As a manifestation of the plastic phenotype of cancer cells, epithelial tumors can undergo a phenotypic switch called an epithelial–mesenchymal transition (EMT) in which cancer cells shift from an epithelial state (cuboidal and clustered, apical/basolateral polarity, extensive cell-to-cell adhesion) to an intermediate or mesenchymal phenotype (spindle-like morphology, loss of polarity, minimal cell-to-cell adhesion). These changes promote the acquisition of migratory and invasive capabilities by the tumor cells while enhancing their resistance to several anticancer agents, including chemotherapies, radiation, small targeted therapies and immunotherapy. Both the phenomenon of EMT and EMT-induced resistance to therapeutics can be reversible.16

TGF-β is considered a master regulator of EMT and has been found to promote a phenotypic transition in many cancer models including breast, prostate, colon and melanoma among others.17 In addition, the transcriptional regulators of the process, overall designated as EMT transcription factors, can upregulate TGF-β expression and secretion by cancer cells, thus creating a positive feedback loop that helps tumor cells maintain their mesenchymal-like phenotype while promoting EMT in surrounding cancer cells.16 In a previous study,18 bintrafusp alfa was shown to efficiently suppress TGF-β1-mediated EMT in human non-small cell lung cancer (NSCLC) models. In vitro, bintrafusp alfa prevented the downregulation of the epithelial marker E-cadherin and the up-regulation of the mesenchymal proteins fibronectin and vimentin when administered simultaneously with TGF-β1 to lung cancer cells (figure 1B). In addition, in cells that had already undergone TGF-β1-induced EMT, subsequent treatment with bintrafusp alfa was able to ‘revert’ EMT, thus restoring E-cadherin expression and decreasing expression of mesenchymal proteins to levels similar to control untreated cells (figure 1B).18 In xenograft models of NSCLC in vivo, bintrafusp alfa treatment was also shown to markedly decrease the expression of the mesenchymal protein vimentin in tumor cells (figure 1C). Because these changes were not induced by anti-PD-L1 treatment alone, the anti-EMT effect of bintrafusp alfa was attributed to its TGF-β ‘trapping’ activity.

By suppressing EMT, bintrafusp alfa may potentially improve antitumor efficacy of other therapeutics such as chemotherapy or radiation. The loss of E-cadherin expression is one of the classical signs of EMT and is associated with increased invasion and drug resistance in several cancers.19 20 While the addition of TGF-β1 to lung cancer cell lines in vitro resulted in decreased susceptibility to the cytotoxic effect of docetaxel, paclitaxel and gemcitabine, the addition of bintrafusp alfa simultaneously or following treatment with TGF-β1 was shown to block or revert, respectively, TGF-β1-induced resistance, thus increasing tumor lysis in response to chemotherapy to levels similar to those of control, untreated cells (figure 1D).18 Moreover, as described below, experiments conducted in vivo also demonstrated the ability of bintrafusp alfa to increase anti-tumor efficacy of chemotherapy in a murine model of colon cancer.

The bifunctional blockade of PD-L1 and sequestration of TGF-β offer a novel approach to alter the immune cell compartment in tumors. TGF-β is a master regulator of the immune system; for an in-depth review of this topic, see Batlle and Massagué.21 For the purpose of this review, we will limit our discussion to the natural killer (NK) and T cell compartments because those have been shown to be the most relevant to the antitumor activity of bintrafusp alfa in preclinical models. There is substantial evidence that TGF-β may alter the phenotype of NK cells and suppress their cytotoxic activity.22 In agreement, a recent study of human NK cells in vitro demonstrated the ability of bintrafusp alfa to prevent changes in gene expression induced by TGF-β1 treatment. Moreover, the anti-PD-L1 moiety of bintrafusp alfa has been found in vitro to mediate antibody-dependent cellular cytotoxicity of multiple human tumor cell lines of breast, cervical and urothelial origin, using human NK cells as effectors.23 24 In several murine models in vivo, bintrafusp alfa has also been shown to improve the recruitment of NK cells to the TME.14 15 For example, in a syngeneic breast murine model (EMT6), bintrafusp alfa treatment increased the population of tumor-infiltrating NK cells including those with an activated phenotype (eg, T-bet⁺; NKG2D⁺; NKp46⁺).14 Moreover, treatment with bintrafusp alfa induced antitumor efficacy in multiple tumor models, which was partially suppressed when NK cells were depleted.14 15 Altogether, the bifunctional blockade of PD-1/PD-L1 and TGF-β was shown to increase NK cell activation, cytotoxicity and recruitment to the TME, thus improving host antitumor immune responses.

