Regulatory Toxicology and Pharmacology 70 (2014) 98–106

Contents lists available at ScienceDirect

Regulatory Toxicology and Pharmacology journal homepage: www.elsevier.com/locate/yrtph

Investigation on baseline toxicity to rats based on aliphatic compounds and comparison with toxicity to fish: Effect of exposure routes on toxicity Jia He, Ling Fu, Yu Wang, Jin J. Li, Xiao H. Wang, Li M. Su ⇑, Lian X. Sheng, Yuan H. Zhao ⇑ State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration, School of Environment, Northeast Normal University, Changchun, Jilin 130117, PR China

a r t i c l e

i n f o

Article history: Received 20 February 2014 Available online 26 June 2014 Keywords: Baseline Exposure route Lethal critical concentration Threshold Bioconcentration Intestinal absorption

a b s t r a c t The aim of this paper was to investigate baseline toxicity to rats and effect of exposure routes on toxicity in rats and fish. In this paper, 1588 industrial chemicals were selected to investigate baseline toxicity to rats. The results showed that rat toxicity varies around a constant for classified compounds or homologues. The toxic contributions of substituted functional groups have been calculated and alkanes were used as baseline toxicity. The toxic contributions, equal to toxic ratios (TR), show that small changes in chemical structure can result in different toxic effect in rat toxicity. However, this situation has not been observed in fish toxicity because the threshold of excess toxicity (e.g. log TR = 1) was too high to distinguish differences in toxicity. Very close critical body residues (CBRs) calculated from percentage of absorption and bioconcentration factors indicate that most of aliphatic chemicals may share the same modes of toxic action between rat and fish species. The high estimation error of bioconcentration factor calculated from computer programs for some compounds suggests that classification of excess toxicity should be based on the CBRs, rather than the TR because the TR is closely related to the exposure routes. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction Estimation of rodent acute toxicity (LD50) is an important task in drug design and toxicological risk assessment of chemicals. Rats and mice are the main species used in these studies. Acute toxicity is considered as the adverse effects occurring within a given time, following a single exposure to a substance (OECD Guideline, 2008). There were a number of QSAR (quantitative structure–activity relationship) models available in the literature for predicting the acute toxicity of chemicals to rats (Tsakovska et al., 2008; Koleva et al., 2011). Some of these predictive methods have been derived from limited data sets of structurally similar chemicals such as alcohols or anilines (Devillers and Devillers, 2009). Recent reviews of existing QSAR methods showed that only a few successful QSAR models were capable of predicting LD50 values for structurally diverse chemicals (Tsakovska et al., 2008; Sazonovas et al., 2010). Among such works were the models reported by Enslein et al. (1989), Zhu et al. (2009) and Lagunin et al. (2011). LD50 after gavage dosing depends on so many variable biological mechanisms that ⇑ Corresponding authors. Fax: +86 431 89165606. E-mail addresses: [email protected] (Y.H. Zhao), [email protected] (L.M. Su). http://dx.doi.org/10.1016/j.yrtph.2014.06.019 0273-2300/Ó 2014 Elsevier Inc. All rights reserved.

reliable predictions are difficult (Sazonovas et al., 2010). Most of published QSARs are local models, i.e. restricted to single classes of chemicals, such as alcohols, phenols and anilines (Tsakovska et al., 2008). The chemicals that are not reactive and do not interact with specific receptors form the so-called baseline (or non-polar narcotics) in aquatic toxicity. In principle, the baseline narcosis is the minimum toxicity that compounds exhibit. The narcosis model for baseline toxicity in aquatic toxicology was based on the earlier works (Ferguson, 1939; Könemann, 1981; Verhaar et al., 1992), who proposed that narcosis was caused when the thermodynamic activity of chemicals reaches a threshold and normal physiological processes were disrupted. It has been traditionally assumed that accumulation of compounds in lipoid membranes of nerve cells plays a key role in narcosis-related toxicity (Wolf et al., 2004). Chemicals acting by a narcosis mechanism achieve their effect once a critical concentration or critical volume has been reached within some biophase sites of action within the organism (Ferguson, 1939). The critical concentration, or called ‘‘critical body residues’’ (CBRs), were found to be equal for all non-reactive chemicals, when expressed on a molar basis (mol/kg) (McCarty et al., 1991; McCarty and Mackay, 1993). The baseline organic compounds cause mortality within a very narrow range of whole-body

99

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

tissue concentrations (2–8 mmol/g wet weight or about 50 mmol/ g lipid) in small aquatic organisms (Meador et al., 2011). Aliphatic and aromatic hydrocarbons, chlorinated substituted compounds, alcohols, ethers, ketones, aliphatic secondary and tertiary amines were classed as baseline or non-polar narcotics compounds (Verhaar et al., 1992). Hydrophobicity was commonly found to correlate well with acute toxicity to aquatic organisms for these compounds. In contrast to the aquatic toxicity, the situation in mammalian toxicity is rather different and even small changes in chemical structure can result in different modes of toxic action (Jäckel and Klein, 1991). The study by Lipnick (1991) utilized a baseline QSAR approach, deriving a simple bi-linear log P correlation on LD50 for chemically simple compounds, in order to identify new toxicological effects for more complex compounds that were identified as outliers in the baseline correlation analysis. However, unsaturated alcohols do not fit the model used for aliphatic alcohols due to different modes of toxic action. Baseline toxicity in mammalian toxicology could be expected for alcohols, acids, ketones, and one-ring aromatics. The aliphatic amines and aldehydes obviously exceed such baseline-toxicity (Jäckel and Klein, 1991; Wolf et al., 2004; Veith et al., 2009). Industrial chemicals, such as alcohols, ethers, amines and other aliphatic compounds, usually do not involve highly specific interactions with receptors (Lipnick, 1999). QSAR models of these industrial chemicals as baseline toxicity for aquatic effects are well developed. However, no systematic efforts have been made to develop QSAR models for narcosis as baseline toxicity in mammals for these industrial chemicals (Veith et al., 2009). It is obvious that toxic effects of a chemical are dependent on the exposure routes. As previously reported, variation in acute toxicity between rats and fish (rainbow trout) was reduced when exposure routes were matched (Delistraty et al., 1998; Delistraty, 2000). The comparison of acute toxicity within and between species over various exposure routes can provide insight into mechanisms of toxic action. In this paper, 1588 well-characterized industrial chemicals, such as alkanes, alkenes, alcohols, ethers, aldehydes, ketones, esters, acids and their derivatives, were selected to investigate the baseline toxicity in mammalian toxicology. The aim of this work is: (1) to explore the relationship between rat acute toxicity and hydrophobicity/substructures for aliphatic compounds; on this basis, (2) to investigate the mammalian baseline toxicity and effect of exposure routes on toxicity; (3) to discuss the factors that influence classification of baseline compounds to rats and fish. The ultimate goal of the paper is to explore if baseline compounds defined in acute fish toxicity can also be defined as baseline compounds in acute rat toxicity.

