic alterations in DLBCL:

Currently, High-Grade B-Cell Lymphoma is not a category in the WHO classification, but is a working group consisting of the “High-Grade B-Cell Lymphomas with MYC and BCL2 or BCL6 rearrangements (double-hit)” and “High-Grade B-Cell Lymphoma-NOS” categories.

Campo ended his part of the session with a succinct summary slide:

The open and unresolved questions that Campo emphasized upon were:

• Why have so many differences been reported in the clinical impact of DHL?
• What is the prognostic impact of BCL2 by not MYC or Double Expressor Lymphoma (DEL)?

Treatment of aggressive B-Cell Lymphomas

The last talk in this Educational Session was given by Sonali M. Smith, MD, a member of our Scientific Advisory Board, from the University of Chicago, Chicago, IL, USA.

She began with a brief historical look; combination chemotherapy and the first demonstration of cure in DLBCL was reported in 1976 (McKelvey et al. Cancer). It has been 40 years since the birth of CHOP chemotherapy.

It has been established that more chemotherapy does not improve outcomes. In a phase III comparison of CHOP, m-BACOD, ProMACE-CytaBOM, and MACOP-B for the treatment of intermediate-grade or high-grade NHL no significant differences in the rates of OR, PR, or CR, the curves for the time to treatment failure (41% of patients in the CHOP and MACOP-B groups were alive without disease at 3 years, and 46% in the m-BACOD and ProMACE-CytaBOM groups), or the estimated OS (50% at 3 years in the ProMACE-CytaBOM and MACOP-B groups, 52% in the m-BACOD group, and 54% in the CHOP group). However, a significant difference was found in the incidence of serious toxicity; fatal toxic reactions occurred in 1% of the CHOP group, 3% of the ProMACE-CytaBOM group, 5% of the m-BACOD group, and 6% of the MACOP-B group. When fatal and life-threatening reactions were combined, significant differences were found between the groups (P = 0.001), with CHOP and ProMACE-CytaBOM being less toxic than m-BACOD and MACOP-B (Fisher et al. N Engl J Med. 1993).

The addition of rituximab to CHOP has improved responses by approximately 15%.

There have been a number of evolutions of R-CHOP for aggressive B-Cell Lymphoma, but R-CHOP has remained the standard of care for the past 15 years. But, can we move past it?

Smith answered this with a “yes” and suggested a number of approaches. Firstly, we could add dose-intensive chemotherapy in a smart way; for example, DA-EPOCH-R has achieved impressive results in GCB-DLBCL. Secondly, we could completely forget chemotherapy and move to anti-CD20 monoclonal antibodies such as obinutuzumab (G). Lastly, consolidation therapy is another option available.

This part of the session the focused on results from the phase III, randomized CALGB 50303 trial (NCT00118209) investigating R-CHOP versus dose-adjusted EPOCH-R in newly diagnosed de novo DLBCL patients.

No significant difference in terms of EFS or OS was found between R-CHOP and DA-EPOCH-R when including all patients. Incidence of toxicity, but not grade 5 toxicities, was higher with DA-EPOCH-R consistent with higher dose-intensity. More patients treated with R-CHOP completed treatment (Wilson et al. ASH. 2016. Abstract #469).

Results of the open-label, randomized, phase III GOYA trial (NCT01287741) comparing efficacy and safety of G-CHOP with R-CHOP in patients with newly diagnosed DLBCL. Overall, 1,418 patients were randomly assigned to G-CHOP (n=706) or R-CHOP (n=712). Baseline characteristics were well balanced between the G-CHOP and R-CHOP arms. GEP-assessed COO distribution was similar in both arms. After a median follow-up of 29 months, no significant difference in investigator assessed PFS was found between G-CHOP or R-CHOP (3-year PFS, 69% vs. 66%; stratified HR, 0.92; 95% CI, 0.76–1.12; P = 0.3868). No difference between the two treatment arms was found for 3-year OS (81.2% vs. 81.4%; HR, 1.00; 95% CI, 0.78–1.28; P = 0.9982). Moreover, grade ≥3 AEs (74% vs. 65%) and SAEs (43% vs. 38%) were more frequently experienced by patients receiving G-CHOP than R-CHOP (Vitolo et al. ASH. 2016. Abstract #470).

In terms of consolidation, much investigation has taken place with “adjuvant” enzastaurin or everolimus; however, the PRELUDE and PILLAR-2 trials found these agents have no impact on Disease Free Survival (DFS) or OS in patients who achieved a CR to R-CHOP or R-chemotherapy (Crump et al. JCO. 2016; Witzig et al. ASCO. 2016. Abstract #7506). One study that did achieve positive results was the double-blind, randomized, phase III REMARC trial (NCT01122472) comparing maintenance with lenalidomide or placebo in responding elderly patients with DLBCL who had received R-CHOP as first-line therapy.

After a 40-month median follow-up, median PFS with lenalidomide was not reached versus 68 months in the placebo group (HR, 0.708; 95% CI, 0.537–0.932; P = 0.0135). Of patients receiving lenalidomide, 18 (21%) improved from PR to CR during maintenance versus 13 patients (14%) in the placebo group. During maintenance, the most frequently reported grade 3–4 AEs in the lenalidomide and placebo groups were neutropenia (56% vs. 22%), rash (5% vs. 1%), infections (8% vs. 6%), and thrombocytopenia (2.5% vs. 0.6%), respectively. Dose adjustments were required in 72% of the lenalidomide patients and 42% of placebo patients. The proportion of patients who discontinued lenalidomide or placebo due to toxicity were 59% and 40% (P < 0.001). Secondary primary malignancies occurred in 33 lenalidomide patients and in 42 placebo patients (Thieblemont et al. ASH. 2016. Abstract #471).

Additionally, PFS has been found to be superior with early consolidative transplantation among patients with high-intermediate-risk or high-risk NHL who responded to induction, although the benefit of consolidative transplant could not be validated in terms of OS potentially due to the effectiveness of salvage transplantation in control patients. Therefore, early and late transplantation result in similar OS in combined risk groups. Early transplantation appears to be beneficial for the small group of patients presenting with high-risk disease (Stiff et al. N Engl J Med. 2013).

Smith then stated that despite all the approaches to improve R-CHOP for DLBCL (such as augmenting the chemotherapy backbone, replacing rituximab with second generation antibody, shortening the cycle length, post-remission ASCT, and post-remission novel agents), R-CHOP-21 remains the standard of care in 2017. Potential reasons for the equivalent outcomes found in these studies include:

• Trials enrolled all-comers with DLBCL
  • Not stratified for GCB and non-GCB
  • Inadvertent inclusion of DHL
  • Mixture of DEL and non-DEL
• Not powered to detect differences based on outcomes of subgroups
• Unexpectedly good outcomes for the control arm

In a multicenter experience of DHL (n=311) published by Petrich et al. (Blood. 2014), R-CHOP was found to be inferior compared to intensive chemotherapy in terms of PFS and OS (HR, 0.53; 95% CI, 0.29–0.98; P = 0.042). It has also been found that immunohistochemical double hit score defined in a large subset of DLBCL patients with double-hit biology strongly associates with poorer outcomes in patients treated with R-CHOP (Green et al. JCO. 2012).

Smith also discussed results of the phase II, open-label, randomized PYRAMID trial (NCT00931918), which aimed to evaluate the efficacy or R-CHOP with or without Velcade® in patients with newly diagnosed non-GCB DLBCL.

Hans et al. (Blood. 2014) reported 2-year PFS in patients treated with R-CHOP was 78% compared to 82% in the VR-CHOP group (HR, 0.73; 95% CI, 0.43–1.24; P = 0.611).

After this, Sonali warned about bias taking place during enrollment onto clinical trials; there is a temptation to exclude less favorable patients due to concerns about delays/risks. Another challenge in trials at the moment is that the subgroups of DLBCL are heterogeneous and overlap meaning that targeting what we consider to be one subset may be much more complex.

So, how do we identify high-risk subgroups of patients in clinical practice. In centers that do not have access to GEP, the Hans algorithm is used but has been found to have a 20–30% error rate compared to GEP, and there is much inter-user variability. However, there are some limitations with using GEP; for example, it requires frozen material which confers a number of challenges in clinical practice. Thus, there is a clear need for an assay that works in real time in FFPET samples. Furthermore, it was asked if there is a need to conduct FISH in all patients to identify DHL? One suggestion for limited resources was:

• Screen for GCB derivation (reduces number of cases by half)
• Test for MYC/BCL2 by immunohistochemistry (20% of DHL will not have MYC/BCL2 overexpression)
• If both are positive, then proceed to FISH

Sonali Smith concluded her part of the session with a concise slide summarizing the suggested approaches to therapy for patients with aggressive B-Cell Lymphoma:

However, there is still a need for adequate biopsies, integrated biologic risk stratification, trials that are powered for specific subsets of patients, and co-operation across groups and with community practices.