TGF-β and PD-L1 are both important regulators of the T cell compartment. Binding of PD-L1 to its receptor PD-1 on activated T cells delivers a negative signal that blocks T cell proliferation, survival and effector functions. In addition, TGF-β promotes T_reg differentiation and inhibits antitumor responses.25 TGF-β has been shown to cause differentiation to the T_reg phenotype by driving FOXP3 expression in vitro.21 26 In melanoma and breast cancer patients, FOXP3 and TGF-β expression are highly correlated, indicating that TGF-β promotes T_reg infiltration in the TME.27 T_regs have also been shown to produce TGF-β1 both as a secreted factor and a surface protein26; secreted TGF-β may help T_regs to maintain a steady presence in tissues and to regulate the activity of other immune cells, while TGF-β1 expressed on the membrane of T_regs has been shown to inhibit cytotoxic T cell activity.28 Thus, bintrafusp alfa also offers a method of inhibiting T_reg activity by sequestering TGF-β. In agreement, treatment with bintrafusp alfa in vitro was shown to partially alleviate T_reg suppression of CD4⁺ T cell proliferation. In the same study, the activity of each component of the bifunctional bintrafusp alfa was compared by using an anti-PD-L1 antibody versus a mutated version of bintrafusp alfa (designated as M7824mut in previous publications) that traps TGF-β but does not bind PD-L1, demonstrating that the blockade of TGF-β alone was responsible for the restoration of effector CD4⁺ T cell proliferation in the presence of bintrafusp alfa.24

In addition to its effects on T_regs, bintrafusp alfa added in vitro to CD8⁺ T cells has been shown to enhance IL-2 production in response to superantigen stimulation14 and to modestly increase antigen-specific T cell-mediated tumor cell lysis.23 In murine models, bintrafusp alfa demonstrated antitumor efficacy and was shown to increase the density of proliferative and cytotoxic CD8⁺ T cells in the TME.15 In syngeneic murine mammary and colon carcinoma models (EMT6, MC38), bintrafusp alfa treatment was shown to significantly reduce tumor growth compared with control and anti-TGF-β treatment alone with the M7824mut agent (figure 2A,B).15 In order to expand on the mechanism by which bintrafusp alfa controls tumor growth, Knudson et al 15 used depletion studies and demonstrated that the antitumor efficacy of bintrafusp alfa was dependent on NK and CD8⁺ T cells. Bintrafusp alfa treatment increased effector and effector memory T cell density in the spleen of tumor-bearing mice and induced an active CD8⁺ T cell phenotype in the TME. A limitation of these studies, however, is the fact that transplanted syngeneic tumor models were used. Given the high expression of TGF-β in the stroma, the use of spontaneous transgenic models with an anatomically intact stroma would be more adequate to evaluate the effects of bintrafusp alfa.

Plasma levels of TGF-β1 were reduced in mice treated with bintrafusp alfa or M7824mut (figure 2C) indicating the ability of these agents to sequester murine TGF-β1 in vivo.15 In addition, bintrafusp alfa significantly reduced intratumoral TGF-β signaling, as shown by decreased levels of phosphorylated SMAD2 in breast tumors (figure 2D 15), thus demonstrating the ability of bintrafusp alfa to accumulate at the site of the tumor via binding to PD-L1. It is important to note that while these experiments did not include an anti-PD-L1 monotherapy group, separate research in the EMT6 or MC38 models from Lan et al 14 extensively compared single anti-PD-L1 or single M7824mut treatment against bintrafusp alfa. In those studies, bintrafusp alfa displayed statistically significantly better tumor control than either anti-PD-L1 or M7824mut alone, and was more efficacious than each individual therapy to increase the density of CD8⁺ T cells in the TME with an activated, proliferative and effector memory phenotype.14

