The prediction of tumor in the TNM staging (tumor, node, and metastasis) stage of colon cancer using the most influential histopathology parameters and to predict the five years disease-free survival (DFS) period using machine learning (ML) in clinical research have been studied here. From the colorectal cancer (CRC) registry of Chang Gung Memorial Hospital, Linkou, Taiwan, 4021 patients were selected for the analysis. Various ML algorithms were applied for the tumor stage prediction of the colon cancer by considering the Tumor Aggression Score (TAS) as a prognostic factor. Performances of different ML algorithms were evaluated using five-fold cross-validation, which is an effective way of the model validation. The accuracy achieved by the algorithms taking both cases of standard TNM staging and TNM staging with the Tumor Aggression Score was determined. It was observed that the Random Forest model achieved an F-measure of 0.89, when the Tumor Aggression Score was considered as an attribute along with the standard attributes normally used for the TNM stage prediction. We also found that the Random Forest algorithm outperformed all other algorithms, with an accuracy of approximately 84% and an area under the curve (AUC) of 0.82 ± 0.10 for predicting the five years DFS.

Colorectal cancer is one of the leading causes of cancer-related death worldwide. The early diagnosis of colon cancer not only reduces mortality but also reduces the burden related to the treatment strategies such as chemotherapy and/or radiotherapy. However, when the microscopic examination of the suspected colon tissue sample is carried out, it becomes a tedious and time-consuming job for the pathologists to find the abnormality in the tissue. In addition, there may be interobserver variability that might lead to conflict in the final diagnosis. As a result, there is a crucial need of developing an intelligent automated method that can learn from the patterns themselves and assist the pathologist in making a faster, accurate, and consistent decision for determining the normal and abnormal region in the colorectal tissues. Moreover, the intelligent method should be able to localize the abnormal region in the whole slide image (WSI), which will make it easier for the pathologists to focus on only the region of interest making the task of tissue examination faster and lesser time-consuming. As a result, artificial intelligence (AI)-based classification and localization models are proposed for determining and localizing the abnormal regions in WSI. The proposed models achieved F-score of 0.97, area under curve (AUC) 0.97 with pretrained Inception-v3 model, and F-score of 0.99 and AUC 0.99 with customized Inception-ResNet-v2 Type 5 (IR-v2 Type 5) model.

Substantial improvements have been made in the management of metastatic colorectal cancer (mCRC) in the last two decades, but disease monitoring remains underdeveloped. Circulating tumor DNA (ctDNA) is a promising prognostic and predictive biomarker; however, ctDNA as a marker for mCRC patients is not well established, and there is still no consensus about how to utilize it most cost-effectively. In this study, we aim to investigate plasma ctDNA levels as a biomarker for therapeutic response of mCRC patients. We performed next-generation sequencing (NGS) by using a 12-gene panel to identify genetic variants in 136 tumor tissue and ctDNA samples from 32 mCRC patients. Genetic variants were detected in approximately 70% of samples, and there was a high concordance (85%) between tumor tissue and plasma ctDNA. We observed ctDNA changes in 18 follow-up patients, including the emergence of new variants. Changes in ctDNA levels significantly correlated with tumor shrinkage (P = 0.041), and patients with a ctDNA decrease >80% after treatment had a longer progression-free survival compared with patients with a ctDNA decrease of 80% was better than those with a ctDNA decrease <80% (OR, 0.026; P = 0.007). In conclusion, this study demonstrates that monitoring of genetic ctDNA variants can serve as a valuable biomarker for therapeutic efficacy in mCRC patients, and that using a moderate-sized 12-gene NGS panel may be suitable for such clinical monitoring. Mol Cancer Ther; 17(10); 2238–47. ©2018 AACR.

Extracellular vesicles (EVs) are valuable sources for the discovery of useful cancer biomarkers. This study explores the potential usefulness of tumor cell-derived EV membrane proteins as plasma biomarkers for early detection of colorectal cancer (CRC). EVs were isolated from the culture supernatants of four CRC cell lines by ultracentrifugation, and their protein profiles were analyzed by LC-MS/MS. Bioinformatics analysis of identified proteins revealed 518 EV membrane proteins in common among at least three CRC cell lines. We next used accurate inclusion mass screening (AIMS) in parallel with iTRAQ-based quantitative proteomic analysis to highlight candidate proteins and validated their presence in pooled plasma-generated EVs from 30 healthy controls and 30 CRC patients. From these, we chose 14 potential EV-derived targets for further quantification by targeted MS assay in a separate individual cohort comprising of 73 CRC and 80 healthy subjects. Quantitative analyses revealed significant increases in ADAM10, CD59 and TSPAN9 levels (2.19- to 5.26-fold, p < 0.0001) in plasma EVs from CRC patients, with AUC values of 0.83, 0.95 and 0.87, respectively. Higher EV CD59 levels were significantly correlated with distant metastasis (p = 0.0475), and higher EV TSPAN9 levels were significantly associated with lymph node metastasis (p = 0.0011), distant metastasis at diagnosis (p = 0.0104) and higher TNM stage (p = 0.0065). A two-marker panel consisting of CD59 and TSPAN9 outperformed the conventional marker CEA in discriminating CRC and stage I/II CRC patients from healthy controls, with AUC values of 0.98 and 0.99, respectively. Our results identify EV membrane proteins in common among CRC cell lines and altered plasma EV protein profiles in CRC patients and suggest plasma EV CD59 and TSPAN9 as a novel biomarker panel for detecting early-stage CRC.

Background. The cancer cell secretome has been recognized as a valuable reservoir for identifying novel serum/plasma biomarkers for different cancers, including colorectal cancer (CRC). This study aimed to verify four CRC cell-secreted proteins (tumor-associated calcium signal transducer 2/trophoblast cell surface antigen 2 (TACSTD2/TROP2), tetraspanin-6 (TSPAN6), bone marrow stromal antigen 2 (BST2), and tumor necrosis factor receptor superfamily member 16 (NGFR)) as potential plasma CRC biomarkers.Methods. The study population comprises 152 CRC patients and 152 controls. Target protein levels in plasma and tissue samples were assessed by ELISA and immunohistochemistry, respectively.Results. Among the four candidate proteins examined by ELISA in a small sample set, only BST2 showed significantly elevated plasma levels in CRC patients versus controls. Immunohistochemical analysis revealed the overexpression of BST2 in CRC tissues, and higher BST2 expression levels correlated with poorer 5-year survival (46.47% versus 65.57%;p=0.044). Further verification confirmed the elevated plasma BST2 levels in CRC patients (2.35 ± 0.13 ng/mL) versus controls (1.04 ± 0.03 ng/mL) (p<0.01), with an area under the ROC curve (AUC) being 0.858 comparable to that of CEA (0.867).Conclusion. BST2, a membrane protein selectively detected in CRC cell secretome, may be a novel plasma biomarker and prognosticator for CRC.