2. Materials and methods 2.1. Rat acute toxicity data (LD50) LD50 values used in this paper were compiled from Zhu et al. (2009). Upon request, a collection of 7385 compounds featured in this publication was kindly provided by the authors in its full format, including all the structures and experimental LD50 values. The values of LD50 for each compound were expressed as log 1/LD50 in mol/kg (mol per kg of rat body weight) according to standard QSAR practices. After aromatic compounds, sulfides, phosphides, and heterocyclic compounds were removed, the remaining acute toxicity data included 1588 aliphatic compounds. These compounds covering well-characterized aliphatic molecular structures were then classified into different series based on chemical functional groups. The name of each functional group and the number of compounds in each class are summarized in Table 1. Details of the classification, together with CAS number can be found in Tables S1 and S2 of Supplementary material.

Table 1 Toxic contributions of functional groups. No.

Functional groups

Number of compounds

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

Alkane Alkene/alkyne Alcohol Ether Aldehyde Ketone Formate Ester Acid Ring EpoxyCarbonate Amine Carbamate Amide Urea Carbamoyl-oxime Acyl chloride Carbonochloridate NitroNitrosoIsocyanatoNitrate Nitrile Oxime Hydrazine Guanidine N@N/C@N FluoroChloroBromoIodoSilane Total compounds

507 348 313 67 116 9 337 109 216 33 10 281 20 74 34 7 11 5 26 63 14 14 115 6 5 7 10 55 252 58 9 84 1588

Toxic contributions 1.670 0.233 0.108 0.080 0.007 0.042 0.112 0.118 0.015 0.101 0.051 0.433 0.404 0.357 0.102 0.206 1.790 0.213 0.708 0.527 0.464 0.269 0.890 0.482 0.275 1.080 0.240 0.475 0.900 0.617 0.901 1.220 0.151 AE = 0 AAE = 0.44 RMSE = 0.57

Note: The toxic contribution of alkanes is obtained from intercept of the regression. The number of compounds illuminates how many compounds contain the functional group. Because compounds contain more than two functional groups, the sum of the compounds in classes 1–33 is more than the number of total compounds 1588.

2.2. Fish 50% lethal concentration (LC50) The acute toxicity data expressed by LC50, the concentration required to kill 50% of fish within 96 h, were taken from Raevsky et al. (2008, 2009). They presented two datasets containing the acute toxicity of chemicals to guppy (Poecilia reticulata), fathead minnow (Pimephales promelas) and rainbow trout (Oncorhynchus mykiss), respectively. They confirmed the well-known good correlations of toxicity between the three fish species (Katritzky et al., 2001) and mentioned that the quality of the experimental data was not perfect for fathead minnow and rainbow trout. This is primarily because data were obtained in different laboratories with different errors of measurements. Therefore, the 96 h-LC50 values in fish for 97 aliphatic compounds used in this paper were based on the toxicity data to guppy. A few data on fathead minnow and rainbow trout were used where data to guppy were missing. Aromatic compounds were excluded from this paper. These data can be found in Table S3 of Supplementary material. 2.3. Fish bioconcentration factor (BCF) and rat intestinal absorption (%Abs.) The internal concentrations were calculated from BCF and %Abs for fish and rats, respectively. The log BCF values were estimated from a log BCF–log KOW relationship (Eq. (1)). This equation is used to estimate the log BCF values for compounds (excluding

100

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

metabolism) with log KOW in the range from 1 to 7 in the Epi Suite (version 4.0) software. This program was developed by the US Environmental Protection Agency’s Office of Pollution Prevention and Toxics and Syracuse Research Corporation (SRC) and can freely be downloaded from their website (http://www.epa.gov/opptintr/ exposure/pubs/ episuite.htm).

log BCF ¼ 0:6598 log KOW  0:333

ð1Þ

%Abs: ¼ 100  ½1  Expð100:7470:340A0:155B Þ

ð2Þ

Eq. (2) is the model used to calculate the percentage of rat intestinal absorption (%Abs.) of aliphatic compounds. Here, A is the overall solute hydrogen bond acidity and B is the overall hydrogen bond basicity. This method was based on the rat intestinal absorption dosed orally by gavage for 105 compounds (Zhao et al., 2003). The predictive ability of the method is quite accurate for compounds with high absorption. 2.4. Molecular descriptors/fragments The logarithm of the octanol/water partition coefficient (log KOW) was obtained from the EPI Suite (version 4.0) program. Where possible measured log KOW values were verified and used in preference to calculated values. Solubility (log S) and Henry law’s constant were also estimated by the program. The toxic contributions of substituted functional groups in molecules in rat toxicity were calculated by use of the Free-Wilson method. The alkane was used as the skeleton structure. The matrix of 36 fragments (or substituted functional groups) for 1588 compounds is given in Table S1 of Supplementary material. 2.5. Statistical analysis The relationship between the molecular descriptors/parameters and toxicity (log1/LD50 in mol/kg or log 1/LC50 in mol/L) was performed using a least-squares linear regression with the Minitab software (version 14). For each regression, the following descriptive information was provided: number of observations used in the analysis (n), coefficient of determination (R2), standard error of the estimate (S). The models were evaluated using the average error (AE = R(Obs  Pred)/n), the average absolute error (AAE = R|Obs  Pred|/n) and the root-mean squared error (RMSE = (R(Obs  Pred)2/n)1/2). The t-test and variance analysis were used to examine differences between two data sets. 3. Results 3.1. Relationship between LD50 and log KOW A linear regression analysis between LD50 and log KOW was carried out for 1588 aliphatic compounds. The result showed that the relationship for the 1588 compounds was very poor; the regression coefficient R2 value was only 0.30 and the standard error S was 0.70. Inclusion of other descriptors calculated in this study (not shown here) did not improve the regression equation. In order to investigate the relationship between toxicity and substructures, the aliphatic compounds were carefully classified into different classes based on their substituted functional groups (Table S1 of Supplementary material), and toxicity values were then examined for the well-classified compounds. Except for the bilinear relationship (log 1/LD50 = a1 log KOW + a2 (log KOW)2 + b) between toxicity and number of carbons or hydrophobicity of mono saturated alcohols as reported by Lipnick (1999) and other author (Koleva et al., 2011), the toxicity varied around a constant was observed for most of classes (see Table S1 of Supplementary material). This can be

