Pulmonary hypertension secondary to pulmonary fibrosis: clinical data, histopathology and molecular insights

In IPF patients, pulmonary function tests are used to evaluate lung function impairment. The FVC and DLCO are the two most common parameters which are assessed in IPF patients. FVC is the quantity of air that can be forcibly exhaled from the lungs after taking the deepest breath possible; and DLCO is the difference of carbon monoxide partial pressure between inspired and expired air. Both parameters are decreased in IPF patients and correlate with disease progression. DLCO can also be decreased in patients with PH without PF. Therefore, in a patient with PF, a decreased DLCO out of proportion to the underlying PF can be a sign of concomitant PH. Until recently, the development of PH in PF patients was predominantly viewed as a result of hypoxic vasoconstriction and destruction of the vascular bed due to the accumulation of scar tissue [57]. However, many clinical reports did not find any correlation between the extent of lung fibrosis and mPAP as measured by high resolution chest computed tomography (HRCT) or FVC in IPF patients [28, 40, 58, 59]. In addition, Nathan et al. [36] reported that DLCO% and FVC% or FVC%/DLCO% ratio were not able to accurately detect PH in IPF patients. Therefore, lung parenchyma destruction and accumulation of scar tissue alone do not explain the development of PH in IPF patients. Interestingly, Zisman et al. [60, 61] have developed a mathematical formula that predicts mPAP from pulmonary function tests using the FVC predicted, DLCO predicted and oxygen saturation. Although, this formula was validated in two distinct small cohorts of IPF patients, the validity of this formula remains to be assessed in a larger cohort of patients. Furthermore, while reduction in DLCO is the most recognized abnormality in pulmonary function tests in patients with PH, this finding is nonspecific given DLCO can also be reduced by other factors such as chronic pulmonary embolism, emphysema, cigarette smoking, pulmonary edema caused by heart failure and anemia. Thus, pulmonary function testing alone has limited predictive value to detect PH in IPF patients [60, 61].

Transthoracic echocardiography (TTE) is currently the most commonly used diagnostic test to screen patients for PH as it is a non-invasive cardiovascular imaging tool that can estimate pulmonary hemodynamics. Commonly used indices to detect elevated PA pressure by TTE include peak tricuspid regurgitant velocity, RV outflow tract acceleration time, and early diastolic pulmonary regurgitant velocity [62]. However, in patients with advanced lung disease, TTE has limited accuracy in detecting early stages of PH [12, 63, 64]. In a study of 374 patients awaiting lung transplant, Arcasoy et al. [12] found that using TTE to measure pulmonary vascular pressure in patients with obstructive or interstitial lung diseases was possible in only 44% of the patients. Pulmonary pressure was found to be inaccurate in 52% of this subset, which if taken as the only diagnostic tool, would lead to an overestimation of PH in this population. Fisher et al. [64] also found that in 48% of their cohort pulmonary pressure estimations as measured by TTE were inaccurate, with equally occurring over-and underestimations. This limited accuracy of TTE in estimating pulmonary vascular pressure may be due to challenges in obtaining satisfactory acoustic windows [63]. Therefore, detection of PH in PF patients by TTE can be confirmed by RHC although it may not change the clinical management given there is no approved PH-targeted therapy currently available for this patient population. TTE is also widely used to visualize right ventricular morphology and detect right ventricular dysfunction [65]. Commonly used RV indices to assess for PH by TTE include RV/LV basal diameter ratio to assess for RV dilatation, and tricuspid annular plane systolic excursion (TAPSE), fractional area change, and RV pulsed tissue doppler S wave velocity to evaluate for RV dysfunction [62]. However, RV dysfunction appears in the later stages of PH and is therefore not a suitable readout for early detection of PH.

High resolution chest computed tomography (HRCT) scans are routinely performed in PF patients to evaluate the lung parenchyma and have also been used to predict concomitant PH by assessing PA size. One study by Devaraj et al. [63] evaluated CT findings in 30 patients with pulmonary fibrosis with PH. The diagnoses in this group included 16 IPF, 10 nonspecific interstitial pneumonia (NSIP), 3 chronic hypersensitivity pneumonitis, and 1 mixed organizing pneumonia and NSIP. While they found that absolute PA diameter measured by CT did not correlate with mPAP or pulmonary vascular resistance index (PVRi) measured by RHC, they showed that correcting for the size of the ascending aorta (ratio of main PA diameter to ascending aorta diameter) significantly strengthened the correlation to mPAP and PVRi in this group of patients. Furthermore, in a retrospective study of 177 IPF patients undergoing RHC, Yagi et al. [54] found that a ratio of > 0.9 between the diameters of pulmonary artery to ascending aorta measured by HRCT can predict a mPAP > 20 mmHg. A larger and more recent study by Chin et al. [66] included 101 patients with ILD, the majority with usual interstitial pneumonia (the CT and histological pattern seen in IPF) or NSIP. This study found that absolute PA diameter was accurate for detection of PH and correlated with mPAP, contrary to the study by Devaraj et al. [63]. However, main pulmonary artery dilation can also occur in IPF patients without PH [67], thus limiting its use for diagnosis of PH. Perez-Enguix et al. [67] proposed to assess enlargement of the segmental arteries as a sign of PH by HRCT. In this study, they found that three out of four segmental pulmonary arteries should be enlarged to be considered as indicative of PH. Indeed, segmental pulmonary artery enlargement in only one or two lobes could merely be a compensatory mechanism to redirect blood flow away from a fibrotic lobe without affecting the pulmonary pressure [63].

The cardiopulmonary exercise test is another non-invasive test that may be used for the detection of PH in patients with PF. One single-center retrospective review of 75 PF patients who had undergone cardiopulmonary exercise test and RHC pre-lung transplant found that PF patients with PH had significantly lower end-tidal and mixed-expired carbon dioxide pressure with a distinctive activity pattern for the ratio of these two parameters compared to PF patients without PH [68]. A follow up study by the same group showed that cardiopulmonary exercise test parameters were able to detect differences between levels of severity of PH in patients with PF using the ratio of minute ventilation to rate of carbon dioxide production and end-tidal carbon dioxide pressure [69].

Cardiac magnetic resonance imaging (MRI) is a more advanced non-invasive cardiovascular imaging tool for evaluation of PH available at more specialized centers. In fact, this technology is considered the gold standard in the quantification of right ventricle volumes and function given the limitations of TTE in obtaining optimal acoustic windows in some patients [63, 70]. With respect to PF patients with suspected PH, Chin et al. found diastolic pulmonary arterial area as measured by magnetic resonance imaging was accurate for detection of PH (AUC 0.897) and correlated well with RHC-derived mPAP [66]. The same study also found similar diagnostic accuracy when using pulmonary artery diameter as measured by HRCT pulmonary angiography. Given the limitations of cardiac magnetic resonance imaging which include cost, availability, and need for specialized expertise, TTE may be a more cost-effective cardiovascular imaging test in the initial evaluation of PF patients with suspected PH. However, magnetic resonance imaging may be better for prognostication and risk stratification given its superiority in assessing right ventricular function.

In summary, there is currently no single non-invasive test that is capable of detection and diagnosis of PH with accuracy in PF patients. However, their implementation can lead to the suspicion of PH, which may then be confirmed by invasive RHC.

Proper angiogenic homeostasis is crucial for pulmonary vascular function and is well-known to be impaired in pulmonary arterial hypertension (PAH). Several studies have demonstrated an imbalance between angiogenic and angiostatic factors in the pathogenesis of PF [57, 85]. For example, angiogenic factors such as interleukin-8 or fibroblast growth factor 2 are decreased while angiostatic factors such as pigment epithelium-derived factor, interferon-γ inducible protein 10 or endostatin have been demonstrated to be increased in the lungs of PF patients.

Murray et al. [86] demonstrated that a reduction of vascular endothelial growth factor (VEGF) expression was correlated with poor survival outcome in PF patients, and that apoptotic endothelial cell-derived mediators lead to epithelial cell injury and reduce wound closure. Recently, it was shown that VEGF was decreased in the fibrotic area of IPF patient lungs concomitant with increased endothelial cells apoptosis, which is consistent with the decreased capillary density found in fibrotic regions. In contrast, VEGF is increased in normal, non-fibrotic areas of the lung and is associated with a proliferative phenotype of endothelial cells [87, 88]. These data are in line with the decreased capillary density in fibrotic region but increased capillary density in non-fibrotic region with preserved architecture in PF lungs.

Animal models of PF have also helped in understanding the relationship between fibrosis, VEGF and pulmonary vascular function. Farkas et al. [89] used transforming growth factor β1 (TGFβ1) to induce lung fibrosis in a female rats. In this model, TGFβ1 up-regulation decreased VEGF expression leading to endothelial cell apoptosis and lower vascular density. Up-regulation of VEGF significantly lowered mPAP in these rats, which was concomitant with an increased capillary density and activation of VEGF receptor 2 signaling. Interestingly, VEGF up-regulation also lead to more severe fibrosis. In contrast, in bleomycin-induced mouse model of PF, VEGF overexpression attenuated lung fibrosis [86]. The conflicting results of these studies can be due to differences in the experimental models of PF induced by adenoviral delivery of a mutant TGF-β1 gene in female rats or by bleomycin in female and male mice. Taken together, elucidating the mechanisms underlying dysregulation of angiogenesis could help us understand the transition from a normotensive to hypertensive pulmonary vascular state in PF-PH patients.

Bone morphogenetic protein receptor type II (BMPR2) is a member of the large TGF-β super-family and regulates growth, differentiation and apoptosis in a diverse number of cell types. BMPR2 plays a major role in PAH wherein its expression is decreased in the lungs regardless of patients carrying a mutation in this gene or not. In PAH, BMPR2 has been shown to play differential roles in pulmonary arterial smooth muscle cell and endothelial cell proliferation, migration and resistance to apoptosis [90]. Furthermore, in PAH patients BMPR2 is downregulated in macrophages, which in turn exhibit increased expression of granulocyte macrophage colony-stimulating factor leading to increased muscularization of the distal pulmonary arteries [91]. Interestingly, Chen et al. [92] found that BMPR2 was also significantly decreased in lung tissue and macrophages of IPF patients, with a greater decrease in IPF-PH group. In this study however, it was not specified if lung tissue was collected from fibrotic or non-fibrotic areas of the lungs. This could be of importance since our group demonstrated that greater histological and molecular differences between PF and PF-PH were found in the non-fibrotic lung areas. Nonetheless, macrophages characterized by decreased BMPR2 signaling and increased expression of the granulocyte macrophage colony-stimulating factor are present in both PF and PH development and could be key players in the vascular changes seen in fibrotic lungs [93].

Epithelial to mesenchymal transition (EMT) is the process by which an epithelial cell differentiates into a mesenchymal cell [68]. This process is thought to play a critical role in response to injury [69] and has been implicated in numerous diseases such as PF [94‐96]. One subtype ofEMT is the transition of endothelial to mesenchymal cells (EndoMT). EndoMT has been implicated in PAH patients and animal models of PH [97‐101]. Ranchoux et al. [97] elegantly described the presence of mesenchymal-like endothelial cells within the intimal layer and vascular lesions of PAH lungs. These cells were characterized by a proliferative and migratory phenotype, accompanied by loss of cell–cell junctions and expression of the well-known EMT transcription factor Twist-1. Interestingly, loss of BMPR2 signaling seemed to be involved in this mesenchymal transition of endothelial cells.

