We addressed the potential limitations of conventional observational epidemiology by utilizing pQTL and eQTL instruments as IVs. Here, we comprehensively evaluated whether circulating levels of several inflammation-related cytokines had a causal effect on BCa risk. These consistent results provide strong evidence in favor of an inverse association between Eotaxin and increased BCa risk. Our findings support the hypothesis that cytokines are potential targets for therapeutic strategies against BCa.
Cytokines are functional proteins that are considered modulators of immune response and inflammation. Accumulating studies are currently available to support the association between dysregulated cytokines and the promotion or inhibition of cancer development29. Cytokine-based therapies have been extensively investigated as potential cancer treatments30. Therefore, it is critical to identify the cytokines that are causally associated with tumors. Previous MR studies have investigated the association between cytokines and different types of cancer. One MR analysis was performed to evaluate the causal inference of 27 cytokines and growth factors on prostate cancer risk and indicated a risk-increasing effect of genetically proxied MCP-1 concentrations on prostate cancer31. Other MR analyses have identified the causal influences of several cytokines on the risk of different cancers, including breast, renal, endometrial, lung, and ovarian cancer20,32,33. However, no MR analyses exist to evaluate the relationship between cytokines and BCa.
Our MR analyses revealed an inverse association between genetically proxied Eotaxin concentrations and BCa risk. Eotaxin, also known as CCL11, is a potent eosinophil chemoattractant cytokine belonging to the CC chemokine family34. Eotaxin is a ligand of the G-protein-coupled receptor CC chemokine receptor (CCR)3 with the highest affinity and is also a ligand of CCR5 but an antagonist of CCR235. CCL11 recruits eosinophils, macrophages, and T helper 2 cells, which play multiple functions and promote the T helper 2-dominant tumor environment through the Eotaxin-CCR3 interaction36,37. Owing to the expression of CCR3 in vascular endothelial cells, Eotaxin directly increases angiogenic activity38. In addition, CCL11 indirectly facilitates angiogenesis by increasing vascular endothelial growth factor expression through the CCL11-CCR3 interacion39. CCL11 promotes the survival of lymphoma cells and facilitates their invasion in prostate cells by activating the CCR3-extracellular signal-regulated kinase pathway40,41. CCL11 also participates in myeloid-derived suppressor cells-induced metastasis by activating extracellular signal-regulated kinase/protein kinase B signaling and inducing epithelial-mesenchymal transition (EMT) in lung cancer42. Eotaxin promotes the migration and invasion of head and neck cancer cells by enhancing cancer stem cell-like properties and inducing EMT43. In addition, Eotaxin exhibits antitumor activity by recruiting eosinophils and inhibiting angiogenesis, which results in necrosis of some parts of the tumor44. Only a few original studies have focused on the concentration of Eotaxin in cancer, particularly as a predictive biomarker, which attests to the fact that the potential of these parameters is not fully understood. Several studies have reported associations between elevated serum Eotaxin levels and cancer outcomes, indicating an essential role of Eotaxin signaling in cancer prognosis. Increased Eotaxin levels have been observed in patients with breast, colorectal, and prostate cancers45–48. Elevated CCL11 expression correlates with an increased risk of gastric cancer49. There is currently no epidemiological or experimental evidence examining circulating Eotaxin levels in patients with BCa, and expanding the knowledge of Eotaxin during cancer development is necessary.
