Programmed death-ligand 1 (PD-L1) is an immune checkpoint molecule that can regulate immune responses in the tumor microenvironment (TME); however, the clinical applications of PD-L1 in early-stage colorectal cancer (CRC) remain unclear. In this study, we aimed to investigate the relationship between PD-L1 expression and survival outcome and explore its relevant immune responses in CRC. PD-L1 expression was evaluated by immunohistochemical staining to determine the tumor proportion score and combined positive score (CPS) in a Taiwanese CRC cohort. The oncomine immune response research assay was conducted for immune gene expression analyses. CRC datasets from the TCGA database were reappraised for PD-L1-associated gene enrichment analyses using GSEA. The high expression of PD-L1 (CPS ≥ 5) was associated with longer recurrence-free survival (p = 0.031) and was an independent prognostic factor as revealed by multivariate analysis. High PD-L1 expression was related to six immune-related gene signatures, and CXCL9 is the most significant overexpressed gene in differential analyses. High CXCL9 expression correlated with increased infiltration levels of immune cells in the TME, including CD8+ T lymphocytes and M1 macrophages. These findings suggest that high PD-L1 expression is a prognostic factor of early-stage CRC, and CXCL9 may play a key role in regulating PD-L1 expression.

Background: Colorectal cancer (CRC) is one of the most prevalent malignant diseases worldwide. Risk prediction for tumor recurrence is important for making effective treatment decisions and for the survival outcomes of patients with CRC after surgery. Herein, we aimed to explore a prediction algorithm and the risk factors for postoperative tumor recurrence using a machine learning (ML) approach with standardized pathology reports for patients with stage II and III CRC. Methods: Pertinent clinicopathological features were compiled from medical records and standardized pathology reports of patients with stage II and III CRC. Four ML models based on logistic regression (LR), random forest (RF), classification and regression decision trees (CARTs), and support vector machine (SVM) were applied for the development of the prediction algorithm. The area under the curve (AUC) of the ML models was determined in order to compare the prediction accuracy. Genomic studies were performed using a panel-targeted next-generation sequencing approach. Results: A total of 1073 patients who received curative intent surgery at the National Cheng Kung University Hospital between January 2004 and January 2019 were included. Based on conventional statistical methods, chemotherapy (p = 0.003), endophytic tumor configuration (p = 0.008), TNM stage III disease (p < 0.001), pT4 (p < 0.001), pN2 (p < 0.001), increased numbers of lymph node metastases (p < 0.001), higher lymph node ratios (LNR) (p < 0.001), lymphovascular invasion (p < 0.001), perineural invasion (p < 0.001), tumor budding (p = 0.004), and neoadjuvant chemoradiotherapy (p = 0.025) were found to be correlated with the tumor recurrence of patients with stage II–III CRC. While comparing the performance of different ML models for predicting cancer recurrence, the AUCs for LR, RF, CART, and SVM were found to be 0.678, 0.639, 0.593, and 0.581, respectively. The LR model had a better accuracy value of 0.87 and a specificity value of 1 in the testing set. Two prognostic factors, age and LNR, were selected by multivariable analysis and the four ML models. In terms of age, older patients received fewer cycles of chemotherapy and radiotherapy (p < 0.001). Right-sided colon tumors (p = 0.002), larger tumor sizes (p = 0.008) and tumor volumes (p = 0.049), TNM stage II disease (p < 0.001), and advanced pT3–4 stage diseases (p = 0.04) were found to be correlated with the older age of patients. However, pN2 diseases (p = 0.005), lymph node metastasis number (p = 0.001), LNR (p = 0.004), perineural invasion (p = 0.018), and overall survival rate (p < 0.001) were found to be decreased in older patients. Furthermore, PIK3CA and DNMT3A mutations (p = 0.032 and 0.039, respectively) were more frequently found in older patients with stage II–III CRC compared to their younger counterparts. Conclusions: This study demonstrated that ML models have a comparable predictive power for determining cancer recurrence in patients with stage II–III CRC after surgery. Advanced age and high LNR were significant risk factors for cancer recurrence, as determined by ML algorithms and multivariable analyses. Distinctive genomic profiles may contribute to discrete clinical behaviors and survival outcomes between patients of different age groups. Studies incorporating complete molecular and genomic profiles in cancer prediction models are beneficial for patients with stage II–III CRC.

To evaluate whether adjusted computed tomography (CT) scan image-based radiomics combined with immune genomic expression can achieve accurate stratification of cancer recurrence and identify potential therapeutic targets in stage III colorectal cancer (CRC), this cohort study enrolled 71 patients with postoperative stage III CRC. Based on preoperative CT scans, radiomic features were extracted and selected to build pixel image data using covariate-adjusted tensor classification in the high-dimension (CATCH) model. The differentially expressed RNA genes, as radiomic covariates, were identified by cancer recurrence. Predictive models were built using the pixel image and immune genomic expression factors, and the area under the curve (AUC) and F1 score were used to evaluate their performance. Significantly adjusted radiomic features were selected to predict recurrence. The association between the significantly adjusted radiomic features and immune gene expression was also investigated. Overall, 1037 radiomic features were converted into 33 × 32-pixel image data. Thirty differentially expressed genes were identified. We performed 100 iterations of 3-fold cross-validation to evaluate the performance of the CATCH model, which showed a high sensitivity of 0.66 and an F1 score of 0.69. The area under the curve (AUC) was 0.56. Overall, ten adjusted radiomic features were significantly associated with cancer recurrence in the CATCH model. All of these methods are texture-associated radiomics. Compared with non-adjusted radiomics, 7 out of 10 adjusted radiomic features influenced recurrence-free survival. The adjusted radiomic features were positively associated with PECAM1, PRDM1, AIF1, IL10, ISG20, and TLR8 expression. We provide individualized cancer therapeutic strategies based on adjusted radiomic features in recurrent stage III CRC. Adjusted CT scan image-based radiomics with immune genomic expression covariates using the CATCH model can efficiently predict cancer recurrence. The correlation between adjusted radiomic features and immune genomic expression can provide biological relevance and individualized therapeutic targets.

The impact of germline variants on the regulation of the expression of tumor microenvironment (TME)-based immune response genes remains unclear. Expression quantitative trait loci (eQTL) provide insight into the effect of downstream target genes (eGenes) regulated by germline-associated variants (eVariants). Through eQTL analyses, we illustrated the relationships between germline eVariants, TME-based immune response eGenes, and clinical outcomes. In this study, both RNA sequencing data from primary tumor and germline whole-genome sequencing data were collected from patients with stage III colorectal cancer (CRC). Ninety-nine high-risk subjects were subjected to immune response gene expression analyses. Seventy-seven subjects remained for further analysis after quality control, of which twenty-two patients (28.5%) experienced tumor recurrence. We found that 65 eQTL, including 60 germline eVariants and 22 TME-based eGenes, impacted the survival of cancer patients. For the recurrence prediction model, 41 differentially expressed genes (DEGs) achieved the best area under the receiver operating characteristic curve of 0.93. In total, 19 survival-associated eGenes were identified among the DEGs. Most of these genes were related to the regulation of lymphocytes and cytokines. A high expression of HGF, CCR5, IL18, FCER1G, TDO2, IFITM2, and LAPTM5 was significantly associated with a poor prognosis. In addition, the FCER1G eGene was associated with tumor invasion, tumor nodal stage, and tumor site. The eVariants that regulate the TME-based expression of FCER1G, including rs2118867 and rs12124509, were determined to influence survival and chromatin binding preferences. We also demonstrated that FCER1G and co-expressed genes in TME were related to the aggregation of leukocytes via pathway analysis. By analyzing the eQTL from the cancer genome using germline variants and TME-based RNA sequencing, we identified the eQTL in immune response genes that impact colorectal cancer characteristics and survival.

Tumor heterogeneity results in more than 50% of hypermutated cancers failing to respond to standard immunotherapy. There are numerous challenges in terms of drug resistance, therapeutic strategies, and biomarkers in immunotherapy. In this study, we analyzed primary tumor samples from 533 cancer patients with six different cancer types using deep targeted sequencing and gene expression data from 78 colorectal cancer patients, whereby driver mutations, mutational signatures, tumor-associated neoantigens, and molecular cancer evolution were investigated. Driver mutations, including RET, CBL, and DDR2 gene mutations, were identified in the hypermutated cancers. Most hypermutated endometrial and pancreatic cancer patients carry genetic mutations in EGFR, FBXW7, and PIK3CA that are linked to immunotherapy resistance, while hypermutated head and neck cancer patients carry genetic mutations associated with better treatment responses, such as ATM and BRRCA2 mutations. APOBEC (apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like) and DNA repair defects are mutational drivers that are signatures for hypermutated cancer. Cancer driver mutations and other mutational signatures are associated with sensitivity or resistance to immunotherapy, representing potential genetic markers in hypermutated cancers. Using computational prediction, we identified NF1 p.T700I and NOTCH1 p.V2153M as tumor-associated neoantigens, representing potential therapeutic targets for immunotherapy. Sequential mutations were used to predict hypermutated cancers based on genomic evolution. Using a logistic model, we achieved an area under the curve (AUC) = 0.93, accuracy = 0.93, and sensitivity = 0.81 in the testing set. The sequential patterns were distinct among the six cancer types, and the sequential mutation order of MSH2 and the coexisting BRAF genetic mutations influenced the hypermutated phenotype. The TP53~MLH1 and NOTCH1~TET2 sequential mutations impacted colorectal cancer survival (p-value = 0.027 and 0.0001, respectively) by reducing the expression of PTPRCAP (p-value = 1.06 × 10⁻⁶) and NOS2 (p-value = 7.57 × 10⁻⁷) in immunity. Sequential mutations are significant for hypermutated cancers, which are characterized by mutational heterogeneity. In addition to driver mutations and mutational signatures, sequential mutations in cancer evolution can impact hypermutated cancers. They characterize potential responses or predictive markers for hypermutated cancers. These data can also be used to develop hypermutation-associated drug targets and elucidate the evolutionary biology of cancer survival. In this study, we conducted a comprehensive analysis of mutational patterns, including sequential mutations, and identified useful markers and therapeutic targets in hypermutated cancer patients.