Idiopathic pulmonary fibrosis (IPF) is a progressive disease of unknown origin characterized by increased matrix deposition, resulting in functional impairment and ultimately respiratory failure 1. Despite considerable progress in understanding the pathophysiology of this disease, there is still no efficient treatment 23. Animal models of pulmonary fibrosis (PF) are critical to investigate pathological mechanisms and are important for preclinical evaluation of novel therapies 45. To date, several hundreds of compounds were identified in various models of experimental PF and proposed to have therapeutic potential for IPF 5. However, only a small number have reached clinical trials and very few have been assessed in a "true" therapeutic setting 56, reproducing the clinical setting where patients with IPF tend to present late, with advanced fibrotic changes in their lungs 7. The vast majority of compounds that were examined in drug intervention studies in experimental PF are only successful in a preventive or prophylactic setting, which means they are given with or prior to administration of the fibrogenic stimulus 5. Assessment tools used in experimental PF are primarily based on histology and quantitative collagen analysis, providing a snapshot of a complex and chronic biological process 8, whereas clinical management of IPF relies on physiologic parameters (lung function, exercise test, echocardiography) and radiology (chest x-ray and high resolution computed tomography). Clinical trials are designed to detect changes in survival and/or lung function parameters 9.

We believe that intervention studies in models of PF would be more meaningful if the compound was administered in the fibrotic phase of the disease and not as prevention 8. The purpose of this study was to characterize a late-stage non inflammatory model of experimental PF in detail 10 by comparing conventional assessment tools to clinically more applicable approaches that do not require euthanasia. The underlying idea is to establish reproducible assessment tools for experimental PF in order to apply these later in drug intervention studies. Morphological changes were quantified with histomorphometry and by analysis of 3D micro-CT images. Lung function was characterized by measuring lung elastic recoil and VO₂max with treadmill exercise.

The data shows that micro-CT combined with invasive, non-terminal lung compliance is equally efficient as histomorphometry to generate a fibrotic grading but has the advantage that it can be done repeatedly in one animal over time. These evaluation tools directly reflect the clinical management of IPF patients. This approach is adapted to follow anatomical and functional changes in living animals over time and is well adapted to perform preclinical drug evaluation studies.

The purpose of this project was to evaluate non-terminal, "clinical" assessment tools in an animal model of experimental PF and compare them with standard methods used in the field. The obvious advantages of non-terminal tools are that severity and progression of fibrotic changes or – in the context of drug intervention studies – their potential reversibility can be followed in individual animals over time. This reduces variability and the number of animals that are needed to create relevant time-course data. In this study, we used an established model of PF which is based on adenovirus mediated overexpression of active TGF-β1 in rat lungs 10. This model incorporates early fibroproliferation while the transgenic protein is still expressed, not associated with significant inflammation (up to 10 days), followed by progressive and persistent fibrosis, which is driven by endogenous processes, likely mediated by locally activated TGF-β 1015.

Current assessment tools for animal models of PF are based on histology and biochemical analysis of lung tissue 1617. It is common practise in experimental models to use one lung for histology while the other lung is frozen and homogenized for biochemical and/or molecular analysis. Histomorphometry is considered as the gold standard. The method of Ashcroft requires multiple photomicrographs randomly acquired across the lung parenchyma and subsequent scoring by several investigators ranging from 0 to 8, where 8 would be a completely obliterated fibrotic field 11. We adapted this approach in the present study with blinded scoring of tissue samples by three different investigators. This method produces solid and reproducible data, but it is time consuming and only a small number of published studies seem to rely on it. Further, the scoring is usually performed on 20 high-power magnification images, which only accounts for a small fraction of the total lung, containing the risk of selection bias.

The quantification of collagen, or its major amino acid hydroxyproline, is extensively used in PF models as readout for disease severity and success of antifibrotic therapy. Compounds are considered unsuitable for fibrosis therapy if they have not shown a substantial decrease in lung hydroxyproline content in an appropriate model. There are a number of methodological issues associated with collagen quantification in the lungs. In experimental PF, the amount of hydroxyproline rarely exceeds a two-fold increase over baseline, and the levels are highly variable 18, narrowing the therapeutic window. This may partly explain why most published studies use a "prophylactic setting", giving the drug prior to initiation of fibrosis (usually by intratracheal bleomycin) 519. With this experimental design it can be assumed that hydroxyproline levels in drug-treated animals were never increased to a similar extend as in control animals. This means, the tested compound interferes with a process that is biologically different from the one in established fibrosis. Further, there is a difference between content and concentration of lung collagen. In our study we have quantified hydroxyproline related to dry lung weight to correct for potential pulmonary fluid accumulation or edema. Some determine collagen per gram wet weight, while others present total content per lung independent of the weight. Finally, it is likely that ultra-structural and qualitative changes in fibrotic matrix, such as collagen fiber stiffness or fiber cross-links markedly impact lung elasticity 20. These changes cannot be determined by quantitative analysis of hydroxyproline or histomorphometry, but may be captured by physiological measurements.

