61 Abnormal collagen structure resulting from lack of contribution of collagen type XIV in lungs of patients with idiopathic pulmonary fibrosis Image analysis The image analysis was performed for 4 different regions within the scanned images for percentage of positive area as previously described [10]. For whole tissue analysis, scanned images were used after removal of artifacts. For the parenchyma analysis, airways and blood vessels were removed from the whole tissue images. Additionally, specific parts of the airways including the airway wall and the epithelium were examined. Airways were extracted using Aperio ImageScope software (Leica Biosystems, Amsterdam, The Netherlands). Based on the localization of the staining, different parts of the airways were selected using Adobe Photoshop (Adobe, San Jose, California, USA). Artifacts and the remaining compartments of the airways which were not used for examination were removed during this step. In the case of IPF airway walls, which have no clear border between the end of the airway wall and the surrounding parenchyma, the distance between the epithelial layer and smooth muscle layer (E-SM distance) was measured. This distance then was multiplied by three to calculate distance from the outside of the smooth muscle layer till the (not visible) parenchyma (SM-P distance). The multiplication by three was calculated based on the ratio between measured E-SM and SM-P distances from IPF airways which were surrounded by still visible parenchyma. The approach is illustrated in Supplementary Figure 1. Next, the total tissue area positive for COL14 was quantified by applying color deconvolution to separate the different staining colors [10]. The quantification of the staining was automatically processed using FIJI Image J. Afterwards, the raw data was sorted in RStudio, from which the percentage of area of the tissue present that stained positive for the protein of interest was calculated. The formula used to calculate the percentage of area stained positive for the protein is as follows: localization of the staining, different parts of the airways were selected using Adobe Photoshop (Adobe, San Jose, California, USA). Artifacts and the remaining compartments of the airways which were not used for examination were removed during this step. In the case of IPF airway walls, which have no clear border between the end of the airway wall and the surrounding parenchyma, the distance between the epithelial layer and smooth muscle layer (E-SM distance) was measured. This distance then was multiplied by three to calculate distance from the outside of the smooth muscle layer till the (not visible) parenchyma (SM-P distance). The multiplication by three was calculated based on the ratio between measured E-SM and SM-P distances from IPF airways which were surrounded by still visible parenchyma. The approach is illustrated in Supplementary Figure 1. Next, the total tissue area positive for COL14 was quantified by applying color deconvolution to separate the different staining colors [10]. The quantification of the staining was automatically processed using FIJI Image J. Afterwards, the raw data was sorted in RStudio, from which the percentage of area of the tissue present that stained positive for the protein of interest was calculated. The formula used to calculate the percentage of area stained positive for the protein is as follows: (%) = ( ) ( ) × 100% Statistical analysis Differences between never-smoker non-IPF, ex-smoker non-IPF and IPF groups were tested using a Kruskal-Wallis test for whole tissue and parenchyma regions. For airway wall and bronchial epithelium regions, a mixed model analysis was used to test the statistical differences between patients incorporating multiple airway wall and bronchial epithelium images per patient. p <0.05 was considered significant. Results Statistical analysis Differences between never-smoker non-IPF, ex-smoker non-IPF and IPF groups were tested using a Kruskal-Wallis test for whole tissue and parenchyma regions. For airway wall and bronchial epithelium regions, a mixed model analysis was used to test the statistical differences between patients incorporating multiple airway wall and bronchial epithelium images per patient. p <0.05 was considered significant. 3
RkJQdWJsaXNoZXIy MTk4NDMw