Immune checkpoint inhibitors (ICI), notably anti-PD-1 antibodies, have drastically changed the management of patients with cancer over the past decade. Indeed, ICI increase survival in several cancers including advanced melanoma and mismatch repair-deficient colorectal carcinoma (CRC) (1, 2). ICI can enhance antitumor immune responses of patients harboring “hot” tumors (3). By contrast, ICI seem ineffective in inducing antitumor immune responses in poorly immunogenic “cold” tumors. Thus, there remains a need for the development of new treatment combinations that would improve the rate of responses to ICI.

An emerging therapeutic strategy consists of combining ICI with locoregional therapies that release massive quantities of antigens following tumor cell destruction, thereby enhancing specific antitumor responses (4). Radiotherapy (5), thermal ablation (6), or transarterial chemoembolization (TACE) (7) have demonstrated synergistic effects with anti-PD-1 therapies. Phase III clinical trials are ongoing to evaluate the efficacy of such combinations, for example, the EMERALD-1 study comparing TACE alone to TACE with the anti-PD-L1 durvalumab (NCT03778957). Moreover, an abscopal effect (i.e. shrinking of tumors that were not the direct target of local therapy) has already been observed following combined treatment with radiotherapy and ICI (8), providing a strong rationale for combined local destruction and ICI.

Conventional TACE (cTACE) is a locoregional therapy used to treat primary liver cancer or liver metastatic colorectal carcinomas (CRC) (9, 10). Cytotoxic drugs are emulsified in Lipiodol (ethyl esters of iodinated fatty acids of poppy seed oil), a radiopaque liposoluble iodinated contrast agent used as a tumor-targeting vehicle, drug delivery system, and microvasculature embolizing agent. Lipiodol allows for a local and slow release of the chemotherapeutic agent into the tumor, leading to less systemic toxicity and increased drug concentration in tumoral liver nodules (11). Additionally, using Lipiodol as a carrier for immunostimulatory molecules such as OK-432, oligonucleotides (ODN) and siRNA can enhance antitumor responses (12–14).

Innate immune stimulators, including toll-like receptor (TLR) agonists or vaccinal adjuvants, can modulate the tumor micro-environment (TME) and therefore potentiate responses to anti-PD-1 therapies (15–17). Such immunomodulators are currently under investigation in solid tumors, notably injected locally in combination with systemic ICI (NCT04401995, NCT05638698).

We aimed to potentialize ICI-driven antitumor immune responses by combining a locoregional treatment with immunomodulators, emulsified in Lipiodol such as in a cTACE procedure, with an already approved systemic anti-PD-1 treatment. Three immunomodulators were selected for their antitumoral adjuvant properties observed in preclinical models and clinical trials: the TLR-3 agonist Poly I:C, TLR-9 agonist CpG-ODN, and saponin QS21. Activation of TLR-3 or TLR-9, both intracellular receptors, leads to improved antigen presentation following dendritic cell stimulation and antitumoral cytokines release (18). The mode of action of saponins as vaccine adjuvants is still poorly understood, but QS21 is thought to induce a Th1 response and enhance antigen capture by dendritic cells (19). It was hypothesized that local administration of an immunomodulator mixed with Lipiodol, used as a radiopaque carrier and tumor-targeting agent, could locally modulate the TME to promote anti-PD-1 antitumoral effects in both treated and distant non-treated nodules in an anti-PD1-resistant murine colorectal cancer model.

Lipiodol, used as an intra-arterial drug delivery system for cytotoxic drugs and for immunomodulating molecules (12–14), is approved as an in situ therapy for liver malignancies (24, 25). Immunomodulators, such as TLR agonists, are mostly injected locally in preclinical studies (26, 27) and in current clinical trials (NCT03865082, NCT04935229), to activate innate and adaptive immune response in the tumor environment, in addition to avoid systemic toxicity. Anti-PD-1 therapies, on the other hand, are approved for the treatment of advanced cancers and are injected i.v. The aim of the study was to assess the efficacy of a combination of a local treatment with a Lipiodol-immunomodulator emulsion, with systemic injections of anti-PD-1.

