Background
The Warburg effect, also termed "aerobic glycolysis", is one of the most remarkable and ubiquitous metabolic characteristics exhibited by cancer cells, representing a potential vulnerability that might be targeted for tumor therapy. Ketogenic diets (KDs), composed of high-fat, moderate-protein and low carbohydrates, are aimed at targeting the Warburg effect for cancer treatment, which have recently gained considerable attention. However, the efficiency of KDs was inconsistent, and the genotypic contribution is still largely unknown.
Conclusions
Our findings identified a metabolic subtype based on glycolysis and KBM that may serve as a promising biomarker to predict the clinical outcomes and therapeutic responses to KDT.
Methods
The bulk RNA-seq data from The Cancer Genome Atlas (TCGA), single cell RNA sequencing (scRNA-seq), and microarray data from Gene Expression Omnibus (GEO) and Cancer Cell Line Encyclopedia (CCLE) were collected. A joint analysis of glycolysis and ketone bodies metabolism (KBM) pathway was performed across over 10,000 tumor samples and nearly 1,000 cancer cell lines. A series of bioinformatic approaches were combined to identify a metabolic subtype that may predict the response to ketogenic dietary therapy (KDT). Mouse xenografts were established to validate the predictive utility of our subtypes in response to KDT.
Results
We first provided a system-level view of the expression pattern and prognosis of the signature genes from glycolysis and KBM pathway across 33 cancer types. Analysis by joint stratification of glycolysis and KBM revealed four metabolic subtypes, which correlated extensively but diversely with clinical outcomes across cancers. The glycolytic subtypes may be driven by TP53 mutations, whereas the KB-metabolic subtypes may be mediated by CTNNB1 (β-catenin) mutations. The glycolytic subtypes may have a better response to KDs compared to the other three subtypes. We preliminarily confirmed the idea by literature review and further performed a proof-of-concept experiment to validate the predictive value of the metabolic subtype in liver cancer xenografts. Conclusions: Our findings identified a metabolic subtype based on glycolysis and KBM that may serve as a promising biomarker to predict the clinical outcomes and therapeutic responses to KDT.