Improvement of risk assessment by integrating toxicological and epidemiological approaches: the case of isoflavones

&nbsp; <strong>Improvement of risk assessment by integrating toxicological and epidemiological approaches: the case of isoflavones </strong> <strong>PhD-thesis Mohammed Ariful Islam </strong> This thesis describes the results of a research project that aimed at the improvement of the risk/benefit assessment of soy isoflavones (SIF) by combining toxicological and epidemiological methods. The toxicological studies were carried out at the Department of Toxicology and part of the results were compared with the outcome of human intervention studies, that were carried out in parallel research project at the Division of Human Nutrition. In Chapter 1 it is explained why we considered such an integrated &ldquo;tox-epi&rdquo; approach to be useful for the prediction of possible effects of SIF in humans on the basis of animal data. SIF are constituents of soy based supplements, which became more and more popular in Western societies over the last decades, because of their putative beneficial health effects, that were related to the SIF present in these supplements. In spite of the long and safe history of soy consumption by the East and the South-East Asian population, the benefit and safety of soy have been challenged in recent years and concerns have been raised about possible adverse health effects. These concerns focussed primarily on the weak estrogenic and proliferative effects of SIF. Chapter 1 also provides some background information on the individual SIF, their structural similarity with the steroid hormone estradiol (E2) and their interaction with the estrogen receptors ER&alpha; and ER&beta;.&nbsp; Chapter 2 describes the differences between rats and humans in the conversion of the three major soy isoflavone glucosides, daidzin, genistin and glycitin, and their aglycones in a series of in vitro models. Results of studies in a Caco-2 transwell model confirmed that deconjugation of the isoflavone glucosides is essential for their transport across the intestinal barrier. It was shown that both rat and human intestinal S9 fractions were able to deconjugate the glucosides, and that intestinal enzymes plaid an important role in this deconjugation reaction. It was demonstrated that in the rat lactase phlorizin hydrolase, glucocerebrosidase, and cytosolic broad-specificity &beta;-glucosidase contribute significantly to this deconjugation, and that in humans deconjugation mainly appeared to occur through the activity of broad-specificity &beta;-glucosidase. Species difference in glucuronidation and sulfation were smaller than for the deconjugation reaction, and it was shown that 7-O-glucuronides were the major metabolites for all the three isoflavone aglycones. The in vitro results also indicated that glucuronidation in rats might be more efficient than in humans, again pointing towards species differences in the metabolism of isoflavone glycosides between rats and humans. It was also shown that the reconjugation reaction has a larger catalytic efficiency than the deconjugation of the glucosides, which corroborates that the detection of aglycones in the systemic circulation is unlikely. It has been reported in literature that following administration of SIF to humans or animals, these compounds are mainly (~98%) present in the systemic circulation in their conjugated form (i.e. as glucuronide and sulphate) of which the estrogenic potency is not yet clear. Chapter 3 provides evidence that in an intact cellular model the major SIF glucuronide metabolites in blood, genistein-7-O-glucuronide (GG) and daidzein-7-O-glucuronide (DG), only become estrogenic after deconjugation. The estrogenic potencies of genistein (Ge), daidzein (Da), GG and DG were determined using stably transfected U2OS-ER&alpha;, U2OS-ER&beta; reporter gene cells and proliferation was tested in T47D-ER&beta; and in T47D breast cancer cells. In all these assays the estrogenic potency of the aglycones was significantly higher than that of their corresponding glucuronides. UPLC analysis revealed that in the in vitro cell line assays, 0.2-1.6% of the glucuronides were deconjugated to their corresponding aglycones. It was also found that, under similar experimental conditions, rat breast tissue S9 fraction was about 30 times more potent in deconjugating these glucuronides than human breast tissue S9 fraction. The results presented in Chapter 3 confirm that SIF glucuronides are not estrogenic as such when tested in an intact cellular model system, and that the small fraction of aglycones account for the observed estrogenic effects. They also provide evidence for a significant species difference in the metabolism of SIF. In Chapters 4 and 5 of this thesis, two rat studies are described, that were performed to further elucidate important modes of action underlying biological effects of SIF and to facilitate an interspecies comparison of the effects observed in rats with those observed in human intervention studies. In these studies inbred ovariectomized Fischer344 rats were used, as an animal model for (post)menopausal women. In the first study described in Chapter 4, two dose levels (i.e. 2 and 20 mg/kg bw) were used to characterise plasma bioavailability, urinary and faecal concentrations of SIF and to investigate changes in gene expression in peripheral blood mononuclear cells (PBMC). The low dose was in line with the type of dosing relevant for human supplement use. Animals were dosed at 0 and 48 hr and sacrificed 4 hr after the last dose. A clear dose dependent increase of SIF concentrations in plasma, urine and faeces was observed, together with a strong correlation in changes in gene expression between the two dose groups. In the transcriptomic analysis, all estrogen responsive genes (ERG) and related biological pathways (BPs) that were found to be affected by the SIF treatment were regulated in both dose groups in the same direction, and indicate possible beneficial effects of SIF. However, most of the common genes in PBMC of rats and of (post)menopausal women, exposed to a comparable dose of the same supplement, were regulated in opposite direction. Thus based on these results no correlation was found between the changes in gene expression in rats and humans, leading to the conclusion that rats might not be a suitable model for humans. In Chapter 5 an animal experiment is described, in which rats received a dose of 2 mg SIF/kg body weight per day for a period of eight weeks. This dosing regimen was similar as that of the parallel human intervention study. Changes in gene expression in different target (i.e. breast (BT), uterus (UT) and sternum (ST)) and non-target (i.e. peripheral blood mononuclear cells (PBMC), adipose (AT) and liver (LT)) tissues were compared. Rank-rank scattered plots did not show any correlation in gene expression changes among different tissues. Out of 87 estrogen responsive genes (ERG), only 19 were found to be significantly regulated (p&lt;0.05) in different tissues. The significantly regulated ERG were mostly found in LT, AT and UT. Surprisingly, no ERG were significantly regulated in BT and ST, although these are considered to be important estrogen sensitive target tissues. No correlation was observed with the changes in gene expression in the PBMC of two rat studies. Correlation was also not seen in the changes of gene expression in PBMC and adipose tissue between rat and humans. In Chapter 6 the results of the research project described in this thesis are evaluated. It was the aim of thesestudies to contribute to theimprovement ofthe risk and/or benefit assessment of SIF for humans, by using in vitro and in vivo animal and human models, and gene expression data in various animal and human tissues, as early biomarkers of effects of exposure to SIF. Although important information has been gathered on the metabolism and the estrogenic activity of SIF and their aglycones, we were not able to predict possible effects in human target tissues based on the results of changes in gene expression in target tissues obtained in the 8 weeks rat study. Possibly aged rats might be a more appropriate model than young ovariectomized rats