ColoFinder: a prognostic 9-gene signature improves prognosis for 871 stage II and III colorectal cancer patients

Discovery and validation of a 10-gene predictive signature for response to adjuvant chemotherapy in stage II and III colon cancer

Identifying patients with stage II and III colon cancer who will benefit from 5-fluorouracil (5-FU)-based adjuvant chemotherapy is crucial for the advancement of personalized cancer therapy. We employ a semi-supervised machine learning approach to analyze a large dataset with 933 stage II and III colon cancer samples. Our analysis leverages gene regulatory networks to discover an 18-gene prognostic signature and to explore a 10-gene signature that potentially predicts chemotherapy benefits. The 10-gene signature demonstrates strong prognostic power and shows promising potential to predict chemotherapy benefits. We establish a robust clinical assay on the NanoString nCounter platform, validated in a retrospective formalin-fixed paraffin-embedded (FFPE) cohort, which represents an important step toward clinical application. Our study lays the groundwork for improving adjuvant chemotherapy and potentially expanding into immunotherapy decision-making in colon cancer. Future prospective studies are needed to validate and establish the clinical utility of the 10-gene signature in clinical settings.

Prognostic impact of tumor size on patients with metastatic colorectal cancer: a large SEER-based retrospective cohort study

Given the poor prognosis of metastatic colorectal cancer (mCRC), this research aimed to investigate the correlation between tumor size and prognosis, and develop a novel prediction model to guide individualized treatment. Patients pathologically diagnosed with mCRC were recruited from the Surveillance, Epidemiology, and End Results (SEER) database between 2010 and 2015, and were randomly divided (7:3 ratio) into a training cohort (n = 5597) and a validation cohort (n = 2398). Kaplan-Meier curves were used to analyze the relationship between tumor size and overall survival (OS). Univariate Cox analysis was applied to assess the factors associated with the prognosis of mCRC patients in the training cohort, and then multivariate Cox analysis was used to construct a nomogram model. The area under the receiver-operating characteristics curve (AUC) and calibration curve were used to evaluate the predictive ability of the model. Patients with larger tumors had a worse prognosis. While brain metastases were associated with larger tumors compared to liver or lung metastases, bone metastases tended to be associated with smaller tumors. Multivariate Cox analysis revealed that tumor size was an independent prognostic risk factor (HR 1.28, 95% CI 1.19-1.38), in addition to the other ten variables (age, race, primary site, grade, histology, T stage, N stage, chemotherapy, CEA level and metastases site). The 1-, 3-, and 5-year OS nomogram model yielded AUC values of more than 0.70 in both the training and validation cohorts, and its predictive performance was superior to that of the traditional TNM stage. Calibration plots demonstrated a good agreement between the predicted and observed 1-, 3-, and 5-year OS outcomes in both cohorts. The size of primary tumor was found to be significantly associated with prognosis of mCRC, and was also correlated with specific metastatic organ. In this study, we presented the first effort to create and validate a novel nomogram for predicting 1-, 3- and 5-year OS probabilities of mCRC. The prognostic nomogram was demonstrated to have an excellent predictive ability in estimating individualized OS of patients with mCRC.

The Expression Profile, Clinical Application and Potential Tumor Suppressing Mechanism of hsa_circ_0001675 in Head and Neck Carcinoma

Purpose

This study sought to identify circular RNAs (circRNA) that participate in the regulation of head and neck cancer (HNC), analyze their clinical application, and predict their molecular mechanism during HNC.

Materials and Methods

High-throughput sequencing was used to analyze circRNA expression in 18 matched HNC and adjacent normal tissues. Target circRNAs with significantly differential expression were obtained. In 103 HNC and adjacent normal tissues, real-time fluorescent quantitative PCR (qRT-PCR) was used to verify the differential expression of target circRNAs. This data was combined with clinicopathological information to analyze the diagnostic value of target circRNA. Bioinformatics was used to find target circRNAs that acted as competitive endogenous RNA (ceRNA) and construct a circRNA-miRNA-mRNA regulatory network. mRNA expression was verified by immunohistochemistry (IHC).

