Fig. 7: Evidence of neoantigen-mediated immune editing leading to a higher fraction of truncal mutation yet with lower quality.

a, Phylogenetic trees based on all high-confidence mutations found across all regions per patient. b, The number of private, shared and truncal mutations in each patient plotted and fraction of truncal mutations calculated per patient (white numbers). For each patient, GrzB expression in tumor subregions based on mIF analysis and the defined CD3⁺CD8⁺ T cell infiltration status is indicated. Smoking status was defined based on deconvolution of the eight different mutational signatures and comparison to known mutational signatures from Alexandrov et al.⁶² with a threshold of >50% for tobacco smoking signature. c,d, Positive correlations found between the TMB and the smoking status (smokers n = 24 macro-regions; nonsmokers: n = 26 macro-regions; one-sided Student’s t-test P = 1.3 × 10⁻⁶) (c), as well as between the expression of GrzB in tumor subregions (smokers: n = 4 patients; nonsmokers: n = 4 patients; mIF, one-sided Student’s t-test P = 0.13) (d). e,f, A higher fraction of truncal (clonal) mutations was found to be significantly associated with smoking status (smokers: n = 4 patients; nonsmokers: n = 4 patients; one-sided Student’s t-test P = 0.019) (e) and with CD3⁺CD8⁺ T cell infiltration (infiltrated: n = 3 patients; excluded: n = 5 patients; one-sided Student’s t-test P = 0.0066) (f). g, Schematic overview of the predicted neoantigen quality model from Łuksza et al.³². h, Neoantigen quality score distributions of private and truncal mutations in each patient (02287: n = 99/121; 02288: n = 26/92; 02289: n = 79/130; 03421: n = 68/24; 02290: n = 21/225; 02671: n = 59/187; 02672: n = 38 of 489; 03023: n = 32/191 (private neoantigens/truncal neoantigens)). i,j, The ratio between the neoantigen quality of truncal versus private mutations in excluded and infiltrated tumors (excluded: n = 5 patients; infiltrated: n = 3 patients; boxplot lines show the mean) (i), as well as in nonsmokers (n = 4 patients) and smokers (n = 4 patients) (j). Unless indicated otherwise, all statistical tests were performed as one-sided Wilcoxon’s nonparametric test and boxplots show the median (line), the IQR between the 25th and 75th percentiles (box) and 1.5× the IQR ± the upper and lower quartiles, respectively. No adjustments were made for multiple testing.