Background
Interstitial lung diseases (ILDs) comprise a wide and heterogenous group of pulmonary pathologies with a variable amount of inflammation and/or fibrosis. They can develop as a result of occupational and/or environmental exposure or as manifestation from other diseases (e.g., connective tissue diseases). In a substantial proportion of patients, no etiology is identified and the ILD is categorised as idiopathic [
1,
2]. Idiopathic pulmonary fibrosis (IPF) is the most common form of idiopathic interstitial pneumonias, with an estimated incidence of 2.1–6.3 cases/10
5 person-years [
1,
3,
4]. Without treatment, the median survival of IPF patients is 2.5–3.5 years [
5]. The diagnosis of IPF is based on a multidisciplinary approach involving clinical, radiological and pathological examinations [
4,
6]. After the exclusion of a secondary cause, a high-resolution computed tomography (HRCT) allows to evaluate the lung parenchyma and to eventually identify patterns of usual interstitial pneumonia (UIP) [
6]. In the absence of an UIP pattern on HRCT, a lung biopsy is recommended for pathological examination and identification of an UIP pattern [
6]. The evolution of the disease is unpredictable. While the majority of patients suffer from a slow degradation of their respiratory function, in some of them, a rapid degradation is observed. Acute exacerbations can also occur and are associated with a poor outcome [
5]. A close follow-up of IPF patients is therefore recommended with a monitoring of clinical symptoms, functional parameters such as the forced vital capacity (FVC), the diffusion capacity of the lungs for carbon monoxide (DLCO), six-minute walking tests and thoracic imaging [
7]. These elements allow the clinicians to calculate a prognostic score: the GAP score [
8,
9]. Various biomarkers were studied to better establish the prognosis of IPF patients. Among them, increased serum concentration of metalloproteinase-7, cancer antigen 19 − 9 and surfactant protein A and D seemed to be associated with a worse prognosis of the disease [
10‐
14]. However, these biomarkers, the GAP score and the HRCT abnormalities mostly reflect the severity of the disease but do not allow the evaluation of the response to antifibrotic treatments nor the intrinsic activity of the fibrotic disease at a selected time point.
A new biomarker candidate of IPF is the fibroblast activation protein-α (FAPα). FAPα is a transmembrane serine protease, expressed at the surface of activated fibroblasts and myofibroblasts [
15,
16]. FAPα detection by immunohistochemistry was previously studied in lung tissues and showed that FAPα was not expressed in normal lung tissue nor emphysema but well in fibroblast foci and fibrotic interstitium of IPF patients [
17]. Recently, a radiotracer based on a specific enzymatic inhibitor of FAPα (FAPI) was developed and showed a high specificity for cancer-associated fibroblasts both in murine and human tumors [
18,
19]. FAPI labelled with Gallium-68 (
68Ga) was then used to evaluate patients with IPF and lung tumors. Different uptake dynamics were observed in IPF lesions and lung tumors and a positive correlation was observed between lung density and FAPI uptake [
20]. These studies indicate that FAPα is a specific marker of lung fibrosis and therefore, the use of FAPα as a biomarker seems to be a promising tool to evaluate fibrogenesis and responses to antifibrotic treatments. To test this hypothesis, we studied FAPα expression in a mouse model of lung fibrosis. More precisely, we evaluated the concentration of FAPα in bronchoalveolar lavage fluids (BALF) and the uptake of FAPI coupled with Fluor-18 ([
18F]FAPI-74) in bleomycin-treated mice by PET/CT scan. We then evaluated the presence of FAPα in BALFs from IPF patients and correlated the results with the severity of the disease.
Materials and methods
Mice and lung fibrosis model
Eight to ten weeks old C57BL/6 female mice (Charles River, Wilmongton, USA) were used for this study. They were kept in a specific-pathogen-free-like environment with access to food and water at will. Lung fibrosis was induced with a single dose of an intratracheal instillation of bleomycin (doses ranging from 0.005U to 0.04U/60μL/mouse) or 60μL of NaCl 0.9% for the controls [
21]. Mice were anaesthetised with a mix of Xylazine (0.07 mg/mL, Rompun, Bayer, Germany) and Ketamine (0.36 mg/mL, Nimatek, Dechra, UK). At various time points, mice were sacrificed by exsanguination for lung and BALFs collection. Some mice received orally once a day 60 mg/kg of nintedanib resuspended in 1% Tween 80 or 1% Tween 80 alone, starting one day after the bleomycin instillation for 9 days. All experimental procedures were reviewed and approved by the local Ethics committee for animal welfare of the Université libre de Bruxelles (ULB) (reference: 201,801).
