Elimination kinetic model for organic chemicals in earthworms

Science of the Total Environment 408 (2010) 3787–3793 Contents lists available at ScienceDirect Science of the Total Environment j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s c i t o t e n v Elimination kinetic model for organic chemicals in earthworms N. Dimitrova a,⁎, S. Dimitrov a, D. Georgieva a, C.A.M. Van Gestel b, P. Hankard c, D. Spurgeon c, H. Li d, O. Mekenyan a a Laboratory of Mathematical Chemistry, University “Prof. As. Zlatarov,” 8010 Bourgas, Bulgaria Department of Animal Ecology, Vrije Universiteit, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands Centre for Ecology and Hydrology, Maclean Building, Benson Lane, Crowmarsh Gifford, Wallingford, Oxon, OX10 8BB, United Kingdom d NERC Centre for Ecology and Hydrology, Lancaster Environment Centre, Bailrigg, Lancaster LA1 4AP, United Kingdom b c a r t i c l e i n f o Article history: Received 31 July 2009 Received in revised form 29 January 2010 Accepted 29 January 2010 Available online 25 February 2010 Keywords: QSAR Elimination PAHs Earthworms a b s t r a c t Mechanistic understanding of bioaccumulation in different organisms and environments should take into account the influence of organism and chemical depending factors on the uptake and elimination kinetics of chemicals. Lipophilicity, metabolism, sorption (bioavailability) and biodegradation of chemicals are among the important factors that may significantly affect the bioaccumulation process in soil organisms. This study attempts to model elimination kinetics of organic chemicals in earthworms by accounting for the effects of both chemical and biological properties, including metabolism. The modeling approach that has been developed is based on the concept for simulating metabolism used in the BCF base-line model developed for predicting bioaccumulation in fish. Metabolism was explicitly accounted for by making use of the TIMES engine for simulation of metabolism and a set of principal transformations. Kinetic characteristics of transformations were estimated on the basis of observed kinetics data for the elimination of organic chemicals from earthworms. © 2010 Elsevier B.V. All rights reserved. 1. Introduction Earthworms live in close contact with the soil and because they represent a major part of the diet of many vertebrate species they could be considered as representatives for the bioavailability of chemicals at contaminated sites (Jager et al., 2000, 2005). Compared to aquatic organisms such as fish, limited bioconcentration data are available for earthworms, which makes modeling of the chemical fate in these organisms a challenging task. Knowledge and experience gained from modeling bioaccumulation in other species could be of help in order to build an easy interpretable and mechanistically sound model for interaction between industrial pollutants and earthworms. Utilisation and adaptation of existing tools may, thus, provide a means of predicting the potential accumulation of organic chemicals in earthworms. Mechanistic understanding of bioaccumulation in different organisms should take into account both the influence of organism and also chemical depending factors on uptake and elimination kinetics. Lipophilicity, metabolism and biodegradation of chemicals in soil are among the factors that may significantly affect the bioaccumulation in soil organisms. Among the organism dependent factors, rate of metabolism is the most important parameter that needs to be considered. This study aims to develop a model that can be used to estimate the elimination kinetics of organic chemicals in earthworms by ⁎ Corresponding author. Laboratory of Mathematical Chemistry, University “Prof. As. Zlatarov,” “Yakimov” St. #1″ 8010 Bourgas, Bulgaria. Tel.: + 359 56 858388. E-mail address: [email protected] (N. Dimitrova). 