Therapeutic Immune Modulation against Solid Cancers with Intratumoral Poly-ICLC: A Pilot Trial

The last decade has ushered in an exciting new age of immunotherapy with the FDA approvals of the first cancer vaccine, Provenge or sipuleucel-T, and checkpoint blockade, e.g., ipilimumab (Yervoy; anti–CTLA-4), pembrolizumab [Keytruda; anti-programmed death (PD1)], and nivolumab (Opdivo; anti-PD1). However, even with advances in immune checkpoint blockade and other systemic chemotherapies, there remains a significant fraction of patients who either fail to respond or become resistant to treatment. Proposed interventions to broaden the fraction of patients benefiting from immunotherapies and increase response rates rely on reversing T-cell exhaustion, reducing immune suppression in the tumor microenvironment (TME), and transforming a noninflamed TME to a “responsive” TME (e.g., immune cell infiltration, upregulation of PD-L1, etc.). Oncolytic viruses in the treatment of cancer work presumably through their effects not only upon tumor cells but by activating innate immunity and inducing tumor-specific immunity (1–4). An intratumoral (IT) approach that mimics viral infection, without associated significant side effects or the complications of inducing dominant antiviral immunity, is one proposed strategy (5).

Polyinosinic-polycytidylic acid-poly-l-lysine carboxymethylcellulose (poly-ICLC, Hiltonol, Oncovir, Inc.) is a synthetic double-stranded RNA viral mimic for a pathogen-associated molecular pattern (PAMP) or “danger signal” that binds to toll-like receptor 3 (TLR3), MDA5, and other pathogen receptors to activate dendritic cells (DC) and subsequently to also trigger natural killer (NK) cells to kill tumor cells. Although initially developed as an IFN inducer, poly-ICLC has been found to have much broader biological effects, including specific antitumor and antiviral actions (6). It activates multiple elements of innate and adaptive immunity, including induction of a “natural mix” of IFNs, other cytokines and chemokines, NK cells, T cells, myeloid DCs, the P68 protein kinase (PKR), and other dsRNA-dependent host defense systems (7, 8). Thus, when properly combined with antigen, poly-ICLC has the potential to generate a “live virus vaccine equivalent” with a comprehensive immune response that includes activation of myeloid DCs, other antigen-presenting cells, and NK cells, and generation of a polyfunctional Th1-polarized and cytotoxic T lymphocyte (CTL) response with increase in CD8 to CD4/regulatory T-cell ratio, which via the induction of specific chemokines can home to tumor or pathogen (9–14).

Although most cancer vaccines are generally designed to utilize known or presumptive tumor antigens, an alternative strategy is “autovaccination,” i.e., the use of the tumor itself as the antigen source, in situ. Poly-ICLC can be given intramuscularly (IM) to induce systemic inflammation and/or IT to induce immune infiltration of tumors. We observed a dramatic response in the first sarcoma case being treated with repeated IT and IM poly-ICLC, an 18-year-old patient with a malignant embryonal rhabdomyosarcoma, who failed eight different regimens of chemotherapy as well as radiotherapy and proton-beam therapy, and was in hospice. Treatment with poly-ICLC resulted in necrosis and regression of facial, oral, retro-orbital, and a large intracerebral tumor. Although the patient eventually succumbed, his life was extended well beyond expectations (15). A phase II trial of single-dose treatment of ultrasound-guided IT poly-ICLC followed by IM poly-ICLC was found to be safe in patients with advanced primary or metastatic liver cancers, with evidence of regression of noninjected metastatic lesions as well as the targeted lesions (16, 17).

Based upon these early indications of clinical response, we conducted a pilot study using this autovaccination strategy with IT and IM poly-ICLC at our institution in advanced treatment refractory head and neck cancers and melanoma. We hypothesize that the therapeutic in situ autovaccination strategy using IT and IM poly-ICLC administration of TLR3 ligand poly-ICLC can reverse DC inhibition in the treated tumor microenvironment, increase the efficiency of antigen presentation to CTLs, prevent tolerization to tumor antigens, and elicit systemic antitumor immunity. Here, we present the first-ever published results of our phase I trial of IT poly-ICLC in treatment of patients with solid cancer.