Results

A total of 714 differentially expressed circRNAs were detected in HNC, and the low expression of hsa_circ_0001675 was particularly significant (fold change [FC] = -4.85, P = 6.305E-05). hsa_circ_0001675 had significantly lower expression in HNC than in normal tissue (P < 0.01). Low hsa_circ_0001675 expression was positively associated with tumor invasion and clinical staging (P < 0.05), and its area under the ROC curve (AUC) was 0.7776. Low hsa_circ_0001675 expression also correlated with the overall survival (OS) rate and the progression-free survival (PFS) rate of HNC patients (P < 0.001). Bioinformatics was used to construct a ceRNA network of hsa_circ_0001675 with six differentially expressed miRNAs (hsa-miR-330-5p, hsa-miR-498, hsa-miR-532-3p, hsa-miR-577, hsa-miR-1248, and hsa-miR-1305) and 411 differentially expressed mRNAs and found that the neuroactive ligand-receptor interaction, and the cAMP and calcium signaling pathways were particularly enriched. Further bioinformatics and IHC analysis showed that miR577/TESC is the likely downstream signaling pathway for hsa_circ_0001675.

Conclusion

This study showed that hsa_circ_0001675 is downregulated in HNC and could be an effective biomarker for HNC diagnosis. In addition, hsa_circ_0001675 may have a potential ceRNA mechanism and suppress HNC disease progression through the hsa_circ_0001675-miRNA-mRNA axis.

Integrative analysis from multi-center studies identities a consensus machine learning-derived lncRNA signature for stage II/III colorectal cancer

Background

Long non-coding RNAs (lncRNAs) have recently emerged as essential biomarkers of cancer progression. However, studies are limited regarding lncRNAs correlated with recurrence and fluorouracil-based adjuvant chemotherapy (ACT) in stage II/III colorectal cancer (CRC).

Methods

1640 stage II/III CRC patients were enrolled from 15 independent datasets and a clinical in-house cohort. 10 prevalent machine learning algorithms were collected and then combined into 76 combinations. 109 published transcriptome signatures were also retrieved. qRT-PCR assay was performed to verify our model.

Findings

We comprehensively identified 27 stably recurrence-related lncRNAs from multi-center cohorts. According to these lncRNAs, a consensus machine learning-derived lncRNA signature (CMDLncS) that exhibited best power for predicting recurrence risk was determined from 76 kinds of algorithm combinations. A high CMDLncS indicated unfavorable recurrence and mortality rates. CMDLncS not only could work independently of common clinical traits (e.g., AJCC stage) and molecular features (e.g., microsatellite state, KRAS mutation), but also presented dramatically better performance than these variables. qRT-PCR results from 173 patients further verified our in-silico findings and assessed its feasible in different centers. Comparisons of CMDLncS with 109 published transcriptome signatures further demonstrated its predictive superiority. Additionally, patients with high CMDLncS benefited more from fluorouracil-based ACT and were characterized by activation of stromal and epithelial-mesenchymal transition, while patients with low CMDLncS suggested the sensitivity to bevacizumab and displayed enhanced immune activation.

Interpretation

CMDLncS provides an attractive platform for identifying patient at high risk of recurrence and could optimize precision treatment to improve the clinical outcomes in stage II/III CRC.

Funding

This study was supported by the National Natural Science Foundation of China (81,972,663); Henan Province Young and Middle‐Aged Health Science and Technology Innovation Talent Project (YXKC2020037); and Henan Provincial Health Commission Joint Youth Project (SB201902014).

A novel prognostic risk score model based on immune-related genes in patients with stage IV colorectal cancer

PURPOSE

The aims of the present study were to explore immune-related genes (IRGs) in stage IV colorectal cancer (CRC) and construct a prognostic risk score model to predict patient overall survival (OS), providing a reference for individualized clinical treatment.

METHODS

High-throughput RNA-sequencing, phenotype, and survival data from patients with stage IV CRC were downloaded from TCGA. Candidate genes were identified by screening for differentially expressed IRGs (DE-IRGs). Univariate Cox regression, LASSO, and multivariate Cox regression analyses were used to determine the final variables for construction of the prognostic risk score model. GSE17536 from the GEO database was used as an external validation dataset to evaluate the predictive power of the model.