Several recent studies demonstrated the importance of TGF-β as a mechanism of resistance to checkpoint inhibitor therapies. One such study, for example, evaluated tumors from patients with metastatic urothelial cancer who were treated with atezolizumab and identified a signature of TGF-β signaling positively associated with lack of response.29 The same study evaluated a combination of anti-PD-L1 and anti-TGF-β in the EMT6 murine breast cancer model, demonstrating that the combination of both blocking antibodies could afford better antitumor effect than that observed with each single treatment alone. In another study,30 blockade of TGF-β signaling was shown to improve the antitumor activity of anti-PD-L1 treatment in a murine model of colon cancer characterized by low mutation burden, T cell exclusion and a TGF-β activated stroma, thus supporting the role of stromal TGF-β as a mechanism of resistance to PD-1/PD-L1 blockade. In light of these studies and the results with bintrafusp alfa, one important question is whether using a dual, bifunctional molecule such as bintrafusp alfa could afford a better outcome as compared with the combination of single anti-PD-L1 and anti-TGF-β agents. Using the MC38 model, Lan et al have shown a modest advantage of using bintrafusp alfa over a combination of anti-PD-L1 or anti-TGF-β alone.14 Studies performed with a similar bifunctional agent that blocks PD-L1 and sequesters TGF-β via a TGF-βRII (designated as Y-trap) have also recently shown that the fusion protein is more efficient than the combination of anti-PD-L1 and IgG-TGF-βRII agents in achieving antitumor activity against human cancer xenografts in humanized mice.27

By sequestering TGF-β, bintrafusp alfa has the ability to modulate tumor cell plasticity, proliferation and susceptibility to other agents. One potential combination that has shown promise in preclinical studies is the use of bintrafusp alfa with tumor-associated antigen (TAA)-targeting vaccines. In a syngeneic breast model, bintrafusp alfa and the TAA-targeting vaccine Ad-TWIST combination significantly increased overall survival above that induced by each monotherapy (figure 2E), demonstrating that bintrafusp alfa may be used in combination with other IO agents. Another potential scenario for combination therapies involving bintrafusp alfa is in the context of standard of care treatment. Both chemotherapy and radiation have been shown to induce the upregulation of TGF-β and PD-L1 expression in various carcinoma models.31–34 Thus, bintrafusp alfa may be particularly effective in combination with these therapies. In preclinical studies,14 bintrafusp alfa was shown to improve antitumor responses in combination with radiation in the syngeneic colon cancer model MC38 in vivo (figure 3A–C). In this model, the combination therapy also demonstrated significant systemic immune activation, shown by an increase of activated CD8⁺ T cells in the spleen above either treatment alone. Concurrent bintrafusp alfa and radiation treatment were also shown to confer antitumor efficacy at the primary tumor site that received radiation and at a secondary location (immune-mediated abscopal effect, figure 3D–F). These data suggested that bintrafusp alfa may be used in combination with local radiation to generate a systemic antitumor immune response that may target metastasis.

In addition to radiation, combination of bintrafusp alfa with chemotherapy (oxaliplatin/5-fluorouracil) has also been shown in preclinical models to be effective at improving antitumor efficacy, where concurrent treatments synergistically induced CD8⁺ T cell activation and effective immune-mediated antitumor effect (figure 3G–I).14 In addition to enhancing CD8⁺ T cell activation, it could also be hypothesized that at least a fraction of the antitumor efficacy observed with combinations of chemotherapy and bintrafusp alfa may be due to the ability of bintrafusp alfa to directly impact the phenotype of the tumor cells. As previously discussed, tumor cells that are highly plastic or mesenchymal are more resistant to cytotoxic lysis, including cell death induced by chemotherapy. Since TGF-β induces EMT, bintrafusp alfa may increase tumor cell susceptibility to chemotherapeutics by depleting TGF-β in the TME and reducing the phenomenon of EMT. This hypothesis is supported by preclinical evidence as shown in figure 1 that bintrafusp alfa effectively decreases mesenchymal features of tumor cells both in vitro and in vivo and increases tumor susceptibility to chemotherapy in vitro.

Numerous inhibitors targeting the TGF-β pathway have progressed to clinical trials including small molecule inhibitors, antibodies and receptor-based TGF-β traps.35 Among the small molecule inhibitors that have been evaluated in preclinical and clinical studies is galunisertib, a small molecule inhibitor targeting the TGF-βRI (ALK5) kinase. This agent has been tested in several cancer types with mixed results. In phase II clinical trials, galunisertib demonstrated some antitumor efficacy in patients with unresectable pancreatic cancer and in a subset of patients with advanced hepatocellular carcinoma;36 37 however, no meaningful antitumor activity was observed in a phase II trial of recurrent glioblastomas.38 Galunisertib has exhibited an acceptable safety profile in phase I testing and in phase II studies.39

