ToxSci Advance Access originally published online on July 17, 2007
Toxicological Sciences 2007 99(2):446-454; doi:10.1093/toxsci/kfm183
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Comparison of Mode of Action of Four Hepatocarcinogens: A Model-Based Approach
* Central Unit Biostatistics
Department of Cellular and Molecular Pathology, German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, D-69009 Heidelberg, Germany
Department of Toxicology, Institute of Pharmacology and Toxicology, University of Tuebingen, D-72074 Tuebingen, Germany
1 To whom correspondence should be addressed at German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, D-69009 Heidelberg, Germany. Fax: +49-6221-42-2397. E-mail: kopp{at}dkfz.de.
Received March 20, 2007; accepted July 11, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Within the scope of the Rat Liver Foci Bioassay the model carcinogens N-nitrosomorpholine (NNM), 2-acetylaminoflouren (2-AAF), phenobarbital (PB), and clofibrate (CF) were analyzed concerning their potency and dose–response relationship to induce foci of altered hepatocytes (FAHs), which are known to be precursor lesions of liver adenoma and carcinoma. The medium-term experiment follows an initiation–promotion protocol using diethylnitrosamine (DEN) as initiator. The present report deals with the application of two biologically based models for hepatocarcinogenesis, the two-stage clonal expansion model (TSCEM), and a color-shift model with beta distributed growth rates (CSMbeta). Both models yield similar conclusions concerning the mode of action of the carcinogens. However, the fit of CSMbeta appears closer to the observations than the fit of TSCEM. The analysis shows that application of a single dose of DEN has a persistent effect on the rate of FAH induction, especially in female rats. Overall, striking differences in the effect of the carcinogens were observed between male and female animals. 2-AAF shows a strong promoting effect in males, whereas in females the initiating effect dominates. NNM has both initiating and promoting effect, but in females, the rate of FAH formation seems to reach saturation at high dose. In the doses applied in the present experiment, PB has the weakest carcinogenic effect. Although PB alone does not induce FAH during the observation period, it increases the rate of FAH formation when applied following initiation with DEN. CF reduces the number and area fraction of GSTP-stained FAH, probably because it suppresses the placental form of glutathione S-transferase–positive phenotype.
Key Words: foci of altered hepatocytes; carcinogenesis model; initiation; promotion.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Quantitative carcinogenic risk assessment for chemicals requires knowledge about the mechanism of carcinogenesis. In general, the formation of malignant tumors from normal tissue is assumed to proceed over intermediate stages (Bannasch, 1968
; Foulds, 1969). Specifically, the occurrence of foci of altered hepatocytes (FAHs) has been shown to be a valuable indicator of a hepatocarcinogenic effect. FAH are characterized by specific morphological and biochemical changes. They can be visualized by staining liver sections. In appropriate animal models, FAH can be generated by treatment with chemicals, hormones, hepadnaviruses, transgenic oncogenes, Helicobacter hepaticus, and ionizing radiation (Bannasch et al., 2001). Several types of FAH can be identified which are assumed to alter their phenotype successively until malignancy (Bannasch, 1968).
The influence of a chemical substance on the formation and size development of FAH can be classified into two modes of action: on one hand, the substance may induce FAH in normal liver tissue by genetic or epigenetic events (Feo et al., 2000; Herath et al., 2006; Pogribny et al., 2006) and on the other hand it may increase the size of existing FAH. The denotation initiation and promotion for the two modes of carcinogenic action was first suggested by Rous and Kidd concerning dermal tissue and is extensively used to date (Friedewald and Rous, 1944; Rous and Kidd, 1941). Peraino et al. (1971) applied this concept to hepatocarcinogenesis experiments, whereupon several initiation–promotion protocols where proposed, in which the test substance is given following one-time treatment with an initiating carcinogen and is administered alone to study its potential to induce FAH (Pereira, 1982; Pitot et al., 1978). It still remains a matter of debate whether chemical carcinogens can be classified according to qualitative differences in their mode of action into initiators and promotors (Bannasch, 1986; Pitot et al., 1989; Schulte-Hermann et al., 1989; Williams, 1989) or whether the effects of these compounds differ quantitatively, promoters acting in accord with the initiators as weak carcinogens.
The evaluation of FAH is associated with stereological problems because measurements are made in two-dimensional liver sections but inference is wanted about the three-dimensional liver. According to Delesse (1847) the FAH area fraction represents an unbiased estimate of the volume fraction of FAH as long as foci are distributed homogeneously in the liver. However, evaluation of area fraction alone can only show the carcinogenic potential of the chemical. It cannot be decided whether the chemical alters the number and/or the size distribution of the foci, because the probability of a focus to be cut increases with its size. As demonstrated by Kopp-Schneider (2003), the analysis of the number of focal transections and their size distribution necessitate the use of a model which incorporates hypotheses about the formation and growth of FAH.
