Quantitative Micro-CT Analysis of Lung Parenchymal and Airway Remodeling in a Ferret Model of Pulmonary Fibrosis

Summary

Here, we present a protocol for using micro-CT to quantitatively assess the characteristics of bleomycin-induced lung fibrosis in a ferret model. This ferret model demonstrates strong translational potential for understanding pulmonary fibrosis and testing new therapies.

Abstract

Pulmonary fibrosis (PF) is a chronic and progressive lung disease characterized by repetitive alveolar injury leading to parenchymal thickening, scarring, and respiratory impairment. Current rodent models, particularly those utilizing bleomycin (BLEO), have limited translational relevance to human idiopathic pulmonary fibrosis (IPF) due to anatomical differences such as the absence of respiratory bronchioles. To address this limitation, we developed a large-animal model of pulmonary fibrosis in ferrets, which possess respiratory bronchioles similar to those in humans. Three-month-old wild-type ferrets received intratracheal instillation of BLEO. Lungs were harvested 8 weeks post-treatment for micro-computed tomography (micro-CT) and histological evaluation. Owing to the limited spatial resolution of multidetector CT (MDCT) in visualizing the ferret small airways, excised lungs were imaged ex vivo using a micro-CT scanner at controlled airway pressures (0 cmH₂O and 25 cmH₂O) to simulate expiratory and inspiratory conditions.

Micro-CT imaging revealed distinct structural differences between normal and BLEO-treated ferret lungs. Normal lungs displayed low attenuation and preserved architecture, whereas BLEO-treated lungs showed markedly increased attenuation consistent with fibrosis. Features characteristic of human fibrotic lung disease, including honeycombing with cystic spaces, thickened interlobular septa, and ground-glass opacities indicative of early fibrotic or inflammatory changes, were evident. Using quantitative micro-CT analysis, it was observed that airways in BLEO-treated ferret lungs exhibited significantly reduced radial expansion, longitudinal stretching, and volume change at both applied air pressures. Furthermore, generation-matched analysis revealed significant thickening of airway walls in regions with visually apparent fibrosis. Histological assessment, including Masson's trichrome staining, confirmed extensive collagen deposition and fibrosis.

This study demonstrates that BLEO-treated ferrets develop radiologic, mechanical, and histopathologic features resembling human pulmonary fibrosis. Furthermore, micro-CT enables high-resolution assessment of fibrotic distribution, airway dynamics, and lung volume changes. These findings highlight the potential of the ferret as a translational model for studying PF pathogenesis and evaluating novel therapeutic strategies.

Introduction

Pulmonary fibrosis (PF) is a chronic and progressive interstitial lung disease in which repeated alveolar injury drives parenchymal thickening, fibrosis, and progressive respiratory dysfunction¹. Pulmonary fibrosis can be induced by various exposures, including medications, environmental toxins, infections, underlying autoimmune disorders, or it can be idiopathic¹. While medications have been developed to slow the progression of disease, particularly for patients with idiopathic pulmonary fibrosis (IPF), pulmonary fibrosis remains incurable, with a survival rate of 3–5 years after diagnosis^1,2. In most cases, the only definitive treatment is a lung transplant; therefore, there is an urgent need to develop new and more effective therapeutic options.

Bleomycin is a medication that is commonly used in humans for the treatment of malignancy; however, its use is limited by pulmonary toxicity, which can rapidly progress to pulmonary fibrosis³. Following the discovery of this side effect, bleomycin was used in research to develop animal models of pulmonary fibrosis. Mice have been commonly used for this purpose^4,5,6 with various methods of Bleomycin administration, including IV and intratracheal instillation^7,8. While fibrosis does develop in these models, none of them encompasses all the cardinal features of pulmonary fibrosis seen in humans⁸. Rodent models of pulmonary fibrosis fail to reflect key features of advanced disease, including honeycomb cyst formation and bronchiolization of the distal alveoli⁹. In addition, studies have shown that bleomycin-induced lung injury begins to resolve spontaneously in rodents approximately 28 days after bleomycin treatment¹⁰. Additionally, none of the agents shown to inhibit fibrosis in these models have had an equivalent effect in humans, limiting our therapeutic options¹¹.

