Use of pathway information in molecular epidemiology

Abstract

Candidate gene studies are generally motivated by some form of pathway reasoning in the selection of genes to be studied, but seldom has the logic of the approach been carried through to the analysis. Marginal effects of polymorphisms in the selected genes, and occasionally pairwise gene–gene or gene–environment interactions,are often presented, but a unified approach to modelling the entire pathway has been lacking. In this review, a variety of approaches to this problem is considered, focusing on hypothesis-driven rather than purely exploratory methods. Empirical modelling strategies are based on hierarchical models that allow prior knowledge about the structure of the pathway and the various reactions to be included as ‘prior covariates’. By contrast, mechanistic models aim to describe the reactions through a system of differential equations with rate parameters that can vary between individuals, based on their genotypes. Some ways of combining the two approaches are suggested and Bayesian model averaging methods for dealing with uncertainty about the true model form in either framework is discussed. Biomarker measurements can be incorporated into such analyses, and two-phase sampling designs stratified on some combination of disease, genes and exposures can be an efficient way of obtaining data that would be too expensive or difficult to obtain on a full candidate gene sample. The review concludes with some thoughts about potential uses of pathways in genome-wide association studies.

Figure 1

Biochemical diagram of folate metabolism (reproduced with permission from Reed et al. [14]). AICART, aminoimidazolecarboxamide ribonucleotide transferase; BHMT, betaine-homocysteine methyltransferase; CBS, cystathionine b-synthase; DHFR, dihydrofolate reductase; DNMT, DNA-methyltransferase; dTMP, thymidine monophosphate; FTD, 10-formyltetrahydrofolate dehydrogenase; FTS, 10-formyltetrahydrofolate synthase; GAR, glycinamide ribotid; G-NMT, glycine N-methyltransferase; HCOOH; formic acid; H₂C = O, formaldehyde; HCY, homocysteine; MAT, methionine adenosyl transferase; MS, methionine synthase; MTCH, 5,10-methylenetetrahydrofolate cyclohydrolase; MTD, 5,10-methylenetetrahydrofolate dehydrogenase; MTHFR, 5,10-methylenetetrahydrofolate reductase; NE, nav-enzymatic; PGT, phosphoribosyl glycinamide transferase; SAH, S-adenosylhomocysteine; SAHH, SAH hydrolase; SAM, S-adenosylmethionine; SHMT, serine hydroxymethyltransferase; THF, tetrahydrofolate; 5m-THF, 5-methylTHF; 5,10-CH₂-THF, 5,10-methyleneTHF; 10f-THF, 10-formylTHF; TS, thymidylate synthase.

Figure 2

Schematic representation of simplified one-compartment model.

Figure 3

Hierarchical clustering of folate genes based on 184 GO terms.

Figure 4

Top-ranking topologies without incorporating priors: left, gene only; right, genes and exposures. With no priors, the two topologies have posterior probabilities 3.9 per cent and 2.3 per cent, respectively. Using a topology derived by hierarchical clustering of the A matrix from simulated data, the top-ranked gene-only topology was identical to that shown on the left, with posterior probability of 9.5 per cent. Using the GO topology shown in Figure 3, the same genes were included, but reordered as (((MTHR, SAHH), MTD), SHMT)with a posterior probability of 6.4 per cent.