Upon exposure or injury, inflammation is first observed, whereas fibrosis and cancer are typically chronic manifestations. Inflammation is defined as the tissue response to harmful stimuli, such as injury, pathogens, and irritants. The purpose of this response is to eliminate insults, mitigate lesions, and repair damaged tissue, which is accomplished mainly through the actions of innate immune cells, blood vessels, and molecular mediators. The initial event of an inflammatory response to a harmful stimulus is characterized by the increased movement of plasma and leukocytes (especially granulocytes) from the blood into injured tissue. This acute innate immune response to injury is highly conserved through evolution and is reminiscent of the immediate rapid massing of inflammatory cells in starfish upon stimulation by a splinter observed by Ilya Ilyich Mechnikov in the late 19th century (Kaufmann 2008). Acute inflammation propagates through cascades of events involving the local vasculature, immune functions, and various cells within the injured tissue. If injury or infection becomes persistent or the lesion is overwhelming and exceeds the capacity of tissue repair, the wound fails to heal and inflammation propagates to a chronic state. During chronic inflammation, there is a progressive shift in the type of cells present at the site of inflammation, often marked by increasing mononuclear cells, and the simultaneous destruction and healing of the tissue from the inflammatory process. In this regard, chronic inflammation may reflect a homeostatic imbalance of physiological systems and functions resulting from the malfunction of tissue, rather than a direct response to classical instigators of inflammation, such as injury and infection. Importantly, chronic inflammation is causatively associated with or is an integral component of many chronic diseases, exemplified by chronic infection, autoimmune dysfunction, atherosclerosis, neurodegeneration, and, pertinent to this review, fibrosis, and malignancy.

Fibrosis is a common pathologic outcome of many chronic inflammatory diseases. In these scenarios, inflammatory and fibrogenic signals interact with each other to stimulate an inflammatory wound healing process, which goes awry and evolves into a progressive and irreversible fibrotic process, if the injury is severe or repetitive, or when the repair response becomes dysregulated. Diseases in which fibrosis is a major cause of morbidity and mortality encompass both organ-specific and multi-systemic illnesses. These include chronic renal disease from infection and diabetes; liver fibrosis and cirrhosis from viral and parasitic infection, alcoholic and non-alcoholic hepatitis, and drug-induced liver injury; pulmonary fibrosis from idiopathic pulmonary fibrosis (IPF), chronic obstructive pulmonary disease (COPD), and inhalation of fibrogenic particles; myocardial infarction; and systemic autoimmune diseases, such as lupus and systemic sclerosis. Despite this vastly diverse etiology and clinical presentation, fibrosis in many diseases appears to follow a common path of development, wherein fibrogenic pathways converge to boost the activation and migration of fibroblasts, and the differentiation of fibroblasts to myofibroblasts (Dong and Ma 2016b; Duffield et al. 2013). Myofibroblasts are rich in cellular machineries for protein synthesis and secretion, which enables cells to produce copious amounts of collagens in fibrosing tissues. Myofibroblasts also synthesize α-smooth muscle actin (α-SMA) that incorporates into the contractile stress fibers to strengthen contraction by myofibroblasts during scar formation (Dong and Ma 2016b; Tomasek et al. 2002).

The influence of inflammation on fibrosis is multifold and is inducer-, time-, and context-dependent (Dong and Ma 2018b; Gieseck, Wilson, and Wynn 2018; Wynn and Ramalingam 2012). Upon infection and wounding, inflammatory cells and injured tissue cells secrete soluble mediators, among which transforming growth factor (TGF)-β1 appears to serve as a common mediator of fibro-blast activation and transformation. In the presence of bacterial or viral infection, toxins, or biliary obstruction, type 1 inflammation is elicited where macrophages, typically, classically activated (M1) macrophages, and neutrophils secrete interleukin(IL)-1 and IL-6, which in turn, induce IL-17. IL-17 signaling then stimulates the production and activation of TGF-β1, forming an IL-1—IL-17—TGF-β1 axis to stimulate sustained tissue repair or organ fibrosis. On the other hand, type 2 inflammation ensues upon exposure to allergens, helminths, fungi, or fibrogenic particles and fibers. In these scenarios, alarmins, such as IL-25, IL-33, and thymic stromal lymphopoietin (TSLP), are produced and released by injured structural cells, such as epithelial cells. Alarmins recruit type 2 inflammatory cells, such as basophils, eosinophils, mast cells, and type 2 innate lymphoid cells (ILC2s), which provide the early production of type 2 cytokines, IL-4 and IL-13. IL-4 and IL-13 induce the formation of T helper (Th) 2 cells that amplify type 2 inflammation by producing type 2 cytokines. IL-13 appears to serve as a major signal to stimulate fibroblast activation and the fibroblast-to-myofibroblast transformation to result in fibrosis. In all cases, fibrosis and granuloma formation reflect the maladaptation of tissue to chronic injury and infection, the purpose of which is to preserve tissue integrity and limit invading pathogens or foreign bodies to a local environment, which is, albeit, at a considerable expense of tissue functions that are lost due to tissue destruction and distortion from fibrosis.