Between 2013 and 2015, 8 patients [7 head and neck squamous cell cancers (HNSCC) and 1 melanoma] were enrolled in pilot phase of clinical trial. This trial is now accruing as a phase II multicenter clinical trial (NCT02423863). Eligible patients had unresectable recurrent or metastatic disease that had failed prior systemic therapy and was radiologically or visually measurable disease at least 10 mm in longest dimension. At least one accessible primary or metastatic tumor site was necessary for IT injection with poly-ICLC with or without ultrasound guidance. All patients had Eastern Cooperative Oncology Group performance status of ≤2 and acceptable hematologic, renal, and liver function per laboratory parameters. Exclusion criteria included bulky intracranial metastatic disease, history of active immunotherapy in the previous month, AIDS defined as CD4 count <200, and life expectancy of less than 6 months in the judgment of the study physician. Written consent was obtained from patients, and the study was conducted in accordance with the provisions of the Declaration of Helsinki. The study protocol and all amendments were approved by the Institutional Review Board at Mount Sinai Hospital.

Patients were to receive two cycles of poly-ICLC treatment, each cycle including a priming and boosting treatment course (Fig. 1), with dosing and frequency based on prior preclinical and phase I trials (18, 19). For cycle 1, patients were treated with 1 mg of poly-ICLC thrice weekly during weeks 1 and 2 (priming treatment course, a total of 6 IT injections) into the same lesion. During weeks 3to 9, patients were treated with IM maintenance boosters biweekly (1 mg per dose, boosting course, total 14 IM injections), followed by a rest week (week 10) without treatment. This 10-week cycle was repeated in cycle 2. Weeks 20 to 26 were a “no treatment rest period” to allow for evaluation of response in the absence of inflammation at week 26. At week 26, patients were assessed and response determined, and those patients with complete response (CR), partial response (PR), or stable disease (SD) were offered option of maintenance therapy from weeks 27 to 36 with administration of 1 mg IM poly-ICLC twice weekly.

Tumor response was evaluated using immune-related response criteria in solid tumors (20): CR, PR, SD, and progressive disease (PD). All the patients with measurable disease at the time of enrollment on the study were eligible for response assessment.

Serial blood samples collected at baseline and at selected time points during and after treatment (Fig. 1) were processed to collect plasma/serum and peripheral blood mononuclear cells (PBMC), and used to evaluate humoral and cellular immunity induced by IT and IM poly-ICLC injections. Tumor biopsies were performed at baseline, week 3, and week 26 (if possible). Pre- and postvaccination tumor biopsies of both IT-treated tumors, and nontreated distant tumors taken when possible, were analyzed by quantitative multiplex immunohistochemistry (IHC) for lymphocyte infiltration [e.g., CD4 and CD8 T-cell subsets, T regulatory cells (FoxP3⁺), PD-1 or CTLA-4 expressing T cells, DC subsets], as well as other immune markers (CD86 antigen-presenting cells, CD68 macrophages/monocytes, CD16 NK cells, and HLA encoding the MHC proteins) using standard IHC techniques (21). Individual antigens were quantified using the CRI Inform software (Perkin Elmer) which applies user-directed antigen thresholds to generate percentages normalized to the total tumor area.

RNA sequencing (RNA-seq) analysis and T-cell receptor (TCR) sequencing (Adaptive, Inc.) were done on PBMC and tumor tissue. PBMCs were isolated from patients at multiple time points prior to and after treatment with poly-ICLC. Paired-end sequencing of cDNA created from isolated mRNA was performed using the Illumina HiSeq 2500 platform at a depth of 30 M–35 M reads. The read quality was assessed using FastQC (22). Corresponding sequence files were processed using publicly available RNA-seq data analysis pipelines (23). Briefly, aligned reads are assigned to genes using the FeatureCounts function of Rsubread. Gene expression in terms of log2-CPM (counts per million reads) was computed and normalized across samples. Genes with low expression (ones not having at least 10 reads per million reads in at least two samples) were filtered out. Differential expression analysis was performed using the limma software package (R, Bioconductor). Gene set enrichment analysis (GSEA) was computed following published recommendations (24, 25). Analysis of blood transcriptome modules (BTM) was done accordingly (26) Comparisons between time points as well as before and after treatment were conducted addressing both gene- and pathway-specific changes (24).

For TCR analysis, genomic DNA was purified from total PBMCs and tumor samples using the Qiagen DNeasy Blood extraction Kit. The TCRβ CDR3 regions were amplified and sequenced using immunoSEQ (Adaptive Biotechnologies), as previously described (27).

