Background: Programmed death - ligand 1 (PD-L1) expression is a dominant immune escape mechanism across human cancers. Ubiquitous copy gains of chromosome 9p24.1 - a region containing the PD-L1 and PD-L2 loci - lead to PD-L1 expression on Hodgkin-Reed-Sternberg cells in classical Hodgkin lymphoma. PD-L1 gene alterations also occur in diffuse large B cell lymphoma (DLBCL), albeit less commonly. Hypothesizing that PD-L1 gene alterations identify DLBCLs in which potent anti-lymphoma immune responses have been activated, we aimed to characterize the immune landscape of PD-L1 gene-altered DLBCLs, describe the key clinical features of these patients, and determine the degree to which PD-L1 gene alterations predict for response to PD-1 blockade therapy in DLBCL.

Methods: FISH was performed on 105 formalin-fixed, paraffin-embedded (FFPE) DLBCL specimens with DNA probes to PD-L1, a region centromeric to PD-L2, and centromere 9. Lymphoma-infiltrating T cell numbers (68 cases), HLA class I/II expression (74 cases), and PD-L1 protein expression (93 cases) were assessed by IHC. RNA sequencing (seq) was performed on FFPE samples utilizing the Illumina HiSeq 2000 platform. Differentially-expressed genes were identified between PD-L1-altered and not altered DLBCLs using a false discovery rate (FDR) of 0.05 and log2 fold change of >1.5. Whole exome seq (WES) was performed on DNA from DLBCL specimens and 22 matched blood samples using the Illumina HiSeq exome kit in collaboration with Theragen Etex Bio Institute.

Results:PD-L1 gene alterations were identified by FISH in 28/105 samples (27% - 16% relative PD-L1 copy gains, 7% PD-L1 amplifications, 2% PD-L1 translocations, 2% chromosome 9 polysomy), were enriched in non-germinal center (GC) DLBCLs (75% non-GC vs 25% GC; p=0.001), and were associated with robust PD-L1 protein expression (Fig 1A-B). PD-L1 gene-altered DLBCLs were more heavily infiltrated by CD8+ T cells compared to PD-L1 not altered DLBCLs (22.6 vs 13.8 CD8+ T cells/hpf; p<0.05), and TCR-β seq revealed a more clonal TCR repertoire among PD-L1-amplified DLBCLs (Fig 1C). Consistent with heightened T cell surveillance, HLA class I expression was absent or decreased more frequently among DLBCLs with vs without PD-L1 gene alterations (42% vs 20%; p<0.05). Furthermore, RNAseq of 12 PD-L1 gene-altered and 12 not altered DLBCLs (controlled for COO) identified differential expression of ~140 genes (Fig 1D). Gene ontology (GO) terms enriched in PD-L1 gene-amplified DLBCLs (p<0.05) included those related to negative regulation of T cell activation, while ingenuity pathway analysis (IPA) identified that a TH1 response and NF-κB activation were significantly enriched in PD-L1 gene-altered vs not altered DLBCLs (z-score p-value <0.05). Pathway-level up-regulation of NF-κB-associated genes in PD-L1 gene-altered DLBCLs was confirmed by GSEA (NES=1.65; p=0.009). WES performed on the same 24 cases revealed a 2-3-fold increase in somatic mutations in genes involved in antigen presentation (HLA, B2M, CIITA) and T cell co-stimulation (CD70, CD83) in PD-L1 gene-altered vs not altered DLBCLs (p=0.02), consistent with the activation of additional genetic mechanisms of immune escape in these lymphomas. Somatic TP53 and TNFAIP3 (A20) mutations, among others, were enriched in DLBCLs with PD-L1 gene alterations (p<0.05) (Fig 1E). Total tumor mutational burden and predicted neo-antigen numbers were similar in PD-L1 gene-altered and not altered DLBCLs (p=ns), arguing that PD-L1 gene-altered DLBCLs are not inherently more antigenic. Patients with PD-L1 gene-altered DLBCLs had higher-risk clinical features and inferior progression-free survival following initial treatment (Fig 1F). Finally, in the relapsed/refractory setting, PD-L1 gene alterations in an exploratory retrospective analysis were associated with response to PD-1 blockade therapy with pembrolizumab among 29 DLBCL patients enrolled in the KEYNOTE-013 study with tissue available (50% objective responses in PD-L1 gene-altered DLBCLs vs 8% in not altered cases), suggesting that PD-L1 gene alterations may be a useful predictive biomarker of response to anti-PD-1 antibody therapy in DLBCL.

Conclusions: Our results indicate that PD-L1 gene alterations identify a unique biological subset of DLBCL in which endogenous anti-lymphoma immune responses have been activated, and which is associated with responsiveness to PD-1 blockade therapy.

Disclosures

Kline:Merck: Honoraria, Research Funding; iTeos: Research Funding. Orlowski:Merck & Co., Inc.: Employment. Smith:Portola: Honoraria; BMS: Consultancy.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution