Classification Of Colorectal Cancer Research Articles

Abstract LB251: Widespread epigenetic deregulation and its dependency on tumor sidedness in colorectal cancer

Abstract Clinical evidence suggests that right-sided colorectal cancer (CRC) tends to have a worse prognosis compared to left-sided CRC. The distinct functional mechanisms underlying this “sidedness” remain cryptic and poorly understood. Thus, molecular classification of CRC is of paramount in targeting specific subtypes and personalizing treatment for patients. To identify targetable molecular features to improve treatment regimens, we performed multi-faceted network analyses of a large number of CRC patients. In our network model, differential expression of transcript abundance, epigenetic features, and mutational frequency were used to identify widespread gene dysregulation and perturbed pathways to reveal disrupted networks distinct to right or left sided colorectal cancer. We discovered histone related genes and distinct epigenetic signals are at play in right-sided CRC. Additionally, our studies reveal “at risk” patient characteristics suggestive of survival differences, to improve diagnostic predictors based on CRC sidedness. Overall, our integrative approach serves to link molecular profiles distinct between left and right sided colorectal cancer, while paving the way towards identifying more effective treatment alternatives from a previously overlooked, readily characterized subtype of colorectal cancer. Citation Format: Brandon Burgman, Xingxin Pan, Anjali Agrawal, Nidhi Sahni, S. Stephen Yi. Widespread epigenetic deregulation and its dependency on tumor sidedness in colorectal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr LB251.

Abstract 3112: Colorectum metabolome map revealed etiological and survival heterogeneity in cancers of different subsites: Challenging the proximal versus distal classification

Abstract Objective: Colorectal cancer (CRC) is usually classified as proximal, distal and rectal cancer. clinicopathological, molecular features and risk factors do not change abruptly and variations exist even among the refined subsites, which may contribute to the failure of mechanism-targeted therapies for CRC and poor accuracy of potential CRC biomarkers. We generated a CRC metabolome map describing etiological and survival heterogeneity in cancers of different subsites of the colorectum. Design: Utilizing 372 tumor tissues with matched normal mucosa liquid chromatography-mass spectrometry was applied to examine metabolomics profiles along 7 subsites of the colorectum: cecum (n=63), ascending colon (n=44), transverse colon (n=32), descending colon (n=28), sigmoid colon (n=75), rectosigmoid colon (n=38) and rectum (n=92). Linear regression showed distribution of metabolites along the subsites. Survival heterogeneity along subsites were statistically tested by cox regression analysis. Metabolome profiles were obtained from invitro cultures of 310 human microbial species. Results: 40 and 76 significantly altered metabolites (including bile acids, lysoPCs and LysPEs) among tumors and normal mucosa showed metabolic heterogeneity existed among CRC subsites. Gradual changes in metabolites abundance with significantly linear trend from cecum to rectum: 23 metabolites in tumor, 30 metabolites in normal mucosa and 15 metabolites in both tumor and normal mucosa, revealed concentration gradient depends on the disease status. Exclusively significant alteration of metabolites only in one subsite showed the differences in cancers among the 7 subsites of colorectum. We identified the metabolites associated with survival were different and unique to each subsite. Integration of microbial metabolome with CRC metabolome revealed microbial metabolites contributed to etiological and survival heterogeneity among CRC subsites. Conclusions: Gradual changes of metabolites from cecum to rectum and the association of different metabolites with CRC risk and survival in each subsite of colorectum, revealed the differences in cancers along the colorectum and challenge the simple classification of CRC in LCC, RCC and rectum. Human microbial metabolites contributed to etiological and survival heterogeneity among CRC subsites. Citation Format: Abhishek Jain, Caroline Johnson, Domenica Berardi, Sajid Khan, Montana Morris. Colorectum metabolome map revealed etiological and survival heterogeneity in cancers of different subsites: Challenging the proximal versus distal classification [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3112.

