Fig. 6

A diagnostic prediction model for subgroups was developed and validated via the LASSO regression. (a) In the PTMETA cohort (n = 64), the determination of the optimal λ was obtained when the binomial deviance reached the minimum value, and further generated LASSO coefficients of the most useful marker genes; (b) Coefficients of 17 marker genes finally obtained in LASSO regression; (c) Receiver operating characteristic (ROC) curves and the associated areas under curves (AUCs) of the diagnostic prediction model in the PTMETA cohort; (d) Comparison of the tumor immune microenvironment across different immune state according to the diagnostic prediction model in the PTMC-TCGA cohort using two methods: ESTIMATE immune scores, GSVA using Bindea et al.’s combined immune gene set; (e) Comparison of the immune cell fraction distinguished by the different immune state in the PTMC-TCGA cohort; (f) Comparison of IFN-γ scores, PDL1, PDCD1, and CTLA4 expression by the different immune state in the PTMC-TCGA cohort. p-values from the Wilcoxon rank-sum tests; (g) Kaplan–Meier curves of PFS according to the diagnostic prediction model in the PTMC-TCGA cohort. p-value from the Log-rank test