Kaplan–Meyer survival curve analyses were performed using publicly available on-line server Tumor Immune Estimation Resource (TIMER, https://cistrome.shinyapps.io/timer/).

The primary objective of this phase I pilot study was to evaluate the safety of IT plus IM poly-ICLC for treatment of patients with advanced accessible solid tumors. Secondary objectives were to investigate the changes in humoral and adaptive cellular immunity induced by autovaccination with poly-ICLC.

Eight patients with treatment refractory solid tumors (1 melanoma and 7 head and neck cancer) were enrolled in this study between January 2014 and July 2014. Two subjects completed 2 cycles of IT treatment, though one of these patients did not complete a second IM boosting cycle; the remaining 6 subjects completed 1 cycle or less of IT poly-ICLC due to progression of disease (Table 1).

One of 2 patients who completed 2 cycles of IT poly-ICLC achieved clinical benefit. Patient 002 was a 54-year-old man with metastatic Epstein–Barr virus (EBV)-positive nasopharynx HNSCC, previously treated and refractory to chemotherapy. He tolerated treatment to completion of 2 cycles (12 IT and 28 IM injections in total) over a 30-week period and was the one patient who achieved clinical benefit with SD (progression-free survival of 6 months; Fig. 2). At 6 months, he developed diplopia due to new brain metastasis; patient treatment was switched to chemotherapy with radiotherapy with control of disease. Patient was still alive on subsequent line of chemotherapy 18 months out from start of poly-ICLC treatment.

The other 7 patients had PD while on poly-ICLC treatment (Table 1). Patient 001 was the only other patient (HNSCC) in our pilot trial to complete 2 cycles of IT poly-ICLC treatments (12 IT injections) though only had 20 of 28 IM boosting treatments. Patient was initially thought to have clinical improvement on exam. CT scans at 10 weeks suggested pseudoprogression, and as the patient was clinically stable, he remained on study for cycle 2. Unfortunately, repeat imaging at week 15 showed clear progression of disease with innumerable pulmonary nodules, and patient was taken off study in the middle of second IM boosting cycle (after 12 IT and 20 IM poly-ICLC injections total) and switched to salvage chemotherapy.

Patients 003, 004, 005, and 008 were patients with metastatic HNSCC (range, 70–80 years old) who also stopped treatment shortly after initiation of poly-ICLC treatment due to rapidly PD (Table 1). Patient 003 received only 3 IT injections, whereas patients 004, 005, and 008 received the first course of 6 IT injections and subsequently 3 to 6 IM injections before progressing. As an example, patient 008 was a 70-year-old male patient with metastatic HNSCC whose treatment was halted particularly early, in cycle 1 at week 3 to 4 due to PD (6 IT and 3 IM injections in total). CT scans at 10 weeks showed PD with new lung nodules and enlarging liver and right abdominal wall lymph nodes. Tissue biopsies were obtained at week 3 and end of treatment (week 9) per protocol.

Patient 006 was a patient with BRAF-wild-type subungal melanoma of the right fourth digit, who developed multiple recurrences through chemotherapy, radiotherapy, targeted therapy, and checkpoint blockade. The patient completed 6 IT and 5 IM injections, but treatment was stopped prior to cycle 2 due to rapidly PD. At 9-week scans, he had PD of pulmonary nodules and lower extremity lesions. The patient was initiated on anti-PD1 blockade (2014), but subsequently progressed through treatment and died.

Patient 007 was a patient with advanced HPV-negative, EBV-positive, progressive tumor after induction chemotherapy and radiotherapy. He was consented and enrolled in the poly-ICLC trial, but after one IT poly-ICLC injection, had aspiration pneumonia/pneumonitis, which was temporally related to aspiration and not considered directly related to treatment. Subsequently, patient was taken off study.

Poly-ICLC was generally well tolerated with the principal side effects of treatment, fatigue, and inflammation at primary injection sites, less than grade 2. One case of overt necrosis of tumor (grade 2) was observed. There was a case of grade 3 pneumonitis in 1 patient (007), but this was temporally related to aspiration pneumonia and not considered directly related to poly-ICLC treatment. Full range of toxicities is listed in detail in Table 1.

Baseline, week-3 (post 6 IT injections), and week-26 (post IT and IM injection) biopsies were obtained if possible. Unfortunately, week-3 (and subsequent) biopsy was not obtained in patients 003 and 007 due to complication of bibasilar pneumonia requiring hospitalization and grade 3 aspiration pneumonitis, respectively. Week-26 biopsy was not obtained for patients 001, 004, 005, and 006 due to PD requiring study termination and change in treatment plan. Except for patients 002, 004, and 008, although the week-3 biopsy was sufficient for pathology diagnosis and IHC analysis, tissue was insufficient for RNA-seq due to inadequate sample. A summary of treatment courses and biopsies (including if tissue adequate for IHC and RNA-seq analysis) for each subject is outlined in Table 1.

Tumor biopsies were obtained of injected site for IT Poly-ICLC at baseline and at week 3 (after 6 IT injections) and week 26 if possible. Quantitative IHC from tumor biopsies of patients with PD (001, 004, 005, and 008) showed unchanged or decreased levels of CD4, CD8, PD-1, and PD-L1 over the course of the treatment period (Fig. 3; Supplementary Table S1). In contrast, IHC analysis of tumor biopsies of patient 002, the one patient who had clinical benefit (SD), showed increased levels of CD4 (60-fold), CD8 (10-fold), PD-1 (20-fold), and PD-L1 (3-fold). In addition, tumor biopsies obtained in patient 002 showed marked increases in immune cells (CD86⁺ antigen-presenting cells, CD68⁺ macrophages/monocytes, CD16⁺ NK cells, HLA encoding the MHC expression) after treatment (Fig. 3).

RNA-seq was used to characterize immune response in the blood compartment (PBMC) versus the tumor site. Analysis of over 18,000 transcripts revealed that poly-ICLC treatment resulted in changes in gene expression in IFN-related genes both at the tumor and PBMC level indicative of local and systemic immune activation.

Nonsupervised hierarchical clustering of patients’ PBMC revealed three major groups of clustering (Fig. 4A). First, similar blood RNA expression profiles were observed between patient 002 pretreatment and off-treatment samples, and screening samples of patients 004 and 008 who developed PD along with one sample during IM poly-ICLC injection. We assigned these samples as baseline cluster, “minimal or absence of immune activation.” Second, similar expression clustering was observed in patient 002 who had SD, indicative of systemic inflammation resulting in a clinical benefit response (inflamed cluster). A clustering of “intermediate” signaling or activation of immune system was observed in blood samples derived from patients 004 and 008 undergoing IT poly-ICLC injections and patient 002 undergoing IM poly-ICLC injections (intermediate cluster).

In order to gain insight on expression alterations with poly-ICLC treatment, we calculated differential gene expression considering samples within clusters as biological parallels. Resulting scatter plot of expression fold changes was inferred from two comparisons: patient 004 and 008 samples from intermediate cluster versus baseline (PD, Fig. 4B) and patient 004 samples from inflamed cluster versus baseline (SD, Fig. 4B) revealed several trends. The majority of gene expression changes upon poly-ICLC IT injections were similar between patients 004, 008, and 002. However, specific gene sets were upregulated in patient 002, but repressed or unchanged in patients 004 and 008 (624 genes, upper-left quadrant, Fig. 4B). These included genes regulating antitumor activity: stimulatory cytokines and chemokines: IL1 beta, CXCL8, antigen processing-related marker CD83, and T-cell activation–related protein IFN gamma (see Table 2A). Fewer genes were specifically upregulated in patients 004 and 008 (66 genes, lower-right quadrant, Fig. 4B), with some related to an IFN response: ISGs OAS and DDX60 (see Table 2B).

To better understand these trends in PBMC of the patient 002, we performed GSEA conditioned upon two independent sources of relevant published gene sets: BTM (23) and the Broad Institute immune gene collection (ref. 21; C7). GSEA using BTM demonstrated upregulated presence of activated DCs in the blood of patient 002 with IT poly-ICLC treatment, as well as upregulation of stimulatory cytokine expression. Interestingly, enrichment of these gene sets was specific for the patient 002, whereas NK/T-cell gene expression was similar between patients with SD and PD responses (Fig. 5). Notably, immune gene expression changed minimally in PBMC upon IM poly-ICLC treatment in both SD and PD patients, but strongly upon IT poly-ICLC treatments, suggesting the importance of IT poly-ICLC injection to achieve systemic immune cell activation. Analysis of GSEA of immune gene sets from the Broad Institute yielded similar trends (Supplementary Fig. S1). Taken together, our observations suggest systemic activation of the immune system including T-cell activation, elevated antigen-presentation cells, and inflammatory cytokine expression in patient 002 blood upon IT injections of poly-ICLC as defined by BTM and Broad gene sets.

