Fig 1: Dysregulation of peroxidasin and ADAMTS16 in NSCLC tumour cells. (A) Label-free quantification of peroxidasin (left panel) and ADAMTS16 (right panel) in non-tumour and tumour ECM samples quantified by MS-based proteomics. Statistical analysis, paired two-sided Student’s t-test with Benjamini–Hochberg correction (n = 34 matched non-tumour–tumour paired tissue samples). (B) Immunohistochemistry (IHC) anaysis of peroxidasin (left panels) and ADAMTS16 (right panels) for exemplar lung cores from the tumour microarray. Scale bars, 100 μm. (C) Quantification of TMA immunostaining for non-tumour-derived lung cores (normal cells) and tumour-derived lung cores (tumour stroma and tumour cells). IHC scores of 0, 1, 2 and 3 correspond to no, mild, moderate and strong immunostaining, respectively. Not interp., not interpretable. Statistical analysis, chi-square test. (D, E) Differential expression analysis of PXDN (left panels) and ADAMTS16 (right panels) in cohorts of lung adenocarcinoma (D) or lung squamous cell carcinoma (E) patients from integrated RNA-seq datasets. Statistical analysis, Wilcoxon signed-rank test (n = 57 and 49 matched adjacent normal–tumour paired tissue samples for adenocarcinoma and squamous cell carcinoma patients, respectively). For (A, D, E) black circle, median; black bar, 95% confidence interval; silhouette, probability density. (F, G) Kaplan–Meier curves of patient survival associated with degree of expression of PXDN (left panels) and ADAMTS16 (right panels) in cohorts of lung adenocarcinoma (F) or lung squamous cell carcinoma (G) patients from TCGA RNA-seq datasets. The lower quartiles (low) and upper quartiles (high) of gene expression were compared. HR, hazard ratio (with 95% confidence interval). Statistical analysis, log-rank test (n = 513 and 501 patients for adenocarcinoma and squamous cell carcinoma cohorts, respectively).