全部
logo

Unsupervised machine learning-based clustering identifies unique molecular signatures of colorectal cancer with distinct clinical outcomes

RAPID COMMUNICATION

Unsupervised machine learning-based clustering identifies unique molecular signatures of colorectal cancer with distinct clinical outcomes

Manish Pratap Singh
Sandhya Rai
Sarvesh K. Gupta
Nand K. Singh
Sameer Srivastava
Genes & Diseases第10卷, 第6期pp.2270-2273纸质出版 2023-11-01在线发表 2023-03-24
124203

Colorectal cancer (CRC) is known to harbor considerable heterogeneity. Consequently, it could be hypothesized that similar-appearing tumors might exhibit substantial genetic differences while diverse-appearing tumors may have a similar genetic landscape. Due to these differences at the molecular level, they behave or respond differently to therapies as well. CRC progression is a multistep process and involves the accumulation of substantial genetic and epigenetic events in a stage-dependent manner. Alterations in Wnt, DNA repair, RAS-RAF-MAPK, and PIK3CA-AKT pathways have been well-established to play a role in the etiology of CRC. In the era of personalized medicine, it becomes essential to identify the molecular subtype of CRC so that the predictive and prognostic potential of CRC could be established. Molecular subtyping has its importance and limitations in the clinical management of CRC and is very complex to comprehend. Thus, there is still a need to create robust and reliable clustering methods that could precisely identify unique molecular signatures and help in the prediction of the clinical response of the patients who share certain clinical as well as molecular characteristics. Frequent mutations have been reported in RAS, BRAF, NRAS, and PIK3CA genes that are major regulators of the above pathways and significantly promote tumor initiation and progression in CRC. Similarly, the clinical significance of epigenetically deregulated MLH1, RASSF1, DAPK1, IFG2, SLITRK5, and IGFBP3 genes is also well-established and documented in CRC. In the present investigation, we have comprehensively analyzed the mutation status of KRAS, BRAF, NRAS, and PIK3CA, and methylation status of MLH1, RASSF1, SLITRK5, DAPK1, IGFBP3, and IGF2 genes in 70 CRC tumor samples and 40 matched normal, which have a significant role in CRC initiation and progression. Further, the patients were stratified according to their shared molecular heterogeneity and the prognostic value of each heterogeneous cluster was evaluated.

pic