Intratumoral injection of the seasonal flu shot converts immunologically cold tumors to hot and serves as an immunotherapy for cancer

Significance Immunotherapy has revolutionized cancer treatment, yielding unprecedented long-term responses and survival. However, a significant proportion of patients remain refractory, which correlates with the absence of immune-infiltrated (“hot”) tumors. Here, we observed that FDA-approved unadjuvanted seasonal influenza vaccines administered via intratumoral injection not only provide protection against active influenza virus lung infection, but also reduce tumor growth by increasing antitumor CD8+ T cells and decreasing regulatory B cells within the tumor. Ultimately, intratumoral unadjuvanted seasonal influenza vaccine converts immunologically inactive “cold” tumors to “hot,” generates systemic responses, and sensitizes resistant tumors to checkpoint blockade. Repurposing the “flu shot” may increase response rates to immunotherapy, and based on its current FDA approval and safety profile, may be quickly translated for clinical care.

Abstract Reprogramming the tumor microenvironment to increase immune-mediated responses is currently of intense interest. Patients with immune-infiltrated “hot” tumors demonstrate higher treatment response rates and improved survival. However, only the minority of tumors are hot, and a limited proportion of patients benefit from immunotherapies. Innovative approaches that make tumors hot can have immediate impact particularly if they repurpose drugs with additional cancer-unrelated benefits. The seasonal influenza vaccine is recommended for all persons over 6 mo without prohibitive contraindications, including most cancer patients. Here, we report that unadjuvanted seasonal influenza vaccination via intratumoral, but not intramuscular, injection converts “cold” tumors to hot, generates systemic CD8+ T cell-mediated antitumor immunity, and sensitizes resistant tumors to checkpoint blockade. Importantly, intratumoral vaccination also provides protection against subsequent active influenza virus lung infection. Surprisingly, a squalene-based adjuvanted vaccine maintains intratumoral regulatory B cells and fails to improve antitumor responses, even while protecting against active influenza virus lung infection. Adjuvant removal, B cell depletion, or IL-10 blockade recovers its antitumor effectiveness. Our findings propose that antipathogen vaccines may be utilized for both infection prevention and repurposing as a cancer immunotherapy.

The tumor microenvironment represents a significant barrier that restricts immune responses against tumors and limits the efficacy of currently available immunotherapies as treatments for cancer. However, immune infiltration of tumors, especially by CD8+ T cells, has been shown to correlate with augmented responses to immunotherapy and improved survival (1⇓⇓⇓–5). An immunologically inflamed (“hot”) tumor microenvironment exhibits robust antigen presentation and T cell activation, contributing to the development of tumor-specific CD8+ T cell functionality that can acutely eliminate cancer cells, generate systemic tumor-specific immunity, and form long-term antitumor memory responses (5, 6). However, a significant proportion of patients harbor an immunologically “cold” tumor microenvironment that is either devoid of immune cell infiltration (an “immune desert”) or that is predominantly infiltrated by suppressive regulatory cell subtypes (including regulatory T cells [Tregs], regulatory B cells [Bregs], and myeloid-derived suppressor cells [MDSCs]) (7⇓–9). In both environments, cancer growth is immunologically unchecked and recruitment of inflammatory immune cells into such tumors is imperative for antitumor responses. Recently, cancer immunotherapy, including blockade of inhibitory immune checkpoints (such as PD-1/PD-L1 and CTLA-4), has emerged as an unprecedented breakthrough for the treatment of cancer that can induce long-term tumor regression (10⇓–12). However, responses to such therapies have been demonstrated to be effective only in select patients, particularly those who harbor a hot tumor microenvironment (13). Therefore, to increase response rates to immunotherapy, innovative solutions are needed to convert cold tumor microenvironments to hot by increasing infiltration of inflammatory immune cells that can serve as targets for immunotherapies in tumors devoid of immune infiltration and can overcome local immunosuppression in tumors infiltrated by regulatory cells.