seen from a plot of toxicity (log1/LD50) against the octanol/water partition coefficient log KOW (Fig. 1). 3.2. Relationship between LD50 and substructures It was obvious that different homologous series had different toxicities. However, examination of the LD50 values showed that the toxicity was not related to the number of substituted groups but the type of groups. In other words, compounds with mono-, di-, tri- and multi- substituted groups did not show a significant difference in toxicity. For example, the mean toxicity of monoalcohols (mean of log 1/LD50 = 1.55) was close to that of di-alcohols (mean of log LD50 = 1.34), and the mean toxicity of mono-acids is 1.71 whereas for di-acids it was 1.69. The same situation was found for all other homologous series/classes. To investigate the toxic contributions of functional groups, a substructural indication variable (Fi, used to describe presence and absence of certain functional group by values of 1 or 0) was given to mono-, di- or multi substituted groups for each homologous series (see Table S2 of Supplementary material) and the toxic contributions of the functional groups were calculated from a multi-linear regression analysis between log 1/LD50 of all the compounds and the indication variables in Table S2 by using Eq. (3). The toxic contribution values (Ci) of the functional groups, as well as the number of compounds containing the functional groups, were listed in Table 1.

Log1=LD50 ¼ 1:670 þ

X

Ci F i

ð3Þ

where, 1.670 is the toxic contribution of skeleton compounds (i.e. alkanes) obtained from intercept of the regression. The toxic contribution values (Ci) in Table 1 showed that cyclo compounds, alcohols, ethers, aldehydes, ketones, esters (including formates) and acids exhibited nearly the same toxicity as the skeleton compounds, alkanes. Almost no significant toxic contributions were observed for hydroxy, ether, carbonyl, ester and carboxyl substituted groups. The toxicities of unsaturated compounds, epoxy compounds, amides, ureas, acyl chlorides and guanidines were slightly higher than that of alkanes. Amines showed excess toxicity but no difference was found in the toxicities of primary amines (see compounds 988–1008 in Table S1 of Supplementary material), secondary amines (compounds 1029–1043) and tertiary amines (compounds 1053–1062). In the halogenated compound series, iodo compounds exhibited the strongest toxicity with a toxic contribution of 1.22 log units. Bromo and fluoro compounds showed nearly the same toxic contribution (0.901 and 0.900). They had a higher toxicity than chloro compounds (0.617) to rats. It is noteworthy that, like other multi- substituted functional groups, no additive effect was found in the multi- halogenated compounds. Therefore, the toxic contributions were given to the multi- halogenated compounds as the sequence of iodo (1.22), bromo/fluoro (0.900) or chloro (0.617). The average error (AE), average absolute error (AAE) and the root-mean squared error (RMSE) for all the 1588 compounds were listed in Table 1 (see No. 34 in Table 1). Compared with the average residual of LD50 of two data sources (AAE = 0.31, see Table S4 of Supplementary material), the AAE of log 1/LD50 (0.44) was slightly higher. However, this method leaded to the same accuracy as found in previous publication (Lagunin et al., 2011). 4. Discussion 4.1. Baseline toxicity to rats The toxic ratio (TR) of the calculated baseline or minimum toxicity (Tpred) and the experimentally determined value (Tobs) is a tool for probing a chemical’s mechanism of toxic action (Verhaar

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

101

Fig. 1. Plot of log 1/LD50 against log KOW for some classified compounds.

et al., 1992; Neuwoehner et al., 2010; Schramm et al., 2011). A TRvalue close to 1 indicates baseline toxicity. A TR-value significantly greater than 1 indicates excess toxicity due to the existence of a more specific mechanism of action. It should be borne in mind that the toxicity used in Eq. (4) is lethal concentration expressed in LC50 or LD50. It can be easily converted into logarithmic form expressed as log1/LC50 or log1/LD50 (see Eq. (5)).

TR ¼ Tpred ðbaselineÞ=Tobs

ð4Þ

log TR ¼ log 1=Tobs  log 1=Tpred ðbaselineÞ ¼ Residual

ð5Þ

In general, LC50 in fish for less inert chemicals (class 2, or polar narcotics) were lower than that predicted by baseline toxicity QSAR equations (Verhaar et al., 1992). These chemicals were usually characterized as possessing hydrogen bond donor acidicity, e.g. primary alkylamines, phenols and anilines. However, effective concentrations of reactive chemicals (class 3), as well as those for specifically acting chemicals (class 4) were between 10 and 104 times (log TR = 1–4) (Verhaar et al., 1992). Inspection of the toxic contributions of the functional groups in Table 1 revealed that alkanes and alcohols or ketones had nearly the same toxicity values to rats. If we used alkanes as a baseline reference (i.e. predicted baseline toxicity = 1.67) and log TR = 1 as reference threshold of excess toxicity, the reactive and specifically acting chemicals could be identified from the threshold. Examination of the TR values (i.e. toxic contributions in Table 1) of 33 functional groups listed in Table 1 showed that three classes (carbamoyl-oximes, hydrazines and iodo substituted compounds) could be identified as acting by reactive or specific mechanisms of action, with an average log TR greater than 1 (see Table 1). All the other 30 classes fell within the baseline or polar narcosis domain, with average log TR values lower than 1. It is worth mentioning that the primary alkylamines were classified as less inert chemicals and secondary and tertiary alkylamines were baseline narcosis for fish (Verhaar et al., 1992), but no difference was observed for toxicity to rats, with an average log TR = 0.40 (Table 1). It should be borne in mind that the reference threshold of excess toxicity used in the fish toxicity cannot completely separate reactive chemicals from baseline narcotics. The distribution of log TR showed that not all of the reactive compounds had log TR values greater than one (Verhaar et al., 1992). A similar situation was