For the observation and evaluation of FAH a Rat Liver Foci Bioassay (RLFB) has been suggested (Ittrich et al., 2003). The RLFB uses 80 animals per experimental group and a medium test period of 12 weeks. During this period rats are hardly affected by the treatment, so that a low drop out is observed. Following the application scheme illustrated in Figure 1, a putative carcinogen is administered at two different dose levels with and without previous exposure to diethylnitrosamine (DEN). To prevalidate the RLFB the four model hepatocarcinogens N-nitrosomorpholine (NNM), 2-acetylaminofluorene (2-AAF), phenobarbital (PB), and clofibrate (CF) were tested in five laboratories in a total of 1600 male and female rats. For microscopic identification of FAH, liver sections were stained by the marker GSTP (placental form of glutathione S-transferase).
|
The present report shows the application of two different models to the data, the classical two-stage clonal expansion model (TSCEM), and a modification of the color-shift model (CSM), the color-shift model with beta distributed growth rates (CSMbeta) (Groos and Kopp-Schneider, 2006
; Kopp-Schneider et al., 1998). Both models involve parameters which allow for inference about the effect of a tested carcinogen on the rate of formation and size development of FAH. The aim of the present study is to analyze sex-specific and substance-specific differences in mode of action of four carcinogens. |
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Experimental Setup
The model carcinogens NNM, PB, 2-AAF, and CF were tested following the application scheme illustrated in Figure 1 in five different laboratories. The study design is shown in Table 1. The participating laboratories were Product Safety-Regulations, Toxicology and Ecology, BASF AG, Ludwigshafen; Toxicology-Rodent Studies and Genetic Toxicology, Bayer AG, Wuppertal; Department of Cellular and Molecular Pathology, German Cancer Research Center, Heidelberg; Institute of Toxicology, GSF-National Research Center for Environment and Health, Neuherberg; Department of Toxicology, Institute of Pharmacology and Toxicology, University of Tübingen, Tübingen. All five laboratories used standard operating procedures for animal experiments and for detection and evaluation of FAH. Each chemical was administered to 40 male and 40 female juvenile (22 ± 1 day old) Wistar rats in five subgroups of eight animals each. Three subgroups received DEN at the start of the experiment. DEN was dissolved in distilled water and applied by gavage at a dose of 30 mg/kg body weight (bwt), whereas controls received distilled water alone. The treatment of test carcinogens ran from the third week until sacrifice 12 weeks after start of the experiment. NNM was given daily 5 days/week by gavage dissolved in distilled water in doses of 1 and 5 mg/kg bwt, while 2-AAF was suspended in olive oil and administered daily on 5 days/week at doses of 2 and 10 mg/kg bwt, respectively. Controls received the respective solvents. PB (0.01 and 0.05%) and CF (0.1 and 0.5%) were mixed into the diet.
|
Liver sections were treated immunohistochemically for the demonstration of GSTP which is a reliable marker for FAH (Ito et al., 1992
; Sato, 1989). Semiautomatic morphometrical evaluation resulted in data files containing the total area of each liver section and the area of each FAH. A total area of 2445-cm2 GSTP-stained liver was evaluated and almost 170,000 GSTP-positive FAH with an area larger than 0.003 mm2 were recorded. Out of the 1600 rats of the present study (40 per sex and substance, four substances, five laboratories) only eight rats died before the end of the study, in no case more than one animal per experimental group. For further details concerning the treatment of animals, the preparation of liver sections and quantitative evaluation of FAH see Ittrich et al. (2003).
Models
The TSCEM.
The TSCEM describes the formation and growth of FAH on the basis of individual hepatocytes (Moolgavkar and Venzon, 1979). The model assumes that single cells alter their phenotype through mutation. The first cell in a clone of intermediate cells is assumed to be generated through a Poisson process with parameter µ and intermediate cells are assumed to be subject to a linear birth death process with birth rate ß and death rate 
. Under the assumption that cells are full packed in space and that FAH are spherical, a focus containing n intermediate cells has a radius of 
where rc is the radius of a single cell. Moolgavkar et al. (1990) derived the following expression for the density of radius distribution of FAH, R(t):
 | (1) |
 |
The TSCEM regards each focus cell as independent from all other cells with the ability to divide or to die or differentiate. As a consequence, it assumes that FAH act independently of each other. It allows for stochastic growth and in particular for extinction of FAH.
Color-shift model with beta distributed growth rates.
The CSM introduced by Kopp-Schneider et al. (1998) describes colonies of intermediate cells as entities of spherical shape which change their size by increase of radius. This approach allows for a size increase not only occurring through clonal expansion but also through recruitment of neighboring cells. Centers of spherical foci are generated according to a homogeneous Poisson process in the liver with parameter µ. Starting from a fixed size, r0, foci grow exponentially with deterministic growth rate, ß. The assumption of deterministic growth rate was found to be too restrictive and an approach with heterogeneous growth was developed (Groos and Kopp-Schneider, 2004, 2006). The CSMbeta allows the growth rate to differ from focus to focus, so that the growth rate ß of each FAH is a realization of a beta distributed random variable B(p,q,a) p,q,a > 0. Since the support of beta distribution with parameters p and q is the interval [0,1] and the daily growth rates are much smaller than one cell division per day, an additional parameter a was introduced to constrict the support to [0,a], a << 1 [day–1]. The following expression was derived by Groos and Kopp-Schneider (2006) for the density of radius distribution of FAH:
 | (2) |
 |
Extinction of FAH is not described by CSMbeta, and correlation between FAH cells is incorporated as it does not treat FAH cells as independent units but rather considers FAH as entities. Originally, CSM was formulated to describe phenotypic shifts in FAH pathogenesis. As the RLFB data set contains data about FAH marked by one enzyme alteration, the full potential of the CSM model is not exploited here and the difference in TSCEM and CSMbeta concerns FAH growth solely.
For details concerning model assumptions and detailed derivation of the density functions 1 and 2 see Moolgavkar et al. (1990) and Groos and Kopp-Schneider (2006), respectively.
Application of the models to data.
The Wicksell-transformation (Wicksell, 1925) is used to derive the two-dimensional size distribution of FAH transections from the three-dimensional FAH size distribution. Equations 1 and 2 describing the 3D-density of the size distribution fR(t) are transformed into the 2D-density of the size distribution
 |
 |
At time point t, the expected number of observable focal transections per unit area is given by 2 x µt x µ
.
Each liver section contributes additively to the log-likelihood with a Poisson term describing the number of focal transections and a sum of 2D-size contributions from each focal transection.
A focus in CSMbeta is assumed to start from a single cell. Altered hepatocytes have an average radius of about rc = 0.014 mm (Jack et al., 1990). Therefore, r0 = 0.014 mm, likewise. Foci with area below 0.003 mm2 were not recorded. This restriction translates to = 0.031 mm.
Separate applications for the five laboratories showed qualitatively similar results, and hence, the data sets for each experimental group were pooled for all five laboratories. Due to the lack of a hierarchical structure of the two applied models, likelihood ratio tests are not applicable, but evidence exists that the likelihood value at the maximum may be taken as a measure for model fit (Kopp-Schneider and Portier, 1991). The MATLAB environment was chosen for implementation of the likelihood because it provides the opportunity to develop the application in a modular structure. The minimization of the negative log-likelihood was executed by the MATLAB intrinsic function fmincon where the Fortran–Mex Interface is used to include Fortran NAg routines for numerical double integration (Nag, Ltd., NAg Fortran Library Manual, Mark 18, 1997).
Model-Based Dose–Response Analysis
The intention of the model-based evaluation is to identify the mode of action of the tested carcinogens. Both models have the advantage that they use easily interpretable parameters. The Poisson parameter µ = µ(dose) implemented in both models is a measure for the potential of a given substance to induce FAH in normal liver tissue. In TSCEM the value of ß = ß(dose) is interpretable as measure for the potential of the carcinogen to increase the size of preexisting FAH, whereas the ratio /ß of death rate to birth rate is kept constant to 0.98 for dose (Kopp-Schneider et al., 2006; Moolgavkar et al., 1990).
In CSMbeta the expectation of the beta distributed growth rate is given by b = ap/(p+q). The expected growth rate ß = ß(dose) describes the effect of the tested substance on size increase.
The analysis of dose dependency of the parameters allows for inference about influence of dose level on the effect of agents concerning FAH formation and size increase. Application of two distinct models allows to study the sensitivity of the conclusions about the mode of action of the carcinogens to model misspecification.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Goodness of Model Fit
Figure 2 shows empirical and theoretical cumulative FAH size distributions from TSCEM and CSMbeta exemplarily for male rats in the 2-AAF experiment. Both models predict FAH size distribution well. Fit of CSMbeta cannot be distinguished from the observed data for all groups except control, whereas the fit of TSCEM slightly deviates from the data in the high dose groups. The application of the models to all other experimental groups leads to similar model fits. Both models, TSCEM and CSMbeta, perfectly fit the observed FAH number (data not shown).
|
Application of TSCEM
Figures 3A and 3B show the dose dependency of maximum-likelihood parameter estimates of the rate of formation µ (Fig. 3A) and the rate of cell division ß (Fig. 3B) in TSCEM for all four substances. Supplementary Table 1 shows the maximum-likelihood parameter estimates for the best fitting TSCEMs for all four substances. Supplementary Table 2 provides maximal likelihood values for combinations of parameter dose dependencies. Using this table, nested models can be compared via likelihood ratio tests.
|
Diethylnitrosamine.
Initially, the acute exposure to DEN was expected to result in a singular increase in formation of FAH. Therefore, a modification of TSCEM allowing for an acute effect of DEN exposure represented by an additional parameter
was applied to the data. The maximum-likelihood parameter estimate for was near to or equal zero in all treatment groups. Identical likelihood values with and without show that inclusion of an acute effect of DEN does not improve model fit, which confirms the assumption that acute DEN exposure has a persistent effect on the formation of FAH.
The results of application of TSCEM show that DEN treatment leads to an additive shift in the formation parameter µ in all treatment groups for males and females, suggesting that the DEN effect on the rate of formation, µ, lasted for the whole duration of the experiment. A similar observation has been made by other researchers (Luebeck et al., 2000). Female rats seem to respond stronger than males to DEN exposure, which is reflected in larger shifts in parameter µ. For 2-AAF treatment a small additive shift in the growth parameter ß for males and particularly for females is observed. However, for the three other substances the DEN effect on ß is less clear.
N-nitrosomorpholine.