One approach that could be utilized involves inducing a strong immune response, unrelated to the immune response against the cancer, within the tumor microenvironment that could then serve as a catalyst for a strong tumor-specific immune response. This concept employs a basic tenet of immunology, that responses against foreign antigens are strong and that responses against self-antigens are inherently weak. Toward avoiding autoimmunity, the immune system has developed many tolerance mechanisms by which strong responses to self-antigens are prevented or eliminated (14⇓–16). Because tumors develop from initially normal cells, many of the antigens of the tumor are self-antigens or antigens similar to self-antigens, and mounting an effective immune response against such antigens is a challenge. This undertaking is made even more difficult by the immunosuppressive nature of the tumor, which increases the immune-activation threshold necessary to be reached before tolerance is broken and potent responses to tumor antigens are mounted. However, when recognizing foreign components (like those associated with pathogens), the immune system is capable of developing strong responses even within the tumor microenvironment (17), and thus, components of pathogens (which can engage receptors associated with innate immunity) may be able to help break tolerance to tumor antigens and improve cancer outcomes.

Here, we show that indeed, pathogens and their components can augment an antitumor immune response within the tumor microenvironment, ultimately converting immunologically cold tumors to hot. This results in inflammatory responses at the injection site that reduce local tumor growth, in augmented systemic antitumor immunity that decreases metastases, and in sensitization of resistant tumors to immune checkpoint blockade. Importantly, we demonstrate that such outcomes can be achieved by intratumoral (i.t.), but not intramuscular, injection of FDA-approved unadjuvanted seasonal influenza vaccines (i.e., “flu shots”), and we elucidate immune mechanisms underlying our observations in the context of multiple mouse and human cancers.

