Immune checkpoint inhibitors, such as anti–PD-1/PD-L1, have emerged as promising therapies for advanced non–small cell lung cancer (NSCLC). However, approximately 80% of patients do not respond to immunotherapy given alone because of intrinsic or acquired resistance. Radiotherapy (XRT) can overcome PD-1 resistance and improve treatment outcomes, but its efficacy remains suboptimal. The tyrosine phosphatase SHP-2, expressed in some cancers and in immune cells, has been shown to negatively affect antitumor immunity. Our hypothesis was that SHP-2 inhibition in combination with anti–PD-L1 would enhance immune-mediated responses to XRT and synergistically boost antitumor effects in an anti–PD-1–resistant mouse model. We treated 129Sv/Ev mice with anti–PD-1–resistant 344SQ NSCLC adenocarcinoma with oral SHP099 (a SHP-2 inhibitor) combined with XRT and intraperitoneal anti–PD-L1. Primary tumors were treated with XRT (three fractions of 12 Gy each), whereas abscopal (out-of-field) tumors were observed but not treated. XRT in combination with SHP099 and anti–PD-L1 promoted local and abscopal responses, reduced lung metastases, and improved mouse survival. XRT also increased SHP-2+ M1 tumor-associated macrophages in abscopal tumors (P = 0.019). The addition of SHP099 also associated with a higher M1/M2 ratio, greater numbers of CD8+ T cells, and fewer regulatory T cells. This triple-combination therapy had strong antitumor effects in a mouse model of anti–PD-1–resistant NSCLC and may be a novel therapeutic approach for anti–PD-1–resistant NSCLC in patients.
Lung cancer remains the leading cause of cancer-related death across the world. Eighty-five percent of lung cancers are of the non–small cell type [non–small cell lung cancer (NSCLC); ref. 1]. For patients with locally unresectable advanced NSCLC, concurrent chemoradiotherapy is currently the standard of care (2). However, the success of this treatment continues to be hampered by the appearance of distant metastases.
Use of immune checkpoint inhibitors has become more widespread, especially anti–PD-1/anti–PD-L1 therapy. However, fewer than 20% of patients with NSCLC respond to immunotherapy alone, and a large proportion develop resistance (3, 4). However, radiotherapy (XRT), especially stereotactic body XRT, can enhance antitumor immunity by releasing tumor-associated antigens and recruiting CTLs to the tumor microenvironment (5–7). Therefore, combinations of XRT with immunotherapy have been investigated in efforts to better achieve antitumor efficacy and potentially attenuate resistance to immunotherapy.
To explore the mechanisms underlying resistance to immunotherapy, our laboratory developed an anti–PD-1–resistant 344SQR NSCLC cell line (8). We previously found that XRT overcame PD-1 resistance in part by upregulating IFNβ and the MHC class I molecule. However, treatment effectiveness remains unsatisfying, as XRT can also induce negative effects on the immune microenvironment such as upregulating M2 tumor-associated macrophages (TAM) and regulatory T cells (Treg), both of which promote an immunosuppressive antitumor phenotype and augment tumor growth (9, 10).
SHP-2 is a nonreceptor, ubiquitous protein tyrosine phosphatase with a relatively conserved structure and function (11). It has usually been described as an oncogene that regulates cancer cell survival and proliferation, primarily by activating the RAS–ERK signaling pathway (12). In lung cancer, SHP-2 is required for the growth of KRAS-mutant and ALK-rearranged NSCLC (13, 14). In T cells, SHP-2 is an intracellular molecule activated downstream of the PD-1 signaling pathway that suppresses T-cell activation (15, 16). In myeloid and natural killer (NK) cells, SHP-2 may decrease the IFNγ-induced activation of STAT1 phosphorylation, thereby suppressing antitumor immunity (17, 18).