This study showed that in a mouse model of CRC, i.t. injection of Lipiodol-CpG emulsion, combined with systemic anti-PD-1, induced an antitumor immune response in injected and distant subcutaneous nodules.

In a first study, a single i.t. injection was performed to select the adjuvant. The Lipiodol-CpG emulsion, especially when combined with systemic anti-PD-1, induced both an increase in pseudo-survival and a decrease in treated tumor growth. No difference was observed between the control group, Lipiodol alone or anti-PD-1 alone in terms of survival and tumor growth.

Lipiodol itself does not have antitumoral properties (28). Besides, the CT26 model, which may be considered as a microsatellite-stable CRC model, has already been characterized as non-responsive to anti-PD-1 monotherapy (29). One of the main mechanisms thought to be the cause of non-responsiveness to PD-1 blockade is that it binds predominantly to immune cells within the TME. The TLR-9 agonist CpG is widely known for its ability to modulate the innate and adaptive immune systems. CpG is an activator of effectors and dendritic cells (30, 31). When emulsified in Lipiodol, it reverses the resistance to anti-PD-1 treatment, presumably by transforming CT26 “cold” tumors into “hot” tumors. Surprisingly, no antitumoral effect of CpG alone was observed, unlike in other published studies on the CT26 model (22, 32), but the latter used multiple injections for a higher total dose of CpG.

Similar to other emulsions used as vaccine adjuvants and known for their immunomodulatory properties (33), using Lipiodol as a carrier may potentiate CpG antitumoral effect thanks to the emulsion formulation. Indeed, Lipiodol-CpG is a stable oil-in-water emulsion that allows longer CpG retention within tumors. Lipiodol, as a drug delivery system, may extend the immune stimulation induced by CpG, leading to a greater antitumoral effect, as it is known for cytotoxic drugs in cTACE. Moreover, since oil-in-water emulsions are more stable with CpG, we can hypothesize that CpG may act as a surfactant, with a similar effect to poly lactic-co-glycolic acid particles (34). However, our Lipiodol-CpG emulsions are direct oil-in-water and not water-in-oil emulsions, recommanded for a cTACE procedure. The water-in-oil emulsions are supposed to improve the delivery of hydrophilic drugs; therefore, our emulsion may not be optimized yet. The stability of Lipiodol-CpG emulsions has been verified only for 24 hours but is not guaranteed beyond 1 day. One limitation of emulsions is that they need to be made extemporaneously. Also, emulsions formulations can, in case a great stability, release only a proportion of the total dose of loaded drug, and not the whole dose (34). Working on the formulation of Lipiodol-CpG emulsions to increase their therapeutic effect is of great interest.

Some studies have suggested a correlation between Lipiodol deposition pattern and treatment response to cTACE in liver malignancies (35, 36). Here, Lipiodol distribution within the tumor was found to be heterogenous, and in responsive mice, Lipiodol was more abundant in the tumor rim. These results may differ from human studies, since Lipiodol emulsions were not injected intra-arterially.

The third step consisted in repeated i.t. injections. This time, CpG alone had a significant and local antitumoral effect. Interestingly, in this model, i) Lipiodol did not modify the effect of CpG, ii) the local antitumoral effect of CpG seemed to be greater than the systemic anti-PD-1 effect, as no difference being observed between the Lipiodol-CpG group and the combination one. We can hypothesize that with this repeated intratumoral administrations protocol, the increased intratumoral retention of the CpG allowed by Lipiodol do not enhance the antitumoral effect during the first eight days post treatment. Moreover, addition of anti-PD-1 did not improve the antitumoral effect of CpG during the first days post treatment.