Discussion Clinical successes utilizing immunotherapy to improve and prolong the lives of patients with cancer have demonstrated a vital role for the immune system in the treatment for cancer. However, thus far immunotherapies have been able to produce durable responses only in a limited proportion of patients. Therefore, to make the next great leap forward, innovative means of engaging the immune system are needed. In our described studies, we have focused on utilizing pathogens to augment inherently weaker antitumor immune responses to generate improved local and systemic cancer outcomes. Importantly, we observed that active influenza virus injection in the lung reduces tumor growth in the lung (even when a melanoma cell line was used that does not undergo productive infection by active influenza virus). However, active influenza virus injection in the skin did not reduce growth of that same melanoma cell line present in the skin. A difference between the lung and the skin is that the lung inherently contains natural cell targets for active influenza virus infection, while the skin does not. Active influenza A viruses bind to specific sialic acid residues on epithelial cells in the upper respiratory tract and subsequently gain entry into the cell and replicate, while cells in the skin lack the specific sialic acid residues necessary for productive influenza virus infection (42). Thus, in lung tissue, productive infection leads to a potent immune response to influenza virus by creating an immunologically inflamed hot microenvironment in the same tissue as the tumor. Without a major natural target for active influenza virus in the skin, the cells most likely to be affected are dendritic cells, which rather than boosting an immune response are dysregulated when active influenza virus is injected in the skin (28), further decreasing the ability of the tumor microenvironment to become immunologically hot. In recent years, viral infection has been harnessed as a vehicle to augment antitumor immune responses, and in particular, oncolytic virus (OV) therapy has been employed as a tool in the clinic. Oncolytic viruses preferentially lyse tumor cells and consequently release tumor antigens and danger-associated molecular patterns (DAMPs) (43). However, in the context of oncolytic viruses, productive infection of the tumor cells themselves is the focus. In this setting, the overexpression of specific proteins by tumor cells (but not normal adjacent cells) is hijacked by oncolytic viruses, which use these overexpressed proteins as entry receptors or to facilitate their own replication. Normal cells with less expression of these proteins do not serve as the major target and are spared, or they utilize IFN signaling (a pathway that is dysregulated in cancer cells) to limit infection. Thus, a major focus in this field has centered on the importance of direct infection of the tumor cell as a prerequisite for generating antitumor immunity. However, the idea that tumor cell lysis by the pathogen is essential has been recently challenged by evidence demonstrating that an inactivated oncolytic virus is capable of initiating antitumor immunity via the STING pathway and may support better immunity than its active oncolytic virus counterpart (44). Additionally, our data indicate that TLR activation via interaction with viral-derived PAMPs is increased in the context of inactivated virus, which may initiate an innate immune response and thereby remodel the tumor microenvironment. Further in support of this is previous research demonstrating that pathogen vaccines can substitute for synthetic TLR agonists to stimulate dendritic cells (45). Our data demonstrate that inactivation of a nononcolytic virus, such as influenza, can augment an antitumor immune response when administered via intratumoral injection, even when the corresponding virus (in active form) is incapable of such activity (as in our setting of active influenza virus administration within a skin melanoma). This indicates that the field of microbial-based cancer therapies (MBCTs), which has experienced a recent resurgence of interest (46, 47), is not limited to the oncolytic class of pathogens or even to the use of active pathogens. Furthermore, in terms of clinical translation, inactivated influenza virus injection can be made available to immunosuppressed patients who are not eligible for active pathogen-based therapies and to patients concerned about sequalae that may result from active pathogen administration. Studies have reported that pathogen-specific (e.g., cytomegalovirus [CMV], influenza virus, Epstein-Barr virus [EBV], etc.) CD8+ T cells infiltrate mouse and human tumors and comprise a significant fraction of intratumoral CD8+ T cells (48⇓⇓⇓–52). The impact of such antiviral immune responders on antitumor immunity demands further investigation and may have important implications for the use of MBCTs in the clinic. Interestingly, patients whose tumors harbor the tetrapeptide, ESSA, a sequence shared by CMV, have been shown to exhibit increased survival in the context of CTLA-4 blockade (53). Further, recently in mouse models, virus-specific memory T cells have been shown to halt tumor growth when their cognate antigens are injected within the tumor to create an immune-“alarming” effect (49). In contrast to this strategy, which requires previous immunity against a specific pathogen, our work suggests that pathogen-related therapies can be harnessed for antitumor immune responses