Domestic ferrets (Mustela putorius furo) are increasingly used in respiratory biomedical research because of their anatomical similarities to the human respiratory system¹². In addition, they are susceptible to similar respiratory pathogens that infect humans, including respiratory syncytial virus (RSV) and influenza, showing their physiological similarities¹². Ferret models of bleomycin demonstrate bronchiolization, the formation of fibrotic foci, and the presence of aberrant basaloid-like (KRT7⁺/KRT17⁺/KRT5^–/TP63⁺) cells⁹. A single dose of bleomycin induced sustained lung fibrosis in ferrets, with restrictive physiology and fibrotic lung abnormalities persisting for at least 22 weeks¹³. This finding differs substantially from the widely used mouse bleomycin model, where fibrosis is generally self-limiting and begins to resolve after approximately 28 days, with substantial regression occurring by 6–8 weeks, suggesting that the ferret model may better recapitulate the chronic and progressive nature of human pulmonary fibrosis. Moreover, ferrets have recently been used in cystic fibrosis research, a systemic disease that most commonly affects the respiratory system, because of their similarity to humans¹⁴. Therefore, these animals may serve as better models for PF than prior rodent models.

This current study aimed to develop an animal model of pulmonary fibrosis that more closely resembles the human condition. We administered intratracheal bleomycin to domestic ferrets. After 8 weeks, these ferrets demonstrated radiographic and histological changes similar to those observed in humans with pulmonary fibrosis.

Protocol

All animal procedures were conducted in accordance with the guidelines and regulations of the Institutional Animal Care and Use Committee (IACUC) at Michigan State University (Animal protocol approval no. PROTO202300066). Animal care and experimental protocols were approved by Michigan State University and performed in compliance with institutional standards for the ethical use of animals in research.

1. Bronchoscopic instillation of bleomycin in ferrets

1. Acquire male ferrets that are within the age range of 3–4 months. The average weight of a male ferret at this age should be approximately 0.9–1.6 kg. House the animals at 21 °C at 40%–50% humidity.
2. Administer 3%–5% isoflurane gas to the ferret using a precision vaporizer anesthesia machine.
3. Maintain anesthesia by covering the nose of the ferret with a nose cone and continue administering isoflurane gas at a concentration of 3%–5%.
4. When the ferret is adequately sedated and recumbent, initiate the bronchoscopy using a 2.7 mm outer diameter bronchoscope.
5. Instill a mix of bleomycin (2 U/kg body weight) and saline via the working channel into the distal trachea to allow for diffuse dispersion of the bleomycin. Ensure that the total volume instilled does not exceed 1 mL. Return the ferrets to their respective housing units and monitor for 8 weeks.
   NOTE: The instillation should take no longer than two min to complete.
6. Administer isoflurane using a precision vaporizer gas anesthesia machine. Cover the nose of the ferret with a nose cone and administer 1%–5% isoflurane until the ferret is recumbent. Continue with gas exposure until apparent clinical death can be confirmed. Administer intravenous euthanasia solution at a dose of 90 mg/kg and monitor vitals until death is confirmed.
7. Harvest the lungs 8 weeks post-treatment for micro-computed tomography (micro-CT) and histological evaluation.

2. High-resolution micro-CT imaging of ferret lungs

1. Prepare excised lung tissue for imaging by removing extraneous tissue and blood via the inferior vena cava incision to allow blood drainage. Perform the CT immediately after preparing the lungs.
2. Obtain tomographic micro-CT images using a micro-CT system with the following image acquisition settings: 90 keV, 88 µA, power setting with a 60 µm Copper and 500 µm Aluminum X-ray collimator filter, using a 72 mm field of view (FOV) at 144 µm isotropic image resolution.
3. Acquire a total of 5 images at 18 s standard gantry times using an automated stitching protocol to overlap, stitch, and reject duplicate slices to create a final stitched image for airway pipeline analysis; this will result in an 68 mm cylindrical diameter FOV x 186 mm total stitched image depth (z plane) at 144 µm voxel resolution, allowing for the complete capture of the entire ex vivo ferret lung block.
   1. Repeat this imaging protocol for each set of ex vivo ferret lungs at both 0 cmH₂O and 25 cmH₂O pulmonary pressures.
   2. Use a pressure-monitoring system connected directly to the lung to precisely control and maintain airway pressure during imaging.
   3. Couple the pressure meter to a regulated airflow source to allow continuous adjustment of the intrapulmonary pressure.
   4. Before imaging, set the system to the desired pressure level (either 0 cmH₂O or 25 cmH₂O) and inflate the lung using a steady airflow.
   5. While imaging, use the pressure meter to provide real-time feedback on the airway pressure and compensate for any minor pressure fluctuations using the airflow source.