RESULTS

A total of 770 candidate DE-IRGs were obtained, and a prognostic risk score model was constructed by variable screening using the following 12 genes: FGFR4, LGR6, TRBV12-3, NUDT6, MET, PDIA2, ORM1, IGKV3D-20, THRB, WNT5A, FGF18, and CCR8. In the external validation set, the survival prediction C-index was 0.685, and the AUC values were 0.583, 0.731, and 0.837 for 1-, 2- and 3-year OS, respectively. Univariate and multivariate Cox regression analyses demonstrated that the risk score model was an independent prognostic factor for patients with stage IV CRC. High- and low-risk patient groups had significant differences in the expression of checkpoint coding genes (ICGs).

CONCLUSION

The prognostic risk score model for stage IV CRC developed in the present study based on immune-related genes has acceptable predictive power, and is closely related to the expression of ICGs.

A machine learning-based gene signature of response to the novel alkylating agent LP-184 distinguishes its potential tumor indications

BACKGROUND

Non-targeted cytotoxics with anticancer activity are often developed through preclinical stages using response criteria observed in cell lines and xenografts. A panel of the NCI-60 cell lines is frequently the first line to define tumor types that are optimally responsive. Open data on the gene expression of the NCI-60 cell lines, provides a unique opportunity to add another dimension to the preclinical development of such drugs by interrogating correlations with gene expression patterns. Machine learning can be used to reduce the complexity of whole genome gene expression patterns to derive manageable signatures of response. Application of machine learning in early phases of preclinical development is likely to allow a better positioning and ultimate clinical success of molecules. LP-184 is a highly potent novel alkylating agent where the preclinical development is being guided by a dedicated machine learning-derived response signature. We show the feasibility and the accuracy of such a signature of response by accurately predicting the response to LP-184 validated using wet lab derived IC50s on a panel of cell lines.

RESULTS

We applied our proprietary RADR® platform to an NCI-60 discovery dataset encompassing LP-184 IC50s and publicly available gene expression data. We used multiple feature selection layers followed by the XGBoost regression model and reduced the complexity of 20,000 gene expression values to generate a 16-gene signature leading to the identification of a set of predictive candidate biomarkers which form an LP-184 response gene signature. We further validated this signature and predicted response to an additional panel of cell lines. Considering fold change differences and correlation between actual and predicted LP-184 IC50 values as validation performance measures, we obtained 86% accuracy at four-fold cut-off, and a strong (r = 0.70) and significant (p value 1.36e-06) correlation between actual and predicted LP-184 sensitivity. In agreement with the perceived mechanism of action of LP-184, PTGR1 emerged as the top weighted gene.

CONCLUSION

Integration of a machine learning-derived signature of response with in vitro assessment of LP-184 efficacy facilitated the derivation of manageable yet robust biomarkers which can be used to predict drug sensitivity with high accuracy and clinical value.

Gene dysregulation analysis builds a mechanistic signature for prognosis and therapeutic benefit in colorectal cancer

The implementation of cancer precision medicine requires biomarkers or signatures for predicting prognosis and therapeutic benefits. Most of current efforts in this field are paying much more attention to predictive accuracy than to molecular mechanistic interpretability. Mechanism-driven strategy has recently emerged, aiming to build signatures with both predictive power and explanatory power. Driven by this strategy, we developed a robust gene dysregulation analysis framework with machine learning algorithms, which is capable of exploring gene dysregulations underlying carcinogenesis from high-dimensional data with cooperativity and synergy between regulators and several other transcriptional regulation rules taken into consideration. We then applied the framework to a colorectal cancer (CRC) cohort from The Cancer Genome Atlas. The identified CRC-related dysregulations significantly covered known carcinogenic processes and exhibited good prognostic effect. By choosing dysregulations with greedy strategy, we built a four-dysregulation (4-DysReg) signature, which has the capability of predicting prognosis and adjuvant chemotherapy benefit. 4-DysReg has the potential to explain carcinogenesis in terms of dysfunctional transcriptional regulation. These results demonstrate that our gene dysregulation analysis framework could be used to develop predictive signature with mechanistic interpretability for cancer precision medicine, and furthermore, elucidate the mechanisms of carcinogenesis.

