Local failures following radiation therapy are multifactorial, and the contributions of the tumor and the host are complex. Current models of tumor equilibrium suggest that a balance exists between cell birth and cell death due to insufficient angiogenesis, immune effects, or intrinsic cellular factors. We investigated whether host immune responses contribute to radiation-induced tumor equilibrium in animal models. We report an essential role for immune cells and their cytokines in suppressing tumor cell regrowth in two experimental animal model systems. Depletion of T cells or neutralization of IFN-γ reversed radiation-induced equilibrium, leading to tumor regrowth. We also demonstrate that PD-L1 blockade augments T cell responses, leading to rejection of tumors in radiation-induced equilibrium. We identify an active interplay between tumor cells and immune cells that occurs in radiation-induced tumor equilibrium and suggest a potential role for disruption of the PD-L1/PD-1 axis in increasing local tumor control.
Clinically stable disease has several manifestations that are broadly characterized by the failure of tumors to progress; instead, they remain static over variable periods of time (1, 2). Mechanistically, stable disease could result from the induction of cellular dormancy, the failure of tumor cells to proliferate and instead adopt a temporary state of quiescence, or an equilibrium process of balanced cell proliferation and cell death. Tumors that receive radiation therapy (RT) sometimes exhibit a progression-free period of stable disease with subsequent late relapse. However, the mechanisms that contribute to both the induction and maintenance of a stable disease state, as well as late failures following RT, are poorly understood. Evidence from various clinical trials across multiple tumor types indicates that local failures often occur after several years. For example, analysis of randomized trials consisting of 11,000 women with early-stage breast cancer treated with surgery and RT demonstrated that >50% of the local failures occurred after 5 y, with 25% occurring after 10 y (3). Similarly, the long-term results of a phase III Radiation Therapy Oncology Group (85–31) trial in prostate cancer comparing RT with or without androgen deprivation demonstrated that >50% of the local failures occurred 5 y after treatment (4). Other disease types also demonstrate late failures following RT, including uveal melanoma at 6 y (15%) (5) and head and neck cancer at 2 y post-RT (13%) (6). One prevailing view is that the duration of progression-free/stable disease is determined by the number of surviving clonogens and tumor repopulation (e.g., fewer surviving clonogens take longer to repopulate). Therefore, the extent of initial tumor cell or stromal killing by radiation is proposed, in part, to mediate the time to relapse. However, an alternative explanation is the induction of tumor equilibrium. Equilibrium is a state in which tumor proliferation is balanced by cell death, and both angiogenic and immunological mechanisms were demonstrated to mediate tumor equilibrium (7, 8). Therefore, the induction of equilibrium could represent a temporarily stable, yet transitional, disease state during which tumor progression is halted, but tumors eventually escape to relapse locally at variable intervals.
Immunologically based equilibrium was reported in a carcinogen-induced animal tumor model in which the maintenance of small stable tumors was associated with the establishment of an equilibrium state (9). In this article, we report that RT induces prolonged stable or partial clinical responses that are characterized by the coincidence of actively proliferating tumor cells and pronounced immune infiltration, resulting in radiation-induced tumor equilibrium (RITE). Using animal model systems, we identify one possible mechanism for RITE as an active balance of two dynamic processes, tumor cell proliferation and apoptosis, that is mediated by local immune cells. These results suggest that immunotherapy may strengthen limited immunity and tip the balance toward improved eradication of residual cancer cells.
Materials and Methods
Human subjects and clinical trials
The oligometastasis trial was approved by the University of Chicago Institutional Review Board (Approval #13619B), and Institutional Review Board–approved written informed consent was obtained from each patient before protocol therapy. Patients eligible for this trial were described previously (10, 11). Declaration of Helsinki protocols were followed.
