Targeting CAFs to Improve Anti-PD-1 Checkpoint Immunotherapy

Tumors that successfully avoid eradication by the immune system consistently deter the infiltration and activity of cytotoxic T lymphocytes. Antibodies targeting the T-cell inhibitory receptor PD-1 have been able to restore antitumor immunity in some patients to provide evident clinical benefit. However, tumors often acquire diverse mechanisms to inhibit the multiple steps controlling T-cell activation, severely limiting such responses. Therefore, understanding and overcoming these mechanisms is key to improve immunotherapeutic effectiveness (1). T-cell activation and suppression alike are strongly associated with the function of immune cells of the myeloid lineage. Whereas lymphocyte priming and stimulation requires mature, fit antigen-presenting cells such as macrophages and dendritic cells, impaired or corrupted myeloid cells are a strong indication for immune dysfunction. One type of myeloid-derived suppressor cells (MDSC), polymorphonuclear (PMN)-MDSCs, comes from a granulocytic, neutrophil-like origin. Indeed, they are difficult to distinguish from neutrophils by cell surface markers alone. The repressive role of these MDSCs can be coordinated by tumor fibroblasts. In mouse models, blocking cancer-associated fibroblast (CAF)–induced recruitment of PMN-MDSCs was shown to improve the therapeutic efficacy of PD-1 treatment (2).

CAFs have been ascribed a multitude of tumor-promoting and immunosuppressive functions, which would make them ideal therapeutic targets (3). Furthermore, fibroblast-specific levels of activation to stromal reprogramming pathway TGFβ, as well as the resulting abundance of extracellular matrix components, have been linked to poor prognosis and immune evasion in colorectal cancer (4). In line with this, Akiyama and colleagues focused on inhibition of platelet-derived growth factor (PDGF)—a key stimulant of fibroblast biology—to revert the immunosuppressive tumor microenvironment (TME) in gastric cancer (5). Advanced, chemotherapy-resistant gastric adenocarcinoma is a common cause of cancer death, and new treatment strategies to block or treat disease progression and metastasis are urgently needed. However, despite progress in targeted therapies and immuno-oncology (including anti-PD-1 treatment), improvements have been modest or limited to a fraction of patients (6). As with other gastrointestinal cancers, combinatorial (immuno) therapies that overcome immune suppression seem to be the way to increase efficacy (1).

Akiyama and colleagues found that PDGFD protein levels identify patients with poor prognosis, in correlation with the expression of the cognate receptor protein PDGFRβ (5). This signaling axis promoted in vitro proliferation of CAFs isolated from surgically resected gastric cancers and was potentiated in the presence of TGFβ signaling, thus indicating relevant cross-talk of two key CAF-bolstering pathways and offering an explanation for tumoral CAF enrichment. To determine whether this signaling axis could be therapeutically blocked, the authors used PDGF receptor–blocking antibodies. Only the combination of two antibodies, one targeting PDGFRα and the other PDGFRβ, reasonably inhibited PDGFD–stimulated CAF growth in vitro. Inhibition of CAF proliferation was more potent when dual PDGFR-targeting tyrosine kinase inhibitor (TKI) sorafenib or regorafenib was applied at relatively high medium concentrations (5).

Either of these small-molecule inhibitors was found to reduce the expression of immunomodulatory genes in CAFs in vitro, including a number of PMN-MDSC–recruiting chemokines. Coculture of a mouse gastric cancer cell line (GAN-KP) with primary skin fibroblasts indicated that, upon interaction, cancer cells upregulate the transcription of Pdgf genes and the fibroblasts in turn upregulate Cxcl1 and Cxcl3. These chemokines were shown to mediate the attraction of neutrophils, which was inhibited by regorafenib (5). To study this mechanism further in vivo, the authors developed a fibrotic, immunosuppressive tumor model based on serial subcutaneous transplantation of the GAN-KP cells. In this model, regorafenib treatment reduced both CAF and tumoral PMN-MDSC abundance. Moreover, the combination of this TKI with antibodies against PD-1 restored antitumor T-cell–mediated immunity better than either treatment alone, suppressing tumor growth.

This study fits well in the emerging view that the complexity of the immunosuppressive TME requires combination therapies that simultaneously target disparate stromal cell types to surmount immune evasion (1–4). Indeed, given the incriminating evidence of CAFs as central drivers of cancer progression (3), the prospect of a drug that effectively impinges on the root source of CAF stimulation is a very attractive one—even more so if it would synergize with immune checkpoint inhibitors. However, a number of caveats need to be considered.

