Blocking Indolamine-2,3-Dioxygenase Rebound Immune Suppression Boosts Antitumor Effects of Radio-Immunotherapy in Murine Models and Spontaneous Canine Malignancies

This article is featured in Highlights of This Issue, p. 4275

Immunotherapy is revolutionizing metastatic cancer therapy. Patients responding to these therapies can achieve durable long-term remissions. However, most patients fail to respond, and some can experience significant immune-mediated toxicities (1–3). There is growing interest in understanding mechanisms of resistance to immunotherapy and finding combinatorial strategies to further increase efficacy while minimizing toxicity.

Local radiotherapy (RT) is an ideal candidate for combined modality immunotherapy strategies. In addition to debulking tumor and releasing tumor antigens, RT has well-established immunomodulatory effects (4). Preclinical and clinical reports confirm the safety and efficacy of multimodality strategies employing RT and immunotherapy. One particularly promising strategy is RT in combination with the immune-stimulatory toll-like receptor 9 (TLR9) agonist CpG oligodeoxynucleotide (CpG), which has demonstrated significant synergy in preclinical models (5) and the ability to induce regression of systemic disease in clinical trials (6, 7). Systemic response rates were about 20% with disease stability in another 20% of patients with refractory systemic (6) or cutaneous lymphoma (7). However, patients whose tumors induced regulatory T cells (Tregs) responded poorly. The effects of CpG on Treg-mediated immune suppression can be paradoxical. CpG can reduce Tregs by converting them to a T-helper phenotype via IL6 production (8) and can also directly reverse Treg function (9). Conversely, CpG can directly induce Foxp3 expression (10) and upregulate indolamine-2,3-dioxygenase (IDO; ref. 8) which is known to induce and maintain Tregs.

IDO is an immunosuppressive enzyme that catalyzes the rate-limiting step in the catabolism of tryptophan to kynurenine and is expressed by numerous human malignancies (11, 12). IDO expression can induce immune tolerance to malignancies (13) via complex mechanisms (14). In dendritic cells (DC), IDO can activate Treg-suppressive function in a CTLA-4–dependent manner (15). It can also prevent DC IL6 production, thereby preventing Treg conversion and maintaining elevated Treg levels despite inflammatory signals such as CpG (16). Furthermore, IDO can directly induce Tregs from naïve CD4⁺ T cells via 3-hydroxyanthranillic acid, a downstream catabolite of IDO tryptophan metabolism (17, 18). IDO can, through similarly complex mechanisms, inhibit natural killer cells and prevent effector T-cell activation and proliferation (14, 19). IDO expression can be paradoxically upregulated after inflammatory signals (20, 21) presumably as a mechanism to limit inflammation and maintain immune homeostasis. 1-Methyl-Tryptophan (1MT) is a pharmacologic inhibitor of IDO with demonstrated antitumor properties, which is being tested alone or in combination with chemotherapy in human trials (22).

Using a novel immunotherapy strategy combining local RT, intratumoral CpG, and IDO blockade, we tested the hypothesis that IDO upregulation after immunostimulatory therapies, such as RT + CpG, maintains tumor microenvironment immune suppression and limits treatment efficacy. We show that immunostimulatory therapies paradoxically upregulated IDO expression, which we termed “rebound immune suppression.” The addition of 1MT decreased IDO activity, decreased Tregs and other immune-suppressive factors within the tumor, and significantly improved the antitumor effects of RT + CpG in a CD8⁺ T-cell–dependent manner. Although the immune effects were primarily limited to the microenvironment of the treated tumor, systemic antitumor responses were observed. We confirmed these results in a veterinary clinical trial of companion canines since canine cancers closely represent human malignancy. We observed significant responses at the RT-treated primary tumor as well as at untreated sites of metastatic disease in most dogs. There were also significant changes in the immunosuppressive tumor microenvironment, corroborating our mechanistic mouse data. In marked contrast with immune checkpoint inhibitors, little toxicity was observed in mouse or canine studies. These results confirm the potency of combinatorial immunotherapy strategies and suggest that IDO may play a critical role in maintaining the immunosuppressive tumor microenvironment after immunotherapy. The addition of IDO blockade may safely improve the efficacy of immunotherapy by preventing rebound immune suppression.