Several antibodies against TGF-β have also progressed to clinical trials. Of these, fresolimumab has been most extensively studied. Fresolimumab (GC1008) is a fully human anti-TGF-β MAb that sequesters all isoforms of TGF-β. It has completed a phase I clinical trial (NCT00356460) with acceptable safety signals in renal cell carcinoma, melanoma and glioma.40 Fresolimumab-related adverse events (AEs) that were reported included gingival bleeding, headaches, epistaxis and skin disorders (actinic keratosis, hyperkeratosis and reversible cutaneous keratoacanthomas).40 Of the 29 enrolled patients, one partial response (PR) was observed in a patient with advanced malignant disease, and six patients had stable disease (SD). Fresolimumab is currently being investigated in ongoing phase I/II trials in advanced and metastatic solid tumors and mesothelioma.

While anti-TGF-β therapeutics offer a path to treat cancers, as monotherapies these agents have had limited success in clinical application thus far. The limited number of patients who responded indicates the need for further treatment combinations; the bifunctionality of bintrafusp alfa thus offers a novel approach.

Bintrafusp alfa is currently being studied in clinical trials at multiple institutions. The dose-escalation portion of a phase I, open-label clinical trial of bintrafusp alfa (MSB0011359C) has been completed, showing promising antitumor efficacy in heavily pretreated patients with metastatic or locally advanced solid tumors without prior checkpoint inhibitor treatment.41 A total of 19 patients were enrolled in that trial at the Center for Cancer Research at the NCI (Bethesda, MD) with primary tumor types including pancreatic, cervix uteri, colorectal, anal and adenoid cystic carcinoma, among others. Bintrafusp alfa showed a safety profile similar to other anti-PD-1/PD-L1 monotherapies with the addition of skin-related AEs characteristic of TGF-β blockade. Evidence of clinical activity was seen with bintrafusp alfa across all evaluated doses (figure 4A); of the 19 patients enrolled, one case (cervical cancer) demonstrated a durable complete response (figure 4B,C), two patients (pancreatic cancer, anal cancer) had a durable PR, and two cases (pancreatic cancer, carcinoid) experienced prolonged SD. Analyses of plasma TGF-β1, -β2 and -β3 levels demonstrated that bintrafusp alfa reached sufficient levels to completely sequester all three isoforms of TGF-β in circulation for the entire dosing period at doses >1 mg/kg.41 This study also reported results of the analysis of exploratory biomarkers conducted in peripheral blood collected at various time points. Though the study included a small heterogenous group of patients, results from a flow cytometry-based analysis of 131 immune cell subsets in peripheral blood mononuclear cells prior to therapy and following administrations of bintrafusp alfa demonstrated a trend in those patients with evidence of clinical benefit toward an increase in peripheral B cells and CD4⁺ T cells, and decreases in MDSCs expressing CD16 at the time of best overall clinical response versus pretherapy levels. However, the limited number of patients analyzed in the phase I dose-escalation study precluded any conclusions about the potential clinical utility of bintrafusp alfa over other anti-PD-1/PD-L1 therapies. In addition, further studies with larger number of patients of defined tumor types will need to be conducted to determine potential biomarkers of response to bintrafusp alfa.

Currently, numerous clinical studies are ongoing involving the use of bintrafusp alfa in patients with a variety of cancer types including breast, NSCLC, prostate, pancreas, biliary and HPV-associated cancers. Several of these trials are investigating combinations of bintrafusp alfa with other immunotherapeutics or other therapeutic modalities such as radiation or chemotherapy. An important question moving forward is whether certain patients could particularly benefit from receiving treatment with bintrafusp alfa versus the use of other anti-PD-1/PD-L1 agents. As one of the features associated with TGF-β signaling in tumors is the occurrence of phenotypic plasticity, the evaluation of biomarkers of EMT could potentially be used for selection of patients who may be more responsive to bintrafusp alfa. Gene signatures of EMT have been defined for specific tumor types, including gastric, colorectal and bladder tumors in association with worse overall survival42–44 and more recently, a signature enriched in genes involved in EMT was found to be associated with resistance to anti-PD-1 therapy in metastatic melanoma patients.45 The use of such signatures could be prospectively used in future clinical studies to investigate potential associations between the occurrence of EMT in tumors and treatment outcome with bintrafusp alfa; if validated, the approach could also be used for selection of patients to be included in bintrafusp alfa clinical studies.