The effect of NNM treatment on the formation parameter µ is almost linear in dose with a marginal saturation effect for DEN-treated males (Fig. 3A). For DEN-treated female rats this saturation effect is much stronger on a higher level. Without previous exposure to DEN, the dose–response curve for µ is increasing with dose.
Dose dependency of the parameter describing FAH cell division, ß, for DEN-treated males shows gently inclined linearity (Fig. 3B). The ß dose–response curve for DEN-treated females is almost identical to that for males. On the other hand, without DEN treatment the ß dose dependency for females without DEN is flat, suggesting that there is no effect of NNM alone on ß for female rats. The ß-estimate for the male group without any treatment is rather high, which leads to a slight decrease of ß in males high dose only group. However, the number of FAH observed in untreated animals is so small that the ß-estimate for the male group without any treatment is associated with large variability.
Overall, NNM seems to have an effect on both parameters, µ and ß, where the effect on the formation parameter µ is somewhat stronger and shows saturation in high dose particularly for females and the effect on ß seems to depend on DEN initiation.
2-acetylaminofluorene.
The effect of 2-AAF treatment on the rate of FAH formation, µ, for male animals is similar to the effect of NNM with a marginally stronger saturation effect (Fig. 3A). The curve for DEN + 2-AAF exposed female animals shows a strong increasing effect in the µ-estimate. The µ-estimate for the females treated by DEN followed by high dose of 2-AAF has the highest value of the whole experiment. The curves for males and females treated with 2-AAF alone are almost identical and they are similar to the corresponding curves for NNM-alone treated groups.
2-AAF treatment has a strong effect on the growth parameter ß for both males and females (Fig. 3B). The dose–response curve for males shows a strong and very similar initiation with DEN. The ß dose–response curves for DEN + 2-AAF and 2-AAF alone show a distinct increase for female animals, DEN having an additive effect on this parameter, i.e., DEN treatment increases ß by 0.38 [day–1] for vehicle control and high dose. In comparison to DEN + NNM, DEN + 2-AAF has a very similar effect on female animals.
Overall, 2-AAF has an effect on both parameters. For males the ß-effect is dominating, whereas for females there is a stronger effect on µ.
Phenobarbital.
A small increasing effect of PB on the estimate of µ for DEN-exposed males is observed (Fig. 3A). The µ-estimates for DEN + PB–treated females are almost constant in dose. Without DEN treatment the µ-estimates are negligible, both for females and males.
For males and for females treated by DEN + PB the growth parameter, ß, follows a linear dose–response curve with small gradient (Fig. 3B). It is noticeable that ß-estimates for males which received only vehicle or PB alone are higher than for males in the "DEN + PB"-group. The number of FAH observed in both groups without DEN treatment, however, is so small that inference from the parameter estimates is restricted. The ß dose–response for females not exposed to DEN is slightly increasing and again, DEN has an additive effect on this parameter.
Clofibrate.
Administration of CF seems to reduce the formation rate µ of GSTP-positive FAH in DEN exposed males and females (Fig. 3A). The effect of CF treatment alone on the formation parameter µ is negligible.
The effect on FAH cell division, described by ß, is negligible for DEN-exposed males (Fig. 3B). ß-estimates for DEN-treated females increase slightly with dose of CF. ß dose dependencies for males and females exposed to CF alone are almost identical and show a slight increase.
Application of CSMbeta
Figures 3C and 3D show the dose dependency of maximum-likelihood parameter estimate of µ and of the expected growth rate, which is given by the expression b=ap/(p+q). As application of TSCEM showed that acute DEN exposure has a persistent effect on formation of FAH, a CSMbeta model with persistent effect on FAH formation and size increase was used. Supplementary Table 3 shows the maximum-likelihood parameter estimates for the best fitting CSMbetas for all four substances.
N-nitrosomorpholine.
For DEN-treated males the formation rate µ increases in dose with a strong saturation effect (Fig. 3C). Females in the same treatment group show that µ is slightly higher for low dose treatment than for high dose treatment. The µ-estimate for low dose is identical to the corresponding DEN control. Without exposure to DEN males and females show an increase in µ for high dose.
For DEN-exposed animals the expected mean growth rates ß show an increasing dose–response curve, with a saturation effect for females (Fig. 3D). Without DEN treatment ß is nearly constant for females and even descending in dose for males, which is probably due to a high ß-value for the control group.
2-acetylaminofluorene.
For DEN-treated animals 2-AAF has an increasing effect on µ (Fig. 3C). The effect is almost linear for females and shows a saturation effect for males. Males without DEN exposure show an increase with dose. When administered alone 2-AAF seems to have a repressive effect on µ for males.
The dose–response curve for the expected mean growth ß is linear for all treatment groups, with a light saturation effect for DEN-exposed females (Fig. 3D). The value for DEN + high dose of 2-AAF treated males is the largest of the whole experiment.
Phenobarbital.
No effect of PB treatment alone on µ is observed in males and females (Fig. 3C). The effect on DEN-exposed males is small. The dose–response curve for DEN-exposed females is striking because the µ-estimate for the DEN + PB groups is much lower than the corresponding DEN control group. No conclusive explanation can be offered for this phenomenon. However, the µ-estimate value increases from DEN + PB low dose to DEN + PB high dose.
