Large meta-analysis of multiple cancers reveals a common, compact and highly prognostic hypoxia metagene

Abstract

Background: There is a need to develop robust and clinically applicable gene expression signatures. Hypoxia is a key factor promoting solid tumour progression and resistance to therapy; a hypoxia signature has the potential to be not only prognostic but also to predict benefit from particular interventions.

Methods: An approach for deriving signatures that combine knowledge of gene function and analysis of in vivo co-expression patterns was used to define a common hypoxia signature from three head and neck and five breast cancer studies. Previously validated hypoxia-regulated genes (seeds) were used to generate hypoxia co-expression cancer networks.

Results: A common hypoxia signature, or metagene, was derived by selecting genes that were consistently co-expressed with the hypoxia seeds in multiple cancers. This was highly enriched for hypoxia-regulated pathways, and prognostic in multivariate analyses. Genes with the highest connectivity were also the most prognostic, and a reduced metagene consisting of a small number of top-ranked genes, including VEGFA, SLC2A1 and PGAM1, outperformed both a larger signature and reported signatures in independent data sets of head and neck, breast and lung cancers.

Conclusion: Combined knowledge of multiple genes' function from in vitro experiments together with meta-analysis of multiple cancers can deliver compact and robust signatures suitable for clinical application.

Figure 1

Hypoxia gene-expression network in HNSCC (Vice 125 data set). Seeds (yellow) and learnt genes (blue) are shown; circle size is proportional to C score. Genes with top 20% C scores are shown. Solid edges connect cluster members with seeds; length is proportional to membership, colour represents Spearman correlation (blue, −1; red, +1). Green dotted edges connect seeds; their length is proportional to the shared neighbourhood, S. This figure appears in colour in the HTML version.

Figure 2

Hypoxia network mapped onto Reactome pathways (A) coloured by increasing C score from dark blue to bright red; and validation of up-regulated HNSCC (B) and BC (C) signatures by comparison with the literature. The proportion of literature-validated genes is shown as function of the number of top-ranked (by C score) genes considered; standard errors estimated by bootstrap. This figure appears in colour in the HTML version.

Figure 3

Common hypoxia signature of 51 genes. (A) Hypoxia/normoxia expression ratio in endothelial, smooth muscle, human mammalian epithelial, renal proximal tubule epithelial cells (EC, SMC, HMEC, RPTEC); and in (B) HIF1a/HIF2a siRNA experiment. (C, D) Connectivity-ranked forest plots: metastases- and recurrence-free survival (MFS, RFS) hazard ratio (HR) (red) with 95% confidence intervals, and HRs if permuted list (black). Control: random sampling of N=51 genes ( × 100 resampling).