We further performed a single-cell analysis to explore the potential mechanism of action of Eotaxin in BCa and found that Eotaxin (CCL11) was predominantly expressed in fibroblasts. Fibroblasts seem to have a bimodal influence on cancer cells in that they initially prevent malignant transformation; however, as the cancer progresses, they promote tumor growth50. The tumor microenvironment, a complex component divided into immune and non-immune components, is regarded as the soil for cancer cells51. CAFs are the most abundant non-immune components of the tumor microenvironment and play complex and multifaceted roles during tumor development via diverse mechanisms52. In addition to directly interacting with epithelial cells, CAFs secrete various factors, such as growth factors, collagen, cytokines, and chemokines, which interact with immune components, support tumor cell growth, and remodel the extracellular matrix, thereby interfering with key aspects of cell proliferation, invasion, and angiogenesis53–55. CAFs also generate lipids and metabolites which are taken up by surrounding cancer cells for tumor growth in vivo by activating their respective signaling pathways, i.e., WNT/β-catenin, phosphoinositide 3-kinase / mammalian target of rapamycin or Hippo pathway56,57. The association between CAFs and BCa has been extensively studied. CAFs induce EMT by secreting IL-6 and thus promote aggressive phenotypes in non-muscle-invasive bladder cancer cells58. In vitro, the crosstalk between CAFs and BCa cells contributes to enhanced cell migration and invasion59. Tumor growth factor β1 secreted by CAFs induces EMT and the in vitro invasion through the TGFβ1-ZEB2NAT-ZEB2 axis60. Solute carrier family 14 member 1+ (SLC14A1+) CAFs promote cancer stemness in BCa cells via the WNT5A paracrine pathway in BCa cells61. A recent study revealed that the high activity of CAF-related genes significantly facilitates the progression and shorter prognosis of BCa62. Compared with the extensively studied tumor-promoting functions, tumor-suppressive CAF functions are less well described. In inflammation-driven model of colorectal cancer, loss of IKKβ in Col6 + fibroblasts hampered tumor growth63. In addition, CD105-negative fibroblasts are highly tumor suppressive in pancreatic cancer64.
We further found that Eotaxin was predominantly expressed in iCAFs but not in mCAFs. Eotaxin expression was significantly higher in normal tissue iCAFs than in BCa iCAFs. The iCAF, which were activated by paracrine factors from cancer cells, were first identified with α-SMAlow IL-6high inflammatory features in pancreatic cancer by Öhlund et al.65. IL-6, IL8, IL-11, CXCL1, CXCL2, and CXCL12 were identified highly expressed and hyaluronic acid synthase 1 (HAS1) and HAS2 were specifically expressed in iCAFs65,66. iCAFs, but not mCAF, are associated with tumor progression and poor prognosis in BCa27. Intercellular adhesion molecule 1+ (ICAM1+) iCAFs are also associated with recurrent BCa67. iCAFs are thought to facilitate tumor progression. However, we found that CCL11 + iCAFs do not promote tumor progression in the same way that other iCAFs do. CCL11 + iCAFs may be a subtype of iCAFs that inhibit tumor progression. The potential functions of iCAFs, particularly CCL11 + CAFs, in BCa have not been elucidated.
Our study has several strengths. The main strengths of this MR study are the wide panel of inflammatory cytokines and the large sample size. The likelihood of horizontal pleiotropy may depend on causal estimates. In our analyses, we used only cis-IVs, which naturally correlate with the corresponding gene more than other genetic variants, for reducing the potential of horizontal pleiotropy. Since cis-pQTLs or cis-eQTLs near the genome position of the corresponding gene influence protein expression, and most drug targets are proteins, MR analyses using cis-pQTLs or cis-eQTLs are likely to have translational relevance.
This study has several limitations. Although the cis-instrument definition approach was used, circulating cytokine concentrations may poorly mimic local cytokine expression. Focusing solely on circulating cytokine concentrations may result in different causal relationships; thus, determining the relationship between circulating cytokines and local tissue expression is necessary. Another important point is that even though a relaxed significance threshold was used for selecting IVs, a small number of IVs, due to the cis-eQTL definition we used, reduced the statistical power. In addition, some important cytokines were not included in our study because cis-eQTLs were unavailable for these cytokines, reducing the statistical power. Future larger-scale and multi-center genome-wide association studies is needed to address these issues. Because we evaluated the causal inference of circulating cytokine concentrations on BCa risk with a linear assumption, nonlinear or time-dependent effects could not be ruled out in the present study. Finally, the participants in this MR analysis were of European ancestry, which restricted generalization to other populations of different ethnicities. Genome-wide association study data from other races are needed to further verify the generalizability of our results.
In conclusion, we highlighted the inverse association between high level of circulating Eotaxin and increased BCa risk. We further found that Eotaxin (CCL11) was predominantly expressed in iCAFs but not in mCAFs, and CCL11 expression was significantly higher in normal tissue iCAFs than in BCa iCAFs. CCL11 + iCAFs may be a subtype of iCAFs that inhibit tumor progression, though further studies are needed to investigate the potential mechanisms of Eotaxin as a drug target for cancer prevention.