For the expected growth parameter ß dose–response for PB treatment alone is descending for males and slightly increasing for females (Fig. 3D). After DEN treatment PB seems to have practically no effect on the size development of FAH in males, whereas for females ß is increasing with dose with a saturation effect.
Clofibrate.
In all treatment groups and for both sexes treatment with CF leads to descending µ-values (Fig. 3C). For the parameter b the dose–response curve for DEN + CF–treated males shows a slight tendency to fall and in all other treatment groups an increasing dose–response is observed (Fig. 3D).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
CF stands out from the other three carcinogens as the number and area fraction of GSTP-stained FAH is reduced in CF-treated groups (Ittrich et al., 2003
), a finding which was not unexpected because similar results have been reported before, probably because peroxisome proliferators suppress the GSTP-positive phenotype (Nishimura et al., 1995). Applications of TSCEM and CSMbeta reflect this effect through a decrease of the parameter of FAH formation, µ, in CF-treated groups (see Fig. 3). Therefore, in the following the effect of the carcinogens on formation and growth of FAH is only compared between NNM, 2-AAF, and PB.
Moolgavkar et al. (1990) applied TSCEM to experimental data from adult female rats treated with eight different concentrations of NNM in drinking water. Animals were killed at different ages and liver sections were examined for ATPase deficient foci (Kunz et al., 1982). One treatment group received a concentration of 40 p.p.m. NNM, which is comparable to the high dose treatment group in the present experiment. (Based on a daily consumption of 100–120 ml drinking water/day and kg bwt, a concentration of 40 p.p.m. NNM approximately corresponds to a daily dose of 5 mg NNM/kg bwt.) Moolgavkar et al. (1990) identified values of µ = 0.4/mm3 and day and ß = 2.1/day for the 40 p.p.m. dose group. In the present experiment values for the parameter of formation µ = 4.8/mm3 and day and growth ß = 0.8/day were estimated for females treated with high dose of NNM. The two experiments are not fully comparable because Moolgavkar and associates used adult animals instead of juvenile rats in our experiment and ATPase instead of GSTP as staining method. The higher µ-value calculated in the present experiment might reflect higher hepatocyte division rates in juvenile as compared to adult rats which are expected to result in higher mutation rates, both spontaneous and NNM induced. In addition, GSTP as a positive stain may yield higher counts particularly of small foci which are more difficult to detect with the ATPase marker, where a decrease in enzyme activity in foci is used for their identification. It is noteworthy, however, that parameter estimates in the two studies do not differ by orders of magnitude.
Figure 4 shows maximum-likelihood parameter estimates for µ and ß in TSCEM for DEN-initiated males and females for NNM, 2-AAF, and PB. Control groups were pooled for each sex. This plot allows for simultaneous analysis of sex and substance-specific effects.
|
Sex-Specific Differences
A clear shift in µ from males to females is observed, suggesting that females respond stronger to DEN than males. Figure 3A confirms the impression that this effect is due to DEN initiation and not to a tendency of female animals to spontaneously develop higher numbers of FAH. Females are often preferred to males in hepatocarcinogenity experiments because of their higher susceptibility toward development of liver tumors, but there is no clear cut gender difference in spontaneous liver tumor incidence when analyzed in various strains of rats (van Ravenzwaay and Tennekes, 2002
). The effect of low dosed 2-AAF on both model parameters is similar for males and females. However, for high dosed 2-AAF the effect on FAH growth (ß) is much more pronounced in males than in females, whereas females respond with a strong increase in rate of FAH formation (µ). Tumoko et al. (1993) observed that orchidectomy decreased the number of 2-AAF induced foci in male rats, whereas ovariectomy had no effect in female rats, suggesting a strong interaction between male sex hormones and mode of action of 2-AAF.
The sex-specific difference in mode of action of NNM is less obvious than for 2-AAF treatment. At high doses, the effect on growth parameter, ß, is similar for females and males. However, a slight decrease in rate of formation, µ, for females is observed, which may be due to a saturation effect.
The effect of PB treatment differs between males and females. Low dosed PB has a stronger effect on µ for males, whereas for females the effect on ß seems to dominate. High dosed PB has a comparable effect on both parameters ß and µ for both genders.
Substance-Specific Differences
Considering the substances separately for the two genders, it is apparent that 2-AAF has the strongest promoting potential for males, whereas NNM and PB behave similarly, NNM being more potent than PB. For females, 2-AAF strikes as a potent initiator. The promotional activity of 2-AAF in rat liver may be related to induction of resistance to apoptosis of initiated cells, presumably by suppression of the mitochondrial proapoptotic pathway by the aromatic amine (Kloehn et al., 2003). Increasing dose from low to high dose NNM in females has no initiating but only a promoting effect. This result is consistent with earlier results by Weber and Bannasch (1994). They evaluated liver sections of male rats at different times of continuous NNM treatment and detected a saturation effect concerning the number of clear/acidophilic FAH. PB is much less potent and seems to have more promoting than initiating potential. Since PB is not able to induce FAH during observation time when administered alone (see Fig. 2) it is remarkable that PB shows an effect on FAH formation rate, µ, when administered following DEN treatment. This observation leads to the still resumed discussion whether a hepatocarcinogen is able to increase FAH formation without direct interaction with DNA (Bannasch and Zerban, 1992).
