A useful approach for studies on the mechanisms of genetic variation in cancer susceptibility is to use intermediary biochemical endpoints with mechanistic relevance to the genes under study. We examined the effects of individual genotype at seven metabolic gene loci on a marker of oxidative DNA damage, 8-oxo-7,8-dihydro-2-deoxyguanosine, in people exposed to polycyclic aromatic hydrocarbons (PAH) from three Central European cities. The GSTT1 homozygous deletion variant was associated with a significant protective effect for exposure to total PAHs and to eight specific PAHs, although the magnitude and significance of the effect varied among these compounds. Categorical sensitivity analysis was used to determine that the frequency of the GSTT1 deletion was significantly higher in people who proved to be more resistant to the DNA damaging effects of PAH exposure than in people who were the most sensitive. There is a growing literature on the protective effect of GSTT1 deletion in both disease and intermediary endpoints related to environmental carcinogenesis. The mechanism for this effect might be related to specific PAH substrate specificities, or could be related to other functions of GSTT1 gene in oxidative stress induced damage pathways. ' 2007 Wiley-Liss, Inc.Key words: biological markers; susceptibility; cohort studyThe paradigm of gene-environment interactions in carcinogenesis holds that genetic variation in loci responsible for the metabolism of carcinogens should be reflected in the well known variability in risk of cancer among exposed populations. This principle has been confirmed in some instances, but not in others, 1-4 mostly using case control approaches. One of the most studied metabolic gene families in this context is the glutathione-S-transferases (GST), which code for a family of phase II conjugating enzymes, generally considered to be involved in detoxification. 5,6 Since many classical electrophilic carcinogens are substrates for the GST enzymes, it has generally been hypothesized that the common human polymorphic variants of these genes, which for GSTM1 and GSTT1 are homozygous deletions, should result in a decrease in conjugation and subsequent elimination of carcinogenic intermediates, with the result of increasing the dose of carcinogen to target tissue. Individual case control studies, as well as pooled and meta analyses have confirmed this hypothesis to some extent, so that for example the deletion of GSTM1 variant was found to be associated with various cancer types, both alone and in combination with other genetic polymorphisms. 7-10 Careful analysis of the literature reveals a much more murky picture for the GSTT1 gene than for GSTM1. Although one group reported GSTT1 to be a risk factor for lung cancer, 11 the majority of studies have found no association between GSTT1 deletion and cancer as shown by a meta and pooled analysis, 12 and in general there are more negative findings for this variant than for the GSTM1 deletion. Even more interesting is the growing body of evidence that ...