Mice and cell lines
C57BL6 and BALB/c mice were purchased from Harlan at 6–7 wk of age. All mice were maintained under specific pathogen–free conditions and used in accordance with the animal experimental guidelines set by the Institute’s Animal Care and Use Committee. This study was approved by the Institutional Animal Care and Use Committee of the University of Chicago. B16-SIY melanoma cells were obtained from Tom Gajewski (University of Chicago) and maintained by R.R.W. and Y.-.X.F. (12). TUBO was cloned from a spontaneous mammary tumor in BALB neu-transgenic (Tg) mice (13). 3T3KB and 3T3NKB were gifted from Dr. Wei-Zen Wei (Wayne State University, Detroit, MI) (13). 3T3KB mouse fibroblasts express H-2Kd and B7.1, and 3T3NKB cells express additional Her-2 or neu.
Tumor growth and treatments
Single–tumor cell suspension was harvested from cultured cells. A total of 5 × 105 cells was injected as described (13). Tumor volumes were measured along three orthogonal axes (a, b, and c) and calculated as tumor volume = abc/2. Tumors were allowed to grow for ∼2 wk (50–150 mm3 in size) before treated with local RT (12). In general, the more radiosensitive TUBO tumors received 15 Gy, and the more radioresistant B16SIY tumors received two doses of 25 Gy. Tumor volumes were measured every 3–4 d. To select stable-equilibrium tumors, we investigated tumors between 50 and 150 mm3 in all of the depletion experiments. For Ab-mediated cell depletions, 200 μg/mouse anti-CD4 (clone GK1.5) and/or anti-CD8 (clone 2.43; Fitch Monoclonal Antibody Facility, the University of Chicago) were delivered three times by i.p. injection at 5-d intervals. For PD-L1–blockade experiments, 200 μg anti–PD-L1 (clone 10F.9G2, Bio X Cell) was administered i.p. every 3 d for a total of four times to mice bearing stable tumors (defined at 3 wk after RT). For IFN-γ–blockade experiments, mice bearing stable tumors were injected i.p. with 500 μg anti–IFN-γ (clone XMG1.2; Bio X Cell) or isotype control every 5 d for a total of three injections.
Frozen mouse tumor tissue was treated with acetone/methanol for 20 min prior to CD8 (clone 53-6.7; BioLegend) staining. Sections were treated with picric acid formalin fixative for 10 min for Ki-67 staining (1:75, clone TEC-3; DAKO). After incubation with biotinylated anti-rat IgG (Vector Laboratories), sections were labeled using the Elite kit (Vector Laboratories) and DAB (DAKO) system. Tissue sections were counterstained and mounted. The slides were scanned at 20× magnification using Aperio ScanScope XT and viewed using Aperio ImageScope. TUBO tumor slides were annotated and analyzed using Aperio ImageScope. Ten random high-power fields were selected for CD8+ and Ki67+ staining. TUNEL signal was analyzed in the same selected area. All staining was performed on sequential slides from three individual mouse tumors.
Ex vivo radiosensitivity assay
Fourteen days after tumor cell implantation, TUBO tumors, which measured 50–200 mm3 in size, were subjected to 15 Gy radiation. Tumors were measured every 3 d, and their responses to RT were characterized as nonresponsive (NR) or responsive (R) after day 6 post-RT. Eight days post-RT, three non-RT control, NR, and R tumors were removed and digested with collagenase into single-cell suspensions. Plated cells from each tumor were allowed to grow for 2 d and were then subjected to 0-, 5, or 10-Gy radiation. Colonies were stained with crystal violet, and those > 50 cells were counted 10 d after plating. The relative surviving fractions after 5 or 10 Gy were calculated as fold change over colony numbers from the 0-Gy control of the same tumor sample. In the same study, stable (S) and late escapers/partial-responding (PR) tumors were excised and subjected to the same radiosensitivity assay, as described above.
