Azacitidine and Romidepsin Synergize to Induce Expression of Cancer Testes Antigens (CTA) and Shift Peripheral T-Cell Lymphoma (PTCL) Cell Lines Toward a Th1-like Phenotype

Introduction: The PTCL appear to be exquisitely sensitive to epigenetic modifiers, including histone deacetylase inhibitors and hypomethylating agents. While the PTCL harbor recurring mutations in genes controlling CpG methylation, including TET2, IDH2 and DNMT3, it is not clear these genetic events portend sensitivity to these classes of drugs. In preclinical models of PTCL, HDACi and HMA exhibit marked class synergy in vivo and in vitro. Early Phase 1 and 2 experiences have confirmed compelling activity across the PTCL, with exceptionally high response rates in patients with angioimmunoblastic T-cell lymphoma. While the precise mechanism of synergy is under investigation, it is clear these drugs can induce a host of immunological effects on both the tumor cell proper as well as the host. We present details of these class effects in models of PTCL.

Methods: We have explored combinations of azacitidine and romidepsin in a panel of 6 TCL cell lines. The cell viability was accessed and synergy was calculated using Excess Over Bliss modeling to determine the efficacy of the combination. We then performed RNAseq analysis on 4 TCL lines. We confirmed the protein expression of multiple targets using western blot to confirm to results obtained from sequencing. TCL lines were exposed to combination for 96 hours, supernatant was collected and assayed for proliferative properties using isolated PBMCs from healthy donors and anti-CD3/anti-CD28 beads.

Results: The combination of azacitidine and romidepsin were potently synergic across all 6 cell lines. Determination of DNA methylation revealed a decrease in %MdC, as a function of concentration. A supervised analysis of the RNAseq revealed almost all of the genes perturbed by the single agent treatments were enhanced by the combination. There were 1215 genes affected by combination treatment with a p-value ≤0.05; 921 of the genes were not found in either single agent cohort. Of the 1215 genes, 936 were upregulated and 279 genes were downregulated. A gene set enrichment analysis (GSEA) of the RNAseq data displayed a significant down regulation in multiple genes involved in cholesterol biosynthesis in the combination, for example, HMGCR (-2.1 FC), FDPS (-2 FC) and IDI1 (-1.8 FC). The RNAseq analysis revealed a marked increase in expression levels in a variety cancer/testis antigens (CTA) that were most amplified by the combination, such as PRAME (4.9 FC) and MAGE-A1 (11.3 FC). As single agents, azacitidine alone was capable of inducing minor expression of some targets while romidepsin did not affect any. There was a radical change in genes involved in viral/interferon response and immune response observed in RNAseq. The expression of a handful of interleukins and interferons associated with viral/interferon and immune responses were induced by single agent azacitidine and amplified the most by the combination, such as IF16 (1.4 FC aza, 2.8 FC combo), IL18 (3.4 FC combo), IL34 (2.4 FC aza, 3.7 FC combo) and IL26 (5.2 combo FC). Noticeably, single agent romidepsin (3 FC) and the combination (6 FC) augmented the expression of CD274 (PD-L1).Most drastically intensified from this subset of immune response genes was TBX21 (6.9 FC) expression that was induced by the combination. We see from a western blot that the combination induces maximum protein expression levels of T-bet (transcription factor encoded by TBX21) in 4/6 TCL cell lines. STAT4, known to interact with T-bet to promote Th1-differentiation, protein expression was confirmed to increase with combination treatment in 4/6 TCL cell lines. Enhanced proliferation was seen when PBMCs were cultured with conditioned medium from 4 different TCL lines treated with the combination.

Conclusions: The generated data presents a strong justification for the use of combination in addition to an immune checkpoint inhibitor, such as a PD-1/PD-L1 inhibitor. With the emergence of members of CTA family becoming a developing marker in immunotherapy, it is of absolute importance to narrow down the possible targets induced by combination therapy. Here we identify two members of the family that are uniquely expressed as a result from exposure to the combination. Interestingly, the combination is also able to induce a Th1-like phenotype that is often associated with a favorable prognosis. These data suggest that azacitidine and romidepsin in TCL is a promising platform treatment that undeniably proves itself advantageous in preclinical models.