Table 1.
Primers sequences for RT-qPCR.
Table 2.
Clinical characteristics in cohorts with different prognosis.
Fig 1.
Cohort screening and analyses workflow.
Fig 2.
Developing CBI for ovarian cancer patients receiving chemotherapy.
(A) The histogram showed the number of genes isolated by Bayesian causal network (strength > 0.85) in 5 enriched biological processes. (B) Multivariate Cox regression analysis of the 49 different chemotherapy node genes to select chemotherapy prognosis-related genes. (C) The causality relationship of ten chemotherapy prognosis-related genes. (D) The expression of chemotherapy prognosis-related genes among CBI subgroups. (E) The overall survival was significantly different among CBI subgroups. (F) Gene expression was detected by RT-qPCR in normal and ovarian cancer cells. Data are presented as mean ± SD of triplicate cultures. N/A indicates that these genes are not expressed in either cell lines (cycles > 35). (**p < 0.01, ****p < 0.0001; unpaired two-sided Student’s t test.).
Fig 3.
Constructing random forest model and verifying CBI.
(A) Learning curve of estimators from 0 to 300. (B) The importance of each chemotherapy prognosis-related gene in random forest model. (C, D, E) Survival analyses of CBI subgroups in the internal validation set (P < 0.001), GSE17260 (P < 0.005), GSE30161 (P < 0.05), GSE26193 (P < 0.005), and GSE32062 (P < 0.005), respectively. (C-G) AUC of the internal validation set, GSE17260, GSE26193, GSE30161 and GSE32062 are 0.94, 0.75, 0.79, 0.74 and 0.68 respectively. (I) Multivariate COX regression analysis confirmed CBI could serve as the independent prognostic factor of advanced ovarian cancer patients receiving chemotherapy.
Fig 4.
Functional enrichment analyses of CBI expanded genes.
(A) The causality relationship of the 346 CBI expanded genes (strength > 0.85). (B) Top 10 GO biological processes (P < 0.001) and (C) top 10 KEGG pathways of CBI expanded genes (P < 0.001). (D) Top up-regulated pathways in CBI-high group and CBI-low group in GSEA (p < 0.05, FDR < 0.25).
Fig 5.
CBI-low group may benefit more from ICB treatment.
(A-B) Immune-related pathways are up-regulated in the CBI-low group (P < 0.05, FDR < 0.25). (C) High expression of immune-checkpoint molecules in CBI-low group. (D-E) The biomarker scores of ICB treatment response are higher in CBI-low group (P < 0.05).
Fig 6.
Mutation analysis of CBI-high and CBI-low groups.
(A-B) The top 10 mutation genes and types in CBI-high group and CBI-low group, respectively. (C) The mutant differential gene between CBI-high group and CBI-low group (P < 0.05).