To gain insight into IT immune cell infiltration profiles, we performed similar gene set enrichment analyses on tumor RNA-seq samples. Both immune gene set collections detected gene signatures of activated DCs, B cells, CD8⁺ T cells, Th17-polarized CD4⁺ T cells, NK cells, monocytes, and neutrophils. Expression of gene signatures related to IFNα, IFNβ, IFNγ, and IL17 stimulation was similarly elevated at the tumor sites of patients with SD and PD responses: at approximately similar levels, before and after poly-ICLC treatment (Supplementary Figs. S2–S4). The latter suggests the immune system is activated at the tumor site, but fails to execute its function due to perhaps different levels and/or distribution of immune cells in tumor and/or exhaustion states as observed in the IHC data.

Combined together, PBMC and tumor RNA-seq gene expression analyses revealed differences in immune profiles between patient 002 and patients 004 and 008. Notably, we detected differential cytokine expression and immune cell activation related to DC–NK cell cross-talk (Fig. 6; Supplementary Table S2). DC infiltration in blood correlated with specific upregulation of XCL1/2 chemokines by NK or T cells in the patient 002 (Fig. 6A). This observation is in agreement with specific upregulation of IL15 and IL23A at the end of the treatment cycle by the same patient, as well as elevated expression of Clec9A. We speculate that XCL1/2 expression may attract the migratory DC subset XCR1-DC, known for its ability to produce IL15, type I IFN, as well as cross-present tumor-derived antigens through the Clec9A receptor. Interestingly, the chemokines CXCL1 and CXCL5 were both elevated at the tumor site of patients 004 and 008 (Fig. 6B) prior to and after poly-ICLC treatment. Both cytokines are ligands of the CXCR2 receptor, activation of which in tumors has been shown to interfere with PD-1/PD-L1 immunotherapies in preclinical studies (28, 29). The trend of increased CXCL1 and CXCL5 expression may identify a protumorigenic inflammatory profile, possibly interfering with the poly-ICLC regimen, whereas expression of IL15, IL23A, Clec9A, XCR1, XCL1, and XCL2 may reflect a better response to immunotherapy.

Our data are limited to the analysis of just a few patients; therefore, we sought verification of these signatures in larger data sets from The Cancer Genome Atlas (TCGA) as a potential approach to define prognostic biomarkers of response to poly-ICLC. Survival analyses of available cancer patient data derived from the TCGA database support our hypothesis (Fig. 6). The high expression of gene markers associated with clinical benefit of poly-ICLC correlates with better prognosis in HNSCC (HNSC) and melanoma (SKMC) patients (Fig. 6C; e.g., DC-associated factors IL15 and XCR1). In contrast, elevated expression of protumorigenic receptor CXCR2 and its ligand CXCL1 correlates with poor prognosis (Fig. 6D). These observations, though clinical trends, suggest that the presence of antigen-presenting cells, IL15 and IL23A, at tumor sites has a beneficial effect on patient survival and perhaps on response to immunotherapy, whereas expression of CXCR2/CXCL1/CXCL5 axis may interfere with the treatment.

No significant findings were noted in analysis of pre- and post-IT poly-ICLC treatment in patients.

TCR sequencing revealed that the patient 002 with clinical benefit (SD) showed increases in TCR clonality (∼20% increase) and density (∼30% increase) after poly-ICLC treatment. Unfortunately, similar analyses of other patients were not possible due to inadequate tissue sampling size or no available tissue after other tests performed (IHC and RNA-seq).

Here, we present the results of a pilot trial testing a strategy of therapeutic in situ autovaccination with IT injections of the dsRNA viral mimic and TLR agonist, poly-ICLC. Poly-ICLC was well tolerated and generated local immune response in tumor microenvironment and systemic immune response as evident in the patient achieving clinical benefit.

