Table 4. Joint effect and interaction between polymorphisms and H. pylori infection on risk of intestinal metaplasia and dysplasia*.
Variant | Genotype | H. pylori infection | OR (95% CI)† | P for trend‡ |
rs2910164 | GG | Negative | 1.00 (referent) | |
GC | Negative | 1.01 (0.68–1.50) | ||
CC | Negative | 1.08 (0.48–2.41) | ||
GG | Positive | 2.44 (1.80–3.31) | ||
GC | Positive | 2.55 (1.86–3.49) | ||
CC | Positive | 2.25 (1.43–3.54) | <0.001 | |
P for interaction§ | 0.35 | |||
rs895819 | TT | Negative | 1.00 (referent) | |
TC | Negative | 0.87 (0.51–1.48) | ||
CC | Negative | 0.93 (0.52–1.66) | ||
TT | Positive | 1.83 (1.09–3.10) | ||
TC | Positive | 2.09 (1.28–3.39) | ||
CC | Positive | 2.77 (1.68–4.56) | <0.001 | |
P for interaction§ | 0.92 |
This analysis was restricted to 2078 participants who had diagnosis for H. pylori infection.
Odds ratios (ORs) and 95% confidence intervals (CI) for the risk of intestinal metaplasia and dysplasia were estimated in reference to superficial gastritis/chronic atrophic gastritis, and adjusted for age, gender, smoking and drinking status. Participants with missing data on any of variables were excluded from the analysis.
P for trend was calculated by assigning an ordinal value (1–6) to each combination of genotype and H. pylori infection (1 = homozygote wild without H. pylori infection, 2 = heterozygote without H. pylori infection, 3 = homozygote variant without H. pylori infection, 4 = homozygote wild with H. pylori infection, 5 = heterozygote with H. pylori infection, 6 = homozygote variant with H. pylori infection) and adding it as a continuous variable to the multivariate model.
Likelihood ratio test was used to calculate the P value for interaction by comparing the two models with and without the product term of genotypes and H. pylori infection.