Interstitial lung diseases (ILDs) comprise a heterogeneous group of lung diseases affecting the interstitium, distinct from obstructive airway diseases such as asthma or chronic obstructive pulmonary disease (COPD). The most extensively studied ILD is idiopathic pulmonary fibrosis (IPF), which is characterized by progressive pulmonary fibrosis. Non-IPF ILDs may include connective tissue disease-related ILDs such as those related to rheumatoid arthritis and other autoimmune diseases, systemic sclerosis (SSc), and polymyositis/dermatomyositis, and ILDs related to chronic sarcoidosis, chronic hypersensitivity pneumonitis, idiopathic non-specific interstitial pneumonia, and exposure-related diseases such as asbestosis and silicosis (Cottin et al., 2019; Kolb and Vašáková, 2019). Up to 40% of patients with these ILDs may develop a progressing fibrotic phenotype.

Progressing fibrosing ILDs are associated with high mortality, with median post-diagnosis survival in patients with IPF estimated at 2–5 years (Raghu et al., 2014). Progression of fibrosing ILD is reflected in various parameters, including decline in pulmonary function, decrease in exercise capacity, deterioration in quality of life, worsening of cough and dyspnea, acute exacerbations, and increase of morphologic abnormalities (Cottin et al., 2019; Kolb and Vašáková, 2019). In patients with IPF, forced vital capacity (FVC) is a well-established predictor of mortality, and acute exacerbations are associated with very high mortality. Although corticosteroids and/or immunosuppressive drugs are sometimes used off-label to treat progressing fibrosing ILDs, currently the only approved treatments to slow disease progression in IPF are nintedanib, a tyrosine kinase inhibitor, and pirfenidone, a pyridone with an unknown mechanism of action (Richeldi et al., 2018). Nintedanib has been approved in the US since 2014 (U.S. Food and Drug Administration, 2020), and in Europe and Japan since 2015 (European Medicines Agency, 2021b), while pirfenidone was approved in Japan in 2008 (BioBusiness Briefs, 2008), in Europe in 2011 (European Medicines Agency, 2021a), and in the US in 2014 (U.S. Food and Drug Administration, 2019). Lung transplantation is the only potentially curative treatment for IPF, and the medical need in IPF and other progressive fibrosing ILDs remains high.

With respect to IPF pathophysiology, there has been a paradigm shift in recent years from a chronic inflammatory disorder to a primarily fibrotic disease. The current paradigm of disease pathogenesis involves sustained alveolar epithelial micro-injury, followed by a disordered repair and wound-healing response. This is characterized by uncontrolled activation of lung fibroblasts and differentiation to myofibroblasts, resulting in excessive extracellular matrix deposition, scarring of lung parenchyma, and (probably irreversible) loss of pulmonary function (Sgalla et al., 2018; Spagnolo et al., 2018). The wound-healing process includes an inflammatory phase, with the involvement of inflammatory cells (in particular macrophages, monocytes, neutrophils, and T lymphocytes) and increased levels of cytokines and growth factors, creating a biochemical environment supporting chronic tissue remodeling. Both drugs effective in the treatment of IPF, nintedanib and pirfenidone, have been shown to have antifibrotic and anti-inflammatory activity (Heukels et al., 2019).

Phosphodiesterases (PDEs) mediate the hydrolysis of the second messengers, cyclic adenosine monophosphate (cAMP) or cyclic guanosine monophosphate (cGMP). PDEs are coded by 11 gene superfamilies containing multiple genes (coding for subtypes A, B, C, etc.) that also give rise to alternative mRNA-splicing variants leading to approximately 100 PDE isoforms (Azevedo et al., 2014).

The PDE4 subtypes A–D are encoded by different genes, PDE4A, B, C, and D, with post-translational processing resulting in N-terminal variant groups (long, short, and super-short form) according to the presence or absence of upstream conserved regions 1 and 2 (UCR1 or UCR2) N-terminal domains (Bolger, 1994; Beard et al., 2000). PDE signaling is highly compartmentalized as PDE4 subtypes can integrate into macromolecular complexes known as signalosomes. For example, in cardiac signalosomes, PDE4D5 interacts with β-arrestin, regulating receptor desensitization, and PDE4B regulates both L-type calcium channels and ryanodine receptor, which controls the process of calcium-induced calcium release (Bolger et al., 2003; Lehnart et al., 2005; Terrenoire et al., 2009; Beca et al., 2011; Leroy et al., 2011; Sin et al., 2011; Maurice et al., 2014). This presents an opportunity to develop subtype-specific PDE4 inhibitors for different conditions and maximize efficacy and tolerability profiles.

PDE4 has traditionally been implicated in the regulation of inflammation and the modulation of immunocompetent cells, and the three selective PDE4 inhibitors currently available support a beneficial role for PDE4 inhibitors in inflammatory and/or autoimmune diseases (Sakkas et al., 2017; Li et al., 2018). Evidence points to the involvement of a range of immune cells and inflammatory responses in IPF. The results of previous negative trials altering inflammation may suggest that inflammatory changes seen in IPF occur independently of the primary fibrotic remodeling process. Observations from human trials and experimental modeling indicate that changes may be needed to the “traditional” model of pulmonary fibrosis, with immune cells, such as macrophages and lymphocytes, modulating existing fibrotic responses (Desai et al., 2018; Heukels et al., 2019).

The first-in-class PDE4 inhibitor, oral roflumilast (Daliresp®, Daxas®), was approved by the U.S. Food and Drug Administration in 2011 to reduce the risk of COPD exacerbations in patients with severe COPD associated with chronic bronchitis and a history of exacerbations (U.S. Food and Drug Administration, 2013). Another compound, oral apremilast (Otezla®), was approved for the treatment of psoriatic arthritis and plaque psoriasis in 2014 (U.S. Food and Drug Administration, 2017). A third PDE4 inhibitor, crisaborole (Eucrisa®), was approved in 2016 for topical treatment of mild-to-moderate atopic dermatitis (U.S. Food and Drug Administration, 2016). None of these show any preferential enzymatic inhibition among the four PDE4 subtypes, A–D.

The general anti-inflammatory potential of PDE4 inhibition, as exemplified by roflumilast, is well established (Hatzelmann et al., 2010), and the use of PDE4 inhibitors in various inflammatory and immune-mediated diseases has been broadly investigated (Sakkas et al., 2017; Li et al., 2018). However, in the last decade, it has become increasingly clear that PDE4 may also play an important role in fibrosis, based on animal studies and on in vitro experiments evaluating the functional role of PDE4 inhibitors in fibroblasts. The attenuation of lung fibrosis by PDE4 inhibitors has been demonstrated under various experimental conditions, most widely in bleomycin-induced fibrosis in rodents. In rat models, rolipram was shown to inhibit fibrotic score, hydroxyproline content, and serum tumor necrosis factor-α (TNF-α) (Pan et al., 2009). In this initial study, the PDE4 inhibitor was administered from the beginning of bleomycin challenge, so it was not clear whether rolipram was primarily active due to inhibition of initial inflammation or inhibition of secondary fibrosis. A second early study in mice and rats, however, showed that oral roflumilast was active both in preventive and in therapeutic protocols in a dose-dependent manner (Cortijo et al., 2009). In lung extracts, roflumilast inhibited histologically assessed fibrosis, hydroxyproline content, and the mRNA expression of TNF-α, transforming growth factor-ß (TGF-ß), connective tissue growth factor (CTGF), α1 collagen, endothelin-1, and mucin 5ac. In bronchoalveolar lavage fluid (BALF), the levels of TNF-α, interleukin (IL)-13, TGF-ß, and mucin 5ac, the formation of lipid hydroperoxides, and the influx of inflammatory cells (e.g. neutrophils and macrophages) were inhibited. Besides fibrosis, right ventricular hypertrophy and vascular remodeling (pulmonary arteries) were positively influenced by roflumilast. The same group later also demonstrated that the metabolome associated with pulmonary fibrosis in bleomycin mice was modulated by roflumilast. Levels of the amino acids (AAs) glycine and proline, involved in collagen formation/structure, were lowered by roflumilast, while lung glutathione and plasma tetrahydrobiopterin were increased, suggesting an alteration in oxidative equilibrium by roflumilast (Milara et al., 2015). Another PDE4 inhibitor, cilomilast, was shown to inhibit late-stage lung fibrosis and tended to reduce collagen content in bleomycin mice, although no effect on TGF-ß1 and collagen type (Col) 1A1 expression was found (Udalov et al., 2010).

Improvement of lung fibrosis by PDE4 inhibition was not limited to the bleomycin model. In a murine model of lung fibrosis targeting type II alveolar epithelial cells in transgenic mice expressing the diphtheria toxin receptor under the control of the murine surfactant protein C promoter, roflumilast lowered lung hydroxyproline content and mRNA expression of TNF-α, fibronectin (FN), and CTGF (Sisson et al., 2018). Interestingly, roflumilast was active both in a preventive and in a therapeutic regimen, and under the latter conditions appeared to be therapeutically equi-effective to pirfenidone and nintedanib. Furthermore, in a mouse model of chronic graft-versus-host disease, lung fibrosis was attenuated by oral roflumilast (Kim et al., 2016). Roflumilast inhibited fibrosis, collagen deposition, hydroxyproline and TGF-ß1 content, cell infiltration, and expression of mRNA for IL-6 and IL-1ß. In addition, in the BALF inflammatory cells (macrophages, lymphocytes, neutrophils, and eosinophils), expression of mRNA for IL-6, IL-1ß, and monocyte chemotactic protein-1 was inhibited by roflumilast. In a rabbit tuberculosis model, pulmonary damage and fibrosis were shown to be inhibited by two PDE4 inhibitors from Celgene, CC-3052 (Subbian et al., 2011) and CC-11050 (Subbian et al., 2016). PDE4 inhibition improved antibiotic therapy and lung fibrosis by positively influencing collagen deposition and mRNA expression of various matrix metalloproteinases, including metalloproteinase 12.

Besides the lung, beneficial effects of PDE4 inhibition on fibrosis have been demonstrated in several other organs including skin, liver, kidney, and colon. For example, in various preclinical mouse models of SSc (bleomycin-induced, topoisomerase I and chronic graft-versus-host disease), skin fibrosis was inhibited by rolipram and apremilast (Maier et al., 2017). This group did not find direct inhibitory effects of PDE4 inhibition on the release of profibrotic cytokines (IL-6, IL-13, TGF-ß1/ß2) in fibroblasts and M2 macrophages purified from peripheral blood of patients with SSc, which may be due to the lack of an exogenous cAMP trigger under the experimental conditions used. In a model of unilateral ureteral obstruction-induced obstructive nephropathy in mice, rolipram was shown to inhibit renal interstitial fibrosis (Ding et al., 2018). In mouse primary tubular epithelial cells in vitro, TGF-ß upregulated PDE4A/B, and rolipram inhibited TGF-ß-induced damage, FN expression, and deficiency of mitochondrial biogenesis (Ding et al., 2018). Roflumilast inhibited diethylnitrosamine-induced liver fibrosis, hydroxyproline deposition, and TGF-ß1 expression in rats (Essam et al., 2019). Similarly, rolipram inhibited collagen deposition, α-smooth muscle actin (α-SMA) staining, and mRNA expression, as well as the expression of TGF-ß1 mRNA and TNF-α protein, in a bile duct ligation-induced hepatic fibrosis model in rats, with upregulation of PDE4A, B, and D (Gobejishvili et al., 2013). In hepatic stellate cells in vitro, rolipram inhibited mRNA expression of α-SMA and Col1A2 (Gobejishvili et al., 2013). With respect to colonic tissue, rolipram inhibited collagen and TGF-ß1 in a model of trinitrobenzene sulfonic acid-induced colitis in rats (Videla et al., 2006), and apremilast inhibited fibrosis in colon, collagen deposition, and the expression of genes related to fibrosis in a model of dextran sulfate sodium-induced colitis ulcerosa in mice (Li et al., 2019). In a murine cecal abrasion model, rolipram inhibited fibrotic reactions, indicating that PDE4 inhibition has the potential to prevent postoperative intra-abdominal adhesions (Eser et al., 2012). Adhesions are assumed to result from laparotomy by abnormal healing. In support of this assumption, rolipram has been shown to be active in a subcutaneous or intraperitoneal polyether-polyurethane sponge implant model in mice by inhibiting intra-implant collagen and TGF-ß1 deposition (Mendes et al., 2009). Thus, in various animal models, the beneficial impact of selective PDE4 inhibition on fibrosis has been proven, most extensively in the lung but also in several other organs. While the specific target(s) of PDE4 inhibitors in fibrotic diseases are largely unknown, it is tempting to speculate that they act either indirectly via inhibition of pro-inflammatory cells and mediators, and/or directly by inhibiting the typical effector cells (fibroblasts, myofibroblasts) mediating fibrosis.