Adding XRT to anti–PD-1/anti–PD-L1 can double the effectiveness of immunotherapy alone (19). Previous findings show that the anti–SHP-2 compound SHP099 triggers antitumor immunity and synergizes with anti–PD-1 (20). However, its effect on PD-1–resistant NSCLC in combination with XRT remains largely unknown. Herein, we hypothesized that anti–SHP-2 would synergize with XRT and anti–PD-L1 therapy to overcome treatment resistance and improve outcomes in our anti–PD-1–resistant 344SQ NSCLC model.
Materials and Methods
Cell lines and drugs
344SQ anti–PD-1–resistant cell line (344SQR) was previously derived from the 344SQ parental NSCLC cell line. (8) The 344SQ parental cell line is a metastatic mouse lung cancer cell line derived from a spontaneous subcutaneous metastatic lesion in p53R172HΔg/+K-rasLA1/+ mice. This cell line was a generous gift from Dr. Jonathan Kurie (The University of Texas MD Anderson Cancer Center, Houston, TX). PANC2 cells were obtained from ATCC. For this study, 344SQR and PANC2 cells were cultured at 37°C in a humidified incubator with 5% CO2 in RPMI1640 medium supplemented with 10% FBS and 100 units/mL penicillin and 100 μg/mL streptomycin. The cells were routinely tested to confirm the absence of Mycoplasma contamination and were cultured for a limited number of generations. The SHP-2 inhibitor SHP099 (Chemietek, clone no. 1801747-11-4) was diluted in methylcellulose (Sigma: 9004-67-5), and anti–PD-L1 (durvalumab; MD Anderson Pharmacy) was diluted in PBS (pH 7.4). For the depletion experiment, TAMs and CD8+ T cells were depleted by anti-F4/80 (clone 53-6.7, BioXCell) and anti-CD8 (CI:A3-1, BioXCell), respectively.
Mice, in vivo studies, and treatments
Establishment of the murine model is described elsewhere (9, 21, 22). Eight- to 12-week-old 129Sv/Ev syngeneic female mice and 8- to 12-week-old female C57BL/6 mice were purchased from Taconic Biosciences and bred at a mouse colony maintained by the Department of Experimental Radiation Oncology at The University of Texas MD Anderson Cancer Center (Houston, TX). All animal procedures followed the guidelines of the Institutional Animal Care and Use Committee.
Primary tumors were first injected subcutaneously with 0.5 × 106 344SQR or PANC2 cells in 100 μL PBS in the right legs, followed by 0.1 × 106 344SQ-R or PANC2 cells in 100 μL PBS injected in the left legs as secondary tumors to evaluate abscopal responses. SHP099 (100 mg/kg) was given orally starting on day 5 post-tumor cell injection and was continued for 5 days on and 2 days off, from day 5 to death. Anti–PD-L1 (10 mg/kg) was given intraperitoneally twice a week from day 5 to death. Radiation (three fractions of 12 Gy each) was given to the primary tumors on days 7, 8, and 9. The radiation dose was selected on the basis of prior experiments (21, 22). For depletion experiments, anti-CD8 (300 μg/mouse) was injected intraperitoneally on day 5 and twice a week thereafter; anti-F4/80 (100 μg/mouse) was administrated daily from day 5 for up to 10 days, alternating between intraperitoneal and intratumoral injections and then twice a week thereafter for maintenance. Mice were euthanized when tumor volumes reached 1,500 mm3 (measured using calipers) or at day 60, tissues were harvested for further analysis and measurement, and lungs were harvested for metastases counts after staining with Bouin's fixative solution (Polysciences Inc., catalog no.16045-1). Survival analysis was continued as independent experiments for indicated days.