CpG ODN 1826 (mouse studies) and 2006 (canine studies) were purchased from Invivogen. 1-methyl-D,L-tyrptophan was purchased from Sigma-Aldrich. CD8-depleting antibody (cloneYTS169.4) was administered prior to the start of therapy and once weekly thereafter. The agonistic anti-mouse CD40 antibody (FGK115.B3) was generated as previously described (23). Rat IgG (Jackson ImmunoResearch Laboratories) was used as a control for anti-CD40. Recombinant human interleukin-2 (rhIL2; Teceleukin; Roche) was provided by the NCI.

All studies were approved by the UC Davis Institutional Animal Care and Use Committee (IACUC), and humane endpoints were used. Female 8- to 12-week-old C57BL/6 or BALB/c mice were purchased from the animal production area at the NCI. All mice were housed at the UC Davis animal facilities under specific pathogen-free conditions. Mouse tumor studies were performed as outlined (Fig. 1D). 4T1 breast adenocarcinoma tumors were grown orthotopically in the mammary fat pad of BALB/c mice, and B16 melanoma tumors were implanted into the flank of C57BL/6 mice. Mouse tumors were irradiated with 8 Gy per fraction delivered with 9 MeV electrons on an Elekta synergy clinical accelerator using a 2-cm diameter electron field and 0.5-cm bolus. Treatment accuracy was confirmed using anthropomorphic phantoms as previously described (24). Mice were sacrificed and tissues harvested at 7 or 14 days after treatment completion for mechanistic studies.

This clinical trial was approved by the UC Davis School of Veterinary Medicine Clinical Trials Review Board and IACUC. Five canines with histologically confirmed metastatic melanoma or sarcoma were enrolled with informed consent of their owners. The trial schema is outlined in Supplementary Fig. S1. Treatment consisted of weekly 8 Gy fractions of RT repeated for a total of four treatments over 4 weeks. RT was delivered to the primary tumor only using a clinical grade linear accelerator at the UC Davis Center for Companion Animal Health. After each RT treatment, canines received a 2-mg intratumoral injection of CpG. Patients also received oral 1MT at 1,200 mg daily for the 28-day treatment period. Radiologic scans, blood, and tissue samples were obtained before treatment and at follow-up visits at weeks 5, 8, and 20. Radiographic response rates were determined using irRECIST criteria (25). Patients were evaluated by a veterinary radiation oncologist for treatment toxicity weekly during therapy and at the 5, 8, and 20-week follow-up visits using clinical exam and blood laboratories. Toxicity was graded according to the VRTOG criteria (26).

Staining procedures and antibodies are detailed in Supplementary Methods. All data were collected using a BD Fortessa flow cytometer equipped with BD FACSDiva software. Data were analyzed using FlowJo v10 (TreeStar).

Quantitative real-time PCR (qRT-PCR) was performed using AB Step-ONE Plus (Applied Biosystems) in the presence of SYBR Green Supermix (Applied Biosystems). Details of RNA isolation, cDNA synthesis, and qRT-PCR can be found in Supplementary Methods.

Kynurenine to tryptophan ratio was determined using mass spectrometry at the UC Davis Metabolomics Center as detailed in Supplementary Methods.

Statistical analysis was performed with Prism software (GraphPad). Data were expressed as mean ± SEM. For analysis of three or more groups, the one-way ANOVA was performed with the Holm–Sidak correction for multiple comparisons. Analysis of differences between two normally distributed test groups was performed with the Student t test. Welch correction was applied to Student t test datasets with significant differences in variance. ^∗, P < 0.05; ^∗∗, P < 0.01; ^∗∗∗, P < 0.001; and ^∗∗∗∗, P < 0.0001.

Immunofluorescence and Immunohistochemistry methods are detailed in Supplementary Methods.