Tumor-promoting inflammation has been recognized as one of the enabling characteristics of tumor development (Hanahan and Weinberg 2011). Chronic inflammation associated with infection or prolonged exposure to environmental insults often precedes and contributes to tumor development by several means, such as oncogenic mutations, genomic instability, and angiogenesis; tumor progression and metastasis; and immunosuppression (Coussens and Werb 2002; Grivennikov, Greten, and Karin 2010; Shalapour and Karin 2015). It is estimated that approximately 25% of the cancers are associated with chronic inflammation caused by infection or physicochemical insults (Balkwill and Mantovani 2012). For example, patients suffering from chronic inflammatory bowel diseases, such as ulcerative colitis and Crohn’s disease, have a 10-fold increased risk of colon cancer (Coussens and Werb 2002). Persistent gastritis caused by Helicobacter pylori increases the risk of gastric cancer by 75% (Eiro and Vizoso 2012), whereas hepatitis B virus (HBV) and hepatitis C virus (HCV) infections are strongly associated with the formation of hepatocellular carcinoma (Ringelhan, McKeating, and Protzer 2017).

At cellular and molecular levels, pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs) are known to be released from microbes and dying cancer or tissue cells, respectively. These initial signals activate myeloid cells that are recruited to the site of the tumor through the action of chemokines, resulting in local inflammation. On the other hand, inflammation promotes tumor initiation by increasing the production of reactive oxygen species (ROS) and reactive nitrogen species (RNS) from inflammatory cells to damage DNA, proteins, and organelles, and by inducing epigenetic changes, both of which favor tumorigenesis. Inflammation also stimulates tumor promotion through multiple means. For instance, tumor-associated macrophages (TAMs) and neutrophils (TANs) secrete cytokines, such as tumor necrosis factor (TNF)-α, IL-1, and IL-6. These pro-inflammatory cytokines evoke inflammatory responses, and act directly on tumor cells leading to the activation of NF-κB (nuclear factor-jB), STAT3 (signal transducer and activator of transcription 3), YAP (Yes-associated protein), and Notch signaling pathways, thereby promoting tumor cell survival and proliferation (Shalapour and Karin 2015). TNF, as its name implies, also kills tumor cells, highlighting a complex interrelation between inflammatory cytokines and tumor cells (Carswell et al. 1975).

Rapidly growing tumors create a hypoxic condition that activates cancer-associated fibroblasts (CAFs) through a hypoxia-inducible factor (HIF)-1-induced TGF-β signaling axis. In this scenario, TAMs may convert from type 1 (TAM1, pro-inflammatory) to type 2 macrophages (TAM2, anti-inflammatory and pro-fibrotic). TAM2 cells produce vascular endothelial growth factor (VEGF) to foster neoangiogenesis both surrounding and within the tumor tissue to support rapid tumor growth. Lymphocytes also play roles in tumor formation. CAFs secrete TGF-β and CXCL13 that recruit lymphotoxin-producing B2 lymphocytes to further support tumor growth, whereas chemokines produced in the tumor recruit tumor-promoting Th17 cells and immunosuppressive regulatory T cells (Tregs). Finally, tumor-infiltrating B cells may undergo class-switch recombination to induce an exhausted or anergic phenotype in cytotoxic T cells (Shalapour and Karin 2015). On the other hand, acute inflammatory reactions often stimulate dendritic cell maturation and antigen presentation that boost anti-tumor immunity, which may be targeted for immunotherapy against cancer. These findings indicate that chronic inflammation influences tumorigenesis and tumor progression and metastasis at several levels via multiple means.

Accumulating evidence supports a close relationship between fibrosis and cancers. It has long been established that certain tumors can arise where scars are formed, giving rise to the term scar carcinoma. Scar carcinoma is most prominent in patients with pneumoconiosis including asbestosis, silicosis, and coal worker’s lung disease (Davis and Cowie 1990; Doll 1955). In other examples, certain fibrotic diseases are associated with increased risks of certain cancers. For instance, patients suffering from cystic fibrosis have an odds ratio of 6.5 for developing digestive tract cancers compared with the general population in North America (Neglia et al. 1995). Nonetheless, a causal relationship between fibrosis and cancer has long been debated, in particular, with regard to whether desmoplasia, which denotes the growth of fibrous connective tissue characteristically associated with malignant neoplasms, precedes, accompanies, or succeeds tumor initiation, progression, and metastasis.

In many tumors, the tumor-associated ECM is strikingly different from the ECM of normal tissues, which, together with infiltrated inflammatory cells and structural epithelial, endothelial, and mesenchymal cells, forms the tumor microenvironment that modulates cancer initiation, progression, and metastasis (Cox and Erler 2016). The initial desmoplastic response to primary tumors by the host tissue is believed to be a defensive one and represents a balance between tumor-promoting and tumor-suppressing cues. This balance eventually tips toward a pro-tumorigenic and pro-metastatic environment. In this scenario, activation of myofibroblasts leads to excessive production of collagen and increased bio-mechanical stiffness, whereas the release of growth factors and cytokines stimulates angiogenesis and cell growth, both of which allow tumor cells to grow unchecked and to invade and metastasize freely. By similar means, fibrotic remodeling at remote sites creates a pro-metastatic microenvironment or niche that favors and supports the colonization of circulating tumor cells or activation of dormant resident tumor cells. These interactions between tumor cells and their stromal microenvironment also affect the response to anti-tumor therapy and, therefore, represent important targets for therapeutic intervention against cancer.