3. Image processing and segmentation

1. Load the reconstructed micro-CT image volumes in DICOM format, which are converted to a neuroimaging informatics technology initiative (NIfTI) format, into the airway segmentation pipeline.
2. Define the region of interest (ROI) in the micro-CT image by manually cropping axial slices at the superior end to exclude the tube and other tracheal inserts (see Figure 1A).
3. Perform airway segmentation using the automated Freeze-and-Grow algorithm^15,16 described in the following major steps.
   1. Place a seed inside the trachea on the topmost axial image slice within the ROI; create an empty image array of input image size to represent the initial confident airway volume (CAV) with all voxels assigned to '0'; and add the seed voxel to CAV by setting its value to '1'.
   2. Set an initial conservative micro-CT intensity threshold t = -1000 HU for segmentation of the airway lumen.
   3. Create an empty image array of input image size to represent the forbidden volume (FV) with all voxels initialized to '0'.
   4. Repeat iterative segmentation until convergence of airway lumen segmentation defined by one of the two flags: 1) the entire threshold range (-1000 to -600 HU) is checked, or 2) all '0' voxels adjacent to '1' voxels in the CAV image are marked as '1' in the FV image.
      1. Compute CAV centerline tree¹⁷ and identify terminal branch endpoints¹⁸.
      2. Apply binary thresholding on the micro-CT image at the current value of t and compute the new lumen connected volume¹⁸ with CAV as seeds over the thresholded region (voxel value ≤ t), while excluding FV.
      3. For each CAV endpoint, compute the associated voxels within the new lumen volume, excluding CAV that are closer to that endpoint than to any other CAV endpoints.
      4. Detect the CAV endpoint with a large volume of associated voxels (>150 voxels, equivalent to 0.5 mm³) as potential leakage sites in the new lumen volume.
      5. Confirm a potential lumen leakage using the automated criteria defined in Nadeem et al.¹⁵.
      6. Remove any leakages in lumen volume by deleting the subtree volume distal to the leakage roots, and add voxels around leakage roots to augment FV using the algorithms described previously¹⁵.
      7. Assign the leakage-removed lumen volume to update the CAV using the algorithms described previously¹⁵.
      8. Increment the threshold t by '1'.
      9. Go to step 3.3.4.
4. Perform computation of a one-voxel thick airway centerline tree using a previously validated automated algorithm¹⁷ described in the following major steps.
   1. Input the segmented airway tree as a binary NIfTI image.
   2. Compute the initial airway centerline tree using a medially preferred minimum-cost path approach¹⁷ described in the following major steps.
      1. Create an empty skeletal image array of input image size with all voxels assigned to '0'; add the automated seed of step 3.3.1 as the initial skeleton voxel with value '1'.
      2. Create an empty image array of input image size with all voxels assigned to '0' to represent the marked airway volume image.
      3. Identify the voxel in the unmarked airway volume that is farthest to the current skeleton and compute the medially-preferred minimum-cost path connecting this point to the current skeleton using the algorithm described previously¹⁷.
      4. Add the resulting path as a new skeletal branch to the skeleton image array, and update the airway-marked volume by applying local scale-adaptive dilation along the branch, as described previously¹⁷.
      5. Repeat steps 3.4.2.3–3.4.2.4 until the marked airway volume is filled.
   3. Prune spurious topological branches while preserving true airway centerlines using a local scale-based selection strategy on branch length, as described previously¹⁷.
   4. Eliminate simple points¹⁹ in the airway centerline that are not branch endpoints (i.e., those adjacent to more than one centerline voxel) to ensure topological consistency and maintain a single-voxel-thick centerline.
5. Perform segmentation of lung parenchyma in micro images of ferret lungs at 0 cmH₂O.
   1. Manually delineate lung parenchyma on every fifth coronal image slice using computational algorithms and graphical interfaces embedded within ITK-SNAP²⁰.
      NOTE: The micro-CT intensity contrast between the lung parenchyma and the surrounding background at 25 cmH₂O inflation state was insufficient, prohibiting reliable segmentation of the lung volume.
   2. Apply image interpolation to generate a segmentation of lung parenchyma at intermediate image slices.
   3. Perform a secondary quality control review to ensure structural continuity of parenchymal segmentation along both axial and sagittal directions.
   4. Exclude airway lumen and wall regions from the lung parenchyma using a local scale-based dilation of lumen volume, and then subtract the dilated volume from the lung parenchyma.
6. Quantify mechanical measures of airways²¹.
   1. Quantify breathing-related radial expansion of individual airways by measuring changes in spatially-matched airway lumen cross-sectional area (CSA) between inspiratory (25 cmH₂O inflation) and expiratory (0 cmH₂O inflation) lung volumes. Compute the radial expansion at an airway branch b, denoted by Δ_air-R(b), as:
          (1)
      where, CSA_ins(b) and CSA_exp(b) denote the airway lumen CSA at inspiratory and expiratory lung volumes. CSA at b is computed using a radial line tracing and locally adaptive half‑max methods over the central half of the branch b²².
   2. Quantify breathing-related longitudinal stretching of individual airways as follows:
          (2)
      where L_ins(b) and L_exp(b) denote geodesic path‑lengths of the airway branch b at inspiratory and expiratory lung volumes, respectively. The geodesic path length of a branch is computed by tracing the centerline path from the carina to the distal endpoint of the branch²¹.
7. Quantify intensity-based measures of lung parenchyma.
   1. Spatially visualize micro-CT intensity distribution of the lung parenchyma in healthy and fibrotic ferrets by generating a "look-through 3D volume rendition" of the lung parenchyma volume in the context of the airway tree. Accomplish the look-through effect using micro-CT intensity-defined transparency mapping and color coding. Apply the same transparency mapping and color-coding scales to both healthy and fibrotic ferret lungs. The following steps are applied to generate the look-through 3D volume rendition.
      1. Use 3D Slicer²³ to open the three dimensional (3D) NIfTI image of airway segmentation at 0 cmH₂O inflation as a "Segmentation".
      2. Generate a 3D airway representation using the Segmentations module. Select the airway volume as the input and, in the Representations tab, click Create under the Closed surface option.
      3. Import the micro-CT intensity image with separated lung parenchyma into 3D Slicer as Volume. In this image, the original micro-CT intensity values are preserved for lung voxels, while all non-lung voxels are assigned a value of -2000 HU.
      4. Configure color and transparency mapping for lung parenchymal intensities using the Volume Rendering module. Map intensity values of -1500 HU, -600 HU, 0 HU, and 1500 HU to black, green, yellow, and red, with opacities of 0, 0.15, 0.3, and 1, respectively.
   2. Use the algorithm developed in this study to estimate the fraction of diseased tissue within the lung parenchyma by analyzing the observed intensity distribution from micro‑CT imaging. The method assumes that the intensity values of healthy parenchymal regions follow a Gaussian distribution. In contrast, fibrotic regions, with characteristically higher intensities, distort the Gaussianity of the right side of the observed histogram. Accomplish this task in the following steps.
      1. Compute histogram for isolated lung parenchymal image in step 3.7.1.3 with 50 bins over the intensity range of -1500 HU to 500 HU.
      2. Compute the mean of the target Gaussian distribution as the mode of the histogram and derive the standard deviation parameter using the intensity values to the left of the mode on the histogram.
      3. Quantify fibrotic tissue volume as the excess area on the right side of the observed histogram that lies above the expected Gaussian distribution curve defined by the mean and standard deviation parameters obtained in step 3.7.2.2.
      4. Compute the diseased tissue fraction as the ratio of the fibrotic tissue volume to the total histogram count representing the parenchymal region volume.