The routine assessment of IPF patients in clinic includes measurement of lung volumes, usually forced vital capacity (FVC) and total lung capacity (TLC). These parameters are difficult to obtain in rodents. In contrast, compliance can easily be measured and it has been shown that this parameter of lung stiffness is reduced in IPF patients and correlates with disease severity 21. We systematically performed pressure volume loops in live rats using an automated ventilator to assess changes in lung elasticity. The parameter K of the Salazar-Knowles equation, a reflection of the curvature of the upper portion of the pressure-volume curve, was reduced by 35 % in animals with PF up to 56 days after AdTGF-β1. There was a strong correlation between this "stiffness" parameter and the fibrosis score. The major advantage of this method is that it can be performed repeatedly in the same animals. We have here used a rat model of PF, but this method is applicable to both mice and rats 222324.

In addition to lung function, human clinical trials are powered to detect differences in exercise tests, mostly the 6 minute walk 25. Therefore, we have investigated the usefulness of VO₂max employing a rodent treadmill system in a contained Plexiglas tube. VO₂max was on average reduced in fibrotic rats by 10% compared to controls, but did not allow us to clearly distinguish between the two groups. It is still unclear why VO₂max alterations were less than expected. The total fibrotic volume defined by CT scan analysis accounted for 15–20% and it is possible that this is not sufficient to reduce VO₂max significantly. It may also be that compensatory lung growth and recruitment of alveolar reserves occurs in this rodent model, thus preventing the progressive functional decline that is so characteristic for lung fibrosis in humans. Further studies are required to better understand the complex changes in cardiopulmonary physiology in animal models of lung fibrosis. However, the presented data do not support VO₂max as a useful tool for assessment of fibrosis severity in this model.

Radiologic examination of lungs including high resolution computed tomography is a mandatory tool to make a clinical diagnosis of IPF/UIP 1. When CT scans are post-processed, three dimensional structures can be displayed with density histograms, making it an ideal candidate to follow total lung parenchymal changes over time. This approach is more widely propagated for chronic disorders characterized by tissue loss such as emphysema 132627, but may be of significant value in fibrotic lung disease. Several software packages are commercially available, allowing extraction of tissue areas of choice which are defined by certain pre-set parameters. After 3-D reconstruction, these areas can be measured and the volume of lung tissue characterized by a specific density range can be quantified. In CT images, fibrotic lung tissue appears denser compared to normal lungs. In contrast, a shift towards lower densities would be associated with air-space enlargement, emphysema or increase in lung volume. While it is relatively easy to "virtually dissect" areas with fibrotic density from normal lung, it is more difficult to clearly distinguish between fibrosis and mediastinal structures such as heart, aorta and vena cava. To circumvent this, we use a semi-automated 3D extraction of lung tissues with help of a "segmentation editor" and a "region growing tool". The volume of fibrotic tissue in our study was in the range of 10–20% of total lung and correlated significantly to conventional fibrosis score and lung function indicating that micro-CT is a promising alternative to predict lung fibrosis as previously described in the bleomycin model using mice 28 or rats 29. It is important to note that this CT technique does not distinguish between inflammatory and fibrotic changes in the lung tissue. Other imaging such as PET 30 or MRI 31 technology might be better suitable to differentiate between inflammation, edema and fibrosis, which is a relevant consideration when inflammatory models of fibrosis are used (e.g. the bleomycin model, in contrast to our non-inflammatory AdTGF-β1 model).

The major benefit of micro-CT combined with lung compliance over the more traditional tools histology and collagen content is that these assessments can be done repeatedly throughout the course of an experiment. It is obvious that validation experiments need to be performed to demonstrate that these assessments do not interfere with the progression of experimental disease. Our approach to future drug efficacy studies with these novel tools would be to obtain baseline evaluation in each animal before fibrosis is induced, then confirmation of fibrosis before the treatment is initiated and subsequent evaluation of potential treatment effects. For this purpose, a minimum number of 3 to 5 animals per group will be required, depending on the severity of fibrosis and the magnitude of treatment effects 8. The overall aim of the new approach is to directly track if documented fibrotic changes at one point of the study can be reversed by the drug intervention at a later time point.

Further, this approach assures that the entire lung is assessed in each animal, while the conventional tools are limited by performing histology on one and biochemical analysis on the other side of the lungs. While we believe that this novel approach eventually will prove to be superior compared to terminal assessment in detecting beneficial drug effects, we acknowledge that a series of drug intervention studies needs to be done to underscore this hypothesis.

In conclusion, we have translated clinically relevant assessment tools into an animal model of severe lung fibrosis and compared them with conventional methods. We show that the extent of fibrosis can be reliably monitored with small animal CT imaging and lung compliance measured with a rodent ventilator. These measurements have the advantage that they can be performed repeatedly in the same animals over time.

The author(s) declare that they have no competing interests.

KA conceived and coordinated the study, carried out the animal work, data analysis and wrote the draft of the paper. RL conceived and coordinated the study, carried out the animal work, data analysis and wrote the draft of the paper. LF participated in the design of the study, helped with the animal work and carried out the histological work. AM helped to carry out the animal work, histology analysis and edit the manuscript. AF helped in the design of the study, and CT analysis of the study. TF helped in the analysis and management of the computed tomography data. GM performed the VO2max analysis and helped in the analysis of the data. MI helped in the design of the study and in the analysis of the pressure volume loops. JG helped to conceive the study, participated in the design of the study and edited the paper. MK helped to conceive the study, participated in the design of the study, provided general supervision and helped to draft and edit the paper. All authors read and approved the final manuscript.