Untransformed human cells would undergo cell cycle arrest with reduced proliferation and mitogenic signaling, and cessation of cell movement if cell density reaches a certain level, a phenomenon known as contact inhibition of proliferation and locomotion of cells that contact each other. This cell contact inhibition is essential to embryonic development, morphogenesis, and tissue repair. In these scenarios, cell-cell contact is essential for contact inhibition but is not sufficient for inhibition of mitotic division of contacting cells. The contact-inhibited cells must also be forced to reduce their cell area under the mechanical stress and constraints for mitotic inhibition to occur. These mechanical constraints are largely imposed by surrounding cells and the ECM (Shraiman 2005). In cancers, transformed cells typically lose contact inhibition and, therefore, would divide and grow over each other in an uncontrolled manner, even when in contact with neighboring cells. This lack of contact inhibition in transformed or cancerous cells is necessary for tumorigenesis, invasion of tumor cells into surrounding tissues, and metastasis of tumor cells to distant organs. The mechanism by which cancerous cells become insensitive to contact inhibition remains largely unclear (Ribatti 2017). Both biochemical and physical mechanisms are involved, including interactions of tumor cells with the matrix. How the tumor microenvironment, chronic inflammation, and fibrosis contribute to the loss of contact inhibition in tumor cells for both mitotic division and locomotion is an intriguing question. In this regard, increased stiffness of the matrix is found to serve as a scaffold to promote tumor cell migration through a process called duro-taxis (Lo et al. 2000), which boosts tumor invasion and metastasis in stiff tissue (Friedl and Wolf 2010).

Pulmonary exposure to CNTs induces a rapid-onset fibrotic response that takes place alongside acute inflammation, forming prominent inflammatory and fibrotic foci during the early phase response. The lesions are characterized by significantly elevated deposition of collagen fibers and increased cellularity consisting of infiltrated inflammatory cells, including a large number of macrophages, with some containing engulfed CNTs, and activated fibroblasts and myofibroblasts, interspersed with CNT fiber singlets, bundles, and aggregates. As CNTs persist in the lung, acute inflammation is resolved to a large extent, but not completely. Instead, it is converted to chronic inflammatory phenotypes, which are relatively mild in intensity but persist throughout the chronic course, leading to progressive chronic fibrosis, indicated by interstitial thickening and formation of fibrotic foci and granulomas, in most animal models. A similar pathogenic process is believed to take place for the development of inflammation and fibrosis in mesothelial tissues, as both mesothelial inflammation and mesothelial fibrotic lesions are consistently observed in animal models with or without mesotheliomas (Table 2). This co-existence and co-development between inflammation and fibrosis in the lung and mesothelial tissues during both the acute and chronic phases indicate a close interaction between inflammation and fibrosis with respect to pathogenic effect development.

The interplay between inflammation and fibrosis plays an important role in the development of a variety of fibrotic diseases. These include diseases caused by infection of bacteria, viruses, fungi or parasites, lesions from toxic chemical agents, such as bleomycin and paraquat, and deposition of fibrogenic foreign bodies, exemplified by silica and asbestos (Borthwick, Wynn, and Fisher 2013; Eming, Wynn, and Martin 2017; Greenberg, Waksman, and Curtis 2007; Lupher and Gallatin 2006; Meneghin and Hogaboam 2007; Mossman and Churg 1998; Pourgholamhossein et al. 2018; Rom, Travis, and Brody 1991; Wick et al. 2010; Wynn and Ramalingam 2012; Xie et al. 2012). In these scenarios, the inciting agents and lesions recruit innate immune cells and activate inflammatory responses. In these inflammatory responses, excessive amounts of pro-inflammatory and pro-fibrotic cytokines, chemokines, growth factors, and ROS are produced and released. The molecular pathways that are activated and mediate these early inflammatory responses may differ, depending on the nature, time, and context of stimulation. For example, activation of the NLRP3 (nucleotide-binding oligomerization domain-like receptor: pyrin domain-containing 3) inflammasome is commonly observed and is required for the maturation and secretion of several pro-inflammatory cytokines, in particular, the pleiotropic IL-1β and IL-18, during acute inflammation (Borthwick 2016; Coll et al. 2015; Hornung et al. 2008). On the other hand, the chronic progression of inflammation and fibrosis depends on pro-fibrotic mechanisms, such as IL-13 and IL-17 signaling cascades (Gieseck, Wilson, and Wynn 2018). These mechanisms exert pathogenic activities to promote fibrosis initiation and progression via multiple means, such as stimulating fibroblast proliferation and inducing fibro-blast-to-myofibroblast differentiation. Reciprocally, the enriched and activated fibroblasts and myofibroblasts during fibrotic responses can produce large amounts of cytokines, chemokines, and ECM proteins that promote the recruitment and activation of immune cells and, thereby, boost and prolong inflammation leading to chronic inflammation (Buckley et al. 2001; Flavell et al. 2008; Kendall and Feghali-Bostwick 2014; Phan et al. 1999). A number of cellular processes and molecular mediators have been identified to mediate the communication between inflammation and fibrosis in human fibrotic diseases and animal models. Some of these factors are induced by CNT exposure and impact the development of CNT-induced inflammation and fibrosis (Figure 3).