Flow cytometry phenotyping
Samples, including tumors, blood, and spleens were collected on day 21 for functional experiments. Tumor tissues were minced into small pieces and digested in Liberase TR (250 μg/mL; Roche) and DNase I (20 μg/mL; Roche) at 37°C for 30 minutes, filtered, and washed with PBS plus 1 μmol/L ethylenediaminetetraacetic acid. The cell suspensions were layered over Histopaque-1077 (Sigma-Aldrich) at 1:1 v/v and centrifuged at 400 × g for 30 minutes (8). The tumor-infiltrating immune cells in the interphase were collected and washed with PBS containing 2% FBS, followed by staining with various markers as described below. Cell surface and intracellular markers were stained with the following fluorochrome-conjugated antibodies: anti-CD45 Pacific blue (catalog no. 103126), anti-CD3 PE-Cy7 (catalog no. 100220), anti-CD4 APC-Fire 750 (catalog no. 100412), anti-CD8 PercpCy5.5 (catalog no. 100734), anti-Foxp3 Alexa 488 (catalog no. 126406), anti-CD11b Alexa700 (catalog no. 101222), anti-CD11c BV510 (catalog no. 117353), anti-CD49 PE/Dazzle 594 (catalog no. 108924), anti-Gr-1 APC (catalog no. 108412), anti-F4/80 BV510 (catalog no. 123135), anti-CD38 PE dazzle 594 (catalog no. 102730), and anti-CD206 PE-Cy7 (catalog no. 141720) from BioLegend and anti-SHP-2 PE (catalog no. 560389) from BD Biosciences. For surface staining, all samples were stained with these antibodies at room temperature for 30 minutes; for intracellular staining of Foxp3 and SHP-2, cells were fixed, permeabilized, and stained according to the manufacturer's instructions. All samples were run on a Gallios (BD Biosciences) Flow Cytometer and analyzed with Kaluza Analysis Software.
A digital multiplexed NanoString nCounter analysis system at MD Anderson (NanoString Technologies) was used for gene expression profiling, with 100 ng total RNA from each sample (from both of primary and abscopal tumor) as input material according to the manufacturer's instructions. Total RNA was isolated from tumor-infiltrating lymphocytes (TILs) by using the RNeasy Mini Kit (Qiagen, catalog no. 74104). Analysis was performed on RNA from TILs (details for TILs isolation was shown in our previous publication; ref. 8). All RNA samples included in this study passed quality control requirements (as assessed by the RNA integrity number) of the NanoString platform. NanoString analysis of 770 immune-related genes was done with the nCounter Mouse PanCancer immune profiling kit on the nCounter analysis system. Details for these 770 genes can be found at https://www.nanostring.com/products/gene-expression-panels/gene-expression-panels-overview/hallmarks-cancer-gene-expression-panel-collection/pancancer-immune-profiling-panel?jumpto=SUPPORT. Quantification of target mRNA in each sample was performed by detection within the nCounter digital analyzer. Data from the NanoString nCounter System were normalized to the internal positive controls and housekeeping genes using the recommended settings in the nSolver Software Normalization Module (NanoString Technologies). Normalized data were exported, and differential expression analysis was performed using a linear model method with the limma package for the R programming language. Then, cell typing and pathway analysis that were differentially expressed in each treatment were identified from the expression data of 770 immune-related genes by nSolver. Analysis and normalization of the raw NanoString data were done with nSolver Analysis Software v1.1 and nCounter Advanced Analysis v2.0.115 (both from NanoString Technologies).
Cell proliferation assay
344SQR cell line was treated with SHP099 at four different concentrations (0, 1, 10, or 100 μmol/L) and proliferation was assessed with a CellTiter-Glo Luminescent Cell Viability Assay Kit (catalog no. G7571, Promega Corp.). Exponentially growing cells were diluted to 3 × 104/mL and seeded into 96-well plates. After 24, 48, or 72 hours, an MTS assay was performed at Thermo Fisher Scientific Multiskan Spectrum and relative cell viability at absorbance (450 mm) was normalized to the 0 μmol/L SHP099 condition.
Colony formation assay
Exponentially growing 344SQR cells were seeded into 6-well plates at different densities (200 cells/well for 0 Gy, 400 cells/well for 2 Gy, 1,000 cells/well for 4 Gy, and 2,000 cells/well for 6 Gy), treated with SHP099 at 0, 10, or 100 μmol/L, and irradiated 2 hours later. Forty-eight hours after, the SHP099 was replaced by fresh RPMI1640, and cells were incubated and monitored for another 8 days. Groups of >50 cells were considered colonies, and colony counts were used for the final analysis.