Color-CADx: a deep learning approach for colorectal cancer classification through triple convolutional neural networks and discrete cosine transform

Colorectal cancer (CRC) exhibits a significant death rate that consistently impacts human lives worldwide. Histopathological examination is the standard method for CRC diagnosis. However, it is complicated, time-consuming, and subjective. Computer-aided diagnostic (CAD) systems using digital pathology can help pathologists diagnose CRC faster and more accurately than manual histopathology examinations. Deep learning algorithms especially convolutional neural networks (CNNs) are advocated for diagnosis of CRC. Nevertheless, most previous CAD systems obtained features from one CNN, these features are of huge dimension. Also, they relied on spatial information only to achieve classification. In this paper, a CAD system is proposed called “Color-CADx” for CRC recognition. Different CNNs namely ResNet50, DenseNet201, and AlexNet are used for end-to-end classification at different training–testing ratios. Moreover, features are extracted from these CNNs and reduced using discrete cosine transform (DCT). DCT is also utilized to acquire spectral representation. Afterward, it is used to further select a reduced set of deep features. Furthermore, DCT coefficients obtained in the previous step are concatenated and the analysis of variance (ANOVA) feature selection approach is applied to choose significant features. Finally, machine learning classifiers are employed for CRC classification. Two publicly available datasets were investigated which are the NCT-CRC-HE-100 K dataset and the Kather_texture_2016_image_tiles dataset. The highest achieved accuracy reached 99.3% for the NCT-CRC-HE-100 K dataset and 96.8% for the Kather_texture_2016_image_tiles dataset. DCT and ANOVA have successfully lowered feature dimensionality thus reducing complexity. Color-CADx has demonstrated efficacy in terms of accuracy, as its performance surpasses that of the most recent advancements.

Abstract 1498: TGFβ signaling in cancer-associated fibroblasts drives a hepatic gp130-dependent pro-metastatic inflammatory program in CMS4 colorectal cancer

Abstract Background: Colorectal cancer (CRC) is characterized by a metastatic pattern that shows tropism towards the liver which is the major cause of CRC-related deaths. The current CRC classification, Consensus Molecular Subtypes (CMS), suggests an association between transforming growth factor (TGF)-β signaling, cancer-associated fibroblasts (CAFs), and the risk of liver metastasis. However, the key downstream effectors of TGFβ signaling in CAFs and their role in hepatic metastases remain poorly understood. Methods: The aim of this study was to investigate the TGF-β/CAF dependent cross talk with the hepatocytes in pre-metastatic niche formation. CAFs from 18 primary CRC tissues and 7 liver-metastasized CRC were stimulated with TGFβ1 and qPCR arrays/ELISAs were performed to identify differentially expressed genes. Results were cross-referenced with RNAseq expression data of CMS-classified CRC and in fresh samples from CRC patients. To investigate CAF-liver crosstalk hepatocytes were stimulated with the TGFβ primed CAF-derived cytokines and the effect on pro-inflammatory gene expression and neutrophil migration was determined. Subsequently, blockade of these pathways via chemical inhibition and CRISPR/Cas9 genetic ablation was performed to study the molecular mechanism involved in the induction of this hepatic pro-inflammatory program. Results: TGFβ signaling in primary CRC-derived CAFs leads to increased expression of different IL-6 cytokine family members, which was also reflected in human CRC samples. CAF-derived IL6 family members induce upregulation of myeloid chemoattractants, including SAA1, in hepatocytes and increased neutrophil-to-hepatocyte migration in vitro. Chemical blockade and genetic ablation of the IL-6 family cytokine signaling pathway showed the critical role of the gp130 co-receptor in the regulation of this inflammatory response in hepatocytes and the potential formation of a premetastatic niche. Conclusion: TGFβ signaling in CAFs actively contributes to the formation of a neutrophil-dependent, pre-metastatic hepatic niche, and this mechanism might play a role in CMS4 subtype CRC. Citation Format: Subinuer Abudukelimu, Tom J. Harryvan, Stef G. Janson, Imke Stouten, Els M. Verdegaal, Lukas JAC Hawinkels. TGFβ signaling in cancer-associated fibroblasts drives a hepatic gp130-dependent pro-metastatic inflammatory program in CMS4 colorectal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1498.

Prognostic and predictive role of immune microenvironment in colorectal cancer.

Colorectal cancer (CRC) represents a molecularly heterogeneous disease and one of the most frequent causes of cancer-related death worldwide. The traditional classification of CRC is based on pathomorphological and molecular characteristics of tumor cells (mucinous, ring-cell carcinomas, etc.), analysis of mechanisms of carcinogenesis involved (chromosomal instability, microsatellite instability, CpG island methylator phenotype) and mutational statuses of commonly altered genes (KRAS, NRAS, BRAF, APC, etc.), as well as expression signatures (CMS 1-4). It is also suggested that the tumor microenvironment is a key player in tumor progression and metastasis in CRC. According to the latest data, the immune microenvironment can also be predictive of the response to immune checkpoint inhibitors. In this review, we highlight how the immune environment influences CRC prognosis and sensitivity to systemic therapy.

Machine learning framework to extract the biomarker potential of plasma IgG N-glycans towards disease risk stratification

Effective management of chronic diseases and cancer can greatly benefit from disease-specific biomarkers that enable informative screening and timely diagnosis. IgG N-glycans found in human plasma have the potential to be minimally invasive disease-specific biomarkers for all stages of disease development due to their plasticity in response to various genetic and environmental stimuli. Data analysis and machine learning (ML) approaches can assist in harnessing the potential of IgG glycomics towards biomarker discovery and the development of reliable predictive tools for disease screening. This study proposes an ML-based N-glycomic analysis framework that can be employed to build, optimise, and evaluate multiple ML pipelines to stratify patients based on disease risk in an interpretable manner. To design and test this framework, a published colorectal cancer (CRC) dataset from the Study of Colorectal Cancer in Scotland (SOCCS) cohort (1999–2006) was used. In particular, among the different pipelines tested, an XGBoost-based ML pipeline, which was tuned using multi-objective optimisation, calibrated using an inductive Venn-Abers predictor (IVAP), and evaluated via a nested cross-validation (NCV) scheme, achieved a mean area under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.771 when classifying between age-, and sex-matched healthy controls and CRC patients. This performance suggests the potential of using the relative abundance of IgG N-glycans to define populations at elevated CRC risk who merit investigation or surveillance. Finally, the IgG N-glycans that highly impact CRC classification decisions were identified using a global model-agnostic interpretability technique, namely Accumulated Local Effects (ALE). We envision that open-source computational frameworks, such as the one presented herein, will be useful in supporting the translation of glycan-based biomarkers into clinical applications.

Pathological image profiling identifies onco-microbial, tumor immune microenvironment, and prognostic subtypes of colorectal cancer.

Histology slide, tissue microbes, and the host gene expression can be independent prognostic factors of colorectal cancer (CRC), but the underlying associations and biological significance of these multimodal omics remain unknown. Here, we comprehensively profiled the matched pathological images, intratumoral microbes, and host gene expression characteristics in 527 patients with CRC. By clustering these patients based on histology slide features, we classified the patients into two histology slide subtypes (HSS). Onco-microbial community and tumor immune microenvironment (TIME) were also significantly different between the two subtypes (HSS1 and HSS2) of patients. Furthermore, variation in intratumoral microbes-host interaction was associated with the prognostic heterogeneity between HSS1 and HSS2. This study proposes a new CRC classification based on pathological image features and elucidates the process by which tumor microbes-host interactions are reflected in pathological images through the TIME.

Integrated meta-analysis of colorectal cancer public proteomic datasets for biomarker discovery and validation.

The cancer biomarker field has been an object of thorough investigation in the last decades. Despite this, colorectal cancer (CRC) heterogeneity makes it challenging to identify and validate effective prognostic biomarkers for patient classification according to outcome and treatment response. Although a massive amount of proteomics data has been deposited in public data repositories, this rich source of information is vastly underused. Here, we attempted to reuse public proteomics datasets with two main objectives: i) to generate hypotheses (detection of biomarkers) for their posterior/downstream validation, and (ii) to validate, using an orthogonal approach, a previously described biomarker panel. Twelve CRC public proteomics datasets (mostly from the PRIDE database) were re-analysed and integrated to create a landscape of protein expression. Samples from both solid and liquid biopsies were included in the reanalysis. Integrating this data with survival annotation data, we have validated in silico a six-gene signature for CRC classification at the protein level, and identified five new blood-detectable biomarkers (CD14, PPIA, MRC2, PRDX1, and TXNDC5) associated with CRC prognosis. The prognostic value of these blood-derived proteins was confirmed using additional public datasets, supporting their potential clinical value. As a conclusion, this proof-of-the-concept study demonstrates the value of re-using public proteomics datasets as the basis to create a useful resource for biomarker discovery and validation. The protein expression data has been made available in the public resource Expression Atlas.

HSPA8 Activates Wnt/β-Catenin Signaling to Facilitate BRAF V600E Colorectal Cancer Progression by CMA-Mediated CAV1 Degradation.

BRAF V600E attracts wide attention in the treatment of colorectal cancer (CRC) as stratifying and predicting a refractory classification of CRC. Recent evidence indicates that Wnt/β-catenin signaling is broadly activated and participates in the refractoriness of BRAF V600E CRC, but the underlying molecular mechanism needs to be elucidated. Here, heat shock 70 kDa protein 8 (HSPA8), an essential regulator in chaperone-mediated autophagy (CMA), is identified as a potential therapeutic target for advanced BRAF V600E CRC. These results show that HSPA8 is transcriptionally upregulated in BRAF V600E CRC, which promotes CMA-dependent degradation of caveolin-1 (CAV1) to release β-catenin into the nucleus and thus activates the Wnt/β-catenin pathway, contributing to metastasis and progression of BRAF V600E CRC. Of note, HSPA8 directly interacts with the KIFSN motif on CAV1, the interaction can be enhanced by p38 MAPK-mediated CAV1 S168 phosphorylation. Furthermore, pharmacological targeting HSPA8 by VER155008 exhibits synergistic effects with BRAF inhibitors on CRC mouse models. In summary, these findings discover the important role of the HSPA8/CAV1/β-catenin axis in the development of refractory BRAF V600E CRC and highlight HSPA8 as a predictive biomarker and therapeutic target in clinical practice.

Reptile Search Algorithm with Deep Convolutional Neural Network for Cloud Assisted Colorectal Cancer Detection and Classification

Cloud-based automatic colorectal cancer (CC) detection involves the usage of cloud computing technology and system to help in the earlier and accurate diagnosis of CC in medical images and patient information. This cloud-based technology aims to improve the efficiency and reliability of CC screening, monitoring, and diagnoses. Automatic CC detection refers to the use of computer-based technology and systems to aid in the earlier and accurate detection of CC in patient data and medical images. This automated system aims to increase the reliability and efficiency of CC monitoring, screening, and diagnosis. Deep learning (DL) methods, especially convolutional neural networks (CNNs), exhibit promising results in automatic CC diagnosis. They can be trained on wide-ranging datasets of medical images to learn patterns and features related to precancerous and cancerous lesion. This study develops a new Reptile Search Algorithm with Deep Learning for Colorectal Cancer Detection and Classification (RSADL-CCDC) technique. The main aim of the RSADL-CCDC method focuses on the automaticclassification and recognition of the CC in the cloud environment. Once the medical images are stored in the cloud server, the detection process is carried out. In the presented RSADL-CCDC approach, the initial stage of preprocessing is performed by bilateral filtering (BF) approach. For feature extraction, the RSADL-CCDC technique applies ShuffleNetv2 model. Besides, the recognition and classification of CC take place using convolutional autoencoder (CAE) model. Finally, the hyperparameter tuning of the CAE technique takes place by utilizing RSA. The experimental validation of the RSADL-CCDC system is performed on benchmark medical database. Extensive results stated the enhanced performance of the RSADL-CCDC technique on CC recognition over other models with respect tovarious actions.

A biospectroscopic approach toward colorectal cancer diagnosis from bodily fluid samples via ATR–MIR spectroscopy combined with multivariate data analysis

In this study, a biospectroscopic approach was reported for the detection of spectral changes and biomarkers for the diagnosis of colorectal cancer (CC) cases from different bodily fluids (blood plasma, blood serum, saliva and colonoscopy disinfection/wash fluids) by using attenuated total reflection–mid infrared (ATR–MIR) spectroscopy. To recognize the molecular level changes in the spectral characteristics of CC and their healthy/control (CH) groups, different multivariate data analyses (HCA, LDA, PCA and SIMCA) were successfully performed over the data of ATR–MIR spectroscopy. Two hundred specimens were characterized in detail over the data of spectral regions (4000–650 cm−1 and regions V–XXII). The findings revealed that significant changes were clearly observed in the concentrations of lipid, protein, nucleic acid and carbohydrate biomolecules for cancer cases based upon their necessity to overcome energy requirements. Supervised multivariate data methodology SIMCA, presented an excellent classification for the studied groups; similarly 100% of the specimens from different bodily fluids were correctly classified by supervised methodology LDA. As a result, the developed ATR–MIR methodology for the classification of CC and their healthy groups highlighted a rapid cancer diagnosis approach from different bodily fluids; therefore, it could be guide to make well decision before histopathological assessment and to screen CC populations existing in society.

Deep immunophenotyping reveals clinically distinct cellular states and ecosystems in large-scale colorectal cancer

Determining the diverse cell types in the tumor microenvironment (TME) and their organization into cellular communities, is critical for understanding the biological heterogeneity and therapy of cancer. Here, we deeply immunophenotype the colorectal cancer (CRC) by integrative analysis of large-scale bulk and single cell transcriptome of 2350 patients and 53,137 cells. A rich landscape of 42 cellular states and 7 ecosystems in TMEs is uncovered and extend the previous immune classifications of CRC. Functional pathways and potential transcriptional regulators analysis of cellular states and ecosystems reveal cancer hallmark-related pathways and several critical transcription factors in CRC. High-resolution characterization of the TMEs, we discover the potential utility of cellular states (i.e., Monocytes/Macrophages and CD8 T cell) and ecosystems for prognosis and clinical therapy selection of CRC. Together, our results expand our understanding of cellular organization in TMEs of CRC, with potential implications for the development of biomarkers and precision therapies.

The circadian clock circuitry deconvolutes colorectal cancer and lung adenocarcinoma heterogeneity in a dynamic time-related framework.

Increasing evidence imputes cancer progression and resistance to therapy to intra-tumor molecular heterogeneity set off by cancer cell plasticity. Re-activation of developmental programs strictly linked to epithelial-to-mesenchymal transition and gaining of stem cells properties are crucial in this setting. Many biological processes involved in cancer onset and progression show rhythmic fluctuations driven by the circadian clock circuitry. Novel cancer patient stratification tools taking into account the temporal dimension of these biological processes are definitely needed. Lung cancer and colorectal cancer (CRC) are the leading causes of cancer death worldwide. Here, by developing an innovative computational approach we named Phase-Finder, we show that the molecular heterogeneity characterizing the two deadliest cancers, CRC and lung adenocarcinoma (LUAD), rather than a merely stochastic event is the readout of specific cancer molecular states which correlate with time-qualified patterns of gene expression. We performed time-course transcriptome analysis of CRC and LUAD cell lines and upon computing circadian genes expression-based correlation matrices we derived pseudo-time points to infer time-qualified patterns in the transcriptomic analysis of real-world data (RWD) from large cohorts of CRC and LUAD patients. Our temporal classification of CRC and LUAD cohorts was able to effectively render time-specific patterns in cancer phenotype switching determining dynamical distribution of molecular subtypes impacting patient prognosis.

Classification of colorectal cancer consensus molecular subtypes using attention-based multi-instance learning network on whole-slide images

Colorectal cancer (CRC) is the third most common and second most lethal cancer globally. It is highly heterogeneous with different clinical-pathological characteristics, prognostic status, and therapy responses. Thus, the precise diagnosis of CRC subtypes is of great significance for improving the prognosis and survival of CRC patients. Nowadays, the most commonly used molecular-level CRC classification system is the Consensus Molecular Subtypes (CMSs). In this study, we applied a weakly supervised deep learning method, named attention-based multi-instance learning (MIL), on formalin-fixed paraffin-embedded (FFPE) whole-slide images (WSIs) to distinguish CMS1 subtype from CMS2, CMS3, and CMS4 subtypes, as well as distinguish CMS4 from CMS1, CMS2, and CMS3 subtypes. The advantage of MIL is training a bag of the tiled instance with bag-level labels only. Our experiment was performed on 1218 WSIs obtained from The Cancer Genome Atlas (TCGA). We constructed three convolutional neural network-based structures for model training and evaluated the ability of the max-pooling operator and mean-pooling operator on aggregating bag-level scores. The results showed that the 3-layer model achieved the best performance in both comparison groups. When compared CMS1 with CMS234, max-pooling reached the ACC of 83.86 % and the mean-pooling operator reached the AUC of 0.731. While comparing CMS4 with CMS123, mean-pooling reached the ACC of 74.26 % and max-pooling reached the AUC of 0.609. Our results implied that WSIs could be utilized to classify CMSs, and manual pixel-level annotation is not a necessity for computational pathology imaging analysis.

Molecular subtyping in colorectal cancer: A bridge to personalized therapy (Review).

Colorectal cancer (CRC) is a malignant tumor and a major cause of morbidity and mortality globally. The classic Tumor-Node-Metastasis staging system, which currently underlies the diagnosis and treatment of CRC, is primarily a 'one drug fits all' model for patients exhibiting the same pathological features. However, a high degree of variability has been established in the long-term survival outcomes of patients with CRC with similar pathological types and stages, which can be partially attributed to tumor-specific molecular biology to some extent. Molecular classification of CRC can further assist with understanding the biological behavior of tumor genesis, development and prognosis, and assist clinicians in improving or customizing the treatment strategy of CRC. In the present study, clinical studies carried out to date are reviewed, and their clinical value is discussed. A multilevel overview of the major molecular types of CRC is provided, in the hope that investigators are encouraged to combine multiple omics studies for interrogating cancer.

Abstract LB177: Tumor sideomics: spatial characteristics and functional deregulation in colorectal cancer

Abstract Clinical evidence suggests that right-sided colorectal cancer (CRC) tends to have a worse prognosis compared to left-sided CRC. The distinct functional mechanisms underlying this “sidedness” remain cryptic and poorly understood. Thus, molecular classification of CRC is of paramount in targeting specific subtypes and personalizing treatment for patients. To identify targetable molecular features to improve treatment regimens, we performed multi-faceted network analyses of hundreds of CRC patients. In our network model, epigenetic features, differential expression of transcript abundance, and mutational frequency were used to identify widespread gene dysregulation and perturbed pathways to reveal disrupted networks distinct to right or left sided colorectal cancer. We discovered histone related genes and distinct epigenetic signals are at play in right-sided CRC. Additionally, our studies reveal “at risk” patient characteristics suggestive of survival differences, to improve diagnostic predictors based on CRC sidedness. Overall, our integrative approach serves to link molecular profiles distinct between left and right sided colorectal cancer, while improving upon previous models of CRC and paving the way towards identifying more effective treatment alternatives from a previously overlooked, readily characterized subtype of colorectal cancer. Citation Format: Brandon Burgman, Xingxin Pan, S. Stephen Yi. Tumor sideomics: spatial characteristics and functional deregulation in colorectal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr LB177.

Blood Biomarkers Panels for Screening of Colorectal Cancer and Adenoma on a Machine Learning-Assisted Detection Platform.

A mini-invasive and good-compliance program is critical to broaden colorectal cancer (CRC) screening and reduce CRC-related mortality. Blood testing combined with imaging examination has been proved to be feasible on screen for multicancer and guide intervention. The study aims to construct a machine learning-assisted detection platform with available multi-targets for CRC and colorectal adenoma (CRA) screening. This was a retrospective study that the blood test data from 204 CRCs, 384 CRAs, and 229 healthy controls was extracted. The classified models were constructed with 4 machine learning (ML) algorithms including support vector machine (SVM), random forest (RF), decision tree (DT), and eXtreme Gradient Boosting (XGB) based on the candidate biomarkers. The importance index was used by SHapely Adaptive exPlanations (SHAP) analysis to identify the dominant characteristics. The performance of classified models was evaluated. The most dominating features from the proposed panel were developed by logistic regression (LR) for identification CRC from control. The candidate biomarkers consisted of 26 multi-targets panel including CEA, AFP, and so on. Among the 4 models, the SVM classifier for CRA yields the best predictive performance (the area under the receiver operating curve, AUC: .925, sensitivity: .904, and specificity: .771). As for CRC classification, the RF model with 26 candidate biomarkers provided the best predictive parameters (AUC: .941, sensitivity: .902, and specificity: .912). Compared with CEA and CA199, the predictive performance was significantly improved. The streamlined model with 6 biomarkers for CRC also obtained a good performance (AUC: .946, sensitivity: .885, and specificity: .913). The predictive models consisting of 26 multi-targets panel would be used as a non-invasive, economical, and effective risk stratification platform, which was expected to be applied for auxiliary screening of CRA and CRC in clinical practice.

A Novel Molecular Analysis Approach in Colorectal Cancer Suggests New Treatment Opportunities

Colorectal cancer (CRC) is a molecular and clinically heterogeneous disease. In 2015, the Colorectal Cancer Subtyping Consortium classified CRC into four consensus molecular subtypes (CMS), but these CMS have had little impact on clinical practice. The purpose of this study is to deepen the molecular characterization of CRC. A novel approach, based on probabilistic graphical models (PGM) and sparse k-means-consensus cluster layer analyses, was applied in order to functionally characterize CRC tumors. First, PGM was used to functionally characterize CRC, and then sparse k-means-consensus cluster was used to explore layers of biological information and establish classifications. To this aim, gene expression and clinical data of 805 CRC samples from three databases were analyzed. Three different layers based on biological features were identified: adhesion, immune, and molecular. The adhesion layer divided patients into high and low adhesion groups, with prognostic value. The immune layer divided patients into immune-high and immune-low groups, according to the expression of immune-related genes. The molecular layer established four molecular groups related to stem cells, metabolism, the Wnt signaling pathway, and extracellular functions. Immune-high patients, with higher expression of immune-related genes and genes involved in the viral mimicry response, may benefit from immunotherapy and viral mimicry-related therapies. Additionally, several possible therapeutic targets have been identified in each molecular group. Therefore, this improved CRC classification could be useful in searching for new therapeutic targets and specific therapeutic strategies in CRC disease.

Tunicate swarm algorithm with deep convolutional neural network-driven colorectal cancer classification from histopathological imaging data

<abstract> <p>Colorectal cancer (CRC) is one of the most popular cancers among both men and women, with increasing incidence. The enhanced analytical load data from the pathology laboratory, integrated with described intra- and inter-variabilities through the calculation of biomarkers, has prompted the quest for robust machine-based approaches in combination with routine practice. In histopathology, deep learning (DL) techniques have been applied at large due to their potential for supporting the analysis and forecasting of medically appropriate molecular phenotypes and microsatellite instability. Considering this background, the current research work presents a metaheuristics technique with deep convolutional neural network-based colorectal cancer classification based on histopathological imaging data (MDCNN-C3HI). The presented MDCNN-C3HI technique majorly examines the histopathological images for the classification of colorectal cancer (CRC). At the initial stage, the MDCNN-C3HI technique applies a bilateral filtering approach to get rid of the noise. Then, the proposed MDCNN-C3HI technique uses an enhanced capsule network with the Adam optimizer for the extraction of feature vectors. For CRC classification, the MDCNN-C3HI technique uses a DL modified neural network classifier, whereas the tunicate swarm algorithm is used to fine-tune its hyperparameters. To demonstrate the enhanced performance of the proposed MDCNN-C3HI technique on CRC classification, a wide range of experiments was conducted. The outcomes from the extensive experimentation procedure confirmed the superior performance of the proposed MDCNN-C3HI technique over other existing techniques, achieving a maximum accuracy of 99.45%, a sensitivity of 99.45% and a specificity of 99.45%.</p> </abstract>

Galactic swarm optimization with deep transfer learning driven colorectal cancer classification for image guided intervention

In this era of “precision” medicine, image-guided intervention (IGI) enables real-time customized and accurate treatment using imaging phenotype-based approaches. Colorectal cancer (CC) is the third most commonly occurring cancer, resulting in nearly 10% of cases over the globe. Colorectal cancer classification (CCC) of histopathological images by artificial intelligence (AI) approaches not only enhances the accuracy and classifier results but also allows physicians to make prompt decisions. In this view, this article introduces a novel Galactic Swarm Optimization with Deep Transfer Learning Driven Colorectal Cancer Classification (GSODTL-C3M) model for IGI. The primary aim of the GSODTL-C3M model is to appropriately categorize the test images into the existence of CC. To accomplish this, the presented GSODTL-C3M model employs image pre-processing using the bilateral filtering (BF) technique to remove noise. Besides, Adam optimizer with the MobileNet model is applied as a feature extractor. Finally, the GSO methodology with long short-term memory (LSTM) is employed to recognize and classify CC. An extensive range of simulations was taken place, and the results stated the advanced performance of the current state of art classification methodologies.