The migration and infiltration of inflammatory cells occur immediately upon CNT exposure and become pronounced within 1 day post-exposure in the lung (Dong et al. 2015; Porter et al. 2010; Shvedova et al. 2005). Infiltrated macrophages constitute the first line of innate immune defense to inhaled CNTs through phagocytosis. Because of their small sizes, most CNTs can be phagocytosed by macrophages. However, phagocytosed CNTs cannot be degraded or are degraded only to a limited extent in macrophages, due to their graphene structure and sp² bonds. For elongate CNTs with a fiber length larger than 15 μm, the length of CNTs exceeds the maximal diameter of macrophages and, as such, these CNTs cannot be phagocytosed easily. In either scenario, frustrated phagocytosis ensues and the involving macrophages release inflammatory mediators, which would amplify inflammation. Elevated inflammatory infiltration and the release of associated mediators would increase the defensive activities against CNT deposition; but, at the same time, these intensified inflammatory events would cause damage to the tissue substantially (Rothen-Rutishauser et al. 2010). Studies have shown that exposure to SWCNTs or MWCNTs commonly induces the production and secretion of cytokines, chemokines, growth factors, and ROS from macrophages in vivo and in vitro, which has been highlighted in a few recent reviews (Dong and Ma 2015; Dong and Ma 2016b; Duke and Bonner 2018; Vietti, Lison, and van den Brule 2016).

CNTs induce the recruitment and infiltration of neutrophils, as well as T and B lymphocytes, in exposed lungs (Dong and Ma 2016c; Dong et al. 2015; Rydman et al. 2015). Notably, CNTs activate the NLRP3 inflammasome to regulate the maturation and secretion of IL-1β and IL-18 in mouse lungs to promote acute inflammation (Rydman et al. 2015), whereas Th2 cell-mediated type 2 immune responses are activated to produce type 2 mediators, such as IL-4, IL-5, IL-13, and TGF-β1 (Dong and Ma 2016a; Dong and Ma 2018a; Dong and Ma 2018b). Some of the factors induced by CNTs are pro-fibrotic and play critical roles in the initiation and propagation of fibrosis, providing a mechanism for the interaction between inflammation and fibrosis in CNT-exposed animals. Functionally, NLRP3—IL-1β pathway is clearly required for the acute inflammatory response, as suppression of the pathway by knocking out the gene encoding IL-1 receptor 1 (IL-1R1) or inhibiting IL-1R1 signaling using anakinra attenuated neutrophilia significantly (Nikota et al. 2017; Rydman et al. 2015). Blockade of the NLRP3—IL-1β pathway did not seem to affect the pro-fibrotic IL-4 and IL-13 signaling. On the other hand, knockout (KO) of STAT6 in mice suppressed both the acute inflammation and the fibrotic lesions induced by CNTs in comparison with wild-type mice (Nikota et al. 2017). These findings support the sequential events in the development of fibrosis (Dong and Ma 2018b), which are sometimes summarized as an adverse outcome pathway (AOP) of lung fibrosis induced by CNTs for risk evaluation of CNT pulmonary fibrogenicity (Labib et al. 2016; Nikota et al. 2017; Vietti, Lison, and van den Brule 2016).

TGF-β1 is a major endogenous pro-fibrotic mediator in most human fibrotic diseases and animal models of fibrosis. Many fibrotic signaling pathways, such as the Th2-driven type 2 inflammatory signaling and the IL-1—IL-17 signaling, converge to induce TGF-β1 expression and activate latent TGF-β1 in tissue. As a negative feedback regulation, TGF-β1 inhibits Th responses by inhibiting T cell proliferation. During type 2 immune responses, M2 macrophages are a major source of TGF-β1 production (Murray and Wynn 2011; Wynn and Ramalingam 2012). TGF-β1, in turn, exerts multiple pro-fibrotic effects on fibroblasts and myofibroblasts, which are the major effector cells for fibrotic matrix deposition, remodeling, and contraction. These include promoting fibroblast proliferation, stimulating fibroblast-to-myofibroblast differentiation, and inducing the expression of ECM proteins (Dong and Ma 2016b; Hinz et al. 2012; Leask and Abraham 2004; Tomasek et al. 2002; Wynn and Ramalingam 2012). Upon exposure to CNTs, TGF-β1 is substantially elevated in the lung and in cultured macrophages (Dong and Ma 2016b; Dong and Ma 2017a; Dong and Ma 2018b).

Knockout of TGF-β1 in mice is lethal by the age of 2–4 weeks due to excessive inflammatory responses in organs such as the heart and the lung (Christ et al. 1994; Kulkarni and Karlsson 1993). As such, it is difficult to investigate the effect of TGF-β1 on organ fibrosis by knocking out or knocking down TGF-β1 in animals. Similarly, direct inhibition of TGF-β1 with neutralizing antibodies or chemical inhibitors in vivo may induce side effects, such as inflammation and abnormalities, and, therefore, is not effective (Akhurst and Hata 2012; Mallat et al. 2001). Since TGF-β1 is expressed in multiple cell types, including macrophages, fibroblasts, and epithelial cells, conditional KO of TGF-β1 in a specific cell type may not be sufficient to reduce the overall level of TGF-β1 in vivo and, therefore, is not a preferred approach either. For these reasons, assessing the fibrogenic effect of modulating TGF-β1 expression and activity has been challenging.