0048-9697/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2010.01.064 accounting for their metabolic potential. The simulator of metabolism in earthworms was built hypothesizing that the principal transformations used by different organisms are similar and the main differences are in their rates of process or enzymes involved in the control of xenobiotic metabolisms (Brown et al., 2004). For example, comparison of metabolism of pyrene in invertebrates and vertebrates showed that the only one major phase I metabolite is 1-hydroxypyrene which is then further metabolised to pyrene-1-glucuronide, pyrene-1-glucoside, pyrene-1-sulfate, or pyrene-1-O-(6″-O-malonyl)glucoside (Giessing et al., 2003; Stroomberg et al., 2004). There is also some evidence that earthworms actually have a limited rate of metabolism compared to other species (Stroomberg et al., 2004). The rates of metabolism for different transformations were accounted for by calculating the probabilities of occurrence for each. In order to fit the kinetic curves of elimination, the probabilities were expressed as a function of time, assuming first order kinetics. 2. Materials and methods 2.1. Elimination kinetic data Elimination kinetic data for 28 structurally similar organic chemicals in earthworms (Eisenia fetida) were extracted from the literature (Belfroid et al., 1994, 1995, Jager et al., 2003, 2005 and Matscheko et al., 2002) and used to build the present model. These chemicals include 12 polycyclic aromatic hydrocarbons (PAHs), two N- and S-heterocyclic polyaromatic chemicals (PACs), two PAC- 3788 N. Dimitrova et al. / Science of the Total Environment 408 (2010) 3787–3793 quinones, seven polychlorinated biphenyls and five cyclic organochlorinated chemicals. The data set also includes one study on lindane elimination from the enchytraeid Enchytraeus albidus (Amorim et al., 2002). Chemical names and relative concentrations in earthworms at different days after transfer from a contaminated soil to a clean soil are presented in Table 1. For further refinement and testing of the model, an additional set of data for elimination of PAHs in earthworms was generated within the framework of the NoMiracle project. NoMiracle (contract no. 003956 under the Sixth Framework Programme) is an international project aiming to support the development and improvement of a coherent series of methodologies that will improve both human and environmental risk assessment procedure. The NoMiracle has 38 partners from 17 European countries competent in human toxicology and epidemiology, aquatic and terrestrial ecotoxicology, environmental chemistry/biochemistry, toxicogenomics, physics, mathematical modeling, geographic informatics, and socio-economic science. The data came from a study in which the epigeic soil dwelling worm Lumbricus rubellus was exposed to a field soil collected from an area adjacent to a factory producing carbon black by partial combustion and pyrolysis of oils. The soil for the study was collected from a site in south-west England (Ordnance Survey Grid Reference ST 539816). Soil from the top 5 cm of the soil profile was collected from 5 sample points representing the corners and centre of a 10 m × 10 m plot and mixed to produce a single batch. Mixed soil was sieved through a 10 mm sieve to remove large roots and stones and then 300 g placed into a series of plastic container. Soils were then wetted by addition of a further 15 ml of water in addition to the existing water present in the field soil and after 24 h a single adult L. rubellus obtained from a field collected population obtained from a commercial supplier (Neptune Ecology, Ipswich, UK) was added to each pot. PAH concentrations in the worms were measured at the start of the experiment. Containers were then sealed with a plastic lid. No food was added. In total 33 separate containers were set-up and left for 168 h at 12± 1 °C under a 16 hr light 8 hr dark regime to allow the added worm to accumulate PAHs from the test soils. At the end of this accumulation phase, the worms were removed from the polluted field soil and placed in a clean clay loam soil (Broughton