The direct modulation of various fibroblast functions by PDE4 inhibitors has been demonstrated in fibroblast cell lines of human origin. Kohyama et al. demonstrated a direct impact of PDE4 inhibitors on fibroblasts in vitro (Kohyama et al., 2004). In human fetal lung fibroblasts (HFL-1), rolipram and cilomilast inhibited FN-induced chemotaxis and contraction of collagen gels. Inhibition of fibroblast function by prostaglandin E₂ (PGE₂) was shifted to the left in the presence of PDE4 inhibitors, and the inhibition of endogenous PGE₂ by indomethacin diminished their effects (Kohyama et al., 2002). The inhibition of the fibroblast cell line HFL-1 functions by cilomilast could be modulated by cytokines like IL-1ß (which upregulated PGE₂ and shifted the cilomilast curve to the left) or IL-4 (which downregulated PGE₂ and shifted the cilomilast curve to the right). The inhibition of HFL-1 functions (FN-stimulated chemotaxis and collagen gel contraction) by rolipram and roflumilast was dependent on cyclooxygenase-2 expression and subsequent PGE₂ synthesis (Kohyama et al., 2002). In addition, TGF-ß1-stimulated FN release was inhibited by the PDE4 inhibitors, paralleled by stimulation of PGE₂ release as a positive feedback mechanism (Togo et al., 2009). In another human fetal lung fibroblast strain (GM06114), roflumilast N-oxide, the active metabolite of roflumilast, in the presence of PGE₂ was shown to inhibit intercellular adhesion molecule-1 and eotaxin release stimulated by TNF-α, proliferation stimulated by basic fibroblast growth factor (bFGF) plus IL-1ß, as well as TGF-ß1-induced α-SMA, CTGF, and FN mRNA expression in the presence of IL-1ß (Sabatini et al., 2010). Importantly, IL-1ß upregulated PDE4 activity.

In normal human lung fibroblasts (NHLFs), the impact of PDE4 inhibitors has been described in several publications. TGF-ß-induced fibroblast-to-myofibroblast conversion assessed by α-SMA expression was shown to be inhibited by piclamilast in the presence of PGE₂ (Dunkern et al., 2007). In subsequent papers, the same group showed the inhibition of IL-1ß plus bFGF-stimulated fibroblast proliferation by piclamilast and the importance of cyclooxygenase-2 and PGE₂ (Selige et al., 2010). By using PDE4 subtype-specific siRNA, the involvement of PDE4B and PDE4A in this response, as well as the involvement of PDE4B and PDE4D in TGF-ß-induced α-SMA expression, were shown (Selige et al., 2011). The importance of a cAMP trigger for the modulation of fibroblast functions by PDE4 inhibition was corroborated by the inhibition by roflumilast of TGF-ß1-induced CTGF mRNA and α-SMA protein expression, and FN in the presence of the long-acting ß2-adrenergic agonist indacaterol (Tannheimer et al., 2012). In addition, various other NHLF responses (collagen synthesis, proliferation, reactive oxygen species and F2-isoprostane formation, NADPH oxidase 4 expression) stimulated by bleomycin or 8-epi-PGF_2α were inhibited by roflumilast N-oxide (Vecchio et al., 2013). In yet another human lung fibroblast model (WI-38), TGF-ß-induced mRNA expression of collagen α1, CTGF, and FN in the presence of the adenyl cyclase activator forskolin was inhibited by roflumilast and another PDE4 inhibitor (compound 1) (Sisson et al., 2018), as well as a dual-selective PDE4/5 inhibitor (compound A) (Muraki et al., 2019).