The OPN (osteopontin, secreted phosphoprotein 1 or Spp1) KO mouse strain provides a partial solution regarding TGF-β1 for the study of CNT-induced lung fibrosis. OPN is both a cytokine and an ECM protein that regulates a number of important biological processes, including inflammatory cell infiltration, tissue remodeling, and organ fibrosis (Berman et al. 2004; O’Regan 2003; Rittling and Singh 2015; Takahashi et al. 2001). In relation to TGF-β1 functions in fibrosis, OPN was markedly induced by CNTs in the lung and modulated a number of functions via TGF-β1 (Dong and Ma 2017a; Khaliullin et al. 2017; Thompson et al. 2015). On days 7 and 28 post-exposure to MWNT-7, the levels of TGF-β1 protein in lung tissue and lung macrophages were significantly lower in OPN KO mice than those in wild-type (WT) mice (Dong and Ma 2017a). Accordingly, activation of Smad-dependent TGF-β1 signaling was significantly attenuated in lung cells, including fibroblasts and myofibroblasts, in OPN KO lungs, compared with WT. Decreased levels of TGF-β1 protein and impaired TGF-β1 signaling correlated with a reduction in fibrosis in OPN KO lungs exposed to MWCNTs, indicated by reduced fibrotic focus formation, fewer fibroblasts and myofibroblasts, and less ECM deposition, in comparison with WT. This study, therefore, demonstrates a critical role of TGF-β1 in promoting the development of fibrosis induced by MWCNTs, which is regulated by OPN. In a separate study, a two-fold induction of TGF-β1 protein in the bronchoalveolar fluid (BAL) from WT lungs exposed to SWCNTs for 7 days was observed. On the other hand, induction of TGF-β1 protein was undetectable in the BAL from OPN KO lungs exposed to SWCNTs for 7 days, which correlated with decreased deposition of collagen in OPN KO lungs on day 28 post-exposure (Khaliullin et al. 2017). These two studies using OPN KO mice provide evidence supporting TGF-β1 as a signaling mediator to link lung inflammatory responses with fibrosis development induced by CNT exposure.

In a reciprocal manner, fibrotic conditions exhibit a strong propensity to boost and propagate inflammatory responses and promote the conversion of acute inflammation to chronic inflammation characteristic of chronic fibrosis, such as the formation of granulomas. During pathologic fibrosis development, fibroblasts and myofibroblasts are markedly enriched through fibroblast recruitment and proliferation and the fibroblast-to-myofibroblast differentiation. These cells function as the major effector cells for fibrosis development, as they mediate the excessive ECM production and remodeling of injured tissues (Katzenstein and Myers 1998; Tomasek et al. 2002; White, Lazar, and Thannickal 2003; Wynn 2008; Wynn and Ramalingam 2012; Zhang et al. 1994). Nevertheless, the effects of the elevated fibroblastic cell activities go beyond ECM reorganization and tissue contraction. In this respect, activated fibroblasts and myofibroblasts synthesize and secrete cytokines, chemokines, and growth factors, as well as express cell surface receptors, constitutively and through induction. Therefore, these cells exhibit certain features of inflammatory cells and have the capability to respond to immune and ECM signals (Finlay et al. 2000; Heino et al. 1989; Mezzano et al. 2000; Phan et al. 1999; Thannickal, Aldweib, and Fanburg 1998b). Myofibroblasts also produce and release ROS and RNS, leading to oxidative stress (Sambo et al. 2001; Sugiura et al. 2006; Thannickal, Aldweib, and Fanburg 1998a; Thannickal and Fanburg 1995; Wu et al. 2013). These pro-inflammatory mediators and ECM proteins generated by fibroblastic cells during fibrosis can, in turn, create a milieu that fosters persistent inflammation through recruiting and activating immune cells, leading to chronic inflammation in fibrotic tissues.

CNT exposure has been demonstrated to trigger significant increases of fibroblasts and myofibroblasts during the acute and chronic phase responses in the lung, especially in fibrotic foci where CNT fibers deposit and aggregate (Dong and Ma 2016b). Meanwhile, in a few studies performed in cultured fibroblastic cells, the fibroblast-to-myofibroblast transformation was observed when primary lung fibroblasts or fibroblast cell lines were treated with CNTs (Dong and Ma 2016b; Dong and Ma 2017a). Furthermore, certain pro-inflammatory mediators were induced in and secreted by, cultured fibroblastic cells when exposed to CNTs, such as cytokines, ROS, and ECM molecules (Alarifi and Ali 2015; Dong and Ma 2016b; Dong and Ma 2017a). As an example, OPN was induced by MWNT-7 in cultured mouse primary fibroblastic cells and was at high levels in fibrotic and granulomatous foci in mouse lungs during the chronic stage, i.e., on day 28 and day 56, post-exposure to MWNT-7 (Dong and Ma 2017a). In another study, a different type of MWCNTs induced high levels of OPN in granulomatous foci, where increased recruitment of macrophages and CD3+ T cells was concurrent, in mouse lungs on day 60 and day 90 post-exposure (Huizar et al. 2011). Thus, OPN may function as a molecular mediator in fibrosis-promoted chronic inflammation.

