Extended Data Fig. 10: Correlations among genomic and immune features and multivariable model for prediction of outcome to immune checkpoint blockade, excluding patients on chemo-immunotherapy and tumors collected at the time of resistance in cohort 1. | Nature Cancer

Extended Data Fig. 10: Correlations among genomic and immune features and multivariable model for prediction of outcome to immune checkpoint blockade, excluding patients on chemo-immunotherapy and tumors collected at the time of resistance in cohort 1.

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

Extended Data Fig. 10

(a) Relationships among different features were assessed by the Spearman’s rho statistic for cohort 1 (N=89 patients) and p values were corrected for multiple comparisons. Four clusters of parameters were identified, related to RTK mutations, HLA genetic variation, tumor aneuploidy and TMB. The color of each dot refers to the Spearman rho coefficient value (darkest blue being 1 and darkest red being -1) and the size of each dot is proportional to the strength of the correlation. Three stars indicate an FDR adjusted p value of <0.05, two stars indicate an FDR adjusted p value of <0.1 and one star denotes an FDR adjusted p value of <0.2. (b-c) Corrected TMB, RTK mutations, molecular smoking signature and HLA germline variation were combined in a multivariable Cox proportional hazards regression model and a risk score was calculated for each case based on the weighted contribution of each parameter. Among 84 patients in total, patients with a higher risk score (N=28 patients) had a significantly shorter overall survival in cohort 1 (13 months vs. 38 months, HR=3.16, 95% CI: 1.68-5.91, log rank p=0.0002; b). The model was trained in cohort 1, fixed and applied in cohort 2 revealing a significantly shorter progression-free survival for high risk patients in cohort 2 (3 months vs. 8 months, HR=2.73, 95% CI 1.15-6.45, log rank p=0.017; c). The second tertile of the risk score was used to classify patients in high risk (top 33.3%, N=28 patients for cohort 1 and N=11 patients for cohort 2) and low risk (bottom 66.6%, N=56 patients for cohort 1 and N=23 patients for cohort 2) groups. 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