To test our hypothesis that IDO-mediated “rebound immune suppression” maintains immune suppression and limits treatment efficacy, we employed the poorly immunogenic and highly metastatic mouse 4T1 orthotopic breast tumor model (27). We first evaluated IDO expression within the tumor microenvironment after immunostimulatory therapies and observed a significant increase in IDO-expressing cells compared with untreated controls in the glandular nests of malignant cells (Fig. 1A and B; P < 0.0001). IDO can be upregulated by tumor-infiltrating immune cells as well as tumor cells. In our model, IDO upregulation appears to be predominately in the neoplastic epithelial cells. IDO upregulation within 4T1 cells was verified in vitro as well (data not shown). We likewise observed a parallel statistically significant 3- to 5-fold increase in IDO1 mRNA expression (Fig. 1C). Importantly, this upregulation of IDO was not limited to RT and CpG but was also observed with other immunotherapy strategies (Fig. 1A–C). These results indicate that immunotherapy can paradoxically upregulate immunosuppressive pathways, such as IDO, which may limit efficacy.

To investigate if IDO-mediated “rebound immune suppression” after RT + CpG limits treatment efficacy, we next tested the effects of adding systemic IDO blockade. Our treatment schema (Fig. 1D) consisted of two 8 Gy fractions of local RT administered to the primary tumor over a 7-day period. Each fraction of RT was accompanied by intratumoral injection of 20 μg of CpG based on dose response data. CpG was administered locally to mirror human clinical trials and minimize systemic toxicity (6, 7). For IDO blockade, 1MT was administered by daily 2 mg i.p. injections throughout the study period. The triple combination decreased IDO enzymatic activity, as measured by serum kynurenine/tryptophan ratio, below the level of untreated controls (Fig. 1E; P = 0.03). We next evaluated the antitumor effects of 1MT, RT, and CpG, alone or in combination. CpG and 1MT alone or in combination had no significant effect on tumor growth (Fig. 1F). Local RT alone significantly reduced tumor growth, but by day 34, there was accelerated tumor outgrowth, and mean tumor size was no longer statistically different than controls (Fig. 1F and G). Although CpG and 1MT had no antitumor effects alone or in combination, when either one was combined with RT, they significantly inhibited tumor growth (Fig. 1F and G). Notably, the triple combination was significantly better than either RT + CpG or RT + 1MT decreasing tumor growth by an additional 3-fold (day 34 mean tumor size: 377 mm³, 1,060 mm³, 1,057 mm³, respectively; Fig. 1F). The triple therapy also significantly improved the survival of 4T1-bearing mice (Fig. 1H, P < 0.001). To ensure these results were not tumor or mouse strain specific, we tested this regimen in C57BL/6 mice bearing B16 melanoma tumors. Again, mice treated with the triple combination had significantly smaller tumors than mice treated with RT + CpG (Fig. 1I, P = 0.042). Importantly, in both models, there was no evidence of autoimmune or other toxicities induced by the therapy. Thus, as hypothesized, the addition of 1MT significantly improved the antitumor effects of RT + CpG in 4T1 and B16 tumors (Fig. 1G and I).

We next examined the systemic antitumor effects of this triple therapy in 4T1-bearing mice. At day 40, heavy burdens of pulmonary metastatic disease could be grossly identified in control but not treated mice (Fig. 1J). The primary tumors (located in the distal mammary fat pad) were treated with 2-cm radiation portals, and the lungs were well outside of the radiation fields. Similar results were also observed by CT imaging (Fig. 1K) and further corroborated in quantifiable fashion using in-vitro lung tumor colony-forming assays with a 6-fold reduction in lung colony-forming units in the triple combination group compared with controls at day 28 (Fig. 1L, P = 0.001). Crucially, lungs from the majority of the triple combination–treated mice grew no tumor colonies.

Taken together, our findings indicate that RT + CpG resulted in antitumor effects but also upregulated IDO expression. The addition of IDO blockade with 1MT decreased IDO activity and significantly improved the antitumor effects. This triple therapy was well tolerated, increased survival, and markedly reduced systemic metastases.