Altogether, previous findings support the notion that CNTs stimulate multiple signaling pathways in time and context-dependent manner, which are principally inflammatory in the early phase and fibrotic in the later. These pathways constitute the underlying mechanisms that interconnect inflammation and fibrosis induced by CNT exposure in the lung (Figure 3).

A close interplay between inflammation and cancer has been recognized in many types of malignancy (Coussens and Werb 2002; Coussens, Zitvogel, and Palucka 2013; Mantovani et al. 2008). The identified interactions involve both intrinsic (tumor cell-initiated) and extrinsic (non-tumor cell initiated) pathways in tumorigenesis. Under chronic inflammatory conditions, cancer initiation and progression are promoted by extrinsic pathways driven by cytokines, chemokines, and growth factors secreted by immune and inflammatory cells. Conversely, tumor cells cause and boost inflammation through intrinsic pathways activated by genetic and molecular events in tumor cells. Altogether, these events render cancer-associated inflammation to serve as a critical hallmark and an enabling factor for tumorigenesis, tumor promotion, and metastasis. Cancer-related inflammation is characterized by a number of activities, including the activation of transcription factors NF-κB, STAT3, and HIF-1α; infiltration of immune cells, such as macrophages, neutrophils, and mast cells; and production of functional molecules, such as cytokines TNF-α, IL-1β, and IL-6, and chemokines CCL2, CXCL1, and CXCR4. These events are capable of initiating and promoting cancer by enhancing cell proliferation and survival, angiogenesis, and tumor cell invasion and metastasis. In the case of occupational and environmental exposure-induced human diseases, such as asbestosis and silicosis, chronic inflammation is commonly observed and is a critical factor in determining the development of mesothelioma and lung carcinoma (Matsuzaki et al. 2012; Mossman et al. 1990; Mossman and Churg 1998; Otsuki et al. 2016).

In response to CNT exposure, inflammation occurs immediately and prolongs over the entire pathologic stage, as discussed above. It is plausible that inflammation precedes tumorigenesis, and tumorigenesis and inflammation co-occur during the chronic stage after exposure. Indeed, a number of studies have demonstrated the co-existence of inflammation and malignancy in CNT-exposed animals, suggesting an association between inflammation and cancer induced by CNTs (Tables 1 and 2) (Kuempel et al. 2017; Sinis et al. 2018). For instance, when directly instilled into the pleural cavity of mice, long MWCNTs with the length of 85% fibers more than 15 μm, which are similar to long fiber amosite asbestos, resulted in chronic inflammation and mesotheliomas in the pleura (Chernova et al. 2017). This study revealed the induction of several markers of cancer-associated inflammation by long MWCNTs, including elevated mRNA levels of the Il6 and the Stat3 genes in CNT-damaged areas, the increased protein level of phosphorylated STAT3, and augmented cell proliferation and oxidative DNA damage, providing direct evidence supporting an interplay between inflammation and cancer. In vitro studies using mesothelial cells indicated that CNTs have a propensity to induce cancer-associated inflammation through intrinsic pathways. Short-term (up to 24-hour) exposure to SWCNTs induced ROS production, DNA damage, and NF-κB activation in human normal and malignant mesothelial cells in a dose-dependent manner (Pacurari et al. 2008). In another study, 4-month chronic exposure to SWCNTs or MWCNTs resulted in increased proliferation, migration, and invasion of human pleural mesothelial MeT5A cells, and elevated expression and activity of matrix metalloproteinase (MMP)-2 (Lohcharoenkal et al. 2013). These findings indicate the tendency for CNTs to generate tumorigenic effects and promote cancer-associated inflammation. Altogether, the initiation of the multilayered genotoxic and inflammatory responses associates inflammation with tumorigenesis in response to CNT exposure, which governs tumorigenesis by CNTs, including the initiation, promotion, and metastasis of tumors in the lung and the chest pleura (Figure 4).

In response to tissue injury or fibrogenic agents and under many disease conditions, fibroblasts and myofibroblasts are enriched and activated to mediate the initiation and progression of tissue fibrosis. Pathologic fibrosis can also be elicited by tumor cells resulting from genetic insults. A close relationship between fibrosis and cancer has been highlighted in recent publications (Kalluri 2016; Ohlund, Elyada, and Tuveson 2014; Rybinski, Franco-Barraza, and Cukierman 2014; Schafer and Werner 2008; Yazdani, Bansal, and Prakash 2017). On one hand, certain factors generated during fibrosis are pro-tumorigenic and contribute to the development of cancer in fibrotic tissues. On the other hand, certain types of tumors induce the formation of fibrotic tissues, i.e. tumor stroma, in which fibroblasts, myofibroblasts, macrophages, and ECM proteins are predominant. The close association between cancer and stromal cells promotes cancer initiation, progression, and metabolism by multiple means and mediators including ROS, cytokines, chemokines, MMPs, and ECM proteins. In this scenario, myofibroblasts and macrophages are major effector stromal cells that interplay with tumor cells to promote tumorigenesis.