PET/CT imaging
FAPI was a gift from the society SOFIE/iTheranostics (Dulles, USA). It was coupled with the radioisotope
18F, produced by the cyclotron of the department of nuclear medicine of Brussels University Hospital, Brussels, Belgium. From 2.05 to 5.68 MBq of [
18F]FAPI-74 diluted in 100 to 200μL of NaCl 0.9% were intravenously injected in mice. PET/CT imaging was then performed under isoflurane anaesthesia, from 0 to 110 min after the injection of the radiotracer using a preclinical PET/CT tomograph (NanoScanPET-CT, Mediso, Hungary). The areas of interest (both lungs) were automatically identified on CT then manually corrected for each mouse (Additional Fig.
1A). The muscle uptake was assessed in the right quadriceps. The mean activity in each area of interest was expressed in Bq/mL and was then divided by the ratio A
0 (injected activity decay-corrected at the start of PET acquisition (Bq) to animal weight (g)) to obtain the mean standardised uptake value (SUVmean). A ratio between the lung uptake and the muscle uptake (ratio lung to muscle or RLM) was also calculated in order to normalise the activity measured in the lung.
Qualitative and quantitative assessment of fibrosis and fibrogenesis in mice
Lungs from control and bleomycin-treated mice were obtained after dissection. Left lungs were fixed with 4% paraformaldehyde and then embedded in paraffin. Tissues sections (4 μm) were stained with Masson’s trichrome. Ashcroft modified score was used to evaluate fibrotic changes [
22]. Eight fields at 20x magnification were analysed. Right lungs were directly stored at -80 °C. Some of these lungs were used to measure hydroxyproline (OH-proline) as previously described [
23]. Others were used to perform quantitative reverse transcription polymerase chain reactions (RT-PCR). Briefly, lungs were homogenised at 4 °C and then RNA was extracted and purified with Rneasy Mini Kit (Qiagen, Hilden, Germany). Samples were then reverse transcripted into cDNA and products were analysed by quantitative RT-PCR. The amplification reaction was performed using SYBR green (Bio-Rad, Hercules, USA) and compared with a standard curve. All expression values were corrected with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as housekeeping gene. Primers are listed in Additional Table
1.
BALFs were obtained at various time points by flushing lungs with 1mL of NaCl 0.9%. Supernatant was collected after a 10-minute centrifugation at 300 g and then stored at -80 °C for further analysis. Measurements of FAPα, monocyte chemoattractant protein-1 (MCP-1) and interleukin 6 (IL-6) on BALF were performed using ELISA kit (R&D systems, Minneapolis, USA). Samples were analysed in duplicates and according to the manufacturer instructions.
Collection of patients bronchoalveolar lavages and clinical data
BALFs were collected from patients in the process of the diagnosis of their ILD. The IPF diagnosis was thereafter confirmed by a multidisciplinary discussion and based on the 2018 ATS/ERS/JRS/ALAT Clinical Practice Guideline [
6]. Control BALF samples were obtained from patients with a pulmonary nodule, from non-infected lung transplant patients with normal HRCT, patient with asthma or chronic cough. Clinical data were gathered from the patient’s medical record including age, gender, body mass index (BMI), tobacco, hypertension, diabetes, and disease progression. The presence of an acute exacerbation was considered as a respiratory deterioration for less than a month associated with new bilateral groundglass opacities or a consolidation on HRCT, not fully explained by cardiac failure or fluid overload [
24]. The progression of the disease was based on the 2022 ATS/ERS/JRS/ALAT Clinical Practice Guideline [
25]. Patients with a progressive disease presented at least two of the following criteria: worsening of the symptoms, worsening of the fibrotic features on HRCT, more than 5% worsening of the FVC (absolute values) or more than 10% worsening of DLCO in 12 months (absolute values). Plasma and BALF FAPα levels were measured by ELISA (R&D systems, Minneapolis, USA). All samples were analysed in duplicates and according to the manufacturer’s instructions.
Lung tissue sample collection and immunohistochemistry
Immunohistochemical staining was performed on formalin-fixed paraffin-embedded lung samples from IPF patients. Tissue sections (4 μm) were subjected to standard immunohistochemistry on a PT Link and an Autostainer Link 48 (Agilent Technologies Belgium S.A./N.V., Diegem, Belgium). Briefly, tissue sections were deparaffinized, rehydrated and were subject to antigen retrieval using EnVision FLEX Target Retrieval High pH Solution (EDTA, pH9; Agilent), for 20 min at 97 °C. Then, slides were incubated with peroxydase blocking solution for 5 min and with the rabbit monoclonal anti-FAPα antibody for 30 min (dilution 1:200, Abcam, Cambridge, UK). The slides were washed and incubated with the EnVision + System-HRAP labelled Polymer anti-rabbit Ig antibody for 30 min (Agilent). Immunostaining was revealed by incubation with diaminobenzidine and hydrogen peroxydase (Agilent). The slides were counterstained with hematoxylin for 5 min, dehydrated and mounted. They were digitized at 20x using a NanoZoomer 2.0 HT (Hamamatsu, Hamamatsu-City, Japan) before assessment.
