Intratumoral Therapy to Make a “Cold” Tumor “Hot”: The Jury Is Still Out

In this issue of Clinical Cancer Research, Babiker and colleagues (1) have published an important article that describes the ILLUMINATE-101 study, a phase I dose-escalation and dose-expansion trial of intratumoral tilsotolimod, an oligodeoxynucleotide Toll-like receptor 9 (TLR9) agonist, in patients with advanced cancer. The work describes the toxicities, clinical impact, and most importantly, correlative marker studies that the authors conclude might justify inclusion of an intratumoral strategy in a regimen of checkpoint inhibition.

The TLR9 molecule is a pattern recognition receptor (PRR) found on immune and other cells, chiefly B cells and macrophages (2, 3). The PRRs function in sensing viral infection by binding distinctive viral structures, such as nucleic acids, and TLR9 binds unmethylated CpG oligodeoxynucleotides. PRR binding promotes downstream signaling via nuclear translocation of transcription factors, including NFκB and phosphorylated IFN response factors (Fig. 1). These assemble as an “enhanceosome” to activate the transcription of innate IFNs and IFN-related molecules known to be lacking in “cold” tumors that are refractory to checkpoint inhibition (4). TLR9 agonists are classified as either type A, B, or C depending on whether they stimulate plasmacytoid dendritic cells, B/natural killer (NK) cells, or both.

The authors have done an excellent job of demonstrating that visceral and nonvisceral intratumoral injections of the type B TLR9 agonist, tilsotolimod, is a practical approach. Impressively, most injected lesions were deep visceral, not skin or nodal, supporting the feasibility of administering treatment that would otherwise be restricted to superficial sites. The intratumoral injections were well tolerated, with a 14.8% rate of grade 3–4 treatment-related toxicities, with only 1 of 54 patients requiring a dose reduction due to toxicity, and no dose-limiting toxicity were observed. Clinically, no antitumor responses were seen by RECIST 1.1, and the findings that only 3 of 16 patients with PD-1 refractory melanoma in the expansion cohort had stable disease with a median survival of 8.3 months would not appear to support further development of this single-agent strategy in melanoma. The authors point out that their goal was achieving a 30% response rate in the Simon two-stage design expansion cohort, which was not observed. Only 35 of 54 patients were evaluable for response, which was not explained, and there was no evident requirement for a minimum duration of stability. There was also little evidence of regression of noninjected lesions, with only 3 of 16 patients with melanoma in the expansion cohort and 6 of 28 from the dose-expansion cohort having some distant regression and none of the total of 9 patents at the 30% reduction level, suggesting that tilsotolimod monotherapy is not clinically useful. The importance of the work, nonetheless, lies in the correlative biomarker assays, which included NanoString and flow cytometry analyses of biopsies at baseline, 24 hours, and 6 weeks. The data indicate that type 1 IFN response genes were upregulated within 24 hours within the injected tumor, although that may be a double-edged sword. STAT1 expression was upregulated, which promotes Th17 cells, which may have both immunosuppressive and immunostimulatory properties (5). The authors described myeloid-derived suppressor cell ingress into tumors, a cell type that is certainly immunosuppressive (6). STAT2 and other IFN response genes were upregulated by intratumoral tilsotolimod, which should promote an inflammatory tumor microenvironment (TME; ref. 7). Taken together, these data clearly demonstrate that tilsotolimod influenced the TME, but neither the overall direction of effect, nor the resulting impact on tumor regression was clear.

The important question that should be asked for tilsotolimod, and other intratumoral strategies is whether combination with programmed-death-1 (PD-1)/PD-L1, CTL-associated protein-4 (CTLA4) and or lymphocyte activating gene (LAG)-3 blockade would have clinical utility. The question may be restated in two parts, both of which must be met for a therapy to be successful; is the generation of an “inflamed” TME with upregulation of IFN signatures within an injected tumor sufficient to overcome resistance to immune checkpoint blockade locally, and if so, would that function similarly in distant tumors? If the answer to the first question is no, then the strategy suggested by the current work might be ineffective, and other intratumoral strategies employing viral vectors, other TLR agonists and additional small molecules may not succeed.

Extensive preclinical animal data suggested that intratumoral viral and TLR9 agonists overcame resistance to checkpoint blockade, increased CD8 and NK-cell infiltration, PD-L1 expression and promoted IFNγ signaling, even in tumors that were JAK-STAT deficient. What data are there regarding the clinical utility of intratumoral injections to generate “inflamed” tumors in patients with cancer? The ILLUMINATE-204 study was a promising but small phase II study of the CTLA4 blocking antibody ipilimumab with tilsotolimod in advanced PD-1 refractory melanoma, in which a promising 22.4% overall response rate (ORR) was observed. Importantly, tumor reduction was noted in both injected and noninjected lesions (8). These data led to a FDA fast track designation. In contrast, in the larger phase III ILLUMINATE-301 study, the ORR for the combination was only 8.8%, insufficient to meet its primary endpoint (9). Prior studies of the intratumorally injected engineered herpesvirus Talimogene Laherperepevac (T-VEC) in advanced, treatment-refractory melanoma led to its FDA approval in 2010 based on progression-free survival (PFS; ref. 10), but the MASTERKEY-265 phase III study of T-VEC with pembrolizumab compared with pembrolizumab alone in previously untreated melanoma showed no significant differences in PFS and overall survival (11). A phase I study investigated the virus-like particle encapsulated, oligodeoxynucleotide type A TLR9 agonist vidutolimod, (CMP-001), alone or with pembrolizumab. In a 40-patient study of escalating doses of intratumorally injected vidutolimod, the RECIST ORR was 20%, whereas intratumoral vidutolimod with pembrolizumab in 98 PD-1 experienced patients demonstrated a 23.5% RECIST ORR (12). Intriguingly, the clinical activity of vidutolimod + pembrolizumab was associated with serum CXCL10 induction and induction of an inflammatory IFNγ (type 2) gene expression profile (13) again demonstrating that TLR9 agonists can modulate the TME.

A phase III study of intratumoral vidutolimod with pembrolizumab versus pembrolizumab alone, and a phase II/III study of vidutolimod with nivolumab versus nivolumab alone have not read out yet. SD-101 is another oligodeoxynucleotide type C TLR9 agonist that has been tested intratumorally alone and in combination with pembrolizumab in melanoma. In a phase I/II study of 86 patients that were PD-1 naïve, an ORR of 76% was achieved (14). When SD-101 was administered with pembrolizumab to 91 PD-1 refractory patients, the ORR was 19% (15). Altogether, based on the data above and from Babiker and colleagues (1), the jury is still out whether an intratumoral injection which makes a cold tumor into an inflamed, “hot” tumor when combined with checkpoint blockade will be clinically meaningful.

The results of the current trial reflect the difficulty of modulating a complex TME; the results of the ILLUMINATE-101 study show an impact on the TME with tilsotolimod, but the negative results of ILLUMINATE-301 “…indicate no role as single agent and questionable efficacy when combined with checkpoint inhibition in this disease.” We conclude that the correlative marker data that Babiker and colleagues have generated are noteworthy and increase understanding of possible mechanisms by which the immunosuppressive nature of the TME develops and how it can be reversed. These tilsotolimod-induced changes shed further light on how to generate an inflamed tumor and are probably necessary but surely not sufficient for promoting a TME by which one can overcome resistance to checkpoint inhibition therapy.