Pulmonary arterial hypertension (PAH) is a life-threatening complication of connective tissue disease (CTD) and has been described in patients with systemic sclerosis (SSc), systemic lupus erythematosus (SLE), mixed CTD (MCTD), primary Sjögren’s syndrome, and rheumatoid arthritis (Sung and Chung, 2015). In Western countries, CTD-associated pulmonary arterial hypertension (CTD-PAH) was the second leading cause of PAH (25%), and almost 75% of CTD-PAH was SSc-related; the 3-year survival within this population was only 56% (Hachulla et al., 2009; Thakkar and Lau, 2016). Similarly, CTD-PAH was the third most common type of PAH (20%) in China (Jiang et al., 2012), and more than half of these cases were SLE-related (58.4%) (Zhao et al., 2017). The estimated 1-year and 3-years survival rates were 85.4 and 53.6%, respectively (Zhang et al., 2011). Recently, PAH-targeted drugs were used in patients with CTD-PAH, but the effect was unsatisfactory (Fisher et al., 2006; Zhang et al., 2011). Immunosuppressive therapy is an indispensable part of treatment for CTD-PAH as well (Kato et al., 2020). The administration of immunosuppressive agents, including systemic glucocorticoids and intravenous cyclophosphamide, may improve the clinical status of PAH patients who have SLE, MCTD, and Sjögren’s syndrome; however, patients with SSc-PAH do not respond well to immunosuppressive therapy (Sanchez et al., 2006; Kato and Atsumi, 2018; Kato et al., 2020).

As mentioned above, an optimal regimen of treatment has not been established, and existing drugs do not show favorable efficiency for all CTD-PAH patients; therefore, novel drugs for this condition are needed. Drug discovery has long been a time- and capital-intensive process, and an unpredictable return on investment may result. Nevertheless, drug repurposing serves as a quicker, and a less costly method to expand indications for already approved drugs (Moosavinasab et al., 2016). A successful example is Pfizer’s sildenafil, which was originally designed for the treatment of angina but then repositioned to treat erectile dysfunction in 1998 (Novac, 2013). Text mining is a feasible strategy to discover drugs with new treatment potential (Andronis et al., 2011). Based on the available literature and biomedical databases, combined with analytical tools, text mining aims to explore drugs with potential value for the treatment of targeted diseases among the currently available drugs. In this study, we discovered that some existing drugs may be used for CTD-PAH treatment by text mining.

CTD-PAH shares similar pathophysiological characteristics with other types of PAH, and dysfunction of multiple cells and molecular processes may contribute to vasoconstriction and inflammation of arterioles, which results in progressive increases in pulmonary vascular resistance and right ventricular afterload, heart failure, and even death (Thenappan et al., 2018b). Currently, all PAH-targeted drugs mainly target pulmonary vasoconstriction rather than vascular remodeling, which underlies the basic pathological characteristics of PAH. By performing text mining, we found six genes of interest and 13 potential drugs for the treatment of CTD-PAH. Notably, the main drivers of vascular remodeling, including endothelial cells (ECs) or pulmonary artery smooth muscle cells (PASMCs), were implicated in the identified target gene–related signaling pathway.

Immunological disturbance and inflammatory mechanisms play a central role during the development of PAH in CTD (Dorfmüller et al., 2003). Elevated serum levels of some proinflammatory cytokines, such as interleukin (IL)-1 and IL-6, were reported in CTD-PAH patients (Dorfmüller et al., 2003). The downregulation of IL-1β and IL-6 expressions could inhibit the development of PAH by suppressing the proliferation and migration of PASMCs in rats (Ou et al., 2020). A study also demonstrated that the expression of IL-6 was increased in the lung, and IL-6 blockade by the monoclonal anti–IL-6 receptor antibody MR16-1 could ameliorate the pulmonary hypertension (PH) of pristane/hypoxia mice (a novel mouse model of PH reflecting the pathological features of CTD-PAH) (Mori et al., 2020). The IL-6 signaling pathway is a promising candidate in the treatment of CTD-PAH, and the use of tocilizumab, an anti-IL-6 receptor antibody was proven to result in clinical improvements in some cases of CTD-PAH (Arita et al., 2010; Furuya et al., 2010; Kadavath et al., 2014). Although siltuximab (targeting IL-6) seems to have a similar effect on CTD-PAH, the related evidence is limited. As for IL-1β blockade, whether it has a positive effect on CTD-PAH requires more evidence.

Pulmonary vascular remodeling is initiated by remodeling of the extracellular matrix resulting from the imbalance of proteolytic enzymes and their inhibitors (Thenappan et al., 2018a). Remodeling of the ECM of the pulmonary artery occurs in the early stage of PAH pathogenesis and even precedes the hemodynamic changes in pulmonary circulation (Thenappan et al., 2018a). As an important type of proteolytic enzyme, matrix metalloproteinase (MMP) shows an increase in activity associated with the proliferation and migration of PASMCs and intimal thickening during the development of PAH (Chelladurai et al., 2012). In the MMP family, the upregulation of gelatinases (MMP-2 and MMP-9) was identified in monocrotaline (MCT)–induced or hypoxia-induced mouse model of PH, and gelatinases were considered possible therapeutic targets for the treatment of PAH (George and D'Armiento, 2011; Liu et al., 2018). We found the possible use of glucosamine, an antagonist of MMP9, in CTD-PAH, but most studies have focused on its clinical use in patients with osteoarthritis thus far. Moreover, contraindications and drug interactions involving glucosamine use are uncommon (Dahmer and Schiller, 2008; Bruyère et al., 2016), therefore, the possible use of glucosamine should be explored.