Statistical analyses were done with GraphPad Prism software (v8.0). Tumor growth curves were compared by using two-way ANOVA. Survival graphs were analyzed by the Kaplan–Meier method and compared with log-rank tests. Student t tests were used to compare bar charts of different treatment conditions, and statistical significance was defined as P < 0.05.
Triple therapy with XRT + SHP099 + anti–PD-L1 has significant antitumor effects
First, we explored whether the SHP-2 inhibitor SHP099, alone or in combination with stereotactic XRT, would have antitumor effects in an anti–PD-1–resistant NSCLC model that involved establishing 344SQR tumors in 129Sv/Ev mice (Fig. 1A). We found that SHP099 alone reduced tumor growth versus the control group (P = 0.002, Fig. 1B; P = 0.001; Supplementary Fig. S1A), and XRT combined with SHP099 significantly increased local tumor control (P = 0.023, Fig. 1B; P = 0.013; Supplementary Fig. S1A) but did not affect out-of-field (i.e., abscopal) tumors relative to SHP099 alone (P = 0.199, Fig. 1C; P = 0.142; Supplementary Fig. S1B). Similarly, XRT + anti–PD-L1 enhanced local tumor control relative to XRT only (P < 0.0001, Fig. 1B; Supplementary Fig. S1C) but did not produce significant abscopal responses versus XRT alone (Fig. 1C; Supplementary Fig. S1D).
Next, we explored whether triple therapy (XRT with SHP099 and anti–PD-L1) would have a greater antitumor effect and generate abscopal responses. Triple therapy induced abscopal effects at the unirradiated sites more than all other combinations (P = 0.0003 vs. XRT + anti–PD-L1; P = 0.012 vs. SHP099 + anti–PD-L1; and P = 0.028 vs. XRT + SHP099; Fig. 1C). Local control of irradiated tumors was also further enhanced relative to XRT + anti–PD-L1 (P = 0.021) or XRT + SHP099 (P = 0.004; Fig. 1C). The survival rate among the mice in the triple-therapy group was 37.5% at 60 days, but all of the mice in the XRT + anti–PD-L1 group died by day 37 (P = 0.0001; Fig. 1D). Lungs were also collected at various times throughout the experiment, sectioned and stained, and metastases were counted. Triple therapy reduced the lung metastases ratio (defined as a ratio between lung metastasis counts and survival days) relative to XRT only (P < 0.0001) and XRT + anti–PD-L1 (P = 0.008; Fig. 1E). Representative images of primary tumors (in the right leg), secondary tumors (in the left leg), and lung metastases in the various treatment groups are shown in Fig. 1F and Supplementary Fig. S1E. At the same time, no apparent toxicities were observed during protocol treatment.
To validate our finding, we used a pancreatic cell line (PANC-02) in C57BL/6 mice to verify the treatment efficacy of this triple therapy. Mice were subcutaneously inoculated with PANC-02 tumor cells in the right hind leg (primary tumors) and the left hind leg (abscopal tumors), and treatments followed the same schedule as PD-1–resistant 344SQ model. The triple therapy extended mouse survival beyond day 40 (P = 0.0042; Supplementary Fig. S2A) and limited the tumor growth at the local (Supplementary Fig. 2B, left) and abscopal tumor sites (Supplementary Fig. 2B, right). To better understand the effect of SHP099 in this triple therapy and explore whether the well-established tumors can be sensitized by this therapy by adding SHP099 at a later stage, mice were divided into three groups, group 1: XRT + PD-L1; group 2: XRT + anti–PD-L1 + SHP099 (starting from day 5); and group 3: XRT + anti–PD-L1 + SHP099 (starting from day 15). We found that adding SHP099 at a later stage (from day 15) could also control the tumor growth and prolong mouse survival compared with double therapy using SHP099 + XRT (P = 0.0034; Supplementary Fig. S1G).