Statistical analysis
All data were tested for normality using the Shapiro-Wilk test and according to the distribution, parametric or non-parametric tests were applied. Differences between the groups were assessed by one-way ANOVA or Kruskal-Wallis test. Holm’s Sidak or Dunn’s tests were used as post-hoc tests. When only two groups were compared, student t or Mann-Whitney tests were performed. Two-way ANOVA was used for the comparison of 2 independent variables followed by Bonferroni’s correction for multiple comparisons. Data are reported as mean ± standard deviation (SD) or median with interquartile range. For correlation analysis, parametric Pearson or non-parametric Spearman correlation were performed and data are reported as Pearson or Spearman r and p-value. Kaplan-Meier survival curves were generated and a log-rank test was applied to compare the two groups of patients. The analysis was done on Graph Pad Prism 8 (San Diego, USA) and all significance levels were fixed at 0.05.
Discussion
The present study aimed to evaluate the role of FAPα assessment either by PET/CT using a
18F-labelled FAPα inhibitor and by dosing FAPα BALF levels by ELISA. Regarding [
18F]FAPI-74 imaging in an in vivo mouse model of lung fibrosis, we showed that this radiotracer is a useful tool to investigate fibrotic activity. Indeed, a higher and significant uptake of [
18F]FAPI-74 is observed at days 10 and 16 post-bleomycin instillation, the period when fibrogenesis occurs in the lung, but neither at day 3, during the early inflammatory phase nor at day 28, during the late fibrotic stage of the model when fibrosis is installed but without further progression. Our results are consistent with recent reports in the literature. Rosenkrans et al. evaluated the lung uptake of [
68Ga]FAPI-46 in a mouse bleomycin model and observed a higher and significant lung uptake of the radiotracer at 7 and 14 days with the highest uptake seen at day 14 [
26]. Another study successfully used [
68Ga]FAPI-46 to discriminate lesions of lung carcinoma and lung fibrosis in ILD patients [
20]. Finally, Yang et al. used [
68Ga]FAPI-04 and observed a higher uptake of the radiotracer in lung from IPF patients compared to controls and the FAPI SUVtotal was negatively correlated with DLCO [
27]. Therefore, [
18F]FAPI-74 PET imaging could provide additional information to respiratory function and HRCT, since these methods evaluate the impact and severity of the disease rather than the intrinsic fibrotic activity. Thus, [
18F]FAPI-74 PET imaging could be an important tool to determine the activity of the disease, its prognosis and to support the decision to prescribe a specific anti-fibrotic therapy, especially in non IPF fibrotic disease (for example ILD secondary to inflammatory disease) helping to decide between the initiation of an antiinflammatory drug alone or in combination with an antifibrotic one.
[
18F]FAPI-74 PET/CT imaging appears to be a better biomarker of the fibrosis activity as compared to [
18F]FDG. Indeed our group previously showed, in the bleomycin model, that [
18F]FDG PET/CT imaging had a significant and higher uptake in lungs from bleomycin-treated mice during the fibrotic phase of the model but also when fibrotic activity is not yet present or already stable, i.e. during the inflammatory and the late fibrotic phases [
23]. Here we show that [
18F]FAPI-74 PET imaging is reflecting the current active fibrogenesis as its uptake is only observed during the fibrogenesis period. Another advantage compared to [
18F]FDG is that [
18F]FAPI-74 PET imaging is unaffected by changes in lung density. Indeed, even if [
18F]FDG uptake in IPF patients is associated with decrease in respiratory function and clinical severity [
28,
29], when a correction for lung density is applied, no significant correlation between [
18F]FDG uptake and one-year prognosis was observed in a cohort of IPF patients [
30]. [
18F]FDG PET/CT was also unable to assess the early response to antifibrotic treatment [
31]. On the contrary, [
18F]FAPI-74 PET imaging appears more specific to fibrogenesis and is not affected by changes in lung density as no uptake is noticed in dense organs such as the brain or the liver in normal condition.
Our study identified one important limitation of [
18F]FAPI-74 PET/CT imaging in the bleomycin model. The determination of regions of interest in mice is challenging due to the presence of non-specific uptake by the thoracic wall and spine (Additional Fig.
1C). The difficulty to discriminate lung uptake from the uptake in the surrounding tissue reduced the precision of measurements. Of note, as the human lung is bigger, this issue is not applicable in patients with fibrotic ILDs. This is probably one of the reasons why we failed to demonstrate a clear dose-response effect on the [
18F]FAPI-74 lung uptake in our mouse model and the fact that only at the highest dose of bleomycin (0.02U) a higher uptake of the tracer was measurable in the lungs.