observed for the toxicity to rats. Some reactive compounds did not show excess toxicity, with log TR values not greater than one (see Table 1). They could not be separated from the baseline compounds by using the threshold value. The above classification revealed that all the baseline compounds for fish toxicity, such as aliphatic alkane, alkene, alcohols, aldehydes, ketones, acids and chloro substituted compounds, could be identified as baseline narcosis for rat toxicity according to the threshold values in fish toxicity. However, some toxic contributions (or log TR) were quite high for some substituted functional groups, e.g. nitrates (0.89), carbonochloridate (0.71), fluoro (0.90) and bromo (0.90), as compared with baseline compounds. The threshold value of log TR = 1 may be too high to distinguish their toxic difference from baseline compounds (i.e. alkanes) for the toxicity to rats. In theory, baseline narcosis toxicity should be close to a constant (or vary in narrow range) or linearly related to a physicochemical property (this will be discussed below). No significant difference in toxicity values should be found from the statistics of the two baseline classes. To examine the toxicity difference between the classes in Table 1, t-test and variance analysis were used for the data analysis. The baseline compounds we selected were the aliphatic alkanes, alcohols, ethers, ketones and acids because they were generally accepted as exhibiting baseline narcosis to aquatic organisms and, more importantly, their average toxicity range to rats was less than 0.11 log units (see Table 1). If we used toxicity values of these aliphatic compounds as a baseline, then a t-test and an F-test could be performed to examine whether the means of other classes are significantly different from the mean of baseline compounds. Inspection of statistical results revealed that the toxicity of chemical classes with toxic contributions greater than 0.2 showed significant differences from baseline compounds, with p-values less than 0.01. It indicated that these classes had greater toxicity than baseline and should not be treated as baseline compounds. Because the experimental uncertainty (the average residual of log 1/LD50 from two data sources) was roughly 0.31 log units for rat toxicity based on the analysis for 141 compounds (see Table S4 of Supplementary material), it was suggested that a value of log TR < 0.30 (TR < 2) should be used as a threshold of excess toxicity for identifying baseline compounds. The same threshold value was used previously for rat toxicity models (Koleva et al., 2011).

102

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

If we used log TR = 0.30 as a threshold value for rat toxicity, the average toxicity of carbonates, amines, carbamates, carbamoyloximes, carbonochloridates, hydrazines, nitrates, nitriles and compounds containing nitro, nitroso, N@N/C@N, fluro, bromo, chloro and iodo functional groups (see Table 1) was greater than baseline narcosis, with an average log TR greater than 0.30. Most of these compounds showed greater toxicity than baseline for aquatic organisms (Verhaar et al., 1992; von der Ohe et al., 2005; Enoch et al., 2011). It is noteworthy that the compounds possessing three-membered heterocyclic rings, a,b-unsaturated aldehydes, ketones, esters and acids were classified as having a reactive mechanism of action for fish. It was also found that as a general trend of this group, log TR decreases with increasing alkyl group size. Inspection of the log TR values for these compounds showed that a few small molecules exhibited excess toxicity with TR values greater than 0.30, but that their average log TR values were less than 0.30. Most of the large molecules did not show excess toxicity. This reduction is probably due to the fact that the alkyl group substitution alters the electron density, and opposes the withdrawing effect of the carbonyl group, which impedes addition to the b-carbon atom of the olefin (Schramm et al., 2011). 4.2. Exposure routes and critical body residues (CBRs) Although most of baseline compounds identified in fish toxicity can also be identified as baseline compounds in rat toxicity, some differences could still be observed between the two species. For example, halogen substituted compounds showed excess toxicity for rat toxicity from the threshold of log TR = 0.3 (see Table 1) but they were identified as baseline compounds (excluding iodine) for fish toxicity from threshold of log TR = 1 (Verhaar et al., 1992). It was obvious that the different thresholds used for fish and rats were one of the reasons. Another reason was the difference of exposure routes. Without a match of the exposure routes, it is very difficult to say whether a chemical has the same or different toxic mechanisms among species. LD50 is defined as a single dose of a chemical that can kill 50% of animals within 24 h and LC50 is defined as the aqueous concentration of a chemical that can kill 50% of aquatic organisms within a certain time (i.e. 96 h-LC50, 48 h-LC50 and 24 h-LC50). Toxic effects expressed as LD50 and LC50 are not only dependent on the interaction ability between chemical and macromolecules at the target site, but also on the exposure routes of the chemical (Delistraty et al., 1998). LD50 and LC50 reflect the toxic effect of a chemical administrated (outside the body), but not the effect of the chemical inside the body. The routes of exposure are very different in rat and fish toxicity. In fish dosing is via the water phase through breathing, while in acute oral rat dosing is via gavage, which may affect the comparison between rats and fish. To overcome this problem, critical concentration in a whole body at the lethal effect (or critical body residue at the lethal effect, CBR) was used in this paper. If no consideration is given to the metabolism and elimination of a chemical, CBRs can be estimated from the percentage of absorption (%Abs.) and bioconcentration factor (BCF) for rats and fish by using Eqs. (6) and (7), respectively.

log 1=CBR ¼ log1=ðLD50 %Abs:Þ ¼ log1=LD50  logð%Abs:Þ

ð6Þ

log 1=CBR ¼ log1=ðLC50 BCFÞ ¼ log1=LC50  log BCF

ð7Þ

It is noteworthy that Eq. (1) used to predict bioconcentration factors is only suitable for compounds with log KOW in the range of 1–7 (Meylan et al., 1999). Significant prediction errors will be observed for the compounds with log KOW < 1 or >7, resulting in larger errors in calculating the CBRs from Eq. (7). Table S1 of Supplementary material listed the percentages of absorption by rat intestine administrated orally. The results