Based on our murine studies, we initiated a pilot clinical trial for outbred companion canines with spontaneous metastatic melanomas and sarcomas. Due to the aggressive nature and poor prognosis of these cancers, the standard of care for these canine patients is palliative RT to the primary tumor consisting of 4 weekly fractions of 8 Gy. Our treatment protocol (Supplementary Fig. S1A) built on this palliative regimen but mirrored our mouse studies with an intratumoral injection of CpG at the time of RT and daily administration of 1MT for the treatment course. CpG was administered at 2 mg/dose to mirror human clinical studies. The activity of CpG ODN 2006 used in human clinical trials was verified in canines with an in-vitro peripheral blood mononuclear cell proliferation assay (Supplementary Fig. S1B). 1MT was administered orally at 1,200 mg per day based on pharmacokinetic studies in canines (28). We verified the ability of 1MT to reduce IDO enzymatic activity in canines by measuring serum kynurenine/tryptophan ratio in samples collected from trial canines (Supplementary Fig. S1C; P = 0.02). We enrolled 5 canines with rapidly progressing metastatic disease to this pilot trial (Supplementary Table S1). Four had mucosal melanomas with pulmonary metastases, and one had soft tissues sarcoma with nodal and pulmonary metastases (Supplementary Table S1). Canines with metastatic melanomas have an extremely poor prognosis (29). The survival of canines in this study (5.8 months, 95% confidence interval: 3.2–9.2 months) exceeded published historic controls (29). All 5 canines responded at the primary site of disease which was within the radiation portals (Supplementary Table S1; Fig. 2A–D). The local response to therapy was robust with a 50% to 100% reduction in tumor mass (Fig. 2A–D). Photographic and CT imaging examples documenting local responses are shown in Fig. 2B–D. Crucially, when examining best systemic responses by immune-related response criteria (25) at distant metastatic sites outside of the radiation portals, we observed one subject with a complete response, two with partial responses, one with disease stabilization, and one with progressive disease (Supplementary Table S1; Fig. 2E–H). CT imaging examples of partial (Fig. 2F–G) and complete responses (Fig. 2H) of index lesions are depicted. Mirroring human cancer immunotherapy trials (1, 3), responses were rapid and robust with >80 to 100% reduction in the volume of index lesions (Fig. 2E) and regression of even large bulky tumors (Fig. 2F and H). Some of the partial responses consisted of a mixed response with regression of some lesions and growth or stability of others (Fig. 2G). This type of mixed response has also been observed in human immunotherapy trials and is associated with a good prognosis (25). Thus, in canines with previously rapidly progressing disease, the local response rate was 100% and the systemic response rate was 60%, with another 20% disease stabilization. Upon cessation of treatment in this trial, all dogs eventually had CT documented or clinical/symptomatic progression. No further treatment was provided beyond the brief treatment course outlined in the schema, and assessment of longer administration, as is used in human trials, is needed.

Canines were monitored closely for toxicity with regular physical examinations and lab work. Importantly, the only adverse effects observed were mild mucositis and skin toxicity within the radiation portals that did not exceed what would be expected from palliative RT alone (Supplementary Table S1). These results demonstrate that this triple combination therapy can be safely administered resulting in significant antitumor effects in metastatic disease and validate the findings of our mouse studies in a model more representative of human cancer.