Anti–SHP-2 affects immune cells in the tumor microenvironment but not cancer cells
To determine the mechanism underlying SHP099-mediated antitumor effects, we assessed relative cell viability and colony formation. At the concentrations used, SHP099 did not inhibit the viability of the anti–PD-1–resistant 344SQ cells in either an MTS assay (Supplementary Fig. S3A) or a colony formation assay (Supplementary Fig. S3B). These results suggested that SHP-2 inhibition decreased tumor load through augmenting antitumor immunity rather than by inhibiting the growth of tumor cells in our anti–PD-1–resistant syngeneic NSCLC model.
Next, we used NanoString immune profiling to evaluate which subgroups of immune populations were affected by SHP-2 inhibition. We profiled tumor-infiltrating immune cells after triple therapy and analyzed the expression of 770 immune-related genes. Cell-type profiling showed that the triple-therapy group had a relatively higher abundance of CD8+ T cells and macrophages in both the primary (Fig. 2A) and abscopal (Fig. 2B) tumors. A pathway Z-score line chart showed that for triple therapy, most of the immune-related pathways were upregulated, and the cancer progression pathways downregulated, in both the primary (Fig. 2C) and abscopal (Fig. 2D) tumors.
To further investigate pathways involving macrophage function, T-cell function, and cancer progression, we selected corresponding Z-scores for these pathways and explored their significance. In the irradiated tumors (Fig. 2E), triple therapy increased macrophage function (P = 0.008 vs. XRT alone; P = 0.028 vs. XRT + anti–PD-L1), T-cell function (P = 0.003 vs. control; P = 0.009 vs. XRT alone), and decreased cancer progression (P = 0.048 vs. control). These observations also held in the unirradiated (abscopal) tumors (Fig. 2F), in which triple therapy increased macrophage function (P = 0.003 vs. control; P = 0.043 vs. XRT alone; P = 0.014 vs. XRT + anti–PD-L1), T-cell function (P = 0.005 vs. control; P = 0.035 vs. XRT + anti–PD-L1), and decreased cancer progression (P = 0.003 vs. control).
Triple therapy increases the M1/M2 ratio
To further explore the mechanisms underlying the effects of triple-combination therapy and to validate the NanoString results on a cellular level, we collected primary and abscopal tumors on day 21 (11 days after XRT) and evaluated myeloid populations, including TAMs (F4/80+CD11b+Gr1int), tumor-associated neutrophils (TAN, CD11b+Gr1high), and conventional dendritic cells (cDC, CD11chighCD11b+) by flow cytometry. In primary tumors (Fig. 3A–C), we found that XRT alone reduced the M1 TAM population (CD11b+Gr1intF4/80highCD38high; P = 0.07; Fig. 3A) and increased the M2 population (CD11b+Gr1intF4/80highCD206high) versus the control group (P = 0.058), although these findings were not statistically significant (Fig. 3B). Triple therapy also upregulated the M1 population relative to the XRT-only group (P = 0.023) but did not affect the M2 population. Representative flow cytometry results for TAMs in a primary tumor are shown in Fig. 3C.
Unlike its effects on primary tumors, in abscopal tumors, triple therapy upregulated the M1 population (P = 0.008; Fig. 3D) and significantly downregulated the M2 population compared with the control group (P = 0.039; Fig. 3E). This effect seemed to be mediated by SHP099 because XRT + anti–PD-L1 in the absence of SHP-2 inhibition did not affect the M2 subpopulation relative to the control or XRT-only group (Fig. 3E). Representative flow cytometry results for TAMs in primary tumors are shown in Fig. 3F, and a representative gating strategy for myeloid subpopulations is shown in Supplementary Fig. S4. In either the primary or the abscopal tumors, triple therapy did not affect the numbers of TAMs (Supplementary Fig. S5A and S5B), TANs (Supplementary Fig. S5C and S5D), cDCs (Supplementary Fig. S5E and S5F), or NK cells (Supplementary Fig. S5G and S5H).