Thus, a multitude of in vitro studies indicate that PDE4 inhibitors are able to directly inhibit various fibroblast functions where either an endogenous or an exogenous cAMP trigger is available in the test system. While the availability of cAMP agonists may be limited under artificial in vitro conditions, it can be expected that in diseased (inflammatory, fibrotic) tissues, cAMP agonists like PGE₂, adenosine, histamine, or adrenalin may be formed. The importance of PDE4 under such conditions may be further enhanced by upregulation of PDE4 activity by cytokines such as IL-1ß.

The modulation of another interesting aspect of fibrosis, epithelial-mesenchymal transition, was addressed in the TGF-ß1-stimulated A549 human alveolar epithelial cell line in vitro. TGF-ß1 stimulated the upregulation of PDE4 subtypes (PDE4A and 4D), and rolipram and siRNA against PDE4A and D inhibited epithelial-mesenchymal transition changes like FN and collagen mRNA expression, although not α-SMA mRNA (Kolosionek et al., 2009).

Emesis is a known class-related side effect of PDE4 inhibitors, hampering their use in humans, and is thought to be associated with PDE4D subtype inhibition. Preferential inhibition of PDE4B may maintain efficacy in treating pulmonary fibrosis whilst avoiding certain adverse events. In the present paper, the preclinical pharmacology of the novel PDE4 inhibitor BI 1015550 is characterized, with results suggesting that this compound has potential for the treatment of IPF and probably other fibrotic ILDs. The data also suggest that BI 1015550 could be an effective combination partner to nintedanib, either by complementary activities or by additive/synergistic effects.

All assays performed comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2015).

The present paper describes the preclinical pharmacology of BI 1015550, a novel oral PDE4 inhibitor that preferentially inhibits PDE4B.

Among the large superfamily of PDE isoenzymes, BI 1015550 selectively inhibits PDE4, and so far, has not shown any other off-target effects in screening against a panel of enzymes and receptors (Cerep high-throughput profile and Cerep non-kinase enzyme profile, Euriofins Cerep, Celle L’Escevault, France). Compared with roflumilast, the only oral PDE4 inhibitor on the market for a lung indication, BI 1015550 shows a unique profile in inhibiting PDE4 subtypes by a preferential inhibition of PDE4B (Table 1). The use of earlier PDE4 inhibitors, including roflumilast and apremilast, has been hampered by gastrointestinal side effects such as nausea, emesis, and/or diarrhea in humans. Although not proven, there is some evidence that these side effects are linked to inhibition of the PDE4D subtype (Giembycz, 2002). In this respect, the roughly 10-fold selectivity of BI 1015550 for inhibition of PDE4B versus PDE4D may be of clinical importance. It may also explain why in Suncus murinus the therapeutic ratio (inhibition of LPS-induced neutrophil influx into the lung versus induction of emesis) of BI 1015550 is superior to roflumilast (Figure 4; Table 2). More importantly, ongoing phase I clinical trials with BI 1015550 support the excellent gastrointestinal tolerability and safety of this compound in humans (data not shown). Another aspect of the use of BI 1015550 in humans relates to toxicity evident in animal studies. The major toxicity of PDE4 inhibitors, especially in the rat, is the induction of a vasculopathy (vasculitis, arteritis) in different tissues, which in some ways is paradoxical given that PDE4 inhibition is strongly linked to anti-inflammatory efficacy (see below). Although induction of vasculitis has never been reported in patients for the two oral PDE4 inhibitors roflumilast and apremilast, which have been on the market for many years, there is still some concern on this potentially serious side effect for the introduction of new PDE4 inhibitors. Dietsch et al. were the first to report stimulation of IL-6 in the rat by the PDE4 inhibitor IC542 (Dietsch et al., 2006). Since IL-6 is a possible candidate for induction of vasculitis in the rat, the lack of stimulatory effect of BI 1015550 on IL-6 in the human setting (Figure 2B) may indicate that, in contrast to rats, the risk for vasculitis induction by BI 1015550 in humans may be quite low. In addition, it is interesting to note that apremilast was reported to be clinically active in Behcet’s disease, a form of vasculitis characterized by inflammation of blood vessels, by resolution of oral ulcers, one of the most common symptoms (Hatemi et al., 2015).