The primary rationale for adding 1MT to RT + CpG was our hypothesis that IDO-mediated rebound immune suppression limited efficacy by maintaining Tregs and an immunosuppressive tumor microenvironment despite inflammatory signals. Clinical efficacy has been observed with RT + CpG, but patients whose tumors induce Tregs are unlikely to respond (6). We examined the fate of immunosuppressive Tregs after the completion of therapy in mice and canines. In mice, Treg levels were maintained within the tumor after RT + CpG (Fig. 3). The addition of 1MT significantly and substantially reduced the Treg subset of CD4⁺ T cells within the tumor microenvironment but not in the periphery (Fig. 3; Supplementary Figs. S2 and S3). The addition of 1MT resulted in a greater than 4-fold decrease in intratumoral Tregs compared with RT + CpG alone 14 days after therapy (Fig. 3A and B; P = 0.02). This triple combination, but not immunotherapy alone (CpG + 1MT), RT alone, or RT + CpG, significantly reduced Tregs compared with control mice (Fig. 3A and B; control: 24.4% ± 2.1% vs. RT + CpG + 1MT 6.2% ± 2.2%; P < 0.01). Similar results were seen if Tregs were analyzed as a percentage of total cells in the tumor (control: 0.093% ± 0.072% vs. RT + CpG + 1MT 0.018% ± 0.003%; P = 0.07, data not shown). IL6 induction after immunostimulatory therapies contributes to the reduction of Tregs. IDO can maintain Tregs in tumors by downregulating IL6 (8). We observed a 9-fold increase in IL6 expression within the tumor microenvironment after triple combination therapy (Supplementary Fig. S4; P = 0.03). We also observed a significant 3- to 4-fold increase in the CD4⁺/Treg ratio, which is a known indicator of immune status and prognosis (11, 30, 31), in RT + CpG + 1MT–treated mice compared with tumor-bearing control mice (P < 0.01), CpG + 1MT–treated mice (P < 0.05), RT-treated mice (P < 0.01), and RT + CpG–treated mice (P < 0.01; Fig. 3C). The addition of 1MT provided no added benefit in terms of Treg reduction and improvement in the conventional CD4/Treg ratio in the dLN (Supplementary Fig. S2A–S2C), spleens (Supplementary Fig. S2D–S2F), or nondraining lymph nodes (data not shown). This is in contrast with the tumor, where the greatest Treg reductions were seen in the 1MT-containing groups (CpG + 1MT and RT + CpG + 1MT; Fig. 3A–C). Similar results were also seen at earlier time points (Supplementary Fig. S3). Thus, although 1MT was administered systemically, its primary action is within the tumor microenvironment where IDO is also upregulated.

Informed by our murine studies, we examined the fate of Tregs in the tumors, dLN, and systemic circulation of canines receiving RT + CpG + 1MT (Fig. 3; Supplementary Fig. S2). When possible, tumor biopsy, tumor-draining lymph node biopsy, and peripheral blood samples were obtained before and after therapy. Peripheral blood was obtained from all dogs. A posttherapy tumor biopsy could not be obtained from one dog due to a complete response. In addition, some dogs displayed massive tissue death in the posttreatment tumor biopsy samples limiting analyses. Lymph node biopsies could not be obtained from two dogs in which dLNs were not accessible with a minimally invasive procedure. Mirroring the results of our mouse studies, the percentage of tumor-infiltrating Tregs was significantly reduced in all dogs after therapy with virtually no Tregs remaining in the tumor microenvironment after therapy (Fig. 3D and E; P = 0.014). Conversely, in the three dLN samples, we did not observe a statistically significant change in Tregs across the cohort with Tregs dropping substantially (3–4-fold) after therapy in some dogs but increasing in others (Supplementary Fig. S2G). Also mirroring our mouse results, no significant change in peripheral Tregs was observed (Supplementary Fig. S2H). To confirm our results from flow cytometric analysis, immunofluorescence staining for FoxP3 expression in canines that had sufficient remaining biopsy samples was performed. Representative fields and quantification from a canine with Treg decreases in the tumor are depicted in Fig. 3F and G confirming marked reduction in Foxp3⁺ cells following triple combination therapy.

In the mouse tumor models, we also observed significant reductions in other immune-suppressive factors in the tumor microenvironment. Tumor associated macrophages (TAM) are known to play a critical suppressive role in the microenvironment of 4T1 tumors (32). Mirroring the reduction seen in Tregs only, the RT + CpG + 1MT triple combination resulted in a statistically significant reduction of TAMs by nearly 5-fold compared with control (Fig. 4A and B; 18.4% ± 10.3% vs. 3.8% ± 1.5%; P < 0.05). At baseline, intratumoral TAMs were M2 polarized as assessed by arginase to inducible nitric oxide ratio (data not shown). We next examined the expression of the TGF-beta gene (TGFB) which both induces and is produced by Tregs and TAMs (reviewed in 33, 34), is produced by IDO-expressing DCs (35), and is also known to block lymphocyte activation (36). All of the therapies reduced TGFB expression relative to control, but this was most pronounced and most significant in the triple combination (Fig. 4C; P < 0.0001). Taken together, these studies indicate that this therapy reduces multiple immune-suppressive factors and restores IL6 signaling within the tumor microenvironment.

