| Tumorigenesis follows a multi-stage course of development,covering normal epithelium,precancerous lesions,and invasive carcinomas.Such dynamic transition involves a continuum of cell states deviating from normal differentiation to malignancy development.Yet the knowledges of cancer initiating events,especially those at the very early onset,still remains uncovered.During precancer stages,some lesions may progress towards cancer,while others would show clinical regression or even disappear.And the differential cell fates with clinical relevance need further illumination.It is thus fundamental to elucidate the underlying molecular and cellular mechanisms driving carcinogenesis.Most of previous studies have relied on established paradigm comparing tumor samples against distant normal tissues.This tumor versus normal approach has nonetheless missed the key pieces of information regarding the accurate alterations in between the two stages.To systematically reveal the natural history of carcinogenesis,we need to faithfully depict the trajectories linking normal,precancer,and cancer.With integrated analysis of genomic,epigenetic,and transcriptomic alterations,we can identify molecular and cellular events that converge on cancer initiation and progression.Esophageal squamous cell carcinoma(ESCC),the 6th leading cause of cancer deaths worldwide,represents a canonical paradigm for stepwise oncogenesis.It is characterized with well-identifiable precancerous lesions that include low and high-grade intraepithelial neoplasia(LGIN and HGIN),collectively known as squamous dysplasia.Thus,ESCC provides an ideal model system to systematically investigate alterations during carcinogenesis.While recent body maps of somatic mutagenesis in morphologically normal esophageal tissues have identified cancer driver mutations,in most cases these events appear to be insufficient for cancer development,indicating that other processes may be causally involved.Instead,oncogenesis may arise from an aberrant dedifferentiated stem-like state driven by silencing of specific differentiation genes.Here we profiled multi-step ESCC develop process with three cohorts.By establishing the CancerStemID framework,we have bult the esophageal-specific regulatory network,consisting of the tissue-specific transcription factors and their target regulons,to reflect the epithelial differentiation process.Based the estimated transcription factor activities,we characterized how normal epithelial cells transition into preneoplastic cells,and ultimately to invasive cancer.Based on computational dissection by transcription factor inactivation load(TFIL),we can isolate a preneoplastic epithelial cell population with significant higher stemness and cancer progression score based on regulatory activities of esophageal specific transcription factors.In summary,these novel insights and the computational CancerStemID framework presented herein could help deliver the much-needed early detection and cancer risk prediction markers for deadly cancers such as ESCC,or alternatively,to help assess the efficacy of early cancer prevention trials. |