Modularity of the metabolic gene network as a prognostic biomarker for hepatocellular carcinoma

TL;DR

This study demonstrates that the modularity of metabolic gene expression patterns in hepatocellular carcinoma correlates with tumor aggressiveness and prognosis, suggesting modularity as a potential biomarker for patient outcomes.

Contribution

It introduces a novel modularity-based metric for metabolic gene expression that predicts prognosis and recurrence in HCC patients, advancing biomarker development.

Findings

Higher modularity correlates with worse prognosis.

Modularity predicts recurrence and survival.

Modularity is linked to glycolytic phenotype and tumor stage.

Abstract

Abnormal metabolism is an emerging hallmark of cancer. Cancer cells utilize both aerobic glycolysis and oxidative phosphorylation (OXPHOS) for energy production and biomass synthesis. Understanding the metabolic reprogramming in cancer can help design therapies to target metabolism and thereby to improve prognosis. We have previously argued that more malignant tumors are usually characterized by a more modular expression pattern of cancer-associated genes. In this work, we analyzed the expression patterns of metabolism genes in terms of modularity for 371 hepatocellular carcinoma (HCC) samples from the Cancer Genome Atlas (TCGA). We found that higher modularity significantly correlated with glycolytic phenotype, later tumor stages, higher metastatic potential, and cancer recurrence, all of which contributed to poorer overall prognosis. Among patients that recurred, we found the correlation of greater modularity with worse prognosis during early to mid-progression. Furthermore, we developed metrics to calculate individual modularity, which was shown to be predictive of cancer recurrence and patients' survival and therefore may serve as a prognostic biomarker. Our overall conclusion is that more aggressive HCC tumors, as judged by decreased host survival probability, had more modular expression patterns of metabolic genes. These results may be used to identify cancer driver genes and for drug design.