Beside PET/CT evaluation of [
18F]FAPI-74 lung uptake, we identified in the BALF of bleomycin-treated mice a significant increase of the FAPα protein as assessed by ELISA. In line with our results, Yang et al. reported an increase of the protein expression of FAPα in homogenised lungs from bleomycin-treated mice as compared to controls [
27]. Interestingly, contrary to what we observed for the pulmonary uptake of [
18F]FAPI-74, a strong and significant dose response effect was observed on FAPα levels in BALF in experiments with variable doses of bleomycin. Furthermore, strong correlations between FAPα concentration in BALF and the Ashcroft modified scale, the concentration of OH-proline in the lungs, and the weight loss were identified. This suggest that BALF concentration of FAPα has a better sensitivity than [
18F]FAPI-74 PET/CT imaging, in the bleomycin model. The lung fibrosis is commonly evaluated by the measurement of OH-proline. However, no dose-response effect was observed in our experiment at day 10 after bleomycin instillation, on the contrary of FAPα BALF concentration. We thus think that FAPα is a better and early marker of fibrosis. BALF FAPα concentration appears also better than the Ashcroft modified scale. Indeed, this latter is a semi-quantitative method to evaluate fibrosis with an inter-observer variability [
22]. Moreover, no significant difference was observed in the Ashcroft modified scale between mice receiving or not nintedanib as antifibrotic treatment, on the contrary of a significant decrease of their BALF FAPα concentration. Altogether, BALF FAPα concentration appears then useful to monitor therapeutic response in preclinical models of lung fibrosis and probably more sensible than the measurement of OH-proline and the Ashcroft modified scale.
In humans, we further confirmed the interest of FAPα measurement in BALF. To our knowledge, we are the first to report that the concentration of FAPα in BALF is significantly higher in IPF patients. Moreover, the higher levels were found in patients with a progressive form of IPF and, in two cases, during an acute exacerbation. Accordingly, our data showed an interesting prognostic role of FAPα measurement in BALF using a cut-off value of 192.5 pg/mL. This confirms the potential of this biomarker in the follow-up of patients affected by a disease characterised by its unpredictable course made of periods of stability, steady progression and acute worsening. The measurement of FAPα in BALF by ELISA seems a useful, easy and cheap way to evaluate the fibrotic activity in the lungs. It could be implemented in the management of ILD patients and have prognostic value at the diagnosis. The immunohistochemistry analysis allowed us to identify the cell populations expressing FAPα, including activated fibroblasts and myofibroblasts, as already described [
17] but also alveolar epithelial cells that appeared hyperplastic. This population could be responsible of the higher FAPα concentration in BALF, following desquamation in the alveolar lumen or through direct secretion. Given that there is a soluble form of FAPα, also known as α2-antiplasmin cleaving enzyme [
32], the increased expression of FAPα on alveolar epithelial cells or activated fibroblasts and myofibroblasts could lead to the higher FAPα concentration in BALF. On the contrary, alveolar macrophages, the main cell population found in BALF [
6] did not seem to express FAPα. Arnold et al. identified a population of cancer-associated macrophages expressing FAPα and that presented characteristics of alternatively activated macrophages [
33]. However, to our knowledge, FAPα is not expressed by alveolar macrophages and based in our data they does not seem to be the source of the FAPα detected in BALF. Further studies should be performed to identify the cell population(s) that lead to higher FAPα BALF levels.
Our study has some limitations. IPF patients and controls were not matched by age and IPF patients were significantly older than controls. However, the impact of age on FAPα levels is probably limited as no strong association has been demonstrated in the limited available literature: in a cohort of patients with oesophageal squamous cell carcinoma, no significant difference of plasma FAPα concentration was observed between patients older or younger than 62 years old [
34]. Also, another group measured FAPα concentration in plasma samples from patients with ST-elevation myocardial infarction (STEMI) and healthy donors. Contrary to our results, the group with the higher FAPα levels in plasma (healthy donors) was younger than STEMI patients [
35]. Another important limitation is that our study was performed on a very small cohort of patients from a single center. Therefore, multicenter studies should be initiated to validate the use of BALF FAPα as a biomarker of lung fibrosis and prognosis of IPF. Unfortunately, bronchoalveolar lavage is an invasive act associated with some risk of exacerbation of the underlying disease [
36,
37]. Therefore, it could not be repeated periodically to assess changes in the fibrotic activity of the disease. Further research could focus on the determination of FAPα in other biological fluids such as plasma, induced sputum or even exhaled breath condensates.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.