This design was then completed by injection of a second “metastasis-like” nodule. This provided evidence of an abscopal effect, with a significant slowdown of the untreated nodule growth. To determine whether an enhanced immune response explains the antitumoral effects, the local and systemic immune cell populations were studied following treatment. Among all groups, no difference in all of the studied cells proportions differed between responsive and non-responsive mice, therefore only groups were compared to each other. Therefore, the immune activity was only associated with the global antitumoral response of a group, and not individually with the response within the same group. For all treated mice, spleen weight increased, as already described following CpG treatment (21). Splenomegaly has been correlated with increased MDSC proportions and is associated with poor prognosis in mouse tumor models, and is considered as a biomarker of response to ICI in the clinic (37, 38). The density of CD8+ T-cells increased after treatment within both treated and untreated nodules, especially in the combination group. Although CD8+ proportions were stable in lymphoid organs, more CD8+ Ki67+ proliferating lymphocytes were found in CpG-based treated mice. Several studies have described an expansion of CD8+ proportions following CpG (22, 39) or anti-PD-1 treatments (3). Additionally, the oil-in-water Lipiodol-CpG emulsion may act similarly to squalene-based oil-in-water emulsions, which significantly enhances the CpG-mediated augmentation of CD8+ T-cell responses, contrary to water-in-oil ones (40). Immunohistochemistry of treated tumors did not show any significant variation in granzyme-B+ CD8+ cells among all groups. However, only small areas of the tumors were sampled for staining and may, therefore, not reflect the real density of immune-infiltrating cells.

CpG-ODN, as TLR-9 agonists, are known for their ability to activate B cells and induce their proliferation, especially the B-class CpG (41, 42). Here, no difference in the global B cell proportion in all studied organs have been observed for all treated groups compared to the controls, in both flow cytometry and IHC. It may be explained by the fact that tissues were sampled 8 days following the first i.t. injection, and the delay may not be optimized to observe an effect of B cells. Also, only global B cells have been studied here, it would be interesting to assess subpopulations of B cells following CpG administrations.

IFN-γ, a proinflammatory cytokine released by Th1 T-cells and NK cells, coordinates innate and adaptative immune responses, promotes T-cell differentiation, and is required for CD8+ T-cells to acquire cytotoxic properties (43–45). According to the data reported here, the expression of IFN-γ increased in T-cells from the combination group, as well as its secretion in all treated animals, confirming the CpG-induced systemic T-cell response. The oil-in-water formulation of the emulsion may explain the increase in the CpG induced IFN-γ expression in the Lipiodol-CpG groups, as described by Zhang, et al. (46), who showed higher immune effects with an oil-in-water formulation than water-in-oil in the context of vaccine adjuvants. IFN-γ was reported to drive CpG-induced anemia (47, 48), which may also have occurred in the treated mice, as suggested by the significantly decreased red blood cell counts.

MDSC are immunosuppressive cells that suppress T-cells and their activity and produce anti-inflammatory cytokines (IL-10, TGF-β, etc.), enabling resistance to antitumor immune responses and immunotherapies (23, 49). Surprisingly, flow cytometry analysis revealed a systemic and local increase in MDSC proportions in all treated groups, despite a significant local antitumoral effect. At the same time, no overexpression of IL-10 was observed in treated mice by flow cytometry. These data suggest that an increased proportion of MDSC may not correlate with a more immunosuppressive environment and activity. This is consistent with previously published data demonstrating that TLR-9 activation increases the proportion of MDSC but inhibits their immunosuppressive role in tumor-bearing mice by inducing their maturation and differentiation (32, 50).

