DEVELOPING A HISTOLOGY-BASED COMPOSITE INDEX TO ASSESS THE DEGREE OF LUNG FIBROSIS USING QUANTITATIVE HALO® MODULES

Abstract

Histological and visual assessment of pulmonary fibrosis are used as a gold standard to determine severity of fibrotic lung disease. The Ashcroft scoring technique and histological stains like Masson’s trichrome and picrosirius red (PSR) have been used for decades as primary readouts, often alongside a biochemical assay estimating the amount of hydroxyproline in the tissue. As pulmonary fibrosis is a heterogeneous disease and collagen content rarely shows more than a two-fold increase between a non-diseased and a diseased lung, therapeutic assessment is challenging due to the variability animal models. To improve the assessment of the severity of fibrotic lung disease, we have developed models using multivariate linear regression and XGBoost machine learning that consider digital quantifications of histological features from 73 clinically diagnosed cases of IPF and 11 control tissues, obtained using the HALO® platform, that correlate with the Ashcroft score and combine them into a composite index that we believe represents a better, more objective way of scoring fibrotic lung tissues. The tissues were stained with H&E for structural assessment, Trichrome (TRI), and Picrosirius Red (PSR) for collagen content, alpha-smooth muscle actin (αSMA) to assess the amount of ECM-producing myofibroblasts (MFBs), and CD68, CD163, and CD206 for macrophage content and activation as they have been correlated with lung fibrosis and are considered a therapeutic target.