A major limitation of the trial was the study population of patients with refractory disease who had failed several other lines of treatment (often with recent chemotherapy; Table 1) and thus already highly immunosuppressed. Certain chemotherapies can suppress T cells proliferating in response to antigen or poly-ICLC, and this effect is seen for as long as 6 months after cessation of chemotherapy (30). In this pilot trial, there were three rapid progressors who did not complete even 1 cycle and may not have received enough treatment with single-agent poly-ICLC to generate an antitumor immune response. Therefore, single-agent poly-ICLC may not be adequate treatment for patients with advanced, rapidly progressing HNSCC; however, information learned from our study can inform combination treatments in the future.

In the patient achieving SD, our pilot results suggest how sequential IT injections of poly-ICLC can potentially induce an effective personalized systemic therapeutic “autovaccination” against tumor antigens in patients. We postulate that a therapeutic in situ autovaccination strategy using poly-ICLC works via three immunomodulatory steps. First, IT Poly-ICLC potentially activates NK cells, triggers TLR3 and MDA-5 receptors present at many APCs, and induces other proapoptotic mechanisms, resulting in local tumor killing and release of tumor antigens (31–33). In prior studies, it has been shown that poly-ICLC can upregulate several hundreds of genes closely representing some 10 canonical innate immune pathways (6, 34, 35). Our RNA-seq analysis detected expression of over 18,000 transcripts and revealed poly-ICLC treatment changed gene expression of IFN-related genes both at the tumor and PBMC level indicative of local and systemic immune activation. Upregulation of genes in response to poly-ICLC treatment included chemokines, interferon-stimulated genes, genes associated with antigen processing, T-cell activation, and apoptosis.

In the second step, poly-ICLC danger signals activate macrophages and DCs at the tumor site, in which they acquire tumor antigens that are cross-presented to CD4 helper cells and to CD8 T cells to generate antigen-specific CTLs. The repeated administration of the poly-ICLC danger signal IT in the context of the patient's own tumor antigens may be critical for optimal priming of the system at this step. GSEA of PBMC and tumor RNA-seq also suggests systemic activation of immune system, increased infiltration of antigen-presentation cells (activated DCs) in the blood, and upregulation of inflammatory cytokine expression upon IT injections of poly-ICLC. In addition, RNA-seq analysis detected gene signatures of activated DCs, B cells, CD8 T cells, Th17-polarized CD4 T cells, NK cells, monocytes, and neutrophils. Importantly, T-cell costimulatory cytokines IL15 and IL23A were highly upregulated at the tumor site of SD patient upon completion of poly-ICLC treatment cycle, indicating the prolonged maintenance of immune activation.

The third step is attraction and maintenance of antigen-specific CTLs via various poly-ICLC–induced chemokines, costimulatory factors, and other mechanisms (14). This was seen systemically in patient 002 with clinical benefit (SD) whose PBMC showed transcriptional upregulation of genes related to cytokines/chemokines, T-cell activation, and antigen processing in response to poly-ICLC treatment. IHC analysis of tumor biopsies of patient 002 showed increased levels of CD4 (60-fold), CD8 (10-fold), PD-1 (20-fold), and PD-L1 (3-fold), consistent with T-cell activation, migration, and consequent upregulation of PD-L1. Other studies have demonstrated that poly-ICLC alone is sufficient to induce expression of the costimulatory molecules B7-H2, CD40, and OX40 (36, 37). Our findings as well as others suggest rationale for potential combination of poly-ICLC with OX40, anti–CTLA-4, anti–PD-L1, FLT3L, and other costimulatory factors for enhanced antitumor activity.

One interesting observation was the increased expression of protumorigenic chemokines CXCL1/5 at the tumor site of patient 004 (PD). CXCL1, expressed by tumor-associated macrophages, neutrophils, and epithelial cells, has been suggested to have tumorigenic and mitogenic properties in cancer cells and shown to be a major component required for serum-dependent melanoma cell growth. Importantly, CXCL1 and CXCL5 elicit its effects by signaling through the chemokine receptor CXCR2 found on MDSCs. Recent preclinical studies have shown that activated CXCR2 signaling modulates the tumor immune microenvironment, reducing CD8⁺ T-cell trafficking and activation, and dampening NK-cell activity and promoting regular T-cell expansion (24, 25). Therefore, in the patient with PD, elevated expression of CXCL1/5 at the tumor site could indicate activated CXCR2 signaling, interfering with poly-ICLC treatment with ensuring tumor outgrowth. AZD5069 is an investigational selective CXCR2 antagonist that inhibits the migration of CXCR2⁺ MDSCs to tumor microenvironment and may enhance immune-mediated tumor attack. It is currently being evaluated for safety and efficacy in human clinical trials. Ongoing clinical trials attempt to explore the potentially synergistic combination of anti–PD-L1 inhibitors (durvalumab) in combination with anti-CXCR2, AZD5069 (AstraZeneca, anti-CXCR2) in HNSCC (NCT02499328), and advanced pancreatic ductal cancers (NCT02583477), respectively. Our results also suggest that a combination of IT poly-ICLC with a CXCR2 inhibitor could be an effective strategy and warrants further investigation (28, 29). Our ongoing adaptive phase II trial attempts to explore how such inflammatory responses can be harnessed to enhance therapeutic efficacy in such effective combinations.