showed that 93% (1482/1588) of compounds exhibited absorption greater than 90% and 98% greater than 80%. 1.6% of compounds had absorption between 50% and 79% and only 4 compounds (4/1588 = 0.25%) had intestinal absorption lower than 50%. The compounds with relatively low absorption were polyols, polybasic acids and polybasic amines. This result was not surprising because all of the compounds were small molecules with few numbers of hydrogen bond donors and acceptors (Zhao et al., 2002, 2003). The above results indicated that LD50 values were roughly equal to the CBRs for almost all the compounds studied in this paper. On the other hand, CBRs are very different for fish toxicity. Fig. 2 is a plot of LC50 and CBRs against log KOW for 71 non-polar narcotic compounds (Table S3 of Supplementary material). As expected, log 1/LC50 was linearly related to the hydrophobicity, expressed as octanol/water partition coefficient. On the other hand, the CBRs were independent of the hydrophobicity (McCarty and Mackay, 1993) and they vary in a narrow range with a mean = 1.87 for baseline compounds (see Fig. 2). The non-polar narcotics used in the fish toxicity were alcohols, halogenated alkanes, ethers and ketones (Verhaar et al., 1992). These homologous series could also be found in the data set for rats. Inspection of data sets for the same homologous series both in fish and rat toxicity showed that CBR values were very close (Table 2). Quite close means of CBRs for the homologous series indicated that these chemicals may share the same mechanisms of toxic action between these two species. The same results have been observed on the investigation of exposure routes in rainbow trout and rats. The ratio of rat oral LD50/trout LC50 indicated that trout were more sensitive than rats. However, when trout and rat toxicity ratios (rat oral LD50/trout oral LC50) were matched on exposure routes, these ratios approached unity, which reflected similar toxicokinetics, despite taxonomic differences (Delistraty et al., 1998). 4.3. The factors that influence the classification of baseline compounds Although the above results show that baseline compounds were the same for the two species and share the same mode of action based on the analysis of CBRs, there were still several questions remaining. For example, the relationship between lethal concentration and toxic response is, or is close to, a thermodynamic process for fish toxicity. On the other hand, the relationship between administered dose and toxic response for rat toxicity is a kinetic process. The critical body residues (CBRs) calculated from log 1/(LD50 %Abs.) from Eq. (6) are not the real critical concentration in rats; CBRs should be less than calculated because absorption of a chemical is not an equilibrium process in rat body. The relationship between toxicity and hydrophobicity is linear for fish, but not for rats. Although the toxicity varies around a constant expressed in mol/kg for each homologous series, a parabolic relationship has been observed for toxicity to alcohols (Koleva et al., 2011). Acids were classified as baseline compounds, but they show very low toxicity to fish. Primary alkylamines exhibit greater toxicity than secondary and tertiary amines to fish but not to rats. Different toxicity sensitivity was observed between rats and fish. Metabolism of chemicals in both fish and rats can also affect identification of baseline to both species. A significant scatter was observed around the mean for each homologous series indicating that many factors can affect toxicity to rats (see Fig. 1). These factors are important and should be considered in the comparison of toxic classification between two species. 4.3.1. Effect of Absorption and bioconcentration on toxicity The relationships between observed toxicity and percentage of absorption or bioconcentration derived from Eqs. (6) and (7) can be expressed by the following equations:

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

103

Fig. 2. Plot of fish toxicity or critical body residue against hydrophobicity for baseline.

Table 2 Critical body residues (CBRs) and number of compounds (N) for fish and rats. Fish

Alcohols Ethers Ketones Amines Chloro alkanes Bromo alkanes

Rats

N

log1/CBR

N

log1/CBR

18 5 16 19 14 5

1.65 1.87 1.92 2.47 2.06 2.25

348 314 116 281 252 58

1.58 1.60 1.73 2.08 2.20 2.43

log 1=LD50 ¼ log 1=CBR þ logð%Abs:Þ

ð8Þ

log 1=LC50 ¼ log 1=CBR þ log BCF

ð9Þ

If we assumed that the toxicity of baseline chemicals inside body was equal to a constant, e.g. log 1/CBR = 2, the toxicity (i.e. log 1/LD50 and log 1/LC50) could be calculated from Eqs. (8) and (9), respectively. Fig. 3 is the theoretical relationship between toxicity and bioconcentration factor based on Eq. (9). Bioconcentration factors of organic compounds studied in this paper vary from 4.41 to 13.30 log units (see Table S1). One log unit increase in bioconcentration can lead to an increase of one log unit in toxicity. The effect of bioconcentration on toxicity is apparent and toxicity expressed as water concentrations known to cause toxic effects is linearly related to the log BCF for fish. In principle, definition of a quantifiable baseline effect should be based on the relationship between toxicity and log BCF, rather than on the relationship between toxicity and log KOW. This is because linear relationship between log BCF and Log KOW is not for all compounds. Discussion in detail can be found below. In contrast to the effect of bioconcentration on fish toxicity, the effect of absorption on rat toxicity was very different. This could be seen from the continuous trend of the toxicity against percentage of absorption based on Eq. (8) (Fig. 4). This plot revealed that absorption did not have a significant effect on rat toxicity if the absorption or bioavailability of compounds was in the range of 50–100%. The observed toxicity will decrease to 1.70 if absorption is 50%, 0.30 log units difference between the toxicity inside (log

1/CBR) and outside body (log 1/LD50), which was less than the experimental error we analyzed above and the threshold value (log TR = 0.3) defined for baseline compounds for rat toxicity. More than half a log unit difference in toxicity will be observed if the absorption was less than 30% (see Fig. 4). A significant difference (more than one log unit) will be observed if absorption of a compound was less than 10% in comparison with the compound with 100% absorption. The absorption values calculated from Eq. (2) showed that absorption was very high for the studied compounds except for a few very hydrophilic compounds (e.g. polyols, polybasic acids and amines). Therefore, absorption was supposed to have little effect on toxicity of the 1588 studied compounds. Because absorption through the gastrointestinal tract was very fast (only takes a few hours) and the average circulation time of blood was 1 min, the CBRs would not be greatly less than the concentrations inside body calculated from absorption. Therefore, it was reasonable to believe that calculated CBRs could roughly reflect the actual critical body residues for rat toxicity. 4.3.2. Effect of hydrophobicity and baseline threshold on toxicity Although the bioconcentration factor (BCF) is a crucial parameter for the study of baseline narcosis, the experimental BCF data is limited and not many data are available in the literature for the studied compounds. It was generally accepted that organic chemical hydrophobicity was the principal driving force of bioconcentration. Linear relationships between bioconcentration factor (BCF) and octanol/water partition coefficient (KOW) have been observed in literature (Meylan et al., 1999) and employed to estimate the bioconcentration factors. If we introduced Eq. (1) into Eq. (9), we could get a relationship between toxicity and hydrophobicity as log KOW.

log1=LC50 ¼ log 1=CBRs þ log BCF ¼ log1=CBRs þ 0:6598 log KOW  0:333

ð10Þ

Eq. (10) indicates that if CBRs vary in a narrow range (approach a constant, e.g. baseline narcosis), log 1/LC50 should be linearly related to hydrophobicity. If CBRs vary in a great range (e.g. reactive compounds), log 1/LC50 depends not only on the log KOW, but also on an extra descriptor used to describe the variation of CBRs to reactive compounds. The Eq. (10) also indicates that linear relationship

104

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

Fig. 3. Effect of bioconcentration on fish toxicity.

