Stokes, Matthew E.

Introduction: Diffuse large B cell lymphoma (DLBCL) patients experience favorable outcomes following frontline immunochemotherapy, however, 30-40% of patients will exhibit primary refractory disease or relapse (rrDLBCL). Recent studies have expanded upon well-defined characteristics of newly diagnosed DLBCL (ndDLBCL) to elucidate genomic landscapes underlying rrDLBCL; yet, high-throughput examination of resistance and relapse mechanisms within rrDLBCL subtypes, including the newly defined dark zone signature (DZsig) addition to cell of origin (COO), have yet to be described. To expand upon the current tumor etiology and define subtype-specific driving mechanisms, we describe a multi-omic analysis leveraging clinical and biologic variables in rrDLBCL. Methods: Tumor biopsies (N=228) collected at the time of relapse were obtained from n=85 participants in the Mayo Clinic/University of Iowa Lymphoma Molecular Epidemiology Resource (MER), n=73 in the Mayo Clinic Biobank, and n=38 from CC-122-DLBCL-001 or n=32 CC-122-ST-001 trials. Tumor samples were analyzed by WES (n=158) or WGS (n=70) and by RNAseq (n=145). Comparative analysis against ndDLBCL MER samples (N=444) utilized WES (n=404) and RNAseq (n=321). Chi Square or Fisher's exact test was used for enrichment analysis of categorical variables, while Wilcoxon or Kruskal-Wallis test was used for continuous variables. Results: rrDLBCL samples were enriched for ABC (46% vs 33% in ndDLBCL, p=0.01), and fewer GCB (27% vs 40%, p=0.006). Amongst rrDLBCL patients, variance in overall survival (from time of original diagnosis; log-rank p=0.001) was observed between COO subtypes, with poor outcomes in patients with DZsig (median 15.0 mo) and GCB (36.5 mo) compared to ABC tumors (74.4 mo). Patients with progression or relapse within 6 months featured a more even COO distribution (ABC - 22%, GCB - 34%, DZsig - 28%) than patients with early (6-24 mo; 51%, 22%, 22%) or late relapses (>24 mo; 62%, 19%, 4%), which were predominantly ABC. Comparing mutation frequency within subtypes, rrDLBCL GCB tumors were enriched for B2M (41% vs 15% in ndDLBCL, p=0.002), EP300 (19% vs 7%, p=0.059), CIITA (19% vs 5%, p=0.02), PRKCB (11% vs 2%, p=0.035), CD79B (11% vs 0%, p=0.004), and CD83 (8% vs 1%, p=0.049) mutations, and CNV events including gains at 8q24.21 (MYC locus; 48% vs 14%, p<0.001) and 18q21.33 (BCL2, TCF4 locus; 36% vs 15%, p=0.021), and losses at 17p13.1 (TP53 locus; 40% vs 19%, p=0.035). rrDLBCL DZsig tumors featured enrichment for BCL6 (22% vs 3%, p=0.025), and SPEN (17% vs 3%, p=0.062) mutations and 2q22.2 losses (LRP1B locus; 20% vs 0%, p=0.019). Despite differential mutation and CNV profiles, gene expression analysis revealed minimal variance within COO subtypes in rrDLBCL and ndDLBCL, however upregulation of lymphoma driving genes including MYBPC2 (LogFC=1.68, FDR<0.001), MYOM2 (LogFC=1.33, FDR=0.007), MYC (LogFC=0.61, FDR=0.04), and CELSR2 (LogFC=0.53, FDR=0.02) were observed in ABC rrDLBCL. Increased frequency of LymphGen classifier A53 was observed in ABC (43% vs 17%, p=0.005) and GCB (40% vs 14%, p=0.009) rrDLBCL. Using CIBERSORTx to infer lymphoma microenvironment (LME) cell type abundance, an LME-depleted class was evident in rrDLBCL tumors, including reduction of CD8+ T cells (effect size (r)=-0.78, p<0.001) and T follicular helper cells (r=-0.48, p=0.012) in GCB rrDLBCL and reduction of T follicular helper cells (r=-0.43, p=0.023) and Tregs (r=-0.32, p=0.065) in DZsig rrDLBCL. Significance: We present an update to the genomic landscape of rrDLBCL, highlighting GCB and DZsig tumors with differential biology and advanced clinical course. Our data supports increased implementation of sequencing efforts in rrDLBCL to more broadly evaluate relapse mechanisms, with optimism that a more detailed understanding of r/r disease may guide trials for targeted treatment strategies. For example, rrDLBCL patients may be strong candidates for MYC pathway targeting agents (BETi), or a Lenalidomide-based regimen (R2 or Tafa-Len), as Lenalidomide has been shown to reprogram the LME via PD-L1 expression in addition to downregulation of MYC and MYC target genes. Or, due to enrichment of variants affecting anti-apoptotic proteins including BCL6 or BCL2 in DZsig and GCB rrDLBCL, a venetoclax-based therapeutic may be considered. Further targeted strategies should be considered as development of a more personalized approach remains an unmet need.