independent of previous exposure; as in the majority of our studies, the hosts had no previous exposure to influenza virus. However, even in such cases, intratumoral administration of inactivated influenza virus increases dendritic cells and antitumor CD8+ T cells and consequently the reduction of tumor growth, without any prerequisite immunity. This may be particularly important for repurposing the seasonal flu shot for cancer immunotherapy and translating it to clinical care, as the seasonal influenza vaccine includes antigens that are altered yearly to match the anticipated predominant strains of the upcoming season. In this context, our lack of the need for previous exposure to the same pathogen and strain is a major advantage. However, it is also important to note that in our studies, previous infection followed by resolution of a particular strain of influenza virus did not prohibit subsequent tumor reduction with intratumoral inactivated influenza virus (i.e., a vaccine) made from the exact same strain. This suggests that patients with or without previous immunity to the influenza virus strain contained within the utilized flu shot may benefit from intratumoral administration of the vaccine. With multiple strains included within each trivalent and quadrivalent flu shot, it may be that an optimal response is achieved when a combination of new and previously experienced antigens is utilized. In this scenario, previously experienced antigens quickly raise inflammatory immune responses, which inherently are likewise quickly quenched with the elimination of the recognized antigen. At the same time, new antigens raise slower responses that are maintained longer and may have sustained positive effects on antitumor immunity. Our study proposes that intratumoral injection of an unadjuvanted seasonal influenza vaccine reduces tumor growth by converting immunologically inactive cold tumors to immune-infiltrated hot tumors, by augmenting DCs (including cross-presenting DCs) and tumor antigen-specific CD8+ T cells within the tumor microenvironment. These findings have important implications for the role of intratumoral seasonal influenza vaccination in priming patients to respond to existing immunotherapies (including, PD-1 and CTLA-4 blocking antibodies). Specifically, our study shows that intratumoral seasonal influenza vaccination 1) can reduce tumors on its own, 2) improves outcomes in the context of tumors that respond to PD-L1 therapy, 3) can reduce tumors even when they are resistant to PD-L1 blockade, and 4) in combination with PD-L1 blockade results in drastic reductions in tumor growth. This suggests that in patients, such vaccination may confer increased efficacy of immune-related therapies, including checkpoint blockade treatments that have dramatically improved survival for a segment of the cancer patient population, but which have not yet been made effective for all patients with cancer (54). Important attention must be paid to the formulation of the vaccine, as some adjuvants may provide improved antipathogen protection, while limiting the ability of the vaccine to improve antitumor outcomes. The adjuvanted seasonal influenza vaccine utilized in our studies has been demonstrated to provide enhanced immunity against active influenza lung infection in persons 65 y of age and older, whose immunity may decrease with increasing age (33, 34). However, this adjuvanted formulation does not augment an antitumor immune response when administered via intratumoral injection, but instead maintains immunosuppressive regulatory B cells within the tumor. Adjuvants play an important role in boosting immune responses. Likely within our unadjuvanted seasonal influenza vaccines natural adjuvants (e.g., host cell proteins and DNA, residual influenza ssRNA, etc.) resulting from the process of manufacturing the inactivated seasonal influenza vaccine likewise improve immunity by interacting with multiple danger-sensing mechanisms (e.g., an RNA-sensing toll-like receptor), an interaction that has been previously shown to improve antitumor immune responses (55). Since the majority of seasonal influenza vaccines currently on the market are inactivated and do not contain manufactured adjuvants (e.g., squalene), and such vaccines have a high safety profile and are FDA approved, the translatability of these as innovative immunotherapies for cancer is high and the barriers to reaching many patients with cancer is low. Although active influenza virus lung infection is a major public health concern, with tens of thousands of deaths documented annually in the United States (56), the Centers for Disease Control and Prevention (CDC) has reported that in the 2017–18 season, only 37.1% of adults received the seasonal influenza vaccine (57). Anecdotally, this percent may be even lower among patients with cancer, in whom infections have also been reported to have greater morbidity and mortality. Importantly, beyond demonstrating that influenza vaccination administered via intratumoral injection can reduce tumor growth, our studies provide evidence that protection against future active influenza lung infection can be provided via intratumoral administration. This suggests that patients receiving intratumoral seasonal influenza vaccination may experience multiple clinical benefits and that seasonal influenza vaccination is a crucial public health tool that may be utilized as both a preventive measure against infection and an immunotherapy for cancer.