In contrast, Tregs were significantly less proliferative in the Lipiodol-CpG and combination groups. Tregs are widely known as immunosuppressive T-cells that inhibit the activation of cytotoxic T-cells and therefore promote tumor proliferation (51). Here, the Lipiodol-CpG emulsion decreased the proportion of proliferative (Ki67+) Tregs without modifying that of Tregs within lymphoid organs, contrary to previous data suggesting that oil-in-water CpG emulsions, more than CpG alone, decrease the proportions of Tregs (40). However, IHC of treated tumors demonstrated a decrease in Tregs in all treated groups, especially in CpG-treated mice. The decrease in the Tregs proportion was also observed, even if not significant, in the lymph node of CpG treated mice only in flow cytometry. This deleterious effect of CpG on Tregs has already been described in a murine model (52). The prognostic impact of Treg cell infiltration in CRC remains controversial. In some studies, high FOXP3+ Tregs infiltration was an unfavorable indicator of patient survival, while other reports reached the opposite conclusion (53, 54). The study reported here demonstrated that antitumoral effects induced by CpG-based treatments are associated with a local decrease in the Tregs population and a systemic decrease in proliferative CD4+ CD25^hi cells.

This study has some limitations. First, intra-tumoral injections do not reflect the complexity of an intra-arterial injection followed by vessel embolization as performed in TACE for humans. Also, subcutaneous tumors are not representative of orthotopic tumors, their microenvironment is different, and they do not simulate the spatial context which is as important as the cellular heterogeneity of the TME (55). The “metastatic” nodule was implanted and not spontaneously generated based on a background disease, therefore it does not reflect the gravity of an advanced disease. Moreover, the “metastatic” nodules were too small to perform both immunohistochemistry and flow cytometry. As it had no antitumoral property alone, no investigation of the immune cells populations has been made following an intra-tumoral injection of Lipiodol. No dose titration of CpG has been performed, but only two retreatments. It would have been interesting to modulate the injected dose of CpG, on one hand to optimize the regimen and dosage in this preclinical CT26 model, and on the other hand to evaluate the response of such cancer type to this immunomodulator.

Nevertheless, these data suggest that local treatment with the TLR-9 agonist CpG, emulsified with Lipiodol and combined with systemic anti-PD-1, induces a systemic antitumoral effect. The oil-in-water Lipiodol emulsion allows efficient loading and controlled release of CpG, which induces favorable immune infiltrate modulations especially when combined with anti-PD-1, even in a “metastatic model”. However, the increased immune infiltration of CD8+ cells in tumors, associated with an increase in IFN-γ T-cell expression and secretion, was not associated with a superior antitumoral effect of the emulsion compared to CpG alone in the metachronous metastatic model. Further studies are required to confirm the potential of this combination. Lipiodol, besides its radio-opacity properties, is of great interest in tumor tagging following intra-tumoral or intravascular injections, as it helps in retaining CpG into the tumor. Intra-arterial injection of Lipiodol-based emulsions, currently performed for liver tumor treatment, may lead to more promising results with the full exploitation of Lipiodol properties.

The raw data supporting the conclusions of this article will be made available by the authors, upon reasonable request.

The animal study was approved by Guerbet ethics committee, N°106. The study was conducted in accordance with the local legislation and institutional requirements. Studies were conducted following the recommendations of the European Convention for the protection of vertebrates Animals use for Experimentation and the local Ethic Committee on Animal care and Experimentation (APAFIS #21714).

A-LG: Conceptualization, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. NF: Conceptualization, Supervision, Validation, Writing – original draft, Writing – review & editing. MS: Formal analysis, Investigation, Methodology, Writing – review & editing. NB: Investigation, Writing – review & editing. VM: Investigation, Writing – review & editing. RS: Investigation, Writing – review & editing. OB: Investigation, Writing – review & editing. MW: Investigation, Writing – review & editing. CCo: Formal analysis, Writing – review & editing. PR: Conceptualization, Formal analysis, Writing – review & editing. JE: Conceptualization, Investigation, Supervision, Validation, Writing – original draft, Writing – review & editing. CCa: Conceptualization, Formal analysis, Investigation, Supervision, Writing – original draft, Writing – review & editing.