Results

Out of the five ferrets, the excised lungs from one healthy ferret and one fibrotic ferret were imaged using a micro-CT scanner at controlled airway pressures of 0 cmH₂O and 25 cmH₂O to simulate expiratory and inspiratory lung inflations, respectively.

Figure 1 illustrates the intermediate results of the image processing pipeline for the micro-CT image (Figure 1A) of an ex vivo ferret lu...

Discussion

This study describes a protocol combining intratracheal bleomycin administration in ferrets with high-resolution ex vivo micro-computed tomography (micro-CT) to characterize structural and mechanical features of pulmonary fibrosis. Micro-CT can detect small airways as small as 0.2 mm and measure volumes, airway thickness, density, and morphology. Increased tissue density, volume loss, and regional heterogeneity closely correspond to histopathology. The primary goal was to establish feasibility and demonstrate th...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
1 mL Luer locking syringe	BD	309628
10 mL Luer locking syringe	BD	303134
24 G ´ 3/4" Exel safelet catheter	EXELINT INTERNATIONAL CO.	26751
3 mL Luer locking syringe	BD	309657
30 mL Luer locking syringe	BD	302832
5 mL Luer locking  syringe	BD	309647
BD Precision glide needle 18 G ´ 1 1/1"	BD	305196
BD Precision glide needle 20 G ´ 1"	BD	305175
BD Precision glide needle 23 G ´ 1"	BD	305145
Dulbecco's phosphate-buffered saline (DPBS -/-)	Gibco	14190250
Endotracheal intubation tubes with cuff oxygen catheter	Caphstion	2.0MM
ET Tube, Uncuffed, 2.0	Teleflex	5-10404
Ethicon Permahand silk suture, Size 2-0, No Needle, 18", 36/Box, A185H	Fisher Scientific	50-209-2809
Invitrogen RNAlater stabilization solution	Fisher Scientific	AM7021
Isoflurane	Covetrus	11695-6777-2
Lubricant PM ointment	AACE Pharmaceuticals	71406-124-35	Eye ointment
MADgic 700 atomizer	Teleflex	MAD700
Manometer, Professional air pressure meter	Leaton	T168
Rodent ventilator model 683	Harvard Apparatus, Inc.	55-3438
Sterile saline, 0.9%	CellPro	DS0329
Veterinary pulse oximeter	uPM60V	UVMI