Loams, Kettering, UK) amended with an additional 3% organic matter (CamBark, LBS Horticultural, Colne, UK). At the end of the 168 hr accumulation and at a further 10 timepoints of 4, 8, 16, 20, 26, 40, 56, 80, 112 and 164 h of the elimination, three worms were removed from the clean soil and snap frozen by immediate emersion in liquid N2 and then stored at −20 °C for later analysis of concentrations of 15 PAHs (naphthalene, acenaphthylene, acenaphthene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, benzo[k]fluoranthene, benzo[a]pyrene, ideno[1,2,3-cd]pyrene, dibenz[a,h]anthracene, benzo[g,h,i] perylene). Time points were spaced with sufficient sampling time at the start of the elimination but also some later sample times to allow the capture of fast and slow kinetic changes in tissue PAH concentrations. The analysis for PAH content of the obtained earthworm samples was conducted by gas chromatography mass spectrometry of dichloromethane extracts following the procedure outline in Hankard et al. (2004) and Meharg et al. (1998). Appropriate standards, spiked samples and blanks were included to ensure the robustness of the analysis in accordance with QA requirements. The data obtained from the field soil PAH kinetic study conducted for L. rubellus are presented in Table 2. These data were used as an external validation set providing information about the interlaboratory variation and model performance. A database with 433 documented biotransformation maps was collected from the literature and stored in electronic format in a metabolism database. The collected metabolic maps were used to develop a library of principal molecular transformations. It contains 43 Phase I and Phase II metabolic transformations. Some principal transformations and their formal kinetic constants are listed in Table 3. Table 1 Relative concentrations in earthworms of training chemicals at different times normalized to the concentration at the beginning of the elimination process (day 0) used to build the model. The elimination kinetic data for 28 structurally similar organic chemicals in earthworms (Eisenia fetida) were extracted from the literature (Amorim et al., 2002a, Belfroid et al., 1994, 1995, Jager et al., 2003, 2005 and Matscheko et al., 2002). # Compound Time of elimination [day] Relative concentration in worm 1 Dibenzo[a]anthracene 2 Dibenzothiophene 3 9,10-Anthraquinone 4 1,2-Benzoanthraquinone 5 Anthracene 6 Indeno[cd]pyrene 7 Benzo[ghi]perylene 8 Carbazole 9 Chrysene 10 Benzo[b]fluoranthene 11 Benzo[k]fluoranthene 12 Benzo[a]pyrene 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 1.00 0.91 0.91 0.85 0.89 0.82 1.00 0.97 0.93 0.90 0.88 0.88 1.00 0.84 0.80 0.78 0.75 0.74 1.00 0.97 0.95 0.94 0.93 0.93 1.00 0.82 0.61 0.48 0.38 0.38 1.00 0.91 0.84 0.62 0.67 0.54 1.00 0.80 0.82 0.66 0.68 0.60 1.00 0.94 0.90 0.90 0.88 0.88 1.00 0.12 0.14 0.02 0.01 0.01 1.00 0.61 0.67 0.45 0.41 0.28 1.00 0.61 0.67 0.45 0.41 0.28 1.00 0.67 0.55 0.37 3789 N. Dimitrova et al. / Science of the Total Environment 408 (2010) 3787–3793 Table 1 (continued) # Compound 13 Phenanthrene 14 Fluoranthene 15 Pyrene 16 Benzo[a]anthracene 17 Lindanea 18 Dieldrin 19 Hexachlorobenzene 20 Pentachlorobenzene 21 1,2,3,4-Tetrachlorobenzene 22 2,3′,4,4′,5Pentachlorobiphenyl 23 2,4,5,2′,4′,5′Hexachlorobiphenyl Table 1 (continued) Time of elimination [day] Relative concentration in worm 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 1 3 6 10 14 0 0.5 1 2 3 5 7 10 0 1 3 7 14 0 3 6 9 13 19 27 33 0 3 6 9 19 27 33 0 0.5 1 2 3 5 9 13 0 1 2 7 14 22 33 47 0 1 2 7 14 22 0.41 0.25 1.00 0.50 0.37 0.30 0.15 0.14 1.00 0.35 0.02 0.00 0.00 0.00 1.00 0.37 0.14 0.05 0.01 0.01 1.00 0.50 0.61 0.37 0.25 0.25 1 0.69 0.40 0.35 0.23 0.20 0.11 0.15 1 0.50 0.20 0.15 0.10 1 0.43 0.10 0.03 0.02 0.008 0.007 0.005 1 0.1 0.006 0.004 0.002 0.0003 0.0006 1 0.531 0.094 0.050 0.028 0.022 0.005 0.002 1 0.31 0.31 0.24 0.15 0.13 0.15 0.09 1 0.95 0.65 0.63 0.45 0.43 (continued on next page) # Compound 24 2,2′,3′,4,4′,5Hexachlorobiphenyl 25 2,2′,3,4,4′,5,5′ Heptachlorobiphenyl 26 2,4,5,2′,5′Pentachlorobiphenyl 27 2,3,3′,4,4′,5Hexachlorobiphenyl 28 2,4,5,3′,4′,5′Hexachlorobiphenyl a Time of elimination [day] Relative concentration in worm 33 47 0 1 2 7 14 22 33 47 0 1 2 7 14 22 33 47 0 1 2 7 14 22 33 47 0 1 2 7 14 22 33 47 0 1 2 7 14 22 33 47 0.18 0.25 1 1 0.74 0.69 0.49 0.46 0..31 0.27 1 1.1 0.78 0.78 0.67 0.72 0.44 0.72 1 0.93 0.68 0.64 0.36 0.15 0.14 0.11 1 0.60 0.45 0.81 0.42 0.57 0.33 0.30 1 0.69 0.63 0.65 0.44 0.44 0.39 0.39 Enchytraeus albidus. 