We next examined the effects of this triple therapy on CD8⁺ T cells and DCs. The percentage of tumor-infiltrating CD8⁺ T cells (expressed as a percentage of total viable cells within the tumor) significantly increased from 0.68% in control mice to 2.71% 7 days after RT + CpG + 1MT (Fig. 5A and B; P = 0.02). The CD8⁺ T-cell/Treg ratio in the tumor was also increased by therapy from 6.3 to 21.7 (Fig. 5C; P = 0.0006). There were no changes in activation/functional/exhaustion markers such as PD-1, CD69, or IFN-gamma in the CD8⁺ T cells after therapy, but the majority of intratumoral CD8⁺ T cells expressed these markers at baseline (data not shown). We also evaluated changes in tumor-infiltrating CD8⁺ T cells in treated canines. By flow cytometric analysis, there was a posttherapy increase in intratumoral CD8⁺ T cells (as a percentage of all CD3⁺ intratumoral cells) in all three evaluable canines (Fig. 5D and E). Overall, the frequency of intratumoral CD8⁺ T cells doubled after therapy, which trended toward statistical significance (Fig. 5D and E; 24% ± 8.5% vs. 47.6% ± 16.4%; P = 0.06). IHC staining likewise demonstrated an increase in intratumoral CD8⁺ T cells after therapy (30 ± 33 vs. 50 ± 39 cells/HPF; Fig. 5F and G; P = 0.09). Mirroring our mouse studies therapy induced a significant increase in the CD8⁺/Treg ratio (Fig. 5H; P = 0.014).

We tested the CD8 dependence of this therapy in mouse models. Depletion of CD8⁺ T cells by intraperitoneal administration of anti-CD8 reduced the number of circulating CD8⁺ T cells by >99% (Supplementary Fig. S5A and S5B). In mice depleted of CD8⁺ T cells, we observed that the antitumor effects of RT + CpG + 1MT were significantly diminished. After treatment with the triple combination, tumors were more than tripled in size in CD8-depleted mice compared with those treated with control IgG (Supplementary Fig. S5C; 2,066 mm³ vs. 684 mm³; P < 0.0001). The survival benefit of the triple therapy was also significantly diminished in CD8-depleted mice (Supplementary Fig. S5D; P = 0.03), indicating the critical role of CD8⁺ T cells in the antitumor effects.

In addition to the effects of CD8⁺ T cells, we also evaluated DC activation and phenotypes following triple therapy. IDO expression has been linked to a tolerogenic DC phenotype, associated with decreased expression of CD80 and MHC II and increased TGF-beta production (35). No change in DC numbers or activation within dLNs was observed (data not shown). Recent data demonstrate that intratumoral DCs, although a minor population, play a critical role in antitumor T-cell responses (37, 38). Given that the major effects of this therapy occur in the tumor microenvironment, we also examined DCs in the tumor microenvironment. There was an increased number of activated DCs within treated tumors, with a 4-fold increase in CD80-expressing DCs (Supplementary Fig. S6A–S6C; P = 0.02) and a near doubling of DC MHCII mean fluorescence intensity (MFI) compared with control mice (Supplementary Fig. S6D and S6E; P = 0.01).

Finally, in three of the canine patients for which sufficient pre- and posttreatment tumor tissue was available, we evaluated gene signatures by qPCR (Fig. 6). Based on our mouse data, we focused on expression of three immunosuppressive markers (IDO1, TGFB, and FOXP3). Two of these patients demonstrated systemic partial responses, and one had systemic progressive disease. Interestingly, in the two responders, expression of all three immunosuppressive genes was markedly decreased after therapy, whereas all three were increased in the setting of progressive disease. These data suggest that this gene signature in the local tumor may represent a predictive biomarker for systemic response to this therapy and requires further evaluation in future studies. Overall, although hypothesis-generating in nature, these canine correlative immune analyses corroborate our mechanistic mouse data and provide rationale for further exploration of these immunologic endpoints in future studies.