EndoMT has also been demonstrated to participate in the fibrotic process in animal models of PF induced by bleomycin [102], irradiation [103] or endotoxemic injury [104]. A total of 16% of the fibroblast-like cells isolated from bleomycin-induced PF mouse lungs were shown to be from endothelial origin, demonstrating a significant contribution of vascular cell dysregulation on lung fibrosis development. Interestingly, as in PAH, Twist-1 was involved in the mesenchymal phenotype of these endothelial cells, in addition to the other characteristic EMT transcription factor Snail [102]. Choi et al. [103] have shown that TGFβR1/Smad-driven EndoMT was present in the early stage of radiation-induced pulmonary fibrotic process, even before the appearance of fibrotic deposits. EndoMT increased along with fibrosis development, suggesting a causal role for EndoMT in the fibrotic process. Finally, inhibition of EndoMT by vildagliptin prevented pulmonary fibrosis development in mice with PF induced by endotoxemic injury [104]. Together, these data support vascular endothelial dysregulation involvement in the pulmonary fibrotic process, creating a fertile ground for PH development in PF patients.

Adenosine is a purine nucleoside, used by cells as a signaling mediator through its binding to four G-protein-coupled receptors. Among these receptors, three are of particular interest: A2AAR and A3AR which are present in the vessel wall [105] and mediate vasodilation [106, 107]; and A2BAR which is expressed by mast cells and macrophages where it participates in the regulation of cytokines expression [105]. In PF, the lung adenosine concentration correlates with fibrosis severity in different animal models of PF [108]. The action of adenosine is mainly mediated by A2BAR in PF and leads to a downstream up-regulation of interleukins 6 and 17 [108]. Interestingly, A2BAR is up-regulated in PF-PH compared to PF patients and its expression correlates with mPAP [107]. On the other hand, adenosine concentration is decreased in the plasma of PAH patients, which participates in vasoconstriction and increases pulmonary arterial pressure [109].

Taken together, these studies demonstrate that PF patients exhibit dysregulation of molecular pathways before the onset of clinical PH. Hence, PF may be viewed as a favorable environment for PH development. However, not all PF patients develop PH, and only a sub-population of PF patients develop PH [28‐30, 32‐35, 37, 39‐41, 43, 44]. Therefore, it is more than likely that other mechanisms aside from fibrotic dysregulation of vascular function participate in PH development in PF patients [9‐11, 46, 84].

Mura et al. [84] are one of the first groups to compare the gene expression of lungs from PF patients with and without PH by microarray. In this elegant study, the authors describe specific gene signatures that differentiate PF and PF-PH patients. The authors found that PF patients without PH mainly had a pro-inflammatory gene signature, while PF patients with severe PH (> 40 mmHg) had a pro-proliferative gene signature. This study demonstrates clear molecular differences between PF patients with and without PH, supporting the hypothesis of specific pathway activation during PH development in PF patients. Interestingly, in patients with mild-to-moderate PH (21–39 mmHg), some patients exhibited a pro-inflammatory gene signature, while others exhibited a pro-proliferative gene signature. This gene signature heterogeneity in the mild-to-moderate group may suggest a transition from pro-inflammatory to pro-proliferative during PH development [54]. On the other hand, Patel et al. [46] performed microarray analysis on distal vasculature from PF and PF-PH, but did not find any significant differences between the two groups. However, their PF-PH patients all belonged to mild-to-moderate PH group, which may still be in line with the study by Mura et al. [84] Another aspect to be considered is the fact that Mura et al. [84] performed their analysis on total lung tissue, while Patel et al. [46] assessed isolated pulmonary distal vasculature. This aspect is of importance, since our group has shown that molecular differences impacting PH development may involve cell types outside the pulmonary vasculature [11]. Therefore, to understand the complex interplay leading to PH development in PF patients, one should consider the lung as a whole rather than focusing only on pulmonary vasculature. In this regard, it could be of interest to use cutting-edge technologies such as single-cell RNA sequencing to better apprehend this complex disease.