First, it is hard to conclude whether the effects of regorafenib–anti-PD-1 therapy on tumor growth are truly synergistic and sufficiently potent. Benefit over monotherapy was statistically significant but modest, and none of the tumors were entirely eradicated 2 weeks after treatment. Perhaps more importantly, although the serial transplantation model used by Akiyama and colleagues may be an adequate recapitulation of murine fibrotic gastric cancer—albeit localized in an anatomically divergent subcutaneous environment—this does not represent the relevant clinical indication. New therapeutic strategies are needed for advanced disease, that is, cancer that has spread to the liver, lungs, or peritoneum wall, which entails a different biology. Therefore, metastatic mouse models based on genetic tumorigenesis or orthotopic injection would be better suited to study immunity in the metastatic microenvironment. Of note, preclinical success of the regorafenib–anti-PD-1 combination in a metastatic setting was previously demonstrated for colon cancer (7). In that study, dual therapy efficacy yielded a marginal, additive improvement over monotherapy on subcutaneous or orthotopic primary tumors, yet had a more promising effect on sustained growth inhibition after treatment discontinuation, as well as a marked decrease in liver metastasis, compared with the TKI alone (7). Unfortunately, in this setting, checkpoint inhibition alone was not included as a control. When it comes to clinical investigation, both regorafenib alone and the regorafenib–anti-PD-1 combination have already been tested in early trials (8). Despite promising results in these initial clinical studies, subsequent real-world clinical colorectal cancer data have been less encouraging (9). Conclusions from an ongoing randomized phase III study (NCT04879368) need to be awaited to assess the efficacy of this dual therapy.

Surprisingly, few if any of the studies into the (pre)clinical therapeutic efficacy of regorafenib reported effects on CAFs. This may be explained by the fact that, like many other multitarget TKIs (mTKI) developed to treat cancer, regorafenib has the ability to inhibit a range of protumoral tyrosine kinases across cell types—although with varying affinities. Originally developed as (B)RAF inhibitors, sorafenib/regorafenib are most commonly seen as antiangiogenic drugs that can also inhibit oncogenic signaling in cancer cells. While this collective, dose-dependent kinase targeting inside disparate cells raises questions about the true potential for and optimal application of TKIs in gastrointestinal cancers (9), the broad inhibitory spectrum offered by mTKIs could—at least in principle—be a partial answer to the complexity of cancer progression and immune evasion. Nonetheless, we have much to learn of the precise mechanism of action of mTKIs.

Finally, not only is the effect of regorafenib on CAFs poorly understood, eliminating these cells from the TME is not without controversy, as CAF biology is even more complex than expected. Recent studies have identified elevated levels of heterogeneity. A number of CAF subtypes are coalescing across cancer types, most prominently the myofibroblast-like CAFs (myCAF)—contractile cells that produce and reorganize tumor extracellular matrix—and immunomodulatory CAFs (iCAF) that secrete various cytokines and chemokines. Moreover, CAFs may also have relevant tumor suppressor or immunostimulatory functions (10). Akiyama and colleagues used spatial transcriptomics to assess the effect of regorafenib on CAF heterogeneity (5). Among three discernable phenotypic CAF clusters, TKI treatment was found to strongly reduce one subtype that expresses iCAF markers and that appears responsible for the immunomodulatory secretory phenotype and PMN-MDSC recruitment. Although this analysis found that a cluster bearing myCAF markers was only mildly diminished by regorafenib treatment, immunofluorescent antibody staining of treated tumor samples indicated a more comprehensive reduction of cells positive for myCAF marker αSMA. Thus, this drug's potentially specific targeting of CAF subtypes or functions regarding immune regulation requires further investigation, including in treated patient biopsies.

In conclusion, the study by Akiyama and colleagues describes a key role for PDGF-stimulated CAFs in gastric cancer progression and immune dysfunction, conferring resistance to T-cell–mediated immunity. Moreover, this work adds relevant insight on the mode of action of regorafenib in targeting CAFs and overcoming immune evasion, increasing our understanding of a therapeutic strategy currently under investigation for gastrointestinal cancers (8) and suggesting PDGF–PDGFR levels as potential biomarkers for response. Future studies should focus on addressing the remaining questions about CAF heterogeneity and the associated therapeutic implications (10), the immunosuppressive mechanism employed by PMN-MDSCs in this context, and whether this treatment combination will succeed in overcoming resistance to checkpoint inhibition in the clinic.

D.V.F. Tauriello reports a patent for US20220089600A1 pending. No other disclosures were reported.