Next, we confirmed the flow cytometry data for TAMs by using NanoString molecular analysis of primary tumors (Fig. 3G) and secondary (abscopal) tumors (Fig. 3H) harvested on day 21. Triple therapy significantly upregulated M1 macrophage markers in primary tumors, such as CD68 (P = 0.023) and CD38 (P = 0.031) versus the XRT-only group and downregulated M2 markers, such as arginase 1 (Arg1, P = 0.002; Fig. 3G). Triple therapy also upregulated the expression of M1 proinflammatory markers in abscopal tumors, such as CD68 (P = 0.0013), CD38 (P = 0.009), and CD80 (P = 0.003) and significantly downregulated Arg1 (P = 0.0086) relative to the control group (Fig. 3H).
We also found that XRT trended to increase M1 TAMs in abscopal tumors. To detect the mechanism of this phenotype, we tested M1 and M2 TAMs in blood and some cytokines and chemokines at the primary and abscopal sites. The flow data showed no obvious changes (and very few myeloid cells) in blood (Supplementary Fig. S6). NanoString data for the expression of cytokines and chemokines indicated that XRT increased Tnfa (P = 0.032), Il6 (P = 0.041), Ifng (P = 0.042), and Ccl2 (P = 0.02) in primary tumors (Supplementary Fig. S7A), while simultaneously increasing Ifng (P = 0.013), Tnfa (P = 0.006), and Cxcl9 (P = 0.042) in abscopal tumors (Supplementary Fig. S7B), all of which could help M1 polarization.
Triple therapy increases antitumor immunity of lymphoid cells
We next collected primary and abscopal tumors on day 21 and evaluated lymphoid subpopulations, including CD4+ T cells (CD45+CD3+CD4+), Tregs (CD45+CD3+CD4+Foxp3+), and CD8+ T cells (CD45+CD3+CD8+) by flow cytometry. The gating strategy is shown in Supplementary Fig. S8. In primary tumors (Fig. 4A), the triple therapy significantly increased CD8+ T cells relative to the control (P = 0.002), XRT-only (P = 0.007), and XRT + PD-L1 groups (P = 0.012). In abscopal tumors (Fig. 4B), the triple therapy also increased CD8+ T cells relative to the control (P = 0.008), XRT-only (P = 0.019), and XRT + anti–PD-L1 groups (P = 0.039). However, in primary tumors, XRT increased the percentage of Tregs versus the control (P = 0.0005; Fig. 4C), but triple therapy decreased the percentage of Tregs relative to the XRT group (P = 0.0074). No differences were found among the various treatment groups in percentages of Tregs in abscopal tumors (Fig. 4D) or CD4+ lymphocytes in either the primary tumors (Supplementary Fig. S9A) or abscopal tumors (Supplementary Fig. S9B). NanoString analysis showed that the triple therapy boosted both the cytotoxicity and activation of immune cells (Fig. 4E and F), as indicated by Gzma (P = 0.001), Gzmb (P = 0.016), and Ifng (P = 0.031) in primary tumors and Gzmb (P = 0.012), Gzmk (P = 0.018), and Prf1 (P = 0.02) in abscopal tumors relative to the control.
SHP-2 is expressed mainly in M1 TAMs, and XRT may induce its expression
Immune cells in the irradiated and abscopal tumors were further analyzed for their expression of SHP-2. On day 21 (i.e., 10 days after XRT), immune cells were isolated and phenotyped by flow cytometric analysis. We found that in our anti–PD-1–resistant NSCLC model, the mean fluorescence intensity (MFI) of SHP-2 at the tumor site appeared mainly in TAMs, followed by TANs, Tregs, CD8+ T cells, and then CD4+ T cells (Fig. 5A and B). Similarly, the percentage of the immune cells expressing SHP-2 was highest for TAMs, followed by TANs and then T cells (Fig. 5C and D). We also found that SHP-2 MFI (Fig. 5E and F) and SHP-2+ subpopulations (Fig. 5G and H) were more common among M1 macrophages than in M2 macrophages (P = 0.0001, Fig. 5F; P = 0.005, Fig. 5H). When we explored possible correlations between XRT and SHP-2 expression, we found that XRT led to significantly increased SHP-2+ M1 TAMs (Fig. 5I) in the unirradiated (abscopal) tumors (P = 0.019) but not in the irradiated (primary) tumors (P = 0.084; Fig. 5J).