It is conceivable that treatment with PB or any other agent decreases the ratio of death to birth rate, /ß. Therefore, the assumption of a constant ratio of death to birth rate is reconsidered. An application of TSCEM with different formation parameter µ for DEN-initiated and uninitiated animals, but independent of the dose of the test substance was implemented in which /ß was dose dependent. The results showed that /ß is decreasing in dose for each of the three substances. The goodness of fit, reflected by the likelihood values at the maximum, was consistent with the corresponding values for the approach with constant ratio /ß shown earlier. However, the estimates for /ß obtained by evaluation of NNM and 2-AAF treatment groups achieve values larger than one, i.e., death rates exceed birth rates, which seems unrealistic. Solely in case of PB treatment /ß values are in a plausible range. Hence, the original approach with constant ratio /ß and dose-dependent µ is preferred for analyzing the mode of action of NNM and 2-AAF showing that both have initiating potential. In case of PB treatment either an interaction of DEN and PB leading to increased FAH formation rate, or a dose-dependent decrease of the ratio /ß may explain the data. PB has repeatedly been demonstrated to decrease apoptosis of hepatoma cells, both in vitro and in vivo (Buchmann et al., 1999; Schulte-Hermann et al., 1990). Inhibition of apoptosis by PB may lead to an increase in foci numbers because the barbiturate thereby attenuates the risk of elimination of spontaneously initiated single cells or small clones, which would otherwise become extinct (Schwarz et al., 1995).
In a similar representation for CSMbeta, Figure 5 shows µ and ß for males and females treated by DEN for NNM, 2-AAF, and PB after pooling control groups for each gender. The conclusions drawn from application of TSCEM model mostly carry over to CSMbeta model. The only striking model specific difference is observed for PB treated females. There is a strong decrease in µ from female DEN controls to females treated with DEN + low dose PB. The effect of DEN + high dosed PB is a pure effect on µ. However, actual values of parameter estimates cannot be compared between TSCEM and CSMbeta as CSMbeta does not allow for FAH to disappear and, therefore, µ-values in CSMbeta are much smaller than corresponding µ-values in TSCEM.
|
As to the comparison of model fit, CSMbeta seems to fit the FAH size distribution slightly more accurately than TSCEM. However, CSMbeta does not incorporate clonal extinction, therefore, the effect of the compound on induction of apoptosis cannot be studied via CSMbeta.
For inference about the characteristic dose–response relationships of the four compounds it suffices to compare changes in parameter estimates. The actual values of the parameter estimates, especially those of the growth rates in TSCEM, depend on the ratio of death to birth rate which was fixed to 0.98. Moreover, a further refinement of the model to include sporadic foci formation prior to the start of the experiment may lead to more plausible estimates of cell division rates but is not warranted since a single observation time point is available. A modification of the TSCEM has been suggested to include the concept that only the "surface" cells of a focus divide instead of all cells as assumed here (Moolgavkar et al., 1996). Application of a surface growth model may result in more realistic cell division rates, but will leave the relationship between rates unchanged.
In summary, the application of TSCEM and CSMbeta to the present experiment allows for detailed inference about the differences in mode of action of the four different test carcinogens and DEN on FAH formation and growth. As both models yield similar results, conclusions may be robust to misspecification of the model. Application of the models suggests a persistent effect of one-time initiation with DEN on FAH formation rate which is stronger for females than for males. NNM was identified as a potent carcinogen with more initiating than promoting activities, i.e., the slope of the dose–response curve is steeper for the rate of FAH formation than for FAH growth. For females a saturation effect for initiating activity was detected. In our study, 2-AAF was found to have the strongest effect on formation and growth of FAH with large sex-specific differences in mode of action, assumedly due to the interaction with male sex hormones. PB showed more promoting than initiating activities and was less potent than NNM and 2-AAF. It is questionable whether PB has a direct effect on the FAH formation rate or contrariwise decreases the ratio of death to birth rate potentially via reduced apoptosis. CF as peroxisome proliferator stands out as it reduces number and area fraction of GSTP-stained FAH in males and females.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Supplementary data are available online at http://toxsci.oxfordjournals.org/.
|
FUNDING |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Bundesministerium für Bildung und Forschung of the Federal Republic of Germany grants (PTJ-BIO 0311834, 0311835, 0312253, 0312254, and 0312261); and DFG grant (KO1886/1-3 and 1-5) to J.G.
|
ACKNOWLEDGMENTS |
|---|
We are grateful to Dres. Susanne Brendler-Schwaab, Erhard Deml, Harald Enzmann, Karin Kuettler, Werner Mellert, Oliver Moennikes, Doris Oesterle, and Ludwig Schladt for letting us use data obtained in their laboratories.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Bannasch P. The cytoplasm of hepatocytes during carcinogenesis. Electron and light microscopic investigations of the nitrosomorpholine-intoxicated rat liver. Recent Results Cancer Res. (1968) 19:1–100.
