Current Cancer Drug Targets - Volume 25, Issue 12, 2025

Gastrointestinal (GI) cancers represent some of the most common and lethal malignancies globally, underscoring the urgent need for improved diagnostic strategies. Traditional diagnostic methods, while effective to some degree, are often invasive and unsuitable for regular screenings.

This review article explores integrating machine learning (ML) with liquid biopsy techniques as a revolutionary approach to enhance the detection and monitoring of GI cancers. Liquid biopsies offer a non-invasive alternative for cancer detection through the analysis of circulating tumor DNA (ctDNA) and other biomarkers, which when combined with ML, can significantly improve diagnostic accuracy and patient outcomes.

We conducted a comprehensive review of recent advancements in liquid biopsy and ML, focusing on their synergistic potential in the early detection of GI cancers. The review addresses the application of next-generation sequencing and digital droplet PCR in enhancing the sensitivity and specificity of liquid biopsies.

Machine learning algorithms have demonstrated remarkable ability in navigating complex datasets and identifying diagnostically significant patterns in ctDNA and other circulating biomarkers. Innovations such as machine learning-enhanced “fragmentomics” and tomographic phase imaging flow cytometry illustrate significant strides in non-invasive cancer diagnostics, offering enhanced detection capabilities with high accuracy.

The integration of ML in liquid biopsy represents a transformative step in the early detection and personalized treatment of GI cancers. Future research should focus on overcoming current limitations, such as the heterogeneity of tumor-derived genetic materials and the standardization of liquid biopsy protocols, to fully realize the potential of this technology in clinical settings.

Cytochrome P450 (CYP) enzymes are involved in the metabolism of xenobiotic and carcinogen. In the study, we evaluated the association of CYP2J2 and CYP2C9 variants with lung cancer.

Five polymorphisms in CYP2J2 and four polymorphisms in CYP2C9 were genotyped in 507 lung cancer patients and 505 controls with Agena MassARRAY platform. The linkage of variants with lung cancer risk was evaluated by odds ratio (OR) and 95% confidence interval (CI) in genetic models and haplotype analyses.

We found that CYP2C9 rs1934967 alleles were associated with lung cancer risk (P < 0.05). In stratified analysis, rs2280274 (women, non-smoker, non-drinker and lymphatic metastasis), rs11207535 (non-smoker and non-drinker), rs10889159 (non-smoker and non-drinker) of CYP2J2, whereas rs1934967 (age ≤ 60, BMI > 24, squamous carcinoma) of CYP2C9 decreased lung cancer risk (P < 0.05). In addition, the results of linkage disequilibrium (LD) analysis showed that rs2280274|rs4388726 - TG (with adjustment: P = 0.042) of CYP2J2 and rs10509679|rs1934967| rs1934968|rs9332220 – GTGG (without adjustment: P = 0.044) of CYP2C9 were linked with a significantly decreased lung cancer risk.

Our results indicated genetic variants in CYP2J2 and CYP2C9 might contribute to the susceptibility of lung cancer in Chinese population.

There is a relative lack of real-world data regarding the treatment of advanced squamous cell lung cancer (SqCLC) with anlotinib.

This study aimed to investigate the clinical efficacy and safety of combining anlotinib with ICIs for the treatment of advanced SqCLC.

Sixty patients with advanced SqCLC from December, 2017 to December, 2022, were retrospectively recruited in the final analysis. The overall survival (OS), adverse events (AEs), and short-term clinical effectiveness, including complete remission (CR), partial remission (PR), stable disease (SD), and progression disease (PD) of the study populations, were observed. Objective response rate (ORR) was defined as the combined percentage of CR and PR in all patients. Disease control rate (DCR) was defined as the combined percentage of CR, PR, and SD in all patients. Subgroups were classified based on age (<65 or ≥65), sex (female or male), and the eastern cooperative oncology group (ECOG) score (0-1 or 1-2).

The median age of the study population was 65 years, and 41 (68.3%) of the participants were male. ORR was 21.67%, and DCR was 83.33%. The median PFS and OS were 7.5 and 19.5 months, respectively. Furthermore, the Kaplan-Meier curves demonstrated no significant difference (P >0.05) in PFS and OS between different age groups, sex, and ECOG scores. Bone marrow suppression was the most common AE, followed by immune pneumonia.

The findings of this study confirmed the effectiveness and safety of anlotinib and ICIs for advanced SqCLC.

Over 50% of lung adenocarcinoma patients have high levels of complement factor H (CFH) expression. Previous studies have reported that CFH inhibits the migration of endothelial cells. In this study, we investigated the mechanism by which CFH affects lung adenocarcinoma development via phosphorylation of STAT3.

Adenovirus expressing mice Cfh gene was used to infect C57 mice for two weeks, and then Lewis Lung Carcinoma (LLC) was injected to develop a subcutaneous tumor. The effect of CFH on human A549 cells was also detected. Moreover, we collected CFH over-expressed conditional medium from HepG-2 cells infected with adenovirus expressing human CFH gene. A549 cells were incubated with the conditional medium, and the effect of the CFH-conditional medium on cell proliferation and migration was detected.

It was found that CFH promoted lung adenocarcinoma growth in vivo, and CFH-conditional medium treatment significantly increased the viability and migration area of A549 cells. CFH-conditional medium increased the phosphorylation of JAK2 and STAT3 in A549 cells. While using STATTIC to block STAT3 phosphorylation, CFH-conditional medium treatment did not affect A549 cell viability or migration compared to the control group.

These data suggested that CFH promoted the proliferation and migration of A549 cells by increasing the phosphorylation level of the JAK2/STAT3 signaling pathway. Furthermore, CFH has the potential to be a target for antitumor therapy.

Colorectal adenocarcinoma (COAD) is a prevalent and lethal form of cancer. Understanding the molecular mechanisms underlying COAD progression is crucial for developing effective diagnostic and therapeutic strategies.

This study aims to explore wound healing-related genes in COAD and their potential roles in tumorigenesis and prognosis using in silico and in vitro methodology.

A set of 70 genes associated with the “wound healing” term ere extracted from the Gene Ontology (GO) database (GO:0042060) and a protein-protein interaction (PPI) network was constructed using the STRING database. The PPI network was analyzed with the CytoHubba plugin in Cytoscape, identifying four major hub genes: MMP2, FN1, NF1, and PTK7. We then analyzed the expression of these hub genes across 16 COAD cell lines and nine normal colon cell lines using RT-qPCR, finding significant overexpression in COAD cell lines. ROC curve analysis confirmed the diagnostic potential of these genes, with MMP2, FN1, and NF1 showing high AUC values. Expression validation using the TCGA COAD cohort, OncoDB, and HPA databases corroborated these findings, highlighting the overexpression and high protein levels of these genes in COAD. Promoter methylation analysis indicated lower methylation levels in COAD samples, suggesting dysregulation through epigenetic mechanisms. Genetic alteration analysis via cBioPortal revealed a spectrum of mutations, with FN1 being the most frequently mutated. Prognostic analysis using a KM plotter showed that high expression of the hub genes is associated with poorer overall survival (OS) and disease-free survival (DFS). Functional state correlations via CancerSEA suggested that these genes promote cell cycle, proliferation, metastasis, and stemness in COAD. Expression analysis in immune cells and drug sensitivity analyses highlighted the roles of MMP2, FN1, and NF1 in macrophages and drug resistance. A miRNA-mRNA network constructed using miRNet identified hsa-miR-200a-3p as a central regulator. Finally, functional assays in HCT116 cells demonstrated that knockdown of MMP2 and FN1 reduced proliferation, colony formation, and wound healing, suggesting these genes as potential therapeutic targets in COAD.

In conclusion, our study identifies MMP2, FN1, NF1, and PTK7 as key wound healing-related hub genes in COAD.

PANoptosis, a novelty mechanism of cell death involving crosstalk between apoptosis, pyroptosis, and necroptosis, is strongly associated with tumor cell death and immunotherapy efficacy. However, its relevance in lung adenocarcinoma (LUAD) remains to be elucidated.

In this study, we acquired 18 PANoptosis-related differentially expressed gene (PRDEG) of LUAD. Based on these genes, LUAD samples were identified with different subtypes by unsupervised clustering. Next, we compared the differences between the subtypes, including clinical features, immune microenvironment, and potentially sensitive drugs. Furthermore, we used machine learning to identify hub prognostic PRDEGs, construct a risk score, and validate it on other external datasets. We incorporated the patient's clinical information and risk score into the proportional hazards model and lasso-cox models to find key prognostic features and constructed five prognostic models. The best model was identified via the area under the curve and validated on an external dataset.

LUAD patients were divided into two clusters named C1 and C2, respectively. The C2 cluster exhibited shorter survival time, more advanced tumor stage, higher suppressive immune cell scores, such as dendritic cells, and higher expression of inhibitory immune checkpoints, such as LAG3 and CD86. TIMP1, CAV1, and CD69 were recognized as key prognostic factors, and risk scores predicted survival with significant differences in the external validation set. Risk score and N-stage were identified as critical prognostic features. The Coxph model outperformed other machine learning clinical models. The 1-, 3-, and 5-year time-ROCs in the external validation set were 0.55, 0.59, and 0.60, respectively.

We demonstrated the potential of PANoptosis-based molecular clustering and prognostic features in predicting the survival of patients with LUAD as well as the tumor microenvironment.