Although limited by a study population of patients with refractory disease who had failed other lines of treatment and were already highly immunosuppressed with significant tumor burden, results of our study should still be informative in the design of future combination studies of poly-ICLC with other systemic therapies including immune checkpoint blockade, or inhibitors of potential protumorigenic signaling pathways, like CXCR2, which are being tested in the clinic. In addition, other areas of ongoing exploration with poly-ICLC treatment are the optimal dose and schedule of treatment. Timed release or structured formulations of poly-ICLC could be adjusted to better mimic a viral infection, especially for deep tumors in which repeated IT administration is logistically more challenging. At the same time, the possibility of overstimulation with a PAMP must also be considered, and the optimal dose and schedule to mediate optimal anticancer effects remains a needed area of our investigation. Immune infiltrate at distal tumors should also be explored to evaluate whether in situ vaccination has systemic therapeutic effect at other metastatic tumor sites. These trials are on-going and under exploration in our current ongoing adaptive phase II trial that attempts to explore these combinations.

In summary, we present a novel IT approach using poly-ICLC, a viral mimic of the double-stranded RNA encountered in viral replication, demonstrating its capacity to induce immunogenic cell death and also an immune stimulation effects through activation of DCs, upregulation of cytokines and chemokines, and T-cell priming. Indeed, other recent clinical studies with engineered herpes simplex virus-1 expressing granulocyte-macrophage colony-stimulating factor (talimogene laherparepvec) also demonstrated response in distal tumors with evidence of enhanced TILs (38). The encouraging results of these studies, IT poly-ICLC, as well as other pattern recognition receptor (PRR agonists such as TLR and STING agonists) suggest how in situ autovaccination can generate a localized antitumor immune response ultimately driving systemic antitumor immunity and provide a strong rationale for clinical exploration of combinations with immune checkpoint antibodies with poly-ICLC as well as other oncolytic viruses. These preliminary findings warrant further investigation, and a larger multicenter phase II clinical trial (clinicaltrials.gov, NCT02423863) is now underway to confirm these findings in advanced solid cancers.

B. Greenbaum reports receiving speakers bureau honoraria from Merck and Bristol-Myers Squibb. A. Salazar reports receiving commercial research grants from and holds ownership interest (including patents) in Oncovir, Inc. P. Friedlander is a consultant/advisory board member for Seattle Genetics and Array. N. Bhardwaj is on the scientific advisory board of Neon, CPS Companion Diagnostics, Genentech, and CureVac. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. Kyi, Y. Saenger, K. Misiukiewicz, A. Salazar, N. Bhardwaj

Development of methodology: C. Kyi, A. Salazar, N. Bhardwaj

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Kyi, R. Sabado, Y. Saenger, T.H. Thin, M. Donovan, P. Friedlander, N. Bhardwaj

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Kyi, V. Roudko, R. Sabado, W. Loging, J. Mandeli, M. Donovan, M. Posner, B. Greenbaum, A. Salazar, P. Friedlander, N. Bhardwaj

Writing, review, and/or revision of the manuscript: C. Kyi, V. Roudko, M. Donovan, M. Posner, K. Misiukiewicz, A. Salazar, P. Friedlander, N. Bhardwaj

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Sabado, Y. Saenger, N. Bhardwaj

Study supervision: Y. Saenger, K. Misiukiewicz, N. Bhardwaj

Other (coordination of care of research subjects and administration of investigational product): D. Lehrer

This research was supported by grants from the Cancer Research Institute, the Melanoma Research Alliance, and NIH. N. Bhardwaj is a member of the Parker Institute for Cancer Immunotherapy, which supported the Mount Sinai Hospital Cancer Immunotherapy Program.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.