Cell sorting and quantitative RT-PCR
Tumors were disassociated, stained, and sorted as described (14). Flow-sorted cells were homogenized in 800 μl TRIzol LS. RNA isolation was performed following the manufacturer’s instructions, with the addition of glycogen. cDNA was synthesized using 10 μl RNA and a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). A mock RT control reaction was also performed using 10 μl RNA. Quantitative PCR was performed on an ABI 7900HT using Power SYBR Green PCR Master Mix (both from Applied Biosystems). Gene fold-change calculations were performed using the comparative Ct method; GAPDH was used as the endogenous gene control. The following primer sequences were used: IFNG: forward, 5′-TGAGCTCATTGAATGCTTGG-3′ and reverse, 5′-ACAGCAAGGCGAAAAAGGAT-3′; STAT1: forward, 5′-AGTCGGAGGCCCTAATGCT-3′ and reverse, 5′-CCATAATGCACCCATCATTCCA-3′; GAPDH: forward, 5′-ACGACCCCTTCATTGAC-3′ and reverse 5′-TCCACGACATACTCAGCAC-3′; IFIT3: forward, 5′-AGTGAGGTCAACCGGGAATCT-3′ and reverse 5′-TCTAGGTGCTTTATGTAGGCCA-3′; and CD274(PDL1): forward, 5′-GCCTGCTGTCACTTGCTACGGG-3′ and reverse 5′-CAGCGAGGCTCTCCCCCTGA-3′.
ELISPOT and CBA assay
Tumor-draining lymph nodes (DLNs) and tumors were removed and prepared into single-cell suspensions, as described (12). Tumor-infiltrating lymphocytes (TILs) were purified using Ficoll-gradient centrifugation (GE). Ninety-six–well HTS-IP plates (Millipore) were precoated with 5 μg/ml anti-IFN–γ Ab (clone R4-6A2; BD Pharmingen) overnight at 4°C. A total of 1–3 × 105 lymph node cells or TILs (harvested at day 40 after RT) was cocultured with 3T3NKB cells at a ratio of 10:1 in the presence (for TILs) or absence of 25 IU/ml IL-2. 3T3KB cells were used as a control cell line for Ag specificity. After 3 d of incubation, cells were removed, 4 μg/ml biotinylated anti–IFN-γ Ab (clone XMG1.2; BD Pharmingen) was added, and the plate was incubated for 2 h at 37°C. Then 0.9 μg/ml avidin-HRP (BD Pharmingen) was added, and the plate was incubated for 45 min at 37°C. The cytokine spots were developed according to product protocol (Millipore). Supernatants from the flat-bottom 96-well plate, where TILs were stimulated with 3T3NKB or 3T3KB cells, as described above, were subjected to BD CBA assay to detect cytokines, according to the instruction manual.
To obtain single-cell suspensions, tumor tissues were digested with 1 mg/ml Collagenase IV and 0.2 mg/ml DNase I (both from Sigma) for 45 min at 37°C. Cells were stained with Abs specific for CD8, CD4, and CD45 (BioLegend). Samples were analyzed on a FACSCalibur Flow Cytometer (BD), and data were analyzed with FlowJo software (TreeStar).
Data were analyzed using Prism 5.0 software (GraphPad). Experiments were repeated two or three times. The p values were assessed using two-tailed unpaired Student t tests.
Stable disease is observed in patients with oligometastasis after stereotactic body RT (ablative radiotherapy)
Results of a phase I trial demonstrated that patients with metastatic cancer in a limited number of organs (oligometatasis; one to five sites) benefited from local high-dose/fraction radiotherapy or stereotactic body RT (SBRT) (10, 11). Among 63 patients with fewer than three oligometastatic tumors that received SBRT (36–48 Gy) (11), 87% exhibited partial and complete responses, resulting in residual tumor nodules that remained progression-free for months or years (range: 4.5–31.4 mo; mean, 20.9 mo, Supplemental Table I). Thirty-one percent of these lesions relapsed in the irradiated site. Of patients with one or two oligometastatic lesions who received SBRT (11), ∼40% had radiographic freedom from progression for relatively long intervals (Supplemental Fig. 1A). The relapses observed are consistent with other reports of late failure, suggesting that SBRT induces stable disease in some oligometastatic tumors in a manner similar to primary tumors. Stable disease or prolonged response with relapse was noted in some imaging studies (10, 11). As an example, Supplemental Fig. 1B shows a representative patient with metastatic lung cancer who was treated with SBRT and underwent a complete response (computerized tomography scan negative). Five years later the tumor recurred locally, rapidly grew to invade the chest wall, and was confirmed by biopsy (Supplemental Fig. 1B). The results from our SBRT trial suggested that ablative RT can induce either a transient or prolonged state of stable disease or cure. Therefore, it is important to study the mechanisms contributing to an arrested state of tumor progression to develop new treatment strategies to prevent relapse and mediate complete cure.