Despite the alleviated inhibition of the PDE4D subtype, the well-known anti-inflammatory and immune-modulating properties of PDE4 inhibitors are still evident for BI 1015550. Qualitatively similar to roflumilast (Hatzelmann and Schudt, 2001), BI 1015550 appeared to be a potent inhibitor of TNF-α and IL-2 in human PBMCs (Figure 1), which, most likely, reflects the impact of BI 1015550 on monocytes and T lymphocytes, respectively. The efficient inhibition of TNF-α was also evident in whole blood (Figure 2), which is of importance because this assay format (ex vivo) can be used as a pharmacodynamic biomarker during clinical studies, as shown first by Timmer et al. for roflumilast (Timmer et al., 2002). Although the higher IC₅₀ value of BI 1015550 for TNF-α inhibition in human whole blood (670 nmol/L) compared with human PBMCs (35 nmol/L) may be explained partly by the moderate plasma protein binding of the compound (77%), the loss of potency is higher than expected. In an effort to mimic the clinical situation, the potent and efficient inhibition of LPS-stimulated TNF-α in whole blood ex vivo was demonstrated for the mouse model in this study (Figure 3). It is clear that the ED₅₀ of oral BI 1015550 (0.04 mg/kg) in this mouse model was much lower than in the mouse lung fibrosis disease models (bleomycin, silica) discussed below. However, in the former ex vivo model, blood is taken close to the maximal plasma concentration of the compound, and, although TNF-α as the pharmacologic read-out may contribute to lung fibrosis pathology, it is not the only relevant mediator. The efficacy and potency of BI 1015550 in the disease models will depend on other pathophysiologic aspects, and on the pharmacokinetics and availability of the compound during 24 h, which will be below the maximal plasma concentration due to the rather short half-life of the compound.

The aberrant wound-healing process in lung fibrosis passes through an inflammatory phase, with the involvement of inflammatory cells (in particular macrophages, monocytes, neutrophils, and T lymphocytes) and increased levels of cytokines (e.g. TNF-α and IL-1ß) and growth factors (e.g. TGF-ß and CTGF), creating a biochemical environment that supports chronic tissue remodeling and fibrosis. Based on this working hypothesis of the pathophysiology of lung fibrosis, the anti-inflammatory (inhibition of TNF-α, inhibition of monocytes, inhibition of neutrophil influx into the lung) and immune-modulatory (inhibition of IL-2, inhibition of T cells) characteristics of BI 1015550 can be expected to contribute to improvement of fibrosis, at least to some extent. Therefore, we were interested to see whether BI 1015550 exerts antifibrotic effects in distinct animal models, and further, if the compound has direct effects on fibroblasts and myofibroblasts.