2.2. Mathematical formalism of the model Within any study that examines the time-dependent pattern of change in body concentration of chemicals in organisms previously exposed, but currently maintained in a clean environment, the decrease of body concentrations may be due to different processes. • Elimination driven by the chemical lipophilicity and concentration gradient, • conversion of the compound into metabolites by biotransformation, and • growth, leading to dilution of the chemical concentration in the organism. Because there was no sufficient experimental information in either the literature or experimental study to account for the effect of growth on body concentrations, only the two first processes were accounted for within the developed model. The elimination kinetics of chemicals were modeled with a first order kinetics model: ½A Š = exp ð−km t Þ exp ½ða + bX Þt Š ½AŠ0 where: [A]0 [A] initial concentration of chemical A, mol/l concentration of chemical A at time t, mol/l ð1Þ 3790 N. Dimitrova et al. / Science of the Total Environment 408 (2010) 3787–3793 Table 2 Data on the elimination kinetics of polycyclic aromatic hydrocarbons from Lumbricus rubellus generated within the framework of the NoMiracle project and used as an external validation set for developing the model. # Compound Time of elimination [hours] Concentration in worm [ng/g wet weight] Relative concentration in worm 1 Phenanthrene 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 269 243 55.4 69.1 50.2 58.7 53.8 41.1 40.8 69.2 32.3 78.8 84.0 49.5 24.7 24.2 32.8 21.6 38.1 28.8 59.2 11.4 282 279 141 191 101 133 97.2 98.5 116 180 126 411 322 156 196 107 127 99.0 100 120 159 130 152 97.5 40.9 81.6 43.7 56.4 30.4 38.8 49.3 211 90.0 282 211 81.5 136 75.7 95.7 58.7 77.5 91.5 225 162 197 149 124 187 102 1 0.90 0.20 0.26 0.19 0.22 0.20 0.15 0.15 0.26 0.12 1 1.07 0.63 0.31 0.31 0.42 0.27 0.49 0.37 0.75 0.15 1 0.99 0.50 0.68 0.36 0.47 0.34 0.35 0.41 0.64 0.45 1 0.78 0.38 0.48 0.26 0.31 0.24 0.24 0.29 0.39 0.32 1 0.64 0.27 0.54 0.29 0.37 0.20 0.26 0.32 1.39 0.59 1 0.75 0.29 0.48 0.27 0.34 0.21 0.28 0.32 0.80 0.57 1 0.76 0.62 0.94 0.51 2 3 4 5 6 7 Anthracene Fluoranthene Pyrene Benz[a]anthracene Chrysene Benzo[k]fluoranthene Table 2 (continued) # Compound 8 Benzo[a]pyrene 9 Ideno[1,2,3-cd]pyrene 10 Dibenz[a,h]anthracene 11 Benzo[g,h,i]perylene km X Time of elimination [hours] Concentration in worm [ng/g wet weight] Relative concentration in worm 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 0 4 8 16 20 26 40 56 80 112 164 134 71.2 105 103 174 177 77.3 82.8 52.0 119 22.2 62.6 28.4 92.5 80.0 177 197 159 6.85 0 10.9 5.58 0 0 0 7.5 61.6 21.1 51.3 198 25.4 119 85.2 0 19.4 29.3 122 95.6 101 85.7 110. 38.4 59.0 23.6 41.2 16.7 19.7 31.1 77.3 55.6 0.68 0.36 0.53 0.52 0.88 0.90 1 1.07 0.67 1.53 0.93 0.81 0.37 1.19 1.04 2.29 2.55 1 0.043 0 0.068 0.035 0 0 0 0.047 0.39 0.13 1 3.86 0.50 2.32 1.66 0 0.38 0.57 2.36 1.87 1.98 1 1.28 0.45 0.69 0.28 0.48 0.19 0.23 0.36 0.90 0.65 formal kinetic constants of metabolism, day− 1 explanatory variables accounting for other mitigating factors such as hydrophobicity (log KOW), molecular size (maximum diameter averaged across all energetically stable conformers, Dmax.aver) and water solubility (SW). The simulation of metabolism is based on the Tissue Metabolism Simulator (TIMES) engine developed in the Laboratory of Mathematical Chemistry of the University “Prof. As. Zlatarov,” Bourgas, Bulgaria, for simulation of