Bannasch P. Preneoplastic lesions as end points in carcinogenicity testing. I. Hepatic preneoplasia. Carcinogenesis (1986) 7:689–695.
Bannasch P, Nehrbass D, Kopp-Schneider A. Significance of hepatic preneoplasia for cancer chemoprevention. IARC Sci. Publ. (2001) 154:223–240.[Medline]
Bannasch P, Zerban H. Predictive value of hepatic preneoplastic lesions as indicators of carcinogenic response. IARC Sci. Publ. (1992) 116:389–427.[Medline]
Buchmann A, Willy C, Buenemann CL, Stroh C, Schmiechen A, Schwarz M. Inhibition of transforming growth factor beta1-induced hepatoma cell apoptosis by liver tumor promoters: Characterization of primary signaling events and effects on CPP32-like caspase activity. Cell Death Differ. (1999) 6:190–200.[CrossRef][Web of Science][Medline]
Delesse M. Procede mecanique pour determiner la composition des roches. Compt.Rend. Acad. Sci. (1847) 25:544–545.
Feo F, Pascale R, Simile M, Miglio MD, Muroni M, Calvisi D. Genetic alterations in liver carcinogenesis: Implications for new preventive and therapeutic strategies. Crit. Rev. Oncog. (2000) 11:19–62.[Web of Science][Medline]
Foulds L. Neoplastic Development. (1969) Vol. 1. London: Academic Press.
Friedewald WF, Rous P. The initiating and promoting elements in tumor production. J. Exp. Med. (1944) 80:101–131.[Abstract]
Groos J, Kopp-Schneider A. Visualization of parametric carcinogenesis models. In: Compstat. Proceedings in Computational Statistics—Antoch J, ed. (2004) Heidelberg, Germany: Physika-Verlag. 189–198.
Groos J, Kopp-Schneider A. Application of a color-shift model with heterogeneous growth to a rat hepatocarcinogenesis experiment. Math. Biosci. (2006) 202:248–268.[CrossRef][Web of Science][Medline]
Herath NI, Leggett BA, MacDonald GA. Review of genetic and epigenetic alterations in hepatocarcinogenesis. J. Gastroenterol. Hepatol. (2006) 21:15–21.[CrossRef][Web of Science][Medline]
Ito N, Shirai T, Hasegawa R. Medium-term bioassays for carcinogens. In: Mechanism s of Carcinogenesis in Risk Identification—Vainio H, Magee PN, McGregor DB, McMichael AJ, eds. (1992) Lyon, France: IARC. 353–388.
Ittrich C, Deml E, Oesterle D, Küttler K, Mellert W, Brendler-Schwaab S, Enzmann H, Schladt L, Bannasch P, Haertel T, et al. Prevalidation of a rat liver foci bioassay (RLFB) based on results from 1600 rats: A study report. Toxicol. Pathol. (2003) 31:60–79.[CrossRef][Web of Science][Medline]
Jack EM, Bentley P, Bieri F, Muakkassah-Kelly SF, Stäubli W, Suter J, Waechter F, Cruz-Orive LM. Increase in hepatocyte and nuclear volume and decrease in the population of binucleated cells in preneoplastic foci of rat liver: A stereological study using the nucleator method. Hepatology (1990) 11:286–297.[CrossRef][Web of Science][Medline]
Kloehn P-C, Soriano ME, Irwin W, Penzo D, Scorrano L, Bitsch A, Neumann H-G, Bernardi P. Early resistance to cell death and to onset of the mitochondrial permeability transition during hepatocarcinogenesis with 2-acetylaminofluorene. Proc. Natl. Acad. Sci. U.S.A. (2003) 100:10014–10019.
Kopp-Schneider A. Biostatistical evaluation of focal hepatic preneoplasia. Toxicol. Pathol. (2003) 31:121–125.[CrossRef][Web of Science][Medline]
Kopp-Schneider A, Haertel T, Burkholder I, Bannasch P, Wesch H, Groos J, Heeger S. Investigating the formation and growth of a-particle radiation-induced foci of altered hepatocytes: A model based approach. Radiat. Res. (2006) 166:422–430.[CrossRef][Web of Science][Medline]
Kopp-Schneider A, Portier CJ. Distinguishing between models of carcinogenesis: The role of clonal expansion. Fundam. Appl. Toxicol. (1991) 17:601–613.[CrossRef][Web of Science][Medline]
Kopp-Schneider A, Portier C, Bannasch P. A model for hepatocarcinogenesis treating phenotypical changes in focal hepatocellular lesions as epigenetic events. Math. Biosci. (1998) 148:181–204.[CrossRef][Web of Science][Medline]
Kunz W, Schaude G, Schwarz M, Tennekes H. Quantitative aspects of drug-mediated tumour promotion in liver and its toxicological implications. Carcinog. Compr. Surv. (1982) 7:111–125.[Medline]
Luebeck EG, Buchmann A, Stinchcombe S, Moolgavkar SH, Schwarz M. Effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin on initiation and promotion of GST-P-positive foci in rat liver: A quantitative analysis of experimental data using a stochastic model. Toxicol. Appl. Pharmacol. (2000) 167:63–73.[CrossRef][Web of Science][Medline]
Moolgavkar SH, Luebeck EG, Buchmann A, Bock KW. Quantitative analysis of enzyme-altered liver foci in rats initiated with diethylnitrosamine and promoted with 2,3,7,8-tetrachlorodibenzo-p-dioxin or 1,2,3,4,6,7,8-heptachlorodibenzo-p-dioxin. Toxicol. Appl. Pharmacol. (1996) 138:31–42.[CrossRef][Web of Science][Medline]
Moolgavkar SH, Luebeck EG, deGunst M, Port RE, Schwarz M. Quantitative analysis of enzyme-altered foci in rat hepatocarcinogenesis experiments—I. Single agent regimen. Carcinogenesis (1990) 11:1271–1278.