The antitumor effects of triple therapy take place mainly through TAMs and CD8+ T cells
To identify the roles of CD8+ T cells and TAMs in the observed antitumor response, we depleted these populations in vivo and analyzed the effects on survival and tumor control at the primary and abscopal tumor sites. Survival was compromised upon CD8 depletion (P = 0.01 vs. triple therapy), F4/80 depletion (P = 0.0041 vs. triple therapy), and combined CD8 and F4/80 depletion (P = 0.0004; Fig. 6A). The CD8-depletion group also had a higher lung metastasis ratio than the triple-therapy group (P = 0.007), as did the F4/80-depletion group (P = 0.003) and the CD8 and F4/80 double–depletion group (P = 0.0006; Fig. 6B). Finally, in accordance with our previous results, triple therapy without CD8 and F4/80 depletion led to suppressed tumor growth at both the primary and abscopal tumor sites (Fig. 6C and D).
In this study, we tested whether the addition of the SHP-2 inhibitor SHP099 to anti–PD-L1 and XRT would improve primary tumor control and abscopal responses in our anti–PD-1–resistant NSCLC model. Indeed, our triple therapy resulted in improved abscopal effects, local control, survival, and lung metastasis. XRT, anti–PD-L1, and SHP099 were all found to be required to drive durable responses.
The two reasons for choosing anti–PD-L1 over anti–PD-1 in this study were: (i) a previous report that XRT could increase PD-L1 expression in cancer cells (6), which meant we would have more targets when we used anti–PD-L1, and (ii) we had concern over the risk of increased toxicity from combining anti–PD-1 and SHP099 because both reagents share the same pathway. Combining two other forms of immunotherapies including, for example, anti–CTLA-4 and anti–PD-1, has shown higher rates of toxic effects, such as colitis (37%) and myocarditis (25%), and early occurrences of fatal toxic effects relative to monotherapy (23).
SHP-2 is a “two birds with one stone” protein because it is an oncogenic gene in cancer cells and a universal target regulating immune cell exhaustion, including T cells, NK cells, and myeloid cells. SHP099 is a unique allosteric inhibitor that acts as “molecular glue” that selectively blocks SHP-2 activity by locking it in an auto-inhibited conformation (12, 24). In our anti–PD-1–resistant model of NSCLC, SHP099 acted only on the immune system and not on cancer cells. Similarly, XRT can be considered a “two-edged sword” with regard to immune function. On the one hand, hypofractionated XRT can enhance antitumor immune responses by releasing immunogenic tumor-associated antigens and chemokines and recruiting antigen-specific CD8+ effector T cells to the tumor microenvironment, which leads to abscopal effects on unirradiated tumors. On the other hand, XRT can also cause rebound immune suppression by increasing PD-L1 expression, Tregs, M2 TAMs, and myeloid-derived suppressor cells (25–29). Collectively, this evidence suggests the outlines of an explanation why triple therapy with XRT, anti–PD-L1, and SHP099 could increase antitumor immunity and produce abscopal effects. In our study, the combination of XRT + SHP099 did not improve the response of abscopal tumors relative to SHP099 alone, suggesting that neither XRT nor SHP099, when used in isolation, has any apparent abscopal effects. However, adding anti–PD-L1 led to abscopal effects, especially compared with anti–PD-L1 + SHP099 or SHP099 alone. In our model, abscopal tumors were implanted subcutaneously and were not spontaneous metastases. However, triple therapy also inhibited spontaneous lung metastases in this model.
We also found that XRT could increase M1 TAMs in abscopal tumors. We have two hypotheses to explain this trafficking effect. First, XRT might help in the transfer of M1 macrophages from the irradiated tumors to the abscopal tumors, but our flow data denied this hypothesis. Then, we next explored the mechanism by NanoString assay and the data indicated that some cytokines and chemokines which could help M1 polarization were higher expressed after XRT either in primary or abscopal tumor sites. In support of the latter hypothesis is that, in our anti–PD-1–resistant NSCLC model, three 12-Gy fractions increased IFNβ secretion (8), a cytokine important in M1 polarization (30–32).