Fig. 4. Effect of absorption or bioavailability on rat toxicity.

between log BCF and log KOW is the fundamental basis of liner relationship between log 1/LC50 and log KOW. Although the linear relationship gave a fair approximation of the BCF, this method was only suitable for the compounds with log KOW in the range of 1–7. Furthermore, log KOW cannot well predict log BCF for all compounds even in the range of log KOW = 1–7. A set of correction factors (Fi) and rules were introduced in the log BCF–log KOW equation for compounds with specific functional groups to improve the accuracy of BCF predictions. Overall, this linear model, using log KOW and correction factors, can estimated the log BCF with a mean error of about ±0.5 log units (Pavan et al., 2006). The high estimation error for log BCF indicated that a high threshold had to be assigned for baseline compounds for fish toxicity. It also indicated the poor probability of classification among baseline, less inert and reactive chemicals based on the threshold of log TR = 1. This can be seen from the definition of TR.

The threshold is defined as TR = Tpred (baseline)/Tobs, which can be expressed as the following equation by introducing Eq. (10) into the definition:

log TR ¼ log 1=Tobs  log 1=Tpred ðbaselineÞ ¼ ðlog 1=CBR þ log BCFÞObs  ðlog 1=CBR þ log BCFÞpred ¼ Dlog 1=CBR þ Dlog BCF

ð11Þ

The classification of a baseline from reactive chemicals should be based on the difference between observed and predicted log 1/CBR values (Dlog 1/CBR), rather than the log TR because the log TR is closely related with the prediction accuracy of bioconcentration factor (see Eq. (11)). First, if a threshold of log TR = 1 was used for the discrimination of excess toxicity from narcotic effect and chemical log BCF was

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

well estimated by log KOW (i.e. the predicted log BCF was equal or close to the experimental log BCF), then log TR would be equal to Dlog 1/CBR. Chemicals with Dlog 1/CBR less than 1 would be classified as baseline narcosis. A value of one log unit used as a threshold indicates poor classification ability. The identified baseline compounds may not have the same TR. In other words, the threshold is too high to distinguish toxic differences by using TR, resulting in a difficulty in discriminating excess toxicity from narcotic effect. For example, fluoro, chloro and bromo compounds may have greater toxicity than baseline compounds, but they cannot be identified from the threshold in fish toxicity. Second, if the threshold log TR = 1 and the observed log BCF was 0.5 log units lower than the predicted log BCF, the chemicals with Dlog 1/CBR in the range of 1–1.5 should be classified as reactive chemicals, but in fact their log TR values were less than 1 based on the calculation of Eq. (11) and they were classified as having a narcotic effect. Third, if the threshold log TR = 1 and the observed log BCF was 0.5 log units higher than the predicted log BCF, then chemicals with Dlog 1/CBR in the range of 0.5–1 should be classified as baseline, but their log TR values were higher than 1 and they were classified as less inert or reactive compounds. There were many examples showing that experimental log BCF values were different from the predicted log BCF values, e.g. log BCF of benzene is 0.63, while the predicted value was 1.07 from EPI Suit program. The primary alkylamine, hexylamine, and the secondary alkylamine (dipropylamine) had nearly the same estimated log KOW (0.82 and 0.79) but with different values of experimental log KOW (2.06 and 1.67), which could result in a 0.26 log units difference in estimating log BCF from the experimental log KOW by Eq. (1) and 0.26 log units difference in predicted toxicity as well. It explains why primary alkylamines show greater toxicity than secondary and tertiary amines. The calculated CBRs for fish toxicity showed that the CBRs were not a constant for the baseline compounds (Table 2) either. The relationship between absorption and hydrophobicity was not straight forward in comparison with that between bioconcentration and hydrophobicity. Absorption is a kinetic parameter and octanol/water partition is a thermodynamic parameter. Although the percentage of absorption was related to hydrophobicity, the absorption determining factors were hydrogen donors and acceptors (Zhao et al., 2003). QSAR analysis showed that the relationship between absorption and log KOW was quite poor. Because absorption is very high for the studied compounds, the toxic effect expressed as LD50 should be roughly the same with the CBRs. The experimental uncertainty was low and a small baseline threshold value could be assigned for rat toxicity. Different toxic effects could be observed for small changes in chemical structure. This explained why chemicals showed higher toxicity sensitivity to rats than that to fish. 4.3.3. Effect of bioavailability, volatility and solubility on toxicity The effect of volatilization on toxicity could be monitored and the concentrations of the chemical being tested could be satisfactorily maintained for testing toxicity of a chemical to fish, but not to rats (OECD Guideline, 2008). Inspection of the Henry’s law constants calculated by HENRYWIN program (EPI Suite 4.0) for the studied compounds showed that volatility was quite high for some compounds (e.g. alkanes, alkenes and ethers). The toxicity can be affected by the volatility from exhalation. The alkyl monoalcohols with a few atoms were less toxic than those with 4–10 carbon atoms. These small volatile alcohols showed reduced toxicity due to the fact that a proportion of the chemical will be lost from the system via exhalation, thus affecting bioavailability (Ellison et al., 2008; Koleva et al., 2011). Solubility was also an important factor governing toxicity. Very often the dosage form does not contain the compounds as a solution or liquid, but rather as solid particles or in suspension. Because