The potential effect of epidermal growth factor receptor (EGFR)–mediated survival of PASMCs has led researchers to test the efficacy of selective targeting EGFR in the PH animal model (Merklinger et al., 2005). Dahal et al. (2010) found that erlotinib (one of the first generation EGFR tyrosine kinase inhibitors; it is also the drug we suggested in Table 1) showed therapeutic benefit in rats with MCT-induced PH, but it did not show therapeutic efficacy in chronic hypoxic mice model. The reason why partial therapeutic benefit was observed may be because different mechanisms were involved in these two commonly used animal models with respect to the involvement of EGF signaling. Most importantly, multiple growth factors, including EGF, platelet-derived growth factor, and fibroblast growth factor, are all involved in PASMC proliferation in response to hypoxia. Therefore, no significant therapeutic efficacy was observed in chronic hypoxic mice. Notably, no significant alteration of EGFR expression in the lung tissues from patients with idiopathic PAH was also reported (Dahal et al., 2010). In Dahal’s study, only patients with advanced-stage idiopathic PAH were enrolled. Therefore, as described in their study, the important role of EGF signaling in the early stage of PAH development could not be excluded. However, the second generation of EGFR tyrosine kinase inhibitors and a new pan-EGFR inhibitor, dacomitinib, showed a significant inhibitory effect on pulmonary vascular remodeling and attenuated pulmonary artery pressure and right ventricular hypertrophy in both MCT and hypoxia-induced mice model of PH (Yu et al., 2019). Other drugs that target EGFR in our list (necitumumab, afatinib, and osimertinib) have not yet been validated. The therapeutic effect of EGFR, a promising target, is required to be elucidated in the future.

Disturbed production of vasoactive, vasoconstrictive, and proliferative mediators from ECs may affect vascular tone and promote vascular remodeling (Zanatta et al., 2019). The vascular endothelial growth factor (VEGF) is the main angiogenic factor and is indispensable for the process of normal angiogenesis. The overexpression of VEGF is associated with proliferation of ECs in severe PAH (Sakao and Tatsumi, 2011). VEGF and its receptors showed increased expressions in animal models (hypoxia or MCT-induced PH), and elevated levels of plasma VEGF were also discovered in PAH patients (Voelkel and Gomez-Arroyo, 2014). However, many studies have revealed that combined with a second hit (high shear stress or chronic hypoxia), VEGF receptor blockade can drive the emergence of apoptosis-resistant ECs with the potential for hyperproliferation, and angio-obliterative PAH may form as a result (Sakao et al., 2009; Voelkel and Gomez-Arroyo, 2014). VEGFA is the most abundant isomer of VEGF in humans and is currently the main target of anti-VEGF treatment. It is unknown whether VEGF receptor antagonists would contribute to the development of PAH, but two patients developed PH were reported in patients who had ovarian cancer received bevacizumab (recombinant humanized monoclonal antibody against VEGF-A) (Garcia et al., 2008). Therefore, VEGF receptor antagonists may not be applicable in the treatment of PAH.

The predominant isoform of transforming growth factor β (TGFβ) in humans is TGFβ-1, and excessive TGFβ-1 signaling is a characteristic of PAH (Sturrock et al., 2006; Kajdaniuk et al., 2013; Calvier et al., 2019). By reducing the expression of PTEN and increasing the activation of PI3K/AKT, TGFβ-1 promotes the proliferation and decreases the apoptosis of PASMCs in lung tissue (Liu et al., 2016). Apoptosis-resistant PASMCs could promote progressive narrowing of the vascular lumen and excessive accumulation of ECM components, which reduced vascular compliance (Thenappan et al., 2018b). Selective trapping of the TGFβ-1 ligand has been demonstrated to improve hemodynamics, pulmonary vessel remodeling, and survival in mice model (Yung et al., 2016). Although hyaluronidase can target TGFβ-1, the indication is as an adjuvant to increase the absorption of other drugs in local use (Table 1). Moreover, hyaluronidase cannot be delivered to the lung by the venous system because the enzyme is rapidly inactivated following intravenous administration. Thus, hyaluronidase is not applicable in the clinical treatment of PAH, but a suitable formulation for delivery may be an alternative consideration.

The first limitation of this study is related to the databases we used. Not all the drug–gene interactions are fully clarified in the present databases. Therefore, the analysis may be more accurate as the databases are updated. Additionally, some candidate genes for targeting might have generalized effects that may not always be desirable. Thus, lack of experiments to validate the targets and drugs that we found is the other limitation of the present study. Experiments and clinical trials are required before these drugs are approved for clinical use.

As discussed above, some drugs that we found (Table 2) may have unfavorable effects depending on the current findings. For example, drugs that target VEGFA may induce PAH, erlotinib (target EGFR) was ineffective in animal model, and hyaluronidase (target TGFβ-1) could not be delivered to lung tissue in its present formulation. With the evolution and improvement of database and analytic tools, this method of data mining will promote drug discovery in CTD-PAH therapy.

In conclusion, 13 drugs targeting six genes that may have potential therapeutic effects on CTD-PAH were discovered, and none of them has been tested in clinical trials for PAH patients. Drug discovery by performing text mining and pathway analysis can help identify existing drugs that have the potential to treat CTD-PAH.