RT-induced stable disease in murine models
To further our understanding of the mechanisms that contribute to the induction of a stable disease state that precedes late failure, we developed two mouse models. First, TUBO cells derived from a spontaneous breast tumor of BALB-neuT–Tg mice were implanted s.c. in wild type (WT) BALB/c mice. TUBO tumors were allowed to establish for 14 d (tumor volume ranged from ∼50 to 200 mm3) before receiving 15 Gy, a dose that was determined to result in stable disease in a high fraction of the mice without inducing tissue necrosis. Of the 193 tumors treated across eight independent experiments (Supplemental Table II), 45% (88/193) of TUBO tumors remained stable and palpable over a 34–60-d period of observation and were classified as stable. The response to RT (15 Gy) could be observed as early as 5–7 d posttreatment, and tumors could be subdivided into R and NR types. Of the tumors that demonstrated an early response to RT, 18% (34/193) completely regressed, 14% (27/193) relapsed over the course of 3 wk (PR), and 45% (88/193) remained stable (S) for >3 wk. The remaining tumors (23%; 44/193) were not controlled by RT treatment (NR).
These different outcomes are summarized in Supplemental Table II and are represented graphically in Fig. 1A and 1B. The spectrum of responses observed in our TUBO mouse model are largely consistent with the range of clinical responses observed after treatment of tumors with RT. Stable TUBO tumors monitored over the course of 50–80 d post-RT can relapse at later time points or remain stable and palpable (Supplemental Fig. 2A, 2B). Induction of stable disease was dependent, in part, on the size of the tumor at the time of local RT delivery and the radiation dose applied. We observed a similar phenotype in larger tumors with average sizes > 200 mm3; however, a higher dose of 20 Gy was needed to induce equilibrium (Fig. 1B). We performed similar experiments with B16-SIY melanoma on the C57/BL6 background, as described previously (12, 14). B16 is radiation resistant in vitro, and a higher dose of RT is required to induce stable disease in vivo. Two doses of 25 Gy were given 10 and 12 d after cell implantation (Fig. 1C). Sixty-two percent (18/29) of the tumors remained stable after RT for ≥30 d, demonstrating that these observations are not unique to a single tumor model or mouse strain. However, the dose required to achieve a high rate of stable disease is likely related to several factors, including the radiosensitivity of the tumor cell lines.
Inherent cellular radiosensitivity is frequently hypothesized to account for the differences observed in the rate of tumor regrowth following treatment of tumors with RT. To address the potential contribution of differential radiosensitivity to the variability in the responses of TUBO tumors to RT, we determined the radiosensitivity of tumor cells taken from mice that were responsive or nonresponsive to RT. We surgically removed TUBO tumors 8 d post-RT and plated tumor cells, which were then assayed for clonogenic survival following delivery of a second dose of RT. The results indicated that at 8 d post-RT, tumor cells derived from nonresponsive tumors in vivo have the same in vitro cellular radiosensitivity as the cells from responsive tumors and control tumor cells from nonirradiated mice (Fig. 2A). These results, coupled with the fact that the parental TUBO cells have the same in vitro radiosensitivity, suggest that differences in radiosensitivity do not account for the disparate regrowth kinetics in vivo following treatment with local RT. Tumor cells derived from stable tumors and tumors that exhibited a partial response followed by early relapse at day 21 post-RT were subjected to the same assay. Unexpectedly, the tumor cells derived from stable tumors exhibited equivalent radiosensitivity to cells from tumors that were only partially responsive (Fig. 2B). These results demonstrate that, in our model, the degree of radiosensitivity of tumor cells cannot account for the differential response of tumors to local RT. The results point toward host factors in the maintenance of radiation-induced stable disease.
