CHARACTERIZATION AND EVALUATION OF CLINICALLY RELEVANT READOUTS IN A PRE-CLINICAL MODEL OF IDIOPATHIC PULMONARY FIBROSIS (IPF)

Idiopathic pulmonary fibrosis (IPF) is a chronic, progressive age-related interstitial lung disease (ILD) with a poor prognosis and very limited therapeutic options (Raghu et al. 2015; Spagnolo et al. 2018). To date, Pirfenidone and Nintedanib are the only two therapies approved for IPF worldwide. However, these drugs can slow-down lung function decline without really stopping or reverting the fibrotic process, and in addition their use is associated with a series of side effects. These characteristics of IPF indicate the ongoing critical importance of drug discovery efforts and highlight as animal models have become an indispensable tool in translational research of human lung disease as IPF, pivotal to test potential new therapeutic drugs. Despite the wide variety of different in vivo model for IPF (Tashiro et al. 2017), there is no consensus on preclinical tests best used to investigate IPF mainly due to some limitation that need to be considered when extrapolating findings from animal model to the human disease (Carrington et al. 2018). In fact one of the most challenging aspects of drug discovery for IPF remains the identification of new therapies that can be translated effectively to the clinic, implying that very few compounds that have shown efficacy in animal models have been successful in human clinical trials and concluding that most of the preclinical models are poorly predictive and scarcely resembling the human disease. Currently the majority of new drugs investigated in preclinical models of IPF are dosed using a prophylactic dosing regimen, whereas patients are almost always treated after the fibrosis is well established but probably, one of the most relevant limitation remains some measurements that should not be easily done in both humans and mice. In this scenario, the main goal of this PhD project was to generate a robust and reliable preclinical model of pulmonary fibrosis, introducing novel readouts, suitable to select and to identify new pharmacological treatments for IPF with an higher translational potential to make the model more impactful to identify new potential treatments for IPF. To achieve this aim, we have introduced in our bleomycin (BLM)-induced mouse model (Stellari et al. 2017), novel readouts such as the Forced Vital Capacity (FVC) and the Diffusion Factor for Carbon Monoxide (DFCO) that are poorly used in preclinical practice but that are respectively the primary and secondary endpoints in clinical trials for IPF (Spagnolo et al. 2013) to monitor patients during the treatments. Additionally, also some emerging biomarkers currently under evaluation in the clinical setting were explored in the model. Finally, we have also worked to refine the histological analysis which still remains an important complementary evaluation to be coupled to the functional readouts. Currently the common histological analysis utilized in preclinical models of lung fibrosis is represented by the Ashcroft scoring system, which revealed some disadvantages such as a time-consuming process, operator-dependent results, limited sensitivity and, most critical, inability to get a direct link to clinics. Therefore, we introduced an automated image analysis by using an artificial intelligence (AI) approach, which improved this analysis recognizing histological features with more accuracy and consistency, reducing significantly the time of the analysis and making the evaluation independent from the operator. All these new readouts, explored firstly in a time course experiment at several time-points up to 49 days from the first bleomycin administration, showed the same profile over time observed with histology in terms of development of fibrotic disease. Next, the relevance of these novel readouts in the bleomycin model will be assessed by the pharmacological validation with Nintedanib, providing evidence of the robustness and the predictivity of these functional readouts with the final aim to enhance the translational and the predictivity values of the model. In summary, this project demonstrated that in the mouse BLM-induced lung fibrosis model it has been possible to explore the same clinically relevant parameters used in IPF patients; in particular lung function tests such as FVC and DFCO, that for their high translational value together with the high sensitivity to assess the efficacy of the compounds has been chosen respectively as the primary and secondary endpoint to support the selection of novel treatments within our internal drug discovery IPF projects. Furthermore, the introduction of these different readouts, that all go to the same direction, has from one side increased the robustness of the model and from the other side has allowed to bring this preclinical model to a level of complexity that mirrors the one observed in human IPF. Overall, this PhD work has enhanced the translational value of the data obtained with the mouse BLM model increasing the chance of selecting promising compounds to advance to clinical trials and has concretely led to significant benefits to drug discovery process in the IPF research, improving the quality and the reliability of the search of novel anti-fibrotic drugs.