Materials and Methods Mice. Mice were housed in specific-pathogen-free facilities and all experiments were conducted in accordance with procedures approved by the Institutional Animal Care and Use Committee (IACUC) and Institutional Biosafety Committee at Rutgers, The State University of New Jersey and Rush University Medical Center, and the Institutional Review Board at Rutgers, The State University of New Jersey. B6 (C57BL/6J), Batf3−/− (B6.129S(C)-Batf3tm1Kmm/J), NSG (NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ; NOD SCID gamma), and BALB/c mice were purchased from The Jackson Laboratory at 6 to 10 wk of age. Active Influenza and Heat-Inactivated Influenza Virus. For experiments utilizing active influenza virus infections, mice were administered 1 × 106 plaque-forming units (pfu) of A/PR8/1934/H1N1 (FLU) (58) or OVA 257–264 SIINFEKL-expressing A/PR8/1934/H1N1 (FLU-OVA) (59) by passive i.n. or i.t. (i.e., at the tumor site) administration (25 to 50 μL). Control mice were administered an equal volume of phosphate-buffered saline (PBS) via the same route. For experiments utilizing heat-inactivated influenza virus (hiFLU or hiFLU-OVA), the virus was inactivated by incubating active A/PR8/1934/H1N1 FLU at 90 °C for 5 min on an IncuBlock Plus heat block (Denville Scientific) prior to injection into mice. For experiments utilizing influenza virus lysate (FLU lysate), active A/PR8/1934/H1N1 FLU was resuspended in RLT buffer (Qiagen) for 1 h to generate a lysate. RLT buffer was then dialyzed using a Slide-A-Lyzer G2 Dialysis Cassette (10-kDa molecular weight cutoff; Thermo Fisher) prior to lysate administration. Vaccines and Adjuvants. FDA-approved 2017–2018 seasonal influenza vaccines were purchased from their respective manufacturers: FLUCELVAX (FluVx1; Seqirus), FLUVIRIN (FluVx2; Seqirus), FLUARIX QUADRIVALENT (FluVx3; GlaxoSmithKline), FLUBLOK (FluVx4; Protein Sciences Corporation), and FLUAD (AdjFluVx; Seqirus). Vaccine details are provided in SI Appendix, Table S1. To mimic adjuvant MF59 (Novartis), AddaVax (Adj; Invivogen) was administered via intratumoral injection (50 μL). Control mice were administered PBS at the same volume via the same route. In experiments in which the adjuvant and vaccine were jointly delivered, 50 µL adjuvant + 50 µL vaccine were mixed and delivered in a total volume of 100 µL via intratumoral injection. In some experiments, MF59, which is primarily composed of squalene, was removed by centrifugal filtration using Amicon Ultra centrifugal filter units with regenerated cellulose filters (with a 30-kDa molecular weight cutoff). MF59-containing AdjFluVx (500 μL) was added to the unit and washed with acetone (250 μL; 3 times) followed by PBS (250 μL; 3 times). The protein component of the vaccine was collected using a pipette, freeze dried, and reconstituted to the original volume using PBS. Tumor Challenge. For tumor challenge experiments, B6 and NSG mice were anesthetized with isoflurane and administered 100,000 to 150,000 B16-F10 melanoma cells (American Type Culture Collection [ATCC]) via i.v. or intradermal (i.d.) injection and BALB/c mice were anesthetized with isoflurane and administered 100,000 to 150,000 4T1 triple-negative breast cancer cells (ATCC) in the mammary fat pad. B16-F10 and 4T1 cancer cell lines were cultured in Dulbecco’s Modified Eagle Medium (Gibco), 10% fatal bovine serum (Sigma-Aldrich), 100 units/mL penicillin (Gibco), 100 mg/mL streptomycin (Gibco), and 0.29 mg/mL glutamine (Gibco) prior to harvesting for tumor injection. Primary tumor growth was monitored by Vernier caliper measurements in 2 perpendicular directions serially after tumor challenge. Mice harboring tumors were killed when the tumor area reached 20 mm in any direction or met other health-related endpoints, as per institutional IACUC policies. To quantify 4T1 lung metastases, 5% India ink (Fisher Scientific) diluted in distilled water was injected into the trachea after euthanasia (60). Lungs were dissected and transferred to Fekete’s solution (40 mL glacial acetic acid, 32 mL [37%] formalin, 700 mL 100% ethanol, and 228 mL double-distilled water) and washed 3 to 4 times in this solution and once in PBS. 4T1 lung surface metastases (white in appearance) and B16-F10 lung surface foci (black in appearance) were manually counted with the use of a magnifying glass. Statistical Analyses. Two-tailed Student t test (for 2 groups) or 1-way ANOVA with Tukey correction (for more than 2 groups) was used to determine statistical significance for data comparisons at a single timepoint. Two-way ANOVA or mixed-effects model with Bonferroni (for 2 groups) or Tukey (for more than 2 groups) correction was used to determine statistical significance for data comparisons with multiple timepoints. Kruskal–Wallis with Dunn’s multiple comparisons test was performed for focused comparisons of 1 group to all other groups at a single timepoint. Mantel–Cox log rank test was performed to determine statistical significance for the comparison of survival curves. Prism version 8.0 (GraphPad) was used for generation of all graphs and performance of statistical and Extreme Studentized Deviate analyses, except for Fig. 1 C and D, where STATA version 15.0 (StataCorp, LLC) was used to perform statistical analyses. Statistical significance shown for survival curves represents a comparison of the 2 survival curves. Statistical significance shown for all other graphs represents comparisons at the indicated timepoint. Statistical significance is denoted as ns, not significant, *P < 0.05, **P < 0.01, and ***P < 0.001. Comparisons with significance at P < 0.001 or P < 0.0001 are listed as ***P < 0.001. Data Availability. Sequencing data are included in the supplementary materials for this manuscript: Datasets S1–S3.

