Identification of Prognostic Gene Signatures by Developing a scRNA-Seq-Based Integration Approach to Predict Recurrence and Chemotherapy Benefit in Stage II-III Colorectal Cancer.

Prospective identification of robust biomarkers related to prognosis and adjuvant chemotherapy has become a necessary and critical step to predict the benefits of adjuvant therapy for patients with stage II-III colorectal cancer (CRC) before clinical treatment. We proposed a single-cell-based prognostic biomarker recognition approach to identify and construct CRC up- and down-regulated prognostic signatures (CUPsig and CDPsig) by integrating scRNA-seq and bulk datasets. We found that most genes in CUPsig and CDPsig were known disease genes, and they had good prognostic abilities in CRC validation datasets. Multivariate analysis confirmed that they were two independent prognostic factors of disease-free survival (DFS). Significantly, CUPsig and CDPsig could effectively predict adjuvant chemotherapy benefits in drug-treated validation datasets. Additionally, they also performed well in patients with CMS4 subtype. Subsequent analysis of drug sensitivity showed that expressions of these two signatures were significantly associated with the sensitivities of CRC cell lines to multiple drugs. In summary, we proposed a novel prognostic biomarker identification approach, which could be used to identify novel prognostic markers for stage II-III CRC patients who will undergo adjuvant chemotherapy and facilitate their further personalized treatments.

Systemically Identifying Triple-Negative Breast Cancer Subtype-Specific Prognosis Signatures, Based on Single-Cell RNA-Seq Data.

Triple-negative breast cancer (TNBC) is a highly heterogeneous disease with different molecular subtypes. Although progress has been made, the identification of TNBC subtype-associated biomarkers is still hindered by traditional RNA-seq or array technologies, since bulk data detected by them usually have some non-disease tissue samples, or they are confined to measure the averaged properties of whole tissues. To overcome these constraints and discover TNBC subtype-specific prognosis signatures (TSPSigs), we proposed a single-cell RNA-seq-based bioinformatics approach for identifying TSPSigs. Notably, the TSPSigs we developed mostly were found to be disease-related and involved in cancer development through investigating their enrichment analysis results. In addition, the prognostic power of TSPSigs was successfully confirmed in four independent validation datasets. The multivariate analysis results showed that TSPSigs in two TNBC subtypes-BL1 and LAR, were two independent prognostic factors. Further, analysis results of the TNBC cell lines revealed that the TSPSigs expressions and drug sensitivities had significant associations. Based on the preceding data, we concluded that TSPSigs could be exploited as novel candidate prognostic markers for TNBC patients and applied to individualized treatment in the future.