CNT-induced fibrosis occurs in fibrosis-prone organs, specifically, the lung, pleura, and liver, where tumors are also identified at a high frequency, supporting a possible interaction between fibrosis and certain cancers (Tables 1 and 2). For example, a single intraperitoneal injection of MWNT-7 (1 mg) induced fibrosis on the liver surface at 1-month post-exposure, whereas intraperitoneal injection of MWNT-7 at 10 mg for two times with 1-week interval resulted in malignant mesotheliomas on the liver surface at 1-year post-exposure, in rats (Nagai et al. 2011). A number of factors, such as TGF-β, PDGF, ECM, MMPs, and ROS, are known to be produced during fibrosis and are pro-tumorigenic. These factors are induced by CNTs in target organs, providing potential molecular links between fibrotic responses and tumorigenesis in CNT-exposed animals.

The tissue inhibitor of metalloproteinase 1 (TIMP1) was rapidly and highly induced in the lung by MWNT-7 in a time- and dose-dependent manner and high levels of TIMP1 were present in the BAL, serum, and lung tissue of MWCNT-exposed mice (Dong and Ma 2017b). Knockout of Timp1 in mice caused significant decreases in fibroblast accumulation, myofibroblast differentiation, collagen fiber deposition, and fibrotic focus formation in lungs exposed to MWNT-7, compared with WT. Importantly, exposure to MWNT-7 significantly increased the expression of the cell proliferation markers Ki-67 (marker of proliferation) and PCNA (proliferating cell nuclear antigen), as well as a panel of cell cycle-controlling genes, in the lung in a TIMP1-dependent manner. Meanwhile, MWNT-7 elicited a significant induction of CD63 and integrin β1 in lung fibroblasts and the formation of the TIMP1/CD63/integrin β1 complex on the surface of fibroblasts, which triggered the phosphorylation and activation of Erk1/2 that mediates cell proliferation (Dong and Ma 2017b). Notably, TIMP1 was identified as an important mediator of tumor cell proliferation and tumor growth with pro-proliferative activities in mouse lung cancer and glioblastoma multiforme models (Friedmann-Morvinski et al. 2016; Xia et al. 2012). The pro-proliferative role of TIMP1 in mouse lung cancer was shown to involve high-level ERK activation (Xia et al. 2012). These studies suggest a possible role of TIMP1 produced during the fibrotic response to CNTs in promoting CNT-induced lung cancer development.

CNTs of varying types commonly induce the excessive production and deposition of ECM proteins, in particular, collagens and fibronectin, markers of tissue fibrosis (Dong and Ma 2016b; Vietti, Lison, and van den Brule 2016). Increased deposition of ECM components can promote cancer progression in multiple ways, such as enhancing the release of growth factors, regulating the behavior of stromal cells, and promoting angiogenesis (Bonnans, Chou, and Werb 2014; Lu, Weaver, and Werb 2012). As discussed above, fibrosis and cancer co-occur in several organs in CNT-exposed animals, including the lung, pleura, and liver. In these scenarios, the newly synthesized fibrotic ECM from fibrosis creates a tumorigenic microenvironment that boosts tumorigenesis in CNT-deposited tissues, and tumor progression and metastasis in nearby and distal tissues, respectively. Notably, MWCNTs that induce severe fibrosis have a higher capability to induce tumorigenesis than those that cannot, or only slightly, induce fibrosis (Kuempel et al. 2017; Sinis et al. 2018). For instance, intraperitoneal injection of MWNT-7 induced fibrosis on the liver surface in rats 1-month post-exposure, whereas injection of tangled MWCNTs did not. Consistently, intraperitoneal injection of MWNT-7 or non-aggregated MWNT-7 resulted in the formation of malignant mesotheliomas on the liver surface in rats at 1-year post-exposure, but injection of tangled MWCNTs did not (Nagai et al. 2011). This study supports a promoting function of fibrosis in tumor development in CNT-exposed animals to some extent.

Nevertheless, the evidence that directly demonstrates the association between fibrosis and cancer in CNT pathology is currently lacking and awaits future studies, which is in part attributable to a lack of appropriate methodology and models for fibrosis and cancer that have long latencies and require prolonged exposure and development. The identification of the mediators that regulate the interplay between CNT-induced fibrosis and cancer would provide new insights into, and the therapeutic targets for, the prevention and treatment of fibrogenic particle and fiber-induced cancer development.