This is the first report of a novel immunotherapy strategy combining local RT + CpG with systemic IDO blockade. The primary purpose of this study was to examine whether IDO-mediated rebound immune suppression after RT + CpG immunotherapy maintains an immunosuppressive microenvironment and limits efficacy. As hypothesized, RT + CpG paradoxically upregulated IDO expression, which we termed “rebound immune suppression.” Importantly, this immunotherapy-induced increase in IDO expression was also observed with other immunotherapies, suggesting that these mechanisms may have broader implications. Other studies also demonstrate the upregulation of IDO after inflammatory signals (20, 21). This rebound immune suppression suggests a physiologic regulatory mechanism, whereby the immune system tempers responses in inflammatory settings to maintain homeostasis. The addition of IDO blockade decreased IDO activity and substantially increased local and systemic antitumor effects. Mechanistically, this therapy reduced multiple immune-suppressive factors, including Tregs, TAMs, and TGF-beta; increased intratumoral CD8⁺ T cells; activated intratumoral DCs; and restored IL6 signaling within the tumor microenvironment.

Although IDO blockade was administered systemically, the immunologic effects of IDO were most pronounced in the tumor microenvironment but limited in the tumor-draining lymph nodes. This suggests that the immune reactivation is likely occurring directly within the tumor microenvironment and that this is sufficient to induce a systemic antitumor immune response. A recent report by Levy and colleagues indicates that reversal of immune suppression triggering an active immune response at a single site is sufficient to induce an effective systemic antitumor immune response (39). Our data are also corroborated by findings demonstrating that after checkpoint inhibition or IDO blockade, little effect is seen in the dLNs, but there are marked effects on CD8⁺ T cells in the tumor (40). Likewise, recent data demonstrate that intratumoral DCs, although a minor population, play a critical role in antitumor T-cell responses (37, 38). Overall, our findings are in line with the published literature and support an emerging concept that augmenting a pre-existent but previously ineffective antitumor immune response directly within the local tumor microenvironment can serve as the nidus for a systemic response.

A principal finding of this study is the utility of RT in combination with immunotherapy. We observed no effect of the immunotherapies alone or in combination but robust effects in combination with RT (Fig. 1). This synergy was seen in both the antitumor effects and the mechanistic immune changes induced by therapy. The reduction in tumor-infiltrating Tregs and TAMs was only seen when RT was combined with immunotherapy (Fig. 3). Likewise, the increase in tumor-infiltrating CD8⁺ T cells induced by RT + CpG + 1MT (Fig. 5) was not seen with RT or immunotherapy alone (data not shown). These results add to the growing body of literature demonstrating the potent synergy of RT and immunotherapy.

The utility of mice to model human disease has recently been called into question (41). Mouse models provide an excellent platform for exploratory studies, but often fail to adequately recapitulate treatment efficacy or toxicity in human clinical trials (42). The introduction of pretranslational confirmatory studies in a more robust model of human disease can help improve the efficiency of the translational pipeline by limiting human testing of therapies that are unlikely to be effective or tolerated. We tested our therapy in such a model by initiating a pilot veterinary clinical trial in companion canines with late-stage metastatic spontaneous melanomas or sarcomas. These cancers arise, grow, and metastasize in the setting of an intact immune system and with patterns akin to human tumors (43). Canine immune systems are genetically and developmentally much more similar to humans than rodent models (44–47). In addition, companion dogs share many environmental risk factors with their human owners. In dogs with rapidly progressive systemic disease, our results demonstrated a significant response at the local RT-treated primary tumor as well as response or disease stabilization in 80% of dogs at untreated sites of metastatic disease. This included substantial reductions of bulky pulmonary metastases which were well outside of the irradiated fields. We also observed significant changes in the immunosuppressive tumor microenvironment confirming the mechanistic data from our mouse models. Overall, the antitumor effects and mechanistic immunologic findings from our mouse models were paralleled in the canine clinical trial.