105

solid particles cannot pass through membranes, a compound has to dissolve to be absorbed. The dissolved part of a compound was generally absorbed faster than the un-dissolved part, thus affecting the bioavailability and resulting in a decrease in toxicity to rats (Zhao et al., 2002). The amount of dissolved chemical was not only related to the solubility of the chemical, but also to the dosage used in the administration. In equilibrium, there was not enough fluid available in the gastrointestinal tract to dissolve the compounds studied in this paper. This can be checked from the dosage (i.e. LD50) and solubility ratio. Examination of the dose (LD50 in mg/ kg) revealed that 156 compounds (10%) were administrated over 10 g/kg, 216 compounds (14%) were administrated between 10 and 5 g/kg and 644 compounds (41%) were administrated between 5 and 1 g/kg. The highest dose was 389 g/kg, which accounts for 39% (389/1000) of body weight for a rat, more than one-third of body weight. The toxicity can be significantly different between LD50 and CBRs for the compounds with very low dosage/solubility ratios. If two compounds had the same CBRs, the compound with a lower solubility would have a lower observed toxicity than that with a higher solubility because of the difference in bioavailability. Unsaturated monoalcohols with more carbon atoms (e.g. number of carbon >10) show lower toxicity than the monoalcohols with 6–7 carbons (see Fig. 1), probably because the bioavailability was affected by lower solubility in water for these large molecules (Koleva et al., 2011). The same situation was found in bioconcentration for fish. A bioavailable fraction was often equated to the truly dissolved fraction. Low solubility for chemicals with log KOW > 5 could cause the dissolved fraction to be reduced (Mackay and Fraser, 2000). 4.3.4. Metabolism and ionization Metabolism may significantly affect the identification of baseline from reactive compounds. BCF is the ratio of a parent chemical concentration in an organism to the concentration in water and absorption is how much parent chemical gets into the blood stream. Metabolisation can not only affect the bioavailability of parent compounds and accuracy of measured BCF, but also the toxic effect of the compounds. Furthermore, metabolism may have greatly different effects on the toxicity to rats and fish. If the metabolites of a baseline compound are more toxic than their parent compound, significant toxic effect will be observed from baseline level. On the other hand, if the metabolites of a reactive compound are less toxic than their parent compound, this reactive compound will be identified as baseline compound because the baseline narcosis is the minimum toxicity that compounds exhibit. For example, alkanes were identified as baseline compounds to fish. No significant difference was observed for the fish toxicity between pentane and hexane. However, the difference of observed rat toxicity for these two compounds was very high with log 1/LD50 = 2.256 and 0.477, respectively. Hexane is much more toxic than pentane and heptane because after metabolisation it will result in 2,5-hexanedione which reacts with critical lysine residues in axonal proteins and cause neurotoxicity. It is well known that pH affects the toxicity of ionisable substances to fish and daphnia. The toxicity of phenols and benzoic acids to Daphnia magna decreases with increasing pH (Cronin et al., 2000). Although both ionized and unionized forms of a weak electrolyte were absorbable, the uptake of the unionized form was generally faster, resulting in a significant decrease of BCF for ionisable compounds, thus significantly affecting their toxicity to aquatic organisms. This can be seen from the toxicity of benzoic acids to fish. Their toxicity was very low in comparison with that of hydrophobic compounds. On the other hand, the toxicity of acids was very close to the toxicity of neural compounds for rats (see Table 1). The reason is the variable physiologic environment in the gastrointestinal tract. A pH value of 1–3 prevails in the stomach

106

J. He et al. / Regulatory Toxicology and Pharmacology 70 (2014) 98–106

owing to the secretion of hydrochloric acid. Luminal pH values increase along the intestine from 5 or 6 up to pH 8 in the lower small intestine and the ascending colon. The acids and bases can be strongly absorbable in the variable environments. 5. Conclusions The toxic contributions calculated for each functional group showed that average toxicity difference was less than 0.11 log units among chemical classes of alkanes, alcohols, ethers, acetones, esters and acids. These compounds can be classified as baseline compounds and be used to identify reactive compounds from baseline level from the toxic ratio (TR) in rat toxicity. Although aldehydes were classified as reactive compounds, they do not show great toxicity to rats. Absorption by rat intestine administrated orally was very fast and also very high for the small molecules. If no consideration is given to metabolism, the toxicity LD50 values nearly reflected the critical body residues (CBRs) of chemicals in rats and toxicity was independent of the hydrophobicity. On the other hand, fish toxicity was closely related to bioconcentration factors, which are linearly correlated to the hydrophobicity expressed as log KOW. Because of the high estimation error of bioconcentration calculated from log KOW, significantly wrong classification of modes of toxic action will be observed from TR for fish toxicity. This also resulted in toxicity sensitivity of fish being lower than rats. The definition of a quantifiable baseline effect for fish toxicity should be based on the relationship between toxicity and log BCF, rather than on the relationship between toxicity and hydrophobicity. The identification of modes of toxic action should be based on CBR-based TR, rather than on the LC50-based TR. There were many factors that could influence the toxic classification to rats and fish, such as metabolism, ionization, solubility and volatility. The CBRs calculated from bioconcentration factors may be different, or significantly different, from the actual CBRs in fish because most of compounds have not reached bioconcentration equilibrium between fish and water on the lethal concentration at the 96 h. Using TR will result in wrong classifications of modes of action if these factors are not considered. Acknowledgments This work is supported by the National Natural Science Foundation of China (21377022 and 21107012) and the Fundamental Research Funds for the Central Universities (12SSXT138). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.yrtph.2014. 06.019. References Cronin, M.T.D., Zhao, Y.H., Yu, R.L., 2000. PH-dependence and QSAR analysis of the toxicity of phenols and anilines to Daphnia magna. Environ. Toxicol. 15, 140– 148. Delistraty, D.A., Taylor, B., Anderson, R., 1998. Comparisons of acute toxicity of selected chemicals to rainbow trout and rats. Ecotoxicol. Environ. Saf. 39, 195– 200. Delistraty, D., 2000. Acute toxicity to rats and trout with a focus on inhalation and aquatic exposures. Ecotoxicol. Environ. Saf. 46, 225–233. Devillers, J., Devillers, H., 2009. Prediction of acute mammalian toxicity from QSARs and interspecies correlations. SAR QSAR Environ. Res. 20, 467–500. Ellison, C.M., Cronin, M.T.D., Madden, J.C., Schultz, T.W., 2008. Definition of the structural domain of the baseline non-polar narcosis model for Tetrahymena pyriformis. SAR QSAR Environ. Res. 19, 751–783.