To further characterize our model system, we stained stable TUBO tumor tissues with Ki-67 and TUNEL. There were clusters of Ki-67+ tumor cells adjacent to large areas of TUNEL+ tumor cells, suggesting that the tumor cells were both proliferating and dying in the stable tumor nodule and raising the possibility that the stable disease state is achieved through equilibrated rates of division and death (Fig. 2C). Furthermore, we noted a prominent T cell infiltration in stable tumors that was dominated by CD8+ T cells. We quantified the number of TUNEL+ cells in randomly selected CD8+ and Ki67+ regions (Fig. 2D). The results show that the TUNEL-staining signal is significantly higher in CD8+ areas than in Ki67+ areas (p = 0.001), suggesting that CD8+ cells could be responsible for the majority of apoptosis in the stable TUBO tumors. Taken together, these data indicate that local ablative RT can induce a stable disease state in two different animal tumor models and that stable disease may result from an equilibrium, which we refer to as RITE, between tumor cell proliferation and tumor cell death. Furthermore, the data also raise the possibility that host immune responses might be involved in the generation of the equilibrium and resulting stable disease state.
CD8+ T cells and IFN-γ are essential to maintain RT-induced equilibrium
To investigate whether host T cells are required in the induction of stable disease, we depleted CD8+ and CD4+ T cells in mice bearing RITE B16-SIY tumors (Fig. 3A). Tumors in the CD4 and CD8 combined-depletion groups showed rapid tumor outgrowth compared with tumors in nondepleted hosts (p = 0.016, V/V0. For changes in absolute volume, see Supplemental Fig. 2D). We observed similar effects when CD8+ T cells alone were depleted from mice bearing RITE TUBO tumors (Fig. 3B, p = 0.015, V/V0. For changes in absolute volume, see Supplemental Fig. 2E). When examining tumors in which CD8+ cells had been depleted, the fraction of Ki-67+ cells was increased, whereas TUNEL+ staining was reduced compared with the same staining in nondepleted tumors (data not shown). These data demonstrate that CD8+ T cells play an essential role in maintaining stable disease, in part through an immunological equilibrium. Consistent with the Ab-depletion experiments, TUBO tumors grown in SCID mice, which lack mature B and T cells, could not be induced into equilibrium by radiation (Fig. 3C).
TUBO cells grow more aggressively in BALB-neuT Tg mice, from which they were derived, compared with WT BALB/c mice (1334 ± 299 mm3 versus 399 ± 183 mm3 day 33 posttumor implanting, p = 0.04). The difference in tumor-growth kinetics is thought to be due to T cell tolerance against tumor-associated Ags, including neu, which is a self- or foreign Ag in BALB-neuT Tg mice and BALB/c mice, respectively. To determine whether a heightened state of T cell tolerance in BALB-neuT Tg mice could affect the induction of radiation-induced equilibrium, we implanted TUBO cells s.c. on the flank of BALB-neuT mice. In 15 d, tumors received 30 Gy locally when the tumor size reached ∼100 mm3. Most tumors remained stable for up to 21 d post-RT, before slowly resuming growth (Fig. 3D). These results demonstrate that radiation-induced equilibrium can be affected by the degree of host T cell tolerance and further implicate T cells as central mediators in RITE.
