Extended Data Fig. 2: Tumor purity correlates with TMB estimates in the immunotherapy treated cohorts.
The correlations of tumor purity with tumor mutational burden and clinical response in 957 TCGA NSCLC samples and in immunotherapy cohorts 1 and 2 are shown in a-f. (a-b) TCGA lung adenocarcinomas-LUAD (N=508 tumors; a) and lung squamous cell carcinomas-LUSC (N=477 tumors; b) with a higher degree of normal contamination had a significantly lower TMB compared to tumors with a tumor purity > 50% (Mann-Whitney U test p=0.06 and p=1.3e-7 for LUAD and LUSC respectively). (c-d) In cohorts 1 and 2, the correlation between obsTMB and tumor purity was particularly pronounced for tumor purity less than 30% (Mann Whitney U test p=0.008 and p=0.08 for overall comparisons of TMB across tumor purity tiers for cohort 1 (N=87 tumors; c) and cohort 2 respectively (N=30 tumors; d). (e-f) Tumor purity was associated with clinical benefit from ICB when mutation-based purity was used, which is most likely attributed to the contribution of TMB in the mutation-based purity calculation; however no difference in tumor purity was found between responding and non-responding tumors when copy-number based tumor purity (N=74 tumors) and adjusted tumor purity (N=87 tumors) was used in cohort 1 (Mann Whitney U test p=0.14 and Mann Whitney U test p=0.22 respectively; e) and cohort 2 (Mann Whitney U test p=0.72 and Mann Whitney U test p=0.6 respectively; f). Seventy four and twenty one tumor samples from cohort 1 and 2 respectively had evaluable copy-number based tumor purity (Methods). The vertical line in each whisker plot signifies the median and the ends the upper and lower quartile, while the whiskers extend 1.5 times the interquartile range from the hinges. P values are based on two-sided testing.