Moolgavkar SH, Venzon DJ. Two-event models for carcinogenesis: Incidence curves for childhood and adult tumors. Math. Biosci. (1979) 47:55–77.[CrossRef][Web of Science]
Nishimura S, Yokoyama Y, Nakano H, Satoh K, Kano H, Sato K, Tsuchida S. Decreased expression of glutathione S-transferases and increased fatty change in peroxisomal enzyme-negative foci induced by clofibrate in rat livers. Carcinogenesis (1995) 16:1699–1704.
Peraino C, Fry R, Staffeldt E. Reduction and enhancement by phenobarbital of hepatocarcinogenesis induced in the rat by 2-acetylaminofluorene. Cancer Res. (1971) 31:1506–1512.
Pereira M. Rat liver foci bioassay. J. Am. Coll. Toxicol. (1982) 1:101–117.[Web of Science]
Pitot H, Barsness L, Goldsworthy T, Kitagawa T. Biochemical characterisation of stages of hepatocarcinogenesis after a single dose of diethylnitrosamine. Nature (1978) 271:456–458.[CrossRef][Medline]
Pitot H, Campbell H, Maronpot R, Bawa N, Rizvi T, Xu Y, Sargent L, Dragan Y, Pyron M. Critical parameters in the quantitation of the stages of initiation, promotion, and progression in one model of hepatocarcinogenesis in the rat. Toxicol. Pathol. (1989) 17:594–611. discussion 611–612.[Medline]
Pogribny IP, Ross SA, Wise C, Pogribna M, Jones EA, Tryndyak VP, James SJ, Dragan YP, Poirier LA. Irreversible global DNA hypomethylation as a key step in hepatocarcinogenesis induced by dietary methyl deficiency. Mutat. Res. (2006) 593:80–87.[Web of Science][Medline]
Rous P, Kidd JG. Conditional neoplasms and subthreshold neoplastic states. J. Exp. Med. (1941) 73:365–390.[Abstract]
Sato K. Glutathione transferases as markers of preneoplasia and neoplasia. Adv. Cancer Res. (1989) 52:205–255.[Web of Science][Medline]
Schulte-Hermann R, Kraupp-Grasl B, Bursch W, Gerbracht U, Timmermann-Trosiener I. Effects of non-genotoxic hepatocarcinogens phenobarbital and nafenopin on phenotype and growth of different populations of altered foci in rat liver. Toxicol. Pathol. (1989) 17:642–649. discussion 649–50.[Medline]
Schulte-Hermann R, Timmermann-Trosiener I, Barthel G, Bursch W. DNA synthesis, apoptosis, and phenotypic expression as determinants of growth of altered foci in rat liver during phenobarbital promotion. Cancer Res. (1990) 50:5127–5135.
Schwarz M, Buchmann A, Stinchcombe S, Luebeck G, Moolgavkar S, Bock KW. Role of receptors in human and rodent hepatocarcinogenesis. Mutat. Res. (1995) 333:69–79.[CrossRef][Web of Science][Medline]
Tokumo K, Umemura T, Sirma H, Gebhardt R, Poirier M, Williams G. Inhibition by gonadectomy of effects of 2-acetylaminofluorene in livers of male, but not female rats. Carcinogenesis (1993) 14:1747–1750.
van Ravenzwaay B, Tennekes H. A Wistar rat strain prone to spontaneous liver tumor development: Implications for carcinogenic risk assessment. Regul. Toxicol. Pharmacol. (2002) 36:86–95.[CrossRef][Web of Science][Medline]
Weber E, Bannasch P. Dose and time dependence of the cellular phenotype in rat hepatic preneoplasia and neoplasia induced by continuous oral exposure to N-nitrosomorpholine. Carcinogenesis (1994) 15:1235–1242.
Wicksell S. The corpuscle problem. A mathematical study of a biometrical problem. Biometrika (1925) 17:87–97.
Williams G. The significance of chemically-induced hepatocellular altered foci in rat liver and application to carcinogen detection. Toxicol. Pathol. (1989) 17:663–672. discussion 673–674.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. V. Bagnyukova, V. P. Tryndyak, B. Montgomery, M. I. Churchwell, A. R. Karpf, S. R. James, L. Muskhelishvili, F. A. Beland, and I. P. Pogribny Genetic and epigenetic changes in rat preneoplastic liver tissue induced by 2-acetylaminofluorene Carcinogenesis, March 1, 2008; 29(3): 638 - 646. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||