To begin to understand the role of host immunity in RITE, we investigated multiple mRNA transcripts of immunological significance in sorted populations of infiltrating CD45+ hematopoietic cells and CD45− tumor cells in stable tumors. CD45− TUBO tumor cells derived from equilibrium tumors expressed high levels of IFN-stimulated genes, as determined by quantitative PCR, suggesting an association between increased IFN signaling within the tumor microenvironment and immune-mediated suppression of tumor regrowth (Supplemental Fig. 2C). Mice bearing RITE tumor nodules were treated with anti–IFN-γ–neutralizing Ab and showed rapid tumor outgrowth (Fig. 4A, p = 0.026. For changes in absolute volume, see Supplemental Fig. 2F). Our results are consistent with an earlier report demonstrating the essential role of IFN-γ in T cell recruitment and control of primary tumor following local RT (15). These data suggest that RT treatment induces changes in the tumor microenvironment that facilitate activation of T cells and production of effector cytokines, such as IFN-γ. To examine Ag-specific T cell responses, tumor Ag–specific IFN-γ ELISPOT assay was performed on the DLNs from untreated control mice, mice with NR tumors or R tumors at 1 wk following RT, and R tumors that remained stable for 5 wk post-RT. To determine neu-specific responses, 3T3KB and 3T3NKB cells were used as control and neu-specific APCs, respectively. ELISPOT assay showed an elevated frequency of neu-specific cytokine-producing effector T cells in mice harboring R tumors compared with untreated (Non-RT) control tumors or NR tumors (p = 0.0013, Fig. 4B). The results demonstrate that the differential response of tumors to RT in individual mice is correlated with the magnitude of the T cell response to tumor Ags.
Although Ag-specific T cell responses were increased significantly in the R group (1 wk post-RT) compared with the non-RT control group (p = 0.0013), T cell responses were significantly lower in animals with stable disease (5 wk post-RT) compared with the animals sacrificed shortly after local radiation (p = 0.012, Fig. 4B). The fact that T cell responses declined over a period of 30 d post-RT suggests that ongoing T cell priming wanes over time, likely as a result of decreased Ag and danger signal release. Compared with the nonirradiated control group, the T cell response from animals with stable disease remained significantly increased in the DLNs (p = 0.02, Fig. 4B), indicating that the relatively modest T cell response is sufficient to mediate equilibrium but not enough to eliminate the tumor completely. In mice bearing stable tumors, we also observed higher levels of TNF-α (Supplemental Fig. 2G). In addition, the frequency of CD8+ T cells was significantly elevated in responsive tumors compared with nontreated controls but similarly became reduced over time (Fig. 4C). To test whether treated mice that had complete tumor regression maintain a prolonged protection against rechallenge, a critical indication for the control of relapse, mice were rechallenged with the same TUBO tumor cells 30 d after the original tumors were not palpable (∼60–70 d post-RT). Rechallenged mice failed to grow tumors, whereas naive control mice exhibited aggressive tumor growth (Fig. 4D). Taken together, our data collectively suggest that ablative RT increases the frequency of tumor Ag-specific CD8+ T cells that infiltrate the tumor, but both overall T cell infiltration and cytokine production decrease over time in the tumor microenvironment, eventually producing RITE.
Immunotherapy can tip the balance to eradicate stable tumors
We hypothesized that the decline in systemic and local T cell function observed over the course of achieving tumor equilibrium might be a result of gradual T cell exhaustion. T cell exhaustion is a state of T cell dysfunction induced by prolonged exposure to antigenic stimulation and is defined by poor effector function, sustained expression of inhibitory receptors, and a transcriptional state distinct from that of functional effector or memory T cells (16). Therefore, we hypothesized that the equilibrium achieved between proliferating tumor cells and T cell–mediated tumor cell death can be tipped to induce complete tumor rejection by either enhancing T cell activation or reducing T cell inhibition to recover the functional capacity of exhausted T cells in the context of tumor equilibrium. Recent studies suggest that PD-L1 expression in human tumor tissues can exhaust T cells (17, 18). We observed an increase in PD-L1 expression in tumor cells from RITE tumors (Supplemental Fig. 2C), concurrent with an increase in systemic effector T cell function in the DLNs (Fig. 4B). Therefore, we hypothesized that blocking PD-L1 using a neutralizing Ab would release suppression of PD-1+ T cells within the tumor microenvironment and result in tumor rejection. Indeed, most RITE tumors were rejected after Ab treatment (Fig. 5A, 5B, p = 0.018). DLNs were harvested from untreated mice, mice with stable tumors, and mice with stable tumors that were treated with anti–PD-L1 Ab 1 wk after Ab treatment (5 wk post-RT) and subjected to IFN-γ ELISPOT assay (Fig. 5C). Ag-specific cytokine release was dramatically increased in mice receiving anti–PD-L1 treatment compared with untreated mice with stable tumors (p = 0.002). These data demonstrate that blockade of a T cell–negative regulator can increase T cell function to facilitate tumor rejection. These results suggest that tumor equilibrium induced by RT can be disrupted to yield complete regression using immunotherapy.
