In this study, we explored the role of mutation gene signature and TMB on survival in patients with iCCA. First, WES data of iCCA from two public database (ICGC and cBioportal) were acquired and frequent mutant genes were identified. Next, univariate, Lasso, and multivariate Cox regression analysis were used to further screen hub prognostic mutant signature and establish a mutation risk model for predicting prognosis. Afterward, the prognostic role of MRS was confirmed, PPI, GO, KEGG and GSEA analysis were conducted to reveal the potential cancer-related crosstalk involved. We also found that higher TMB was associated with poor prognosis. Furthermore, MRS, TMB and TNM stage were confirmed as independent predictors for OS. We constructed a reliable nomogram model based on the risk factors for OS with a satisfactory performance.
Gene mutations are ubiquitous in tumorigenesis and development of iCCA. Previous studies have reported comprehensive molecular alterations in biliary tract cancers [18, 29]. In our study, the most relevant mutation was TP53 (26.7%), followed by TTN (20.7%), KRAS (19.1%), MUC2 (14.5%) and ARID1A (12.9%), which was consistent with Cao et al [31]. We also found some special mutation genes of iCCA in our study, such as IDH1 (7.5%), BAP1 (9.1%), PBRM1 (7.2%) and EPHA2 (7.9%). Jiao et al preformed exome sequencing on iCCA and found that frequent genes (such as BAP1, ARID1A and PBRM1) mutations in chromatin-remodeling pathway. Our results are consistent with previous reports [32]. We also attempt to link genetic alternations to prognosis of patients, and found that patients with some gene mutations exhibited a borderline significance in worse outcomes compared to wide-type patients. Remarkably, on the basis of these prognostic factors, we developed a mutation risk score to predict survival. The MRS of our model was calculated based on 6 hub prognostic mutant genes (CDC27, AAK1, TP53, RBM10, KRAS and IPO5). This MRS model shows high predictive accuracy for OS, and provides a reliable tool for the prognosis predictor, which will be further applied in the clinical practice. To shed light on the potential molecular mechanisms underlying iCCA, the functional pathway and GSEA analysis were performed. The result of functional enrichment pathway indicated these prognostic mutation genes were closely correlated to cancer associated signaling pathways, such as cancer development and immune related pathway. In addition, GSEA also showed that high MRS group enriched with immune related pathway.
In recent years, the extensive indications of immune checkpoint inhibitor (ICI) therapy have developed the treatment of patients with advanced-stage cancers. However, the satisfactory effect of ICI is limited to a minority of patients. TMB, a novel predictive biomarker, is considerable to predict clinical response to ICI and contribute to recognize patients who will obtain therapeutic benefit [19]. The higher the TMB, as generally believed, means more tumor antigen, which is beneficial to activating the body’s immune function. Previous studies involving TMB mostly focused on the predictive capacity to the efficiency of ICIs, and showed robust correlation between higher TMB and better response to ICIs therapy. However, limited researches explored its prognostic value on iCCA. Many studies have indicated a relationship between TMB and survival in cancers. Owada-Ozaki et al from Japan found that higher TMB was correlated with shorter disease-free survival in NSCLC patients [33]. A study from China demonstrated that in HCC patients who had received radical resection, patients with higher TMB tend to have higher recurrence risk rate, and it was an independent risk factors of RFS [34]. We identified the median number of TMB in iCCAs was 1.25 (range 0.03–54.74). A large-scale examination of TMB on iCCA patients has been reported by Cao et al [31]. They analyzed the frequency and type of genetic aberrations in detail by comprehensive genomic profiling, and found the genomic heterogeneity between eastern and western patients of iCCA, but the relationship between TMB and prognosis was not mentioned. Besides, Tian et al investigated the comprehensive genomic features of Chinese patients with iCCA, and explored the relationship between TMB and some gene alternations [35]. It should be noted that TMB of their cohort was more than that in our study. The reason for this is that we only counted non-synonymous variants.
In consistent with previous findings on other tumors, our results showed that higher TMB was correlated with poor prognosis in patients. Therefore, we concluded that TMB had divergent survival difference. In addition, our results indicated that a prognostic model incorporating TMB is more than likely to improve prognostication and risk stratification in iCCA.
We explored the potential prognostic role of MRS and TMB on iCCA in this study, and found that the prognostic performance of the predictive model of TMB or MRS was better than that of TNM stage. Furthermore, the results of multivariate analysis indicated that TMB, MRS and TNM stage are independently prognostic factors in iCCA. Apart from new exploration on prognostic value of MRS and TMB, several drawbacks should be mentioned in our study. First of all, the mutation data of iCCA were extracted from public database, which only included the specimen performed WES. Targeted sequencing data was not considered in our study. more WES data from clinical patients should be incorporated to reduce selective bias. Secondly, the underlying mechanism behind the prognostic MRS and TMB in iCCA should be further investigated. Further experiments both in vitro and in vivo are required to support the present results. Finally, this study didn’t prove the specific mutation genes whether led to the abnormal gene expression, which deserve further exploration.