Bleomycin has been widely used in rodents to model pulmonary fibrosis, to understand mechanisms involved in fibrogenesis, and to evaluate potential new therapies. However, it should be noted that although bleomycin-induced pulmonary fibrosis mimics many features of human disease, fibrosis resolves in rodents whereas in humans it is generally irreversible. Furthermore, whilst human pulmonary fibrosis predominantly impacts the peripheral airways, in rodents it is often found more prominently around the central airways. Despite these restrictions, the bleomycin model is still a good tool to assess the efficacy of potential compounds in general as proof of principle (Moeller et al., 2008; Jenkins et al., 2017). The therapeutic treatment regimen of BI 1015550 attenuated some important aspects of disease pathology, including FVC and Cstat decline (Figure 5). These therapeutic effects were statistically significant at the highest BI 1015550 dose. In addition, lung tissue density was improved to the same extent, although this effect did not reach statistical significance (Table 3). With regards to the Ashcroft score, BI 1015550 only showed a non-significant trend towards inhibition. For other experimental parameters investigated (BALF protein and monocytes, airway compliance), no effects of BI 1015550 were detected. The lack of effect of BI 1015550 on inflammatory mediators in BALF as well as the minimal effects on histologically assessed lung fibrosis in the present study are in contrast to bleomycin mouse studies reported by others (Cortijo et al., 2009; Udalov et al., 2010). We can only speculate that fibrotic changes in the mouse bleomycin model performed under our conditions are not captured by our scoring system, originally designed to assess human pulmonary fibrosis. It might also be possible that the changes in tissue volume do not only reflect fibrotic changes, but are related to other remodeling aspects like neovascularization in our model, as shown by Ackermann et al. (Ackermann et al., 2017). Differences versus previous studies may also be due to differing treatment durations and timings of data collection. Regardless of the nature of such changes, improvement in functional parameters, such as lung function, suggests a positive impact of BI 1015550 in pulmonary tissue remodeling. This assumption is supported by the positive findings for BI 1015550 in the silica-induced lung fibrosis model in the mouse. In contrast to bleomycin, which induces reversible lung fibrosis, silica particles induce a progressive type of lung fibrosis, which resembles the human counterpart of a progressive fibrosing ILD. To the best of our knowledge, a PDE4 inhibitor that preferentially inhibits PDE4B has not been evaluated in this animal model before. As shown in Table 4, BI 1015550 in a therapeutic regimen achieved dose-dependent improvements in semi-quantitative scores of granuloma formation, inflammation, and fibrosis, although these effects did not reach statistical significance. In BALF, macrophages and neutrophils were significantly inhibited at the highest dose, while the inhibition of MPO and KC did not reach statistical significance.

Overall, the beneficial effects of BI 1015550 in two lung fibrosis models are in agreement with preclinical data in the literature that selective PDE4 inhibitors could be clinically effective not only in lung fibrosis but, based on animal data, probably also in fibrosis of other organs (e.g. liver, kidney, colon, and/or skin). This hypothesis is supported by several studies showing that selective PDE4 inhibitors are able to directly target fibroblasts in addition to inflammatory and immune-competent cells. By using relevant cells, namely human lung fibroblasts from patients with IPF, we were able to support this by showing that BI 1015550 inhibits TGF-ß-stimulated transformation into myofibroblasts (Figure 6A). This indicates a differential mode of action for BI 1015550 compared with nintedanib, which was inactive under these conditions and did not enhance the efficacy of BI 1015550. Furthermore, BI 1015550 attenuated TGF-β-induced Col1, Col3, and FN gene expression (Figures 6B–D), which are important constituents of fibrotic extracellular matrix, and the combination with nintedanib at 100 nmol/L showed additive effects in inhibiting Col3 expression. In addition, BI 1015550 inhibited bFGF plus IL-1β-induced cell proliferation. Nintedanib by itself, at a relevant concentration (100 nmol/L), inhibited the proliferation by only 15%, but, most strikingly, the combination of BI 1015550 plus nintedanib resulted in synergistic inhibitory effects and shifted the concentration–response curve to the left by about 10-fold (Figure 6E). Inhibition of human lung fibroblast proliferation and myofibroblast transformation in vitro might contribute to potential clinical efficacy of BI 1015550 in IPF. Furthermore, the combination of BI 1015550 with nintedanib might yield synergistic therapeutic effects in patients with IPF and other fibrotic lung diseases. This kind of synergism between a PDE4 inhibitor that preferentially inhibits PDE4B and nintedanib has not been reported before.

In summary, BI 1015550 is an oral preferential inhibitor of PDE4B with suggested improved tolerability in humans compared with the selective oral PDE4 inhibitors on the market; the preclinical profile suggests that this compound is a promising oral clinical drug candidate for the treatment of IPF and other ILDs. A phase II trial of BI 1015550 versus placebo in patients with IPF is currently ongoing (NCT04419506). Future clinical studies will show whether BI 1015550 will be able to provide substantial efficacy alone, or whether the combination with IPF standard therapies such as nintedanib may be preferred in order to fully exploit the therapeutic potential of BI 1015550.