The interactions among inflammation, fibrosis, and cancer in disease development discussed above are largely based on animal studies. Extrapolation of these laboratory findings to human health is necessary, which is particularly true for evaluation of the adverse health effects of nanomaterials including CNTs, as the human effects of nano exposures remain largely to be defined at the present stage. The recent classification of MWNT-7 as ‘possibly carcinogenic to humans (Group 2B)’ and SWCNTs and all other MWCNTs (excluding MWNT-7) as ‘not classifiable as to their carcinogenicity to humans (Group 3)’ by IARC illustrates the utility and importance of this approach for risk evaluation and regulation of nano exposures where animal studies and mechanistic understanding were emphasized (Grosse et al. 2014). Nonetheless, considerable knowledge gaps exist in elucidating the mechanistic steps involved in CNT carcinogenicity, which hinders further risk assessment of tumorigenic CNTs (Kuempel et al. 2017). As has been demonstrated for many human cancers, the close interplay among inflammation, fibrosis, and cancer is an integral component of cancer development induced by CNTs and, therefore, represents a valuable mechanistic knowledge base for the risk assessment of CNT tumorigenesis. Several recent studies have employed end point-targeted pathway analyses, i.e. AOPs, to outline and integrate major steps toward the development of a specific endpoint phenotype (Labib et al. 2015; Nikota et al. 2017; Vietti, Lison, and van den Brule 2016). In these analyses, interactions among inflammation, fibrosis, and cancer were demonstrated as major events in the development of pulmonary fibrosis and cancer induced by CNTs.

Comparison between animal studies and human epidemiology (where available) is another useful and necessary step in translating animal data to human health. Since human studies of nano exposure and health impact are currently lacking, knowledge obtained from human diseases caused by particles and fibers, such as pneumoconiosis, lung cancer, and mesothelioma, is sometimes used as a comparison in the animal-to-human extrapolation of nanotoxicity. This approach is particularly useful for the analysis of the mode of action and disease pathogenesis, which is well illustrated in the demonstration of the fiber length paradigm for some adverse effects of nanotubes. However, a comparison between animal and human studies can be challenging. In part, this is due to the fact that human disease is often heterogeneous in terms of etiology and pathological and clinical phenotypes, and that epidemiological data are generally limited with regard to the size, genetic variability, and demographic distribution of the populations studied, the uncertainty in exposures (exposure agent, mixture, and the dose, duration, and route of exposure), long latency for many disease phenotypes, and numerous confounding factors, such as co-existing disease conditions and medications. Variations in association analysis of human data have been observed for particle- and fiber-induced diseases. A few examples of such variations are discussed below.

While the fiber pathogenicity paradigm has been well demonstrated in animal studies and is commonly supported by epidemiological studies for asbestos-induced mesotheliomas (Donaldson et al. 2010; McDonald et al. 1989; Rogers et al. 1991), some human data indicated otherwise. In one case, mesotheliomas were not found associated with long asbestos fibers as believed but were probably associated with some low-aspect-ratio fibers (Churg and Vedal 1994).

The association between fibrosis and cancer is generally accepted for silica and asbestos-induced diseases, but variations exist in terms of inducer, dose, and latency. There is a general agreement that, for amosite or crocidolite-induced diseases, mesotheliomas and plagues appear at much lower burdens than does asbestosis; but mesotheliomas were shown to appear at higher fiber burdens than do pleural plaques (Churg 1991; Roggli, 1990; Roggli, Pratt, and Brody 1986; Wagner et al. 1988). Chrysotile-induced mesotheliomas appeared at very high burdens of tremolite and chrysotile that are comparable to that of asbestosis or airway fibrosis (Churg 1991; Roggli 1990; Roggli, Pratt, and Brody 1986). In one study, co-existence of pleural plaque and mesothelioma was found to be 70% for plaques (69 in 103 cases) and 83% for mesotheliomas (69 in 83 cases) (Churg and Vedal 1994). In this case, a higher fiber concentration was found for induction of lung cancer than for asbestosis by chrysotile, while the fiber concentration for induction of mesothelioma was the same as that for induction of plague. For amosite, however, mesotheliomas or lung cancer appeared at much lower burdens than that seen with airway fibrosis or asbestosis. Therefore, the relative fiber burdens required for tumorigenesis and fibrogenesis in the lung and the pleura by asbestos fibers can differ greatly, depending on the location and type of the lesions and the type and properties of asbestos fibers deposited.

Due to the long latency and variable pace of development for fibrosis and mesothelioma, the association between fibrosis and mesothelioma by way of latency and slope of dose-response curve can also be difficult and variable in human studies. In a study on lung tissue samples from 819 individuals, which consist of 419 cases of malignant mesothelioma, 340 cases of lung carcinoma, and 206 cases of asbestosis, all associated with asbestos exposures, the samples were divided chronologically into two groups, with those occurring in the first half in group 1 (median exposure duration: 27 years) and those occurring in the second half in group 2 (median exposure duration: 25 years). It was found that patients with asbestosis appeared to have slightly older ages (64 years in group 1 and 69 years in group 2) than patients with mesothelioma (62 years in group 1 and 65 years in group 2) or lung carcinoma (62 years in group 1 and 67 years in group 2) (Roggli and Vollmer 2008). The source of this slight difference in age between lung fibrosis and malignancy caused by asbestos exposures was not identified but possibly reflects a slightly longer latency for clinically recognized fibrosis than for mesothelioma and lung cancer.

Overall, these associations in humans support a mechanistic connection among inflammation, fibrosis, and cancer induced by particles and fibers, and, at the same time, emphasize the importance of taking into considerations of the multiple factors involved and the specific context examined when analyzing the associations in humans and when extrapolating animal findings to human health.