Enoch, S.J., Ellison, C.M., Schultz, T.W., Cronin, M.T.D., 2011. A review of the electrophilic reaction chemistry involved in covalent protein binding relevant to toxicity. Crit. Rev. Toxicol. 41, 783–802. Enslein, K., Lander, T.R., Tomb, M.E., Craig, P.N., 1989. A predictive model for estimating rat oral LD50 values. Toxicol. Ind. Health 5, 261–387. Ferguson, J., 1939. The use of chemical potentials as indices of toxicity. Proc. R. Soc. London, B 127, 387–404. Jäckel, H., Klein, W., 1991. Prediction of mammalian toxicity by quantitative structure–activity relationships: aliphatic amines and anilines. Quant. Struct. Act. Relat. 10, 198–204. Katritzky, A.R., Tatham, D.B., Maran, U., 2001. Theoretical descriptors for the correlation of aquatic toxicity of environmental pollutants by quantitative structure–toxicity relationships. J. Chem. Inf. Comput. Sci. 41, 1162–1176. Koleva, Y.K., Cronin, M.T.D., Madden, J.C., Schwöbel, J.A.H., 2011. Modelling acute oral mammalian toxicity. 1. Definition of a quantifiable baseline effect. Toxicol. in Vitro 25, 1281–1293. Könemann, W.H., 1981. Quantitative structure-activity relationships in fish toxicity studies. 1. Relationship for 50 industrial pollutants. Toxicology 19, 209–221. Lagunin, A., Zakharov, A., Filimonov, D., Poroikov, V., 2011. QSAR modelling of rat acute toxicity on the basis of PASS prediction. Mol. Inf. 30, 241–250. Lipnick, R.L., 1991. Outliers: their origin and use in the classification of molecular mechanisms of toxicity. Sci. Total Environ. 109–110, 131–153. Lipnick, R.L., 1999. Correlative and mechanistic QSAR models in toxicology. SAR QSAR Environ. Res. 10, 239–248. Mackay, D., Fraser, A., 2000. Bioaccumulation of persistent organic chemicals: mechanisms and models. Environ. Pollut. 110, 375–391. McCarty, L.S., Mackay, D., Smith, A.D., Ozburn, G.W., Dixon, D.G., 1991. Interpreting aquatic toxicity QSARs: the significance of toxicant body residues at the pharmacologic endpoint. Sci. Total Environ. 109 (110), 515–525. McCarty, L.S., Mackay, D., 1993. Enhancing ecotoxicological modeling and assessment: body residues and modes of toxic action. Environ. Sci. Technol. 27, 1719–1728. Meador, J.P., Adams, W.J., Escher, B.I., McCarty, L.S., McElroy, A.E., Sappington, K.G., 2011. The Tissue residue approach for toxicity assessment: findings and critical reviews from a Society of Environmental Toxicology and Chemistry Pellston Workshop. Integr. Environ. Assess. Manag. 7, 2–6. Meylan, W.M., Howard, P.H., Boethling, R.S., Aronson, D., Printup, H., Gouchie, S., 1999. Improved method for estimating bioconcentration/bioaccumulation factor from octanol/water partition coefficient. Environ. Toxicol. Chem. 18, 664–672. Neuwoehner, J., Zilberman, T., Fenner, K., Escher, B.I., 2010. QSAR-analysis and mixture toxicity as diagnostic tools: Influence of degradation on the toxicity and mode of action of diuron in algae and daphnids. Aquat. Toxicol. 97, 58–67. Organisation for Economic Cooperation and Development (OECD) guideline for the testing of chemicals 425, 2008. On acute oral toxicity – up-and-down procedure, Paris, France. Pavan, M., Worth, A.P., Netzeva, T.I., 2006. Review of QSAR models for bioconcentration. European Commission Joint Research centre (EUR 22327 EN). Raevsky, O.A., Grigor’ev, V.Y., Weber, E.E., Dearden, J.C., 2008. Classification and quantification of the toxicity of chemicals to guppy, fathead minnow and rainbow trout: Part 1. Nonpolar narcosis mode of action. QSAR Comb. Sci. 27, 1274–1281. Raevsky, O.A., Grigor’ev, V.Y., Dearden, J.C., Weber, E.E., 2009. Classification and quantification of the toxicity of chemicals to guppy, fathead minnow, and rainbow trout. Part 2. Polar narcosis mode of action. QSAR Comb. Sci. 28, 163– 174. Sazonovas, A., Japertas, P., Didziapetris, R., 2010. Estimation of reliability of predictions and model applicability domain evaluation in the analysis of acute toxicity (LD50). SAR QSAR Environ. Res. 21, 127–148. Schramm, F., Müller, A., Hammer, H., Paschke, A., Schüürmann, G., 2011. Epoxide and thiirane toxicity in vitro with the ciliates Tetrahymena pyriformis: structural alerts indicating excess toxicity. Environ. Sci. Technol. 45, 5812–5819. Tsakovska, I., Lessigiarska, I., Netzeva, T., Worth, A.P., 2008. A mini review of mammalian toxicity (Q)SAR models. QSAR Comb. Sci. 27, 41–48. Veith, G.D., Petkova, E.P., Wallace, K.B., 2009. A baseline inhalation toxicity model for narcosis in mammals. SAR QSAR Environ. Res. 20, 567–578. Verhaar, H.J.M., Vanleeuwen, C.J., Hermens, J.L.M., 1992. Classifying environmentalpollutants. 1: structure–activity relationships for prediction of aquatic toxicity. Chemosphere 25, 471–491. Von der Ohe, P.C., Kühne, R., Ebert, R.-U., Altenburger, R., Liess, M., Schüürmann, G., 2005. Structural alerts – a new classification model to discriminate excess toxicity from narcotic effect levels of organic compounds in the acute Daphnid assay. Chem. Res. Toxicol. 18, 536–555. Wolf, W., Lieder, P.H., Walker, J.D., 2004. Application of QSARs: correlation of acute toxicity in the rat following oral or inhalation exposure. QSAR Comb. Sci. 23, 521–525. Zhao, Y.H., Abraham, M.H., Le, J., Hersey, A., Luscombe, C.N., Beck, G., Sherborne, B., Cooper, I., 2002. Rate-limited steps of human oral absorption and QSAR studies. Pharm. Res. 19, 1446–1457. Zhao, Y.H., Abraham, M.H., Hersey, A., Luscombe, C.N., 2003. Quantitative relationship between rat intestinal absorption and Abraham descriptors. Eur. J. Med. Chem. 38, 939–947. Zhu, H., Martin, T.M., Ye, L., Sedykh, A., Young, D.M., Tropsha, A., 2009. Quantitative structure–activity relationship modeling of rat acute toxicity by oral exposure. Chem. Res. Toxicol. 22, 1913–1921.

Investigation on baseline toxicity to rats based on aliphatic compounds and comparison with toxicity to fish: effect of exposure routes on toxicity.

The aim of this paper was to investigate baseline toxicity to rats and effect of exposure routes on toxicity in rats and fish. In this paper, 1588 ind...
648KB Sizes 0 Downloads 3 Views