Fig. 2: CCDI predicts response to ICI in NSCLC patients. | npj Precision Oncology

Fig. 2: CCDI predicts response to ICI in NSCLC patients.

From: Integrating bulk, single-cell, and spatial transcriptomics to identify and functionally validate novel targets to enhance immunotherapy in NSCLC

Fig. 2

a Box plot showing the difference in immunophenotype score (IPS) and response to ICIs between high- and low-risk subgroups of TCGA-LUAD cohort (n = 514) separated by CCDI. A higher IPS indicates greater tumor immunogenicity and a higher likelihood of responding to IPS. b The proportion of patients predicted by ImmuCellAI to respond to ICIs in high-risk and low-risk TCGA-LUAD subgroups. c The ICI response of high- and low-risk TCGA-LUAD subgroups was predicted using the SubMap algorithm with three NSCLC ICI trials as training sets. d–l Validation of CCDI’s ability to distinguish between ICI responders and nonresponders across various cancer types. Kaplan–Meier survival curves for the high and low-risk subgroups of the IMvigor210 cohort defined by CCDI (log-rank test, P < 0.0001) (Bladder urothelial carcinoma, n = 310) (d) and distribution and average values of CCDI among these patients with complete response (CR), partial response (PR), stable disease (SD), progressive disease (PD), and those not estimated (NE) (e, f). The differences in anti-PD-1/L1 treatment response and CCDI between high and low-risk subgroups across different cohorts: GSE78220 (melanoma, n = 28) (g), GSE67501 (human renal cell carcinoma, n = 11) (h), GSE165252 (esophageal adenocarcinoma, n = 40) (i), GSE126044 (NSCLC, n = 16) (j), GSE135222 (NSCLC, n = 27) (k), and GSE190265 (NSCLC, n = 34) (l). *P < 0.05, **P < 0.01, ***P < 0.001.

Back to article page