Acknowledgments This research was supported by the Biospecimen Repository and Histopathology Service Shared Resource and the Immune Monitoring Shared Resource of Rutgers Cancer Institute of New Jersey (funded in part by NIH P30CA072720). This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the National Cancer Institute; the Office of Research, Development and Information, Centers for Medicare & Medicaid Services; Information Management Services, Inc.; and the SEER Program tumor registries in the creation of the SEER-Medicare database. C.B.C. and J.H.N. were supported in part by a fellowship from the New Jersey Commission on Cancer. J.H.N. also received support from the NIH-Rutgers Biotechnology Training Grant (T32GM008339). N.L.H. is a trainee of the National Institute of General Medical Sciences/NIH through the Rutgers University Pipeline-Initiative for Maximizing Student Development Program (R25GM055145). A.Z., A.L., and A.W.S. are supported in part by an NIH grant (R01CA225993). Support provided by the Robert Wood Johnson Foundation (grant 74260) facilitated work conducted at the Child Health Institute of New Jersey. Illustrations were created with Biorender.com. We thank the reviewers of this manuscript for their in-depth review and constructive comments that have substantially improved the final product.

Footnotes Author contributions: J.H.N., C.B.C., N.L.H., P.K.B., K.H.G., F.J.K., H.L.K., P.G.T., V.G., T.M.K., J.R., E.A.S., J.M., L.K.D., D.B.S., A.B.R., L.Y.L., A.L., J.M.S., M.J.F., E.S.R., A.L.M., J.S.R., A.W.S., and A.Z. designed research; J.H.N., C.B.C., N.L.H., P.K.B., S.M.A., R.P., R.E., M.M.A., S.B., S.T., M.L., S.L., D.J.M., E.F.G., K.H.G., G.G.-A., M.R., C.N., A.A., J.W.G., J.R.B., R.E.R., D.R., S.R.J., S.-J.W., F.J.K., J.M., and J.S.R. performed research; P.G.T., V.G., J.P., M.P.K., E.A.S., J.L., A.L.M., and J.S.R. contributed new reagents/analytic tools; J.H.N., C.B.C., N.L.H., P.K.B., S.M.A., R.P., R.E., M.M.A., S.B., S.T., M.L., S.L., D.J.M., E.F.G., K.H.G., D.R., S.R.J., F.J.K., J.M., J.M.S., and A.Z. analyzed data; and J.H.N. and A.Z. wrote the paper.

Competing interest statement: P.K.B. and H.L.K. are employees of Replimune, Inc. E.A.S. receives research funding from Astellas/Medivation. M.J.F receives research funding from Pfizer and Biodesix, consulting funds from Astrazeneca and Pfizer, and honoraria from AstraZeneca, Merck, and Genentech. All other authors declare no conflicts of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1904022116/-/DCSupplemental.

This open access article is distributed under Creative Commons Attribution License 4.0 (CC BY).