As noted above, previous clinical studies demonstrated late radiation failures. One hypothesis is that robust cytoreduction by effective ionizing radiation induces prolonged regrowth kinetics compared with less cytoreduced states resulting from lower doses or more radioresistant tumor cells. In this article, we suggest an alternative explanation for the stable, or disease-free, intervals that precede late radiation failures as equilibrium states (RITE) generated by a balance between tumor cell proliferation and T cell–mediated killing. Tumors treated with radiotherapy were observed to contain actively proliferating tumor cells and pronounced T cell infiltration. Using animal models of RT-induced tumor equilibrium, we confirmed the essential role of CD8+ T cells in achieving and maintaining progression-free disease, because CD8+ T cell depletion led to rapid tumor growth. Conversely, enhancing T cell function through blockade of inhibitory receptors led to eradication of the remaining tumor. Based on our data, we propose that RITE is dynamically balanced by two opposing mechanisms: aggressive tumor proliferation and CD8+ T cell–mediated tumor cell apoptosis.
In the TUBO tumor model, intrinsic radiosensitivity of cells from individual tumors did not dictate the response to radiation. To consider the contribution of adaptive immune responses in radioresistance, we conducted experiments in SCID mice. The results are shown in Fig. 3C. TUBO tumors in SCID mice were less responsive than were TUBO tumors in immunocompetent BALB/c mice and never reached equilibrium post-RT. As indicated in the ELISPOT assay, levels of host T cell priming correlated with tumor responses. Our data suggest that local radiation can induce tumor Ag–specific T cell responses in a subset of treated mice, but measureable heterogeneity exists between mice in the induction of tumor-specific immune responses that correlate with local tumor control. At this point we cannot decipher the mechanism(s) underlying the discrepancy in responses in genetically identical hosts (inbred WT BALB/c); however, the tumor cells are likely more genetically heterogeneous than the mice as a result of intrinsic genomic instability. Alternatively, limited evidence was recently reported that individual inbred mice have different levels of genomic acetylation/methylation that can influence their behavior (19, 20). Inbred mice also were reported to respond differentially to viral infection (21). Despite the unknown factor(s), the clinical implications of this finding are even greater in humans, who are vastly more heterogeneous with regard to genetic background and tumor immunogenicity.
Clinical and experimental data suggest that the effects of local radiation may extend to tumors outside the treatment field. This abscopal effect was recently highlighted by a case report suggesting the possible role of adaptive immunity (22). These data are consistent with the idea that local RT can induce immune responses to control local tumor cells but might also reduce the outgrowth of distant metastasis. However, more studies are needed to optimize the conditions for combination strategies using RT and immunotherapy.
