AUTHOR=Zhang Wanfeng , Huang Fang , Tang Xia , Ran Longke 

TITLE=The clonal expression genes associated with poor prognosis of liver cancer

JOURNAL=Frontiers in Genetics

VOLUME=13

YEAR=2022

URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.808273

DOI=10.3389/fgene.2022.808273

ISSN=1664-8021

ABSTRACT=<p>The extensive spatial genomic intratumor heterogeneity (ITH) in liver cancer hindered treatment development and limited biomarker design. Early events that drive tumor malignant transformation in tumor founder cells are clonally present in all tumor cell populations, which provide stable biomarkers for the localization of tumor cells and patients’ prognosis. In the present study, we identified the recurrently clonal somatic mutations and copy number alterations (CNAs) (893 clonal somatic mutations and 6,617 clonal CNAs) in 353 liver cancer patients from The Cancer Genome Atlas (TCGA) and evaluated their prognosis potential. We showed that prognosis-related clonal alterations might play essential roles in tumor evolution. We identified 32 prognosis related clonal alterations differentially expressed between paired normal and tumor samples, that their expression was cross-validated by three independent cohorts (50 paired samples in TCGA, 149 paired samples in GSE76297, and 9 paired samples in SUB6779164). These clonal expression alterations were also significantly correlated with clinical phenotypes. Using stepwise regression, we identified five (<italic>UCK2</italic>, <italic>EFNA4</italic>, <italic>KPAN2</italic>, <italic>UBE2T</italic>, and <italic>KIF14</italic>) and six (<italic>MCM10</italic>, <italic>UCK2</italic>, <italic>IQGAP3</italic>, <italic>EFNA4</italic>, <italic>UBE2T</italic>, and <italic>KPNA2</italic>) clonal expression alterations for recurrence and survival model construction, respectively. Furthermore, in 10 random repetitions, we showed strong applicability of the multivariate Cox regression models constructed based on the clonal expression genes, which significantly predicted the outcomes of the patients in all the training and validation sets. Taken together, our work may provide a new avenue to overcome spatial ITH and refine biomarker design across cancer types.</p>