A GWAS on Helicobacter pylori strains points to genetic variants associated with gastric cancer risk

Abstract

Background: Helicobacter pylori are stomach-dwelling bacteria that are present in about 50% of the global population. Infection is asymptomatic in most cases, but it has been associated with gastritis, gastric ulcers and gastric cancer. Epidemiological evidence shows that progression to cancer depends upon the host and pathogen factors, but questions remain about why cancer phenotypes develop in a minority of infected people. Here, we use comparative genomics approaches to understand how genetic variation amongst bacterial strains influences disease progression.

Results: We performed a genome-wide association study (GWAS) on 173 H. pylori isolates from the European population (hpEurope) with known disease aetiology, including 49 from individuals with gastric cancer. We identified SNPs and genes that differed in frequency between isolates from patients with gastric cancer and those with gastritis. The gastric cancer phenotype was associated with the presence of babA and genes in the cag pathogenicity island, one of the major virulence determinants of H. pylori, as well as non-synonymous variations in several less well-studied genes. We devised a simple risk score based on the risk level of associated elements present, which has the potential to identify strains that are likely to cause cancer but will require refinement and validation.

Conclusion: There are a number of challenges to applying GWAS to bacterial infections, including the difficulty of obtaining matched controls, multiple strain colonization and the possibility that causative strains may not be present when disease is detected. Our results demonstrate that bacterial factors have a sufficiently strong influence on disease progression that even a small-scale GWAS can identify them. Therefore, H. pylori GWAS can elucidate mechanistic pathways to disease and guide clinical treatment options, including for asymptomatic carriers.

Keywords: GWAS; Gastric cancer; Helicobacter pylori.

Fig. 1

Neighbour-joining tree based on whole genome sequence alignment of all 173 strains from hpEurope-derived populations. Branches are shaded according to the population determined by fineSTRUCTURE analysis [17]. Labels reveal the patient disease background grouped into three categories: non-atrophic gastritis, progressive towards gastric cancer and gastric cancer. The scale bar represents a genetic distance of 0.02

Fig. 2

Location of genetic elements associated with gastric cancer on ELS37 genome (GCA_000255955.1). GWAS comparing isolates from patients with (a) non-atrophic gastritis to those with gastric cancer and precancerous progression and (b) gastric cancer to those with non-atrophic gastritis and precancerous progression. Two GWAS were performed with bugwas software for each panel, one based on SNPs (upper panels) and the other based on k-mers (lower panels). Positions of the genomic elements are represented on the horizontal axis expressed. Log 10 of p value for each hit is recorded on the vertical axis. The blue line indicates a p value ≤ 10⁻⁵

Fig. 3

Prevalence of genes highlighted by GWAS in H. pylori genomes. a Prevalence of genes containing a GWAS hit with p value < 10⁻⁵ in three groups of isolates: non-atrophic gastritis isolates (red, n = 55 genomes), progressive toward cancer isolates (grey, n = 49) and gastric cancer isolates (black, n = 39), and defined as the ratio of number of isolates in each group harbouring the gene and the total number of isolates in each group. b Matrix of correlation of pairs of gene prevalence patterns in 143 H. pylori genomes. Red indicates that two genes have a high positive correlation of their patterns of presence and absence in all genomes examined and blue indicates a negative correlation. White indicates core genes that did not vary in prevalence in the dataset and for which correlations could not be calculated

Fig. 4

Repartition of risk scores on 173 strains from hpEurope-derived sub-populations, according to patient disease background. Each point corresponds to the risk score associated with a single strain. This risk score was calculated based on the presence of risk or safe genotype for each of the 9 genes considered (Table 2)