Recent studies point to an interesting role for T cells and IFN in achieving equilibrium in carcinogen-induced murine tumor models (9, 23, 24). The underlying conclusions from these studies are that emerging tumors are immunologically selected, leaving behind clones that either lack expression of suitable Ags for T cell–mediated recognition (Ag-loss variants) or have evolved other mechanisms to resist immune attack (acquired resistance). Although immunological selection pressure could drive the selection of poorly antigenic clones (25, 26), tumor outgrowth need not be accompanied by loss of antigenicity, because tumors expressing tumor-specific or tumor-associated Ags have been documented in patients, and spontaneous T cell responses against these Ags can be detected. Thus, established tumors can escape immune-mediated killing through mechanisms other than loss of Ag expression. In particular, many mechanisms of immune suppression are enlisted by the tumor to prevent adequate recognition, including recruitment of suppressor cell subsets and upregulation or overexpression of T cell–negative regulators, such as PD-L1. Disruption of these acquired resistance mechanisms can uncover potent host immunity that can inhibit tumor progression or lead to complete regression, likely depending on various factors that affect the overall strength of the underlying host T cell response. It is clear that local radiation can tip, at least transiently, the balance of suppression back toward host immunity to generate a temporary state of equilibrium, the duration of which is affected by the state of host tolerance to expressed tumor Ags. Thus, tumors might actually proceed through several periods of equilibrium and escape during the outgrowth of nascent tumors and again during treatment of the established tumor with cytotoxic therapies, such as radiation. Our results suggest that radiation-induced equilibrium is likely to be mediated by similar host mechanisms to equilibrium of nascent tumors, with host T cells and IFN being essential. However, without further intervention, most irradiated tumors eventually escape as the result of poorly defined mechanisms. We demonstrate that expression of PD-L1 in the irradiated tumor microenvironment is a critical barrier to host immunity, and therapeutic blockade with anti–PD-L1 Ab could disrupt equilibrium, resulting in complete tumor regression in a high percentage of mice.
Emerging preclinical data are revealing how the host immune system sculpts the development of cancer (25, 26). Preclinical data show that adoptively transferring tumor Ag–specific CD4+ T cells can arrest multistage carcinogenesis in the dormant state through TNF and IFN-γ signaling (27). However, there are very limited data addressing the potential role that antitumor immunity plays in partial response/stable disease that can be induced and maintained by RT. Even fewer studies use tumor equilibrium as a therapeutic window during which host antitumor immunity is most amenable to therapeutic interventions.
T cell exhaustion within the tumor microenvironment may be partially responsible for the stalemate that produces a period of radiographic stability following RT. New reports reveal that the inhibitory receptors PD-1 and Tim-3 are positively correlated with exhaustion of tumor-specific CD8+ T cells in human and murine tumors (28–30). Furthermore, two recent clinical trials concluded that blockade of the PD-1 pathway can restore immune responses, leading to tumor control in some advanced tumors that failed conventional therapies (31, 32). Interestingly, anti–PD-L1 treatment fails to inhibit tumor growth if PD-L1 is not expressed within the local tumor tissue, and 36% of patients with detectable expression of PD-L1 in tumor tissue respond to the treatment. We report that treatment of tumors with RT can induce the expression of PD-L1, and tumor-infiltrating T cells often express PD-1 after RT. Based on the known inhibitory function of PD-1 on CD8+ T cells, we demonstrate in this study that PD-L1 blockade can break the equilibrium stalemate and tip the balance toward tumor regression. Direct activation of T cells within the tumor microenvironment might be another strategy to clear RT-induced equilibrium tumors (18, 33).
Our results identify an active interplay between tumor cells and immune cells that occurs in RITE and suggest that anti–PD-L1 Ab might be useful in the treatment of radioresistant tumors. Furthermore, our results demonstrate that the efficacy of RT can be potentiated by proper immunotherapy.
We thank Drs. S.J. Kron and N.N. Khodarev of the University of Chicago for helpful scientific discussions.
This work was supported in part by National Institutes of Health Grants CA141975 and CA97296 (to Y.-X.F.) and CA111423 (to R.R.W.), a grant from The Ludwig Foundation (to R.R.W.), and a generous gift from The Foglia Foundation (to Y.-X.F. and R.R.W.).
The online version of this article contains supplemental material.
The authors have no financial conflicts of interest.