Development of an oncogenic dedifferentiation SOX signature with prognostic significance in hepatocellular carcinoma

Abstract

Background: Gradual loss of terminal differentiation markers and gain of stem cell-like properties is a major hall mark of cancer malignant progression. The stem cell pluripotent transcriptional factor SOX family play critical roles in governing tumor plasticity and lineage specification. This study aims to establish a novel SOX signature to monitor the extent of tumor dedifferentiation and predict prognostic significance in hepatocellular carcinoma (HCC).

Methods: The RNA-seq data from The Cancer Genome Atlas (TCGA) LIHC project were chronologically divided into the training (n = 188) and testing cohort (n = 189). LIRI-JP project from International Cancer Genome Consortium (ICGC) data portal was used as an independent validation cohort (n = 232). Kaplan-Meier and multivariable Cox analyses were used to examine the clinical significance and prognostic value of the signature genes.

Results: The SOX gene family members were found to be aberrantly expressed in clinical HCC patients. A five-gene SOX signature with prognostic value was established in the training cohort. The SOX signature genes were found to be closely associated with tumor grade and tumor stage. Liver cancer dedifferentiation markers (AFP, CD133, EPCAM, and KRT19) were found to be progressively increased while hepatocyte terminal differentiation markers (ALB, G6PC, CYP3A4, and HNF4A) were progressively decreased from HCC patients with low SOX signature scores to patients with high SOX signature scores. Kaplan-Meier survival analysis further indicated that the newly established SOX signature could robustly predict patient overall survival in both training, testing, and independent validation cohort.

Conclusions: An oncogenic dedifferentiation SOX signature presents a great potential in predicting prognostic significance in HCC, and might provide novel biomarkers for precision oncology further in the clinic.

Keywords: Oncogenic dedifferentiation; Prognostic value; Stem cell-like properties.

Fig. 1

Expression of SOX signature genes in HCC patients. a The normalized expression of SOX signature genes (SOX3, SOX4, SOX11, SOX12, and SOX14) were compared between 50 normal liver tissues and 186 HCC tissues from the TCGA-LIHC Cohort I. b The normalized expressions of SOX signature genes were compared between HCC patient subgroups with different tumor grade. c The normalized expressions of SOX signature genes were compared between HCC patient subgroups with different tumor stage. Independent student’s t test, *, P < 0.05, **, P < 0.01, ***, P < 0.001, ****, P < 0.0001, ns, not significant. The figures were generated using GraphPad Prism 5

Fig. 2

The SOX signature represents an oncogenic dedifferentiation phenotype. a The normalized expressions of liver cancer dedifferentiation markers and liver progenitor cell markers in HCC patients with different SOX signature score. b The normalized expressions of hepatocyte terminal differentiation markers in HCC patients with different SOX signature score. One-way ANOVA test. P value less than 0.05 was considered statistically significant. The figures were generated using GraphPad Prism 5

Fig. 3

Prediction of the SOX signature-regulated transcriptional network. a The Venn diagram show overlapping downstream targets of SOX signature genes. b Prediction of SOX signature gene binding motif. c Network of SOX signature gene downstream targets and their associated genes. d Gene ontology and signaling pathway analysis of SOX signature gene downstream targets

Fig. 4

The prognostic significance of SOX signature genes in multiple HCC clinical cohorts. a The patients in the training set (TCGA-LIHC Cohort I, n = 188) were divided into “High sox group” and “Low sox group” according to the SOX signature score. Kaplan–Meier survival curves of the two risk groups were plotted and the log-rank P value of the survival difference calculated between them (Upper panel). Kaplan–Meier survival curves of HCC patients from subgroups with different SOX signature score (Lower panel). b Similar analysis was down in the testing set (TCGA-LIHC Cohort II, n = 189). c and validated in an independent validation set (LIRI-JP Cohort, n = 232). P value less than 0.05 was considered statistically significant. The figures were generated using SPSS v19