Extended Data Fig. 4: Impact of molecular smoking signature on outcome for immune checkpoint blockade and association with TMB. | Nature Cancer

Extended Data Fig. 4: Impact of molecular smoking signature on outcome for immune checkpoint blockade and association with TMB.

From: Multimodal genomic features predict outcome of immune checkpoint blockade in non-small-cell lung cancer

Extended Data Fig. 4

(a-b) In silico dilution experiments of single base substitutions to evaluate the power to accurately determine contribution of a dominant mutation signature. Mutation signature analyses were performed on whole exome data from 985 NSCLC tumors (N=508 lung adenocarcinomas and N=477 squamous cell carcinomas) obtained through TCGA. Seventy-six NSCLC tumors (N=64 lung adenocarcinomas and N=12 squamous cell carcinomas) had a tumor mutation load >250 and a molecular smoking signature >75% and were further selected for an in silico dilution series. (a) Mutation counts were diluted from maximum count to a minimum of 5 using random resampling (3 re-sampling replicates per dilution level), to evaluate consistency and divergence in the predicted presence of a smoking signature. Connected points in (a) indicate mean values per case per dilution series. On average, 20 mutations were sufficient to predict the presence of a smoking signature at a 50% level. (b) Mutational load below 20 mutations lead to a 30% difference from the original contribution of the C>A transversion rich signature value and therefore represents a threshold beyond which, there is a significant deviation from accurately determining a dominant mutation signature. Canter values indicate the mean values per dilution level and error bars indicate standard error of the mean. (c) Patients with a molecular smoking mutational signature derived durable benefit from immune checkpoint blockade (N=80, log rank p=0.031). TMB did not fully explain the association between molecular smoking signature and ICB response. (d) The number of clonal mutations was plotted against the percent contribution of the molecular smoking signature for each tumor (N=74 tumors in total, N=33 responding tumors and N=41 non-responding tumors). Each dot represents a tumor and each tumor is color coded by response, with responding tumors showed in orange and non-responding tumors showed in blue. While there is a correlation between the number of clonal mutations and a dominant molecular smoking signature, there are tumors with a low clonal TMB (dotted red line indicates a TMB threshold used to classify TMB in high/low categories) but a dominant molecular smoking signature which derive benefit from immune checkpoint blockade. The median point estimate and 95% confidence intervals for survival were estimated by the Kaplan–Meier method and survival curves were compared by using the nonparametric log rank test. Log rank p values are based on two-sided testing.

Source data

Back to article page