Together, these studies support our point of view that PF patients will somehow branch into two different groups: patients who will develop PH and those who will not. Interestingly, Rajkumar et al.[86]have shown that the gene signatures between patients with PF-PH and idiopathic PAH patients are unique. This reinforces our need for a better understanding of PH development in PF-PH patients, as new therapies developed for PAH may not work in this population.

Hypoxia-inducible factor 1 alpha (HIF-1α) is a transcription factor activated in response to hypoxia. When cells sense hypoxia, HIF-1α is translocated to the nucleus and binds to the hypoxia response element to activate the cellular response to hypoxia. In PAH, HIF-1α is activated and participates in the apoptosis-resistant and pro-proliferative phenotype of pulmonary arterial endothelial and smooth muscle cells. In PF patients, HIF-1α is known to be activated throughout the lung parenchyma [110]. However, its activation within the vasculature is more controversial. Garcia-Morales et al. [107] have shown HIF-1α activation in pulmonary smooth muscle cells of PF and PF-PH patients, while Bryant et al. [110] only observed HIF-1α activation in the vasculature of patients with PH. However, in both studies, authors found that the expression of HIF-1α in the vasculature to be significantly increased in IPF-PH patients compared to IPF. Although these two studies report a different pattern of HIF-1α expression in IPF patients pulmonary vasculature, they both clearly demonstrate the presence of HIF-1α within the vasculature of PF-PH patients. HIF-1α activation may be caused by a decrease in the local concentration of oxygen in PF-PH compared to PF. This notion is supported by the study of Shorr et al. [38], which demonstrated that the need of oxygen in PF patients awaiting lung transplant independently correlated with the presence of PH. Another cause for HIF-1α activation in PF-PH may also be dysregulation in oxygen sensing, since it was demonstrated that HIF-1α is activated in normoxic conditions in PAH [111].

However, these hypotheses have yet to be investigated specifically in the context of PH secondary to PF. It is also interesting to note that mice with a specific knock-out of HIF-1α in pulmonary arterial endothelial cells when exposed to bleomycin develop PF similarly to wild-type mice, but do not exhibit elevated pulmonary pressure [110], further supporting a specific role of HIF-1α in the development of PH.

Our group has recently highlighted molecular differences between PF patients with and without PH [11] in addition to the histological characteristics described previously [80]. We found a significant increase of the transcription factor Slug within the lung macrophages of PF-PH patients compared to patients with PF alone [11]. Moreover, we described a remarkable new role for Slug up-regulation within the macrophages, which induces expression of the extra-cellular matrix prolactin-induced protein (PIP). PIP, in turn, promotes proliferation of endothelial and vascular smooth muscle cells. In addition, we demonstrated that bleomycin did not recapitulate the histological features of human in combined PF-PH patients. In the bleomycin-induced PF animal model, vascular remodeling is mainly restricted to fibrotic area of the lung similar to what is observed in PF patients. However, in PF-PH patients vascular remodeling is also evident in non-fibrotic preserved area of the lung. To overcome this translational hurdle in animal models, we developed a new pre-clinical animal model of PF-PH based on the serial administration of bleomycin and monocrotaline. Monocrotaline, an endothelial toxin extracted from Crotalaria spectabilis seeds, has been used for more than four decades to induce PH in rats. Using this combined model, we were able to recapitulate human histological features specific to PF-PH lungs. Furthermore, we demonstrated that this new rat model reproduced the Slug/PIP activation in macrophages as seen in the lungs of patients with PF-PH, in contrast to the model of bleomycin alone. More importantly, we found inhibition of lung Slug was able to prevent PH development in PF rats without affecting lung fibrosis. To the best of our knowledge, this is the first time that a potential treatment was tested in this unique pre-clinical model of PF-PH which could be effective for prevention of PH in PF patients [11].