ABSTRACTDiffuse large B‐cell lymphoma (DLBCL) patients that fail to achieve a complete metabolic response with frontline immunochemotherapy have a poor prognosis. Genomic profiling has led to a broader understanding of the molecular drivers in DLBCL, but it is unknown how well current classifiers identify patients that will experience primary treatment resistance (PTR). Using whole exome and RNA sequencing data from newly diagnosed DLBCL patients, we evaluated the genomic landscape of PTR and compared it to that of non‐PTR DLBCL. We found a significant increase in the frequency of TP53 (34% vs. 15%, p = 0.005) and ARID1A mutations (21% vs. 7%, p = 0.007) in PTR cases, with pathway analysis further demonstrating a downregulation of TP53 and an increase in chromatin modifying pathways. These results suggest that TP53 and ARID1A may be key mediators of PTR and important pathways contributing to the poor outcomes. We found that the current molecular classifiers were unable to identify PTR cases at diagnosis. However, our newly identified high‐risk signature identified 46% of PTR cases at diagnosis. Overall, these results contribute to our understanding of the genomic landscape of patients with primary treatment resistance.

Background Diffuse large B cell lymphoma (DLBCL) is the most common aggressive lymphoid malignancy in adults. Though standard immunochemotherapy regimens can result in clinical remission and cure in a majority of patients, approximately 30% of patients are primary treatment resistant or eventually relapse (rrDLBCL), and their prognosis is very poor. Recent molecular classification by large scale genomic and transcriptomic profiling of newly diagnosed DLBCL tumors has resulted in a deeper understanding of disease drivers, however, the molecular heterogeneity of rrDLBCL is yet to be fully explored. Therefore, in this study, we used an unsupervised approach to better define transcriptional and genetic programs in rrDLBCL samples hypothesizing that molecular stratification within this high-risk group may lead to a better understanding of this disease, which remains an unmet need, and a more personalized therapeutic approach. Methods Data used in this study included RNA (n=143) and whole exome (n=126) sequencing data from available FFPE tumor samples at the time of a relapse (any line of treatment, r1-r10 relapse timepoints included in analysis, one per patient), consented to the Molecular Epidemiology Resource (n=61), banked in the Mayo Lymphoma Biobank (n=50), or consented to the CC-122-ST-001 clinical trial (n=32, NCT01421524). Unsupervised clustering was performed on protein coding gene expression values using the non-negative matrix factorization (NMF) approach. NMF was carried out by performing 200 runs on cluster sizes from 2 to 6. Cell of origin (COO) was determined using the method described by Reddy et al. The tumor microenvironment was analyzed using CIBERSORTx and Lymphoma Microenvironment Classification (LME). Genetic classification was done using LymphGen. Results In order to identify unique transcriptional programs in rrDLBCL, we implemented NMF and consensus clustering to define an ideal quantity of stable patient clusters. We identified 4 rrDLBCL patient clusters (rrC1-rrC4) as the most stable solution, based on consensus clustering selection criteria of iterative NMF runs. While relapse timepoint was not associated with individual clusters, we did observe significant differences in tumor characteristics between the clusters. COO classification found that rrC1 and rrC4 were enriched for GCB-DLBCL while rrC3 was enriched for ABC-DLBCL classification (P < .001). To explore the biologic programs in each cluster, we used pathFindR, which leverages protein-protein interaction networks to identify disease related pathways, rrC1 and rrC4 were defined by pathways involved in cell cycle and mismatch repair; oxidative phosphorylation in rrC2; and MAPK and NF-kB in rrC3. Next, LME classification was done to define patterns in the tumor microenvironment and we found a significant difference between groups (P < .001), with rrC2 and rrC4 tumors enriched for a depleted TME, while rrC1 and rrC3 had a more immune rich or inflammatory TME. This was supported by CIBERSORTx analysis identifying significant differences (P > .001) in abundance of CD8 T cells, memory CD4 T cells, and M1 macrophages between clusters, with rrC4 having the lowest abundance. We then explored genetic patterns in each cluster. Gene level mutation enrichment analysis showed BCL7A mutation exclusively in C4 (6/6, P = .03). The clusters were not associated with LymphGen classification. Lastly, our recently published high risk signature classification that is associated with early clinical failure and is characterized by metabolic dysregulation, a depleted TME, T53 alterations, and poor outcome, was significantly different between the groups (P = .03) with enrichment for high-risk cases in rrC4. In summary, we show for the first time that rrDLBCL patients can be classified into four gene expression clusters that are associated with distinct pathway, TME, and genetic programs. These clusters should now be tested to learn if they can help select patients for newer therapies for rrDLBCL such as CAR-T and bispecific antibodies.