Immunotherapy can provide durable responses in a number of metastatic cancers (1–3). Unfortunately, most patients will fail to respond to these therapies, and these treatments can induce significant toxicities. To increase response rates, combinatorial strategies are being actively investigated. Recent clinical trials demonstrate that combining PD-1 and CTLA-4 checkpoint inhibitors can increase response rates in metastatic melanoma but are also accompanied by an equally marked increase in toxicity with 40% of patients responding but also 53% of patients experiencing grade 3 or 4 toxicities (48). These toxicities were attributed to the immune dysregulation induced by multiple checkpoint blockade. RT + CpG + 1MT employs immune stimulation in combination with blockade of immune suppression limiting overlapping toxicity profiles and leaving multiple checkpoint pathways intact. We observed minimal toxicity with this regimen.

An area for future study is the assessment of longer courses of therapy. Treatment only lasted 2 weeks in mice and 4 weeks in canines. All of our canines eventually recurred after treatment discontinuation. Although robust responses were observed to this short treatment course, most published immunotherapy strategies employ prolonged courses of therapy, and it is likely that prolonging the course of therapy in our models would further increase the magnitude and duration of response.

In summary, this study substantiates the power of combinatorial immunotherapy strategies and confirms that RT can be a potent partner for immunotherapy. Furthermore, it demonstrates that combinatorial strategies designed to minimize overlapping toxicity can be safe and effective. The correlative studies described here demonstrate the importance of posttreatment tumor biopsies, which are rarely employed in human trials, to evaluate the immunologic changes induced by therapy and to further our understanding of the mechanistic underpinnings of therapy response and failure. In addition, they suggest that IDO-mediated rebound immune suppression may be a general mechanism of resistance to some immunotherapy strategies and that addition of IDO blockade may improve the effectiveness of such strategies. They demonstrate the feasibility of preclinical testing of immunotherapy in spontaneous canine malignancies as a pretranslational step in a model more reflective of human disease. Overall, these studies provide a strong rationale for clinical translation of this immunotherapy strategy to substantially improve the already documented efficacy of RT + CpG.

M. Chen reports receiving a commercial research grant from Bayer. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.M. Monjazeb, M.S. Kent, M. Chen, G.D. Sckisel

Development of methodology: A.M. Monjazeb, M.S. Kent, S.K. Grossenbacher, A. Mirsoian, M. Chen, A. Kol, E.E. Sparger

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.M. Monjazeb, M.S. Kent, S.K. Grossenbacher, A. Mirsoian, M. Chen, A. Kol, S.L. Shiao, A. Reddy, J.R. Perks, W.T.N. Culp

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.M. Monjazeb, S.K. Grossenbacher, C. Mall, A.E. Zamora, M. Chen, A. Kol, S.L. Shiao, A. Reddy, W.T.N. Culp, E.E. Sparger, R.J. Canter

Writing, review, and/or revision of the manuscript: A.M. Monjazeb, M.S. Kent, S.K. Grossenbacher, C. Mall, A.E. Zamora, A. Mirsoian, M. Chen, A. Kol, S.L. Shiao, W.T.N. Culp, R.J. Canter, G.D. Sckisel, W.J. Murphy

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.M. Monjazeb, S.K. Grossenbacher, C. Mall, A.E. Zamora, A. Mirsoian, A. Reddy

Study supervision: A.M. Monjazeb, W.J. Murphy

The authors acknowledge Weihong Ma, Shuaib Juma, Hong Chang, Hung Kieu, and Monja Metcalf for their technical support and Teri Guerrero for veterinary clinical trials support.

This study was supported by UC Davis CTSC K12 Scholar Program, NIH 2 KL2 RRO24144-06; American Cancer Society Institutional Research Grant, IRG-95-125-07; Amador Cancer Research Foundation; Christine and Helen Landgraf Award; and NIH R01 CA 095572, NIH R01 CA 072669, and NCI P30CA093373.

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.