We further found that SHP-2 was expressed to different extents in several different immune cells, including (in decreasing order) TAMs, TANs, Tregs, CD8+ T cells, and CD4+ T cells. This may help to explain our NanoString findings showing that adding that SHP-2 inhibitor could increase the abundance of antitumor immune cells, such as cytotoxic cells, macrophages, and CD8+ T cells and activate immune-related pathways and decrease cancer progression pathways. We further found that adding SHP099 could weaken radiation-induced immunosuppressive effects via M2 polarization, decreasing XRT-induced Tregs, and increasing CD8+ T cells. M1 macrophages have proinflammatory effects, and an antitumor subgroup of TAMs could secrete proinflammatory cytokines such as TNFα and IFNβ to recruit T lymphocytes (33, 34), and this may also explain why the triple-therapy group had more CD8+ T cells when SHP-2 was blocked. Our finding that XRT increased SHP-2+ M1 TAMs, especially in abscopal tumors, could be interpreted at two levels. First, because SHP-2 is a negative target in antitumor immunity, SHP-2+ M1 TAMs would not have stronger antitumor effects than SHP-2− M1 TAMs, as is the case for PD-1+ macrophages and PD-1− macrophages (35). Thus, it would be rational to use SHP-2 inhibition to overcome this negative regulation for an antitumor effect. Second, XRT could increase SHP-2+ M1 TAMs and more targets are expressed when we use SHP099. This is another rationale for combining SHP099 and XRT.
In summary, we propose the following structure to summarize the major outcomes of this study in Fig. 7. Irradiation of tumors increased PD-L1 expression, providing a rationale for adding anti–PD-L1 to inhibit this immunosuppressive environment, and increased M2 macrophages, which further contributes to effector cell suppression. SHP099 could reverse this effect and repolarize XRT-induced M2 TAMs to the antitumor M1 phenotype. In abscopal (unirradiated) tumors, XRT increased M1 TAMs, including SHP2+ M1 TAMs. The SHP099 component of the triple therapy further polarized M2 macrophages to the M1 phenotype, synergistically boosting abscopal antitumor effects.
Disclosure of Potential Conflicts of Interest
J.W. Welsh is co-founder of Healios Oncology, MolecularMatch, and OncoResponse; is an advisor for AstraZeneca, Merck, MolecularMatch, Incyte, Aileron, and Nanobiotix; is a scientific advisory board member for Legion Healthcare Partners, RefleXion Medical, MolecularMatch, OncoResponse, CheckMate, Alpine Immune Sciences, and Nanorotbotics; reports receiving commercial research support from GlaxoSmithKline, Bristol-Meyers Squibb, Merck, Nanobiotix, Mavu Pharma, Takeda, and Checkmate Pharmaceuticals; and has the following patents: MP470 (amuvatinib), MRX34 regulation of PD-L1, and XRT technique to overcome immune resistance. MD Anderson Cancer Center has a trademark for RadScopal. No potential conflicts of interest were disclosed by the other authors.
Conception and design: D. Chen, H.B. Barsoumian, J.W. Welsh
Development of methodology: D. Chen, H.B. Barsoumian, A.I. Younes, F. Masropour
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Chen, A.I. Younes, M. Wasley
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Chen, H.B. Barsoumian, A.I. Younes, H. Menon, S. Mosaffa
Writing, review, and/or revision of the manuscript: D. Chen, H.B. Barsoumian, L. Yang, A.I. Younes, V. Verma, Y. Hu, H. Menon, M. Wasley, T. Ozgen, K. Klein, M.A. Cortez
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Chen, L. Yang, S. Mosaffa, T. Ozgen, J.W. Welsh
Study supervision: M.A. Cortez, J.W. Welsh
The authors thank Christine F. Wogan, MS, ELS, of MD Anderson's Division of Radiation Oncology, for reviewing and editing this article. This study was supported in part by Cancer Center Support (Core) Grant P30 CA016672 from the NCI, NIH.
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.