catabolic and metabolic transformations (Jaworska et al., 2002 and Mekenyan et al., 2004). The core part of the metabolism simulator is a set of hierarchically ordered principal molecular transformations. Mathematical formalism, characteristics and hierarchy of transformations, as well as the rules for their application reflect the specificity of the modeled phenomenon. For example, in the recently developed model for predicting bioconcentration in fish (Dimitrov et al., 2005a), one of its important components is the simulator of fish metabolism. Because experimentally determined bioconcentration generally is related to the parent chemicals, the simulation of 3791 N. Dimitrova et al. / Science of the Total Environment 408 (2010) 3787–3793 Table 3 Representative list of principal transformations used to simulate metabolism in earthworms. ki, d− 1 # Type of reaction Transformation 2 Phase I 3 Phase I 28 Phase I 31 Phase II 32 Phase II Mercapturic acid conjugation 0.047 40 Phase I Benzene C-hydroxylation 0.83 Arhene oxidation 0.0070 Arhene oxidation 0.0092 Dehalogenation 0.056 Glutathione conjugation 0.0025 3792 N. Dimitrova et al. / Science of the Total Environment 408 (2010) 3787–3793 metabolism was focused predominantly on the parent chemicals. The same approach was applied in this study for the development of a model for elimination of organic chemicals from earthworms. According to the probabilistic approach, metabolic transformations are characterized by probabilities of their occurrence. In order to reproduce kinetic data these probabilities were expressed as a function of time assuming first order kinetics (Dimitrov et al., 2007): Pi = ð1− expð−ki t ÞÞ ð2Þ ki = − ln ð1−Pi Þ = t ð3Þ where ki day− 1, is a surrogate of the first order kinetic constant of ith transformation. The formal kinetic constant km of the metabolism of a chemical into different products is determined as the sum of kinetic constants of the transformations used in the metabolic process: I km = ∑ ki ð4Þ i=1 The non-linear least square method (Marquardt, 1963) was used to estimate the model parameters a, b and ki by minimizing the sum of squares of residuals: Obs min SSR = ∑ ∑ a;b;ki j t Calc Ajt Ajt − Aj0 Aj0 !2 ð5Þ where j and t are indexes corresponding to different chemicals and time. Depending on the use of various explanatory variables X (see Eq. (1)) different models were obtained based on the fit with the experimental data. Model quality was estimated on the basis of the sum of squares of residuals (SSRs), coefficients of determination (R2) and variances (s2). 2.3. Model applicability domain A stepwise approach for determining the model applicability domain has been proposed recently (Dimitrov et al., 2005b). It accounts for the variation of parameters that could affect the bioavalability of chemicals (general requirements), structural similarity (structural domain), mechanistic background of the model (mechanistic domain), and performance of the simulator of metabolism (metabolism domain). The first two steps of the model applicability domain have been accounted for in the present kinetic model: domain of general parametric requirements and structural domain. The domain of general parametric requirements includes the range of variation of hydrophobicity (log KOW), molecular weight (MW) and water solubility (SW) of the chemicals in the training data set. Chemicals with MW from 167 to 395 Da, log KOW in the range of 3.0 to 8. 