Nomogram Integrating Genomics with Clinicopathologic Features Improves Prognosis Prediction for Colorectal Cancer

The current tumor staging system is insufficient for predicting the outcomes for patients with colorectal cancer because of its phenotypic and genomic heterogeneity. Integrating gene expression signatures with clinicopathologic factors may yield a predictive accuracy exceeding that of the currently available system. Twenty-seven signatures that used gene expression data to predict colorectal cancer prognosis were identified and re-analyzed using bioinformatic methods. Next, clinically annotated colorectal cancer samples (n = 1710) with the corresponding expression profiles, that predicted a patient's probability of cancer recurrence, were pooled to evaluate their prognostic values and establish a clinicopathologic-genomic nomogram. Only 2 of the 27 signatures evaluated showed a significant association with prognosis and provided a reasonable prediction accuracy in the pooled cohort (HR, 2.46; 95% CI, 1.183-5.132, P < 0.001; AUC, 60.83; HR, 2.33; 95% CI, 1.218-4.453, P < 0.001; AUC, 71.34). By integrating the above signatures with prognostic clinicopathologic features, a clinicopathologic-genomic nomogram was cautiously constructed. The nomogram successfully stratified colorectal cancer patients into three risk groups with remarkably different DFS rates and further stratified stage II and III patients into distinct risk subgroups. Importantly, among patients receiving chemotherapy, the nomogram determined that those in the intermediate- (HR, 0.98; 95% CI, 0.255-0.679, P < 0.001) and high-risk (HR, 0.67; 95% CI, 0.469-0.957, P = 0.028) groups had favorable responses.Implications: These findings offer evidence that genomic data provide independent and complementary prognostic information, and incorporation of this information refines the prognosis of colorectal cancer. Mol Cancer Res; 16(9); 1373-84. ©2018 AACR.

A 16-gene signature predicting prognosis of patients with oral tongue squamous cell carcinoma

Background

Oral tongue squamous cell carcinoma (OTSCC) is the most common subtype of oral cancer. A predictive gene signature is necessary for prognosis of OTSCC.

Methods

Five microarray data sets of OTSCC from the Gene Expression Omnibus (GEO) and one data set from The Cancer Genome Atlas (TCGA) were obtained. Differentially expressed genes (DEGs) of GEO data sets were identified by integrated analysis. The DEGs associated with prognosis were screened in the TCGA data set by univariate survival analysis to obtain a gene signature. A risk score was calculated as the summation of weighted expression levels with coefficients by Cox analysis. The signature was used to distinguish carcinoma, estimated by receiver operator characteristic curves and the area under the curve (AUC). All were validated in the GEO and TCGA data sets.

Results

Integrated analysis of GEO data sets revealed 300 DEGs. A 16-gene signature and a risk score were developed after survival analysis. The risk score was effective to stratify patients into high-risk and low-risk groups in the TCGA data set (P < 0.001). The 16-gene signature was valid to distinguish the carcinoma from normal samples (AUC 0.872, P < 0.001).

Discussion

We identified a useful 16-gene signature for prognosis of OTSCC patients, which could be applied to clinical practice. Further studies were needed to prove the findings.

Loss of Myosin Vb in colorectal cancer is a strong prognostic factor for disease recurrence

Background:

Selecting the most beneficial treatment regimens for colorectal cancer (CRC) patients remains challenging due to a lack of prognostic markers. Members of the Myosin family, proteins recognised to have a major role in trafficking and polarisation of cells, have recently been reported to be closely associated with several types of cancer and might thus serve as potential prognostic markers in the context of CRC.

Methods:

We used a previously established meta-analysis of publicly available gene expression data to analyse the expression of different members of the Myosin V family, namely MYO5A, 5B, and 5C, in CRC. Using laser-microdissected material as well as tissue microarrays from paired human CRC samples, we validated both RNA and protein expression of Myosin Vb (MYO5B) and its known adapter proteins (RAB8A and RAB25) in an independent patient cohort. Finally, we assessed the prognostic value of both MYO5B and its adapter-coupled combinatorial gene expression signatures.

Results:

The meta-analysis as well as an independent patient cohort study revealed a methylation-independent loss of MYO5B expression in CRC that matched disease progression. Although MYO5B mutations were identified in a small number of patients, these cannot be solely responsible for the common downregulation observed in CRC patients. Significantly, CRC patients with low MYO5B expression displayed shorter overall, disease-, and metastasis-free survival, a trend that was further reinforced when RAB8A expression was also taken into account.

Conclusions:

Our data identify MYO5B as a powerful prognostic biomarker in CRC, especially in early stages (stages I and II), which might help stratifying patients with stage II for adjuvant chemotherapy.