Innovative strategies are needed to treat the world's most common cause of cancer death, non-small cell lung cancer (NSCLC) [1
]. Less than a quarter of lung adenocarcinomas (LUAD) harbor genetic abnormalities for which targeted therapies have been derived. Early responses are often robust for these but are generally followed by acquired resistance [3
]. Lung squamous cell carcinoma (LUSC) has no approved targeted therapies and few effective chemotherapeutic options beyond the first line of therapy. In the current study, we offer a genomically based, hypothesis-driving analysis to suggest a rationale for a novel combinatorial therapeutic approach to efficacious treatments for advanced NSCLC. The backdrop for the present study comes from our initial clinical trials in our Stand up to Cancer project (SU2C) in which patients with advanced, heavily-pretreated NSCLC received a form of “epigenetic therapy” combining low doses of the DNA hypomethylating agent azacytidine (AZA – Vidaza) and the HDAC inhibitor entinostat [5
]. Only two of now 65 patients treated to date have had RECIST criteria responses to this therapy alone, but these were very robust and durable (5). A group of patients followed for 8 to 26 months responded to multiple different therapeutic regimens given subsequently, suggesting a “priming” effect of epigenetic therapy (5). Twenty-five percent of these patients with both LUAD and LUSC experienced RECIST criteria responses to their subsequent regimens. These subsequent therapies included not only standard chemotherapies but also immunotherapy targeting the PD-1 immune-checkpoint which when given alone has yielded responses in 16 to 17% of patients with advanced NSCLC [6
] (Supp. Fig. 1
). While the number of patients who have received epigenetic therapy followed by immune checkpoint blockade is small, a clinical trial to evaluate potential sensitization to PD-1 immune checkpoint blockade with epigenetic therapy in patients with NSCLC has now begun.
This trial will be biopsy driven and offer the opportunity to examine hypotheses generated in the present pre-clinical work in order to develop biomarker strategies. In this regard, one of the key therapy agents being employed in the trial is AZA, a nucleotide analog DNA demethylating agent which blocks the activity of all three biologically active DNA methyltransferases (DNMT's) and also triggers degradation of these proteins in the nucleus [9
]. With respect to sensitization potential of this drug for immune responses, such targeting of DNMT's is known to induce increased expression of promoter DNA hypermethylated cancer testes antigens and also is reported to up-regulate other individual facets of the tumor immune stimulating profile, including major histocompatibility antigens, and transcription factors IRF7
]. In this regard, we previously reported that elements of such immune pathway activation were produced by low doses of DNA demethylating agents in a genomics based, pre-clinical approach [17
]. These studies demonstrated how low doses of AZA, which avoid early, cytotoxic and off-target effects, can provide a memory for a “reprogramming”-like effect on hematopoietic and selected examples of solid tumor cells [17
]. We hypothesize in this work that these effects may underlie the fact that significantly lowering doses of DNMT inhibitors in the clinic may account for the markedly decreased toxicity, and significant clinical efficacy, which has led to FDA approval of AZA for myelodysplasia (MDS) [18
Initially, we have focused our pre-clinical studies for low dose AZA on NSCLC. By first deriving genomic signatures of gene expression responses and DNA methylation for treated NSCLC lines, we observed in most cell lines a complex, multi-faceted up-regulation, involving hundreds of genes of the immune profile of these cells which includes the target of immune checkpoint therapy, the tumor ligand PD-L1. Moreover, using this extensive genomic signature, we have been able to specifically query hundreds of primary NSCLC samples in the Cancer Genome Atlas project (TCGA) for how basal expression of these immune genes and related DNA methylation events group lung cancers. We define a stark clustering of subsets of primary LUAD and LUSC for an “immune evasion” signature, which relates highly to events for low interferon pathway signaling and includes low levels of PD-L1 [20
]. Low expression of these genes closely matches those up-regulated by AZA treatment of the NSCLC cell lines. We hypothesize that these may be cancers which would benefit from AZA priming together with immune checkpoint therapy and outline a signature that may identify predictive biomarkers from biopsies forthcoming in the current trial.