0 and SW in the range of 1.8·10− 4 to 21 mg/l, respectively were assigned to belong to the domain. EPI Suite Estimation Software (http://www.syrres.com/) was used to calculate log KOW and SW. The structural component of the model domain is based on the structural similarity between chemicals in the training set and the target chemical. The atom-centered fragments (ACFs) accounting for the first neighbors, hybridization and attached H-atoms were used as characteristics of the chemical structure. If atoms from an aromatic ring are first neighbors then the whole ring is added to the ACFs; the same holds for the sequences of three heteroatoms, which are also considered as first neighbors if one of their atoms is among the first neighbors. ACFs extracted from the training set of chemicals were used to define the structural domain of the model. Each target chemical is also partitioned to ACFs. A chemical is classified to belong to the structural domain if all its ACFs are found in the set of ACFs extracted Table 4 Fitted kinetic models, explanatory variables used and corresponding statistics. Model Explanatory variables SSR R2 s2 16 0.00 0.14 1. ½AŠ ½AŠ0 = exp ½ða + b logKOW Þt Š log KOW 2. ½AŠ ½AŠ0 = exp ð−km t Þ km 3.4 0.78 0.03 3. ½AŠ ½AŠ0 = exp ð−km t Þ exp ½ða + b logKOW Þt Š km, log KOW 3.0 0.81 0.03 4. ½AŠ ½AŠ0 = exp ð−km t Þ exp ½ða + bDmax:aver Þt Š km, Dmax.aver 3.0 0.81 0.02 5. ½AŠ ½AŠ0 = exp ð−km t Þ exp ½ða + bSW Þt Š km, SW 3.0 0.81 0.02 from the training chemicals. Both domains (general requirements and structural domain) determine the total model domain. 3. Results and discussion The statistical characteristics of model (1) depending on included explanatory variables are presented in Table 4. As can be seen from the table, hydrophobicity (log KOW) as a single model variable was unable to reproduce simultaneously the kinetics data for all chemicals. However, good statistical fits with this model (not shown in the table) were obtained when each kinetic curve was fitted separately to the available data. This is an indication that log KOW is not a universal parameter for the explanation of chemicals' elimination from organisms and the fits obtained for the separate curve reflect the implementation of chemical specificity in the adjusted model parameters a and b. Model fits demonstrate that metabolism appears to be the most significant factor controlling elimination kinetics (Table 4). Combining metabolism with log KOW, Dmax.aver or SW does not improve statistics significantly. The range of variation of these parameters within the training set cannot be ignored and potential explanation for their non-significance is that for the studied chemicals metabolism dominates over the other factors. From a mechanistic point of view the role of hydrophobicity was expected to be more substantial for non-metabolizing chemicals. In Fig. 1 the concentrations fitted and predicted by models (1) and (3) from Table 4 are presented. The predictions by model (1) are based on log KOW only, whereas the simulated metabolism is added in model (2). Fig. 1. Parity plot of observed and predicted relative concentrations in earthworms: ○ — model (1) from Table 4 and ● — model (3) from Table 4. N. Dimitrova et al. / Science of the Total Environment 408 (2010) 3787–3793 3793 factor for the elimination of studied chemicals. This is even though there are some suggestions from the literature that the rates of metabolism of some organic chemicals may be comparatively low compared to some other groups of organisms (Stroomberg et al., 2004). Within the studied chemical space, the addition of other factors in the model practically did not affect its quality. The significance of these factors is expected to be pronounced when the training set is expanded. Based on 15 additional chemicals tested within the framework of the NoMiracle project and using L. rubellus as the test species it was showed that model predictions were comparable with the variation of experimental data. Due to the small number of the training set chemicals the applicability domain of the current elimination model is relatively restricted. Thus, further toxicokinetic data relating to the elimination of organic chemicals by earthworms are needed to support the development of more comprehensive prediction of organic chemical accumulating and metabolism such has been possible in studies with fish (Dimitrov et al., 2005a). Acknowledgement Fig. 2. Interlaboratory variation of data and model predictions for Chrysene elimination from earthworms: ● — Eisