Molecular diagnosis and outcome prediction in lymphoma

Division Of Basic Sciences - Nci

Investigators

Linked publications & trials

Abstract

On the basis of gene expression profiling, the laboratory proposed that the most common form of lymphoma, diffuse large B cell lymphoma (DLBCL), is a composite of three molecularly distinct diseases that are indistinguishable by standard diagnostic methods. These diseases, termed germinal center B cell-like (GCB) DLBCL, activated B cell-like (ABC) DLBCL, and primary mediastinal B cell lymphoma (PMBL), arise from B lymphocytes at different stages of differentiation by distinct oncogenic pathways. The curative response of patients with DLBCL to chemotherapy is highly variable, and the DLBCL subtype distinction accounts, in part, for this heterogeneity. With CHOP multi-agent chemotherapy, the 5-year survival rates of ABC DLBCL and GCB DLBCL are 60% and 30%, respectively. This clinical disparity likely reflects the host of genetic differences between these DLBCL subtypes. A recurring theme that emerges from our molecular profiling efforts in lymphoma is that the curative response to treatment and the length of survival following diagnosis are dictated by molecular features of the tumors at diagnosis. In DLBCL, we developed a multivariate model of therapeutic outcome based on gene expression signatures, which quantitatively reflected distinct aspects of tumor biology. To bring these findings into the clinic, we investigated methods to utilize formalin-fixed and paraffin-embedded tissue for gene expression profiling since most lymphoma biopsies are routinely stored in this fashion. Together with collaborators in the Lymphoma/Leukemia Molecular Profiling Project (LLMPP), we developed a Nanostring platform for digital gene expression analysis has proved highly effective in distinguishing ABC and GCB DLBCL. This technology has been licensed by Nanostring and Veracyte, which has applied to the FDA for approval to aid in the diagnosis of DLBCL. We have been conducting genomic analysis of patients enrolled in therapeutic clinical trials. In a phase 2 trial of ibrutinib in relapsed/refractory DLBCL, we used gene expression profiling to subdivide the cases into ABC and GCB subtypes. The response rate in ABC DLBCL was significantly greater than in GCB DLBCL (37% vs. 5%), as predicted by our laboratory studies showing addiction to chronic active B cell receptor (BCR) signaling and ibrutinib sensitivity in cell line models of ABC DLBCL. Analysis of recurrent mutations in ABC DLBCL revealed a higher response rate in tumors with mutations affecting the BCR subunit CD79B and especially in tumors with both CD79B and MYD88 mutations. These two mutations are also recurrent in aggressive lymphomas involving certain extranodal sites, such as primary central nervous system lymphoma (PCNSL). This clinical finding suggested that DLBCL can be usefully subdivided further based on genetic abnormalities in order to predict response to targeted agents. To test this, we undertook a multi-platform genomic analysis of 574 DLBCL tumors and integrated gene expression profiling with analysis of DNA copy number alterations, translocations, and mutations, resulting in a genetic taxonomy of DLBCL that has provided unexpected biological and clinical insights. Tumors were classified into the same genetic subtype if they shared multiple recurrent genetic alterations. This approach initially yielded four DLBCL genetic subtypes - termed MCD, BN2, N1, and EZB - that refine and extend the gene expression-based classification of DLBCL. MCD and N1 tumors are primarily subsets of ABC DLBCL and EZB is a subset of GCB DLBCL, but BN2 tumors are drawn from ABC, GCB and Unclassified DLBCL. More recently, we have extended our analysis of this dataset with a new algorithm, LymphGen, which provides a probability that a lymphoma belongs to a particular genetic subtype. This effort led to the discovery of 2 new genetic subtypes, termed ST2 and A53, and the subdivision of the EZB subtype into an EZB-MYC+ and an EZB-MYC- subtype. Strong support for the biological and clinical relevance of the DLBCL genetic subtypes came from analysis of responses to immunochemotherapy. Within ABC DLBCL, the outcomes following R-CHOP chemotherapy were significantly different, with adverse 5-year overall survival rates in the MCD (37%), A53 (33%) and N1 subtypes (22%) compared with a more favorable survival rate in BN2 (76%). Within GCB DLBCL, favorable 5-year survival rates were observed in ST2 (81%), EZB-MYC- (82%), A53(100%), and BN2 (100%) compared with EZB-MYC+ tumors (48%). Each genetic subtype was also characterized by the expression of distinct gene expression signatures, indicating striking differences between the subtypes with respect to B cell differentiation stage and transcription factors, oncogenic signaling mechanisms and the tumor microenvironment, emphasizing that view that this genetic taxonomy identify tumors with shared pathobiological processes. One implication of these findings is that clinical trials in which R-CHOP is combined with a novel agent should ascertain which DLBCL genetic subtypes were enrolled, given the striking differences in outcome following R-CHOP alone among these subtypes. To this end, we have implemented the LymphGen algorithm on a publicly accessible web site. The genetic abnormalities and gene expression signatures that characterize the DLBCL genetic subtype suggest that they will respond differentially to targeted therapy. The MCD and BN2 subtypes are enriched for genetic changes in the BCR-dependent NF-kB pathway, and cell line models of these subtypes are heavily reliant on this oncogenic signaling for survival. The EZB genetic subtype does not rely on this pathway but instead engages PI3 kinase signaling to maintain viability. These distinctions are apparent in our clinical trials. As mentioned above, in relapsed/refractory ABC DLBCL, tumors of the MCD genotype frequently responded to ibrutinib. In a clinical trial in primary central nervous system lymphoma, which belongs to the MCD subtype, we showed that ibrutinib induced objective response in 94% of cases as monotherapy, and together with chemotherapy induced complete responses in 83% of patients. Our phase II trial of ibrutinib monotherapy in relapsed/refractory DLBCL led to a randomized phase III trial ("Phoenix") of ibrutinib plus R-CHOP chemotherapy in newly diagnosed non-GCB DLBCL, which LYMB colleague Wyndham Wilson and I co-led together with Anas Younes and Janssen colleagues. The trial showed an ibrutinib benefit in younger patients (age60), resulting in a 12.3% improvement in 3-year overall survival. To understand the biological basis for this ibrutinib benefit, we applied our LymphGen genetic classifier to data from Phoenix biopsy samples, allowing us to distinguish three main subtypes: MCD, N1 and BN2. Younger patients with the MCD and N1 subtypes had a 100% 3-year event-free survival with ibrutinib plus R-CHOP but only 43% and 50% survival, respectively, with R-CHOP alone. MCD and N1 acquire mutations targeting the BCR-dependent NF-kB pathways in 91% and 57% of cases, respectively, consistent with their sensitivity to ibrutinib. Our findings demonstrate that the survival benefit of ibrutinib in younger non-GCB patients on the Phoenix trial has a strong biological basis, supporting the view that an ibrutinib-containing regimen should be considered for such patients.