A new web-based tool for high-throughput DNA methylation pathway analysis of multiple cancer types
Global cytosine hypomethylation and promoter hypermethylation are characteristic in cancer, but how DNA methylation targets biological pathways and gene sets is unclear. It is hypothesized that during tumorigenesis certain pathways and gene sets are commonly dysregulated via DNA methylation across cancer types. A logistic regression-based gene set enrichment method called LRpath was established to identify important pathways in high-throughput data. LRpath performs well with both small and large sample sizes, does not require subjective selection of a significance cutoff (which many other programs do), and easily compares and visualizes results across multiple studies. The authors discovered that several well-known cancer-related pathways were significantly differentially methylated, while others surprisingly showed fewer differentially methylated genes than expected by chance.
The web-based LRpath tool allows for identification and comparison of pathway signatures across multiple studies and allows both undirectional and directional testing options. The undirectional test distinguishes between “enriched” gene sets (pathways with more genes changed in methylation than expected by chance) and “depleted” gene sets (those with fewer changed). The directional test distinguishes gene sets enriched with hyper- or hypomethylated genes. The authors used LRpath to uncover the pathways that were commonly altered with respect to DNA methylation levels using publicly available data from ten different types of tumors relative to normal tissues. CpG methylation levels were determined using the Illumina HumanMethylation27 BeadChip. A wide spectrum of cancers were represented, including two types of lung cancer (adenocarcinoma and squamous cell carcinoma) and cancers of breast, colon, brain, myeloma, kidney, ovarian, prostate, and stomach.
A surprising level of concordance in differential methylation across multiple cancer types was observed by filtering to gene sets exhibiting significant change (p-value < 0.0001) in at least five (50%) of the ten cancer types. Among commonly hypomethylated groups were immune-related functions (except for prostate cancer, most of whose immune response genes were hypermethylated), genes with peptidase activity, and genes involved in epidermis/keratinocyte development and differentiation, suggesting DNA methylation-driven cancer cell invasion and tumorigenesis across various types of cancer. Commonly hypermethylated groups included homeobox and other DNA-binding genes, genes involved in nervous system and embryonic development, and voltage-gated potassium channels. In particular, the promoter of KCNA3, a gene involved in voltage-gated potassium channel activity, was hypermethylated in 8 out of the 10 cancers, identifying it as one of the most prevalent events in tumorigenesis. A higher proportion of Polycomb Repressive Complex 2 (PRC2) target genes than genes not targeted by PRC2 in developmental pathways were differentially methylated in multiple tumor types. Interestingly, none of the PRC2 target genes involved in dermal development were differentially methylated in multiple myeloma, highlighting a difference between blood and solid cancers. For many gene sets, the authors found significant overlap in the specific subset of differentially methylated genes across multiple cancer types. Interestingly, fewer DNA repair genes were differentially methylated than expected by chance.
The authors concluded that DNA methylation changes in cancer tend to target a subset of the known cancer pathways affected by genetic aberrations. Further studies are needed to elucidate consistent methylation differences between solid and non-solid tumors; to identify driver genes of each cancer type; and to expand the study to more methylation sites beyond CpG islands and promoters.
Kim JH, Karnovsky A, Mahavisno V, Weymouth T, Pande M, Dolinoy DC, Rozek LS, & Sartor MA (2012). LRpath analysis reveals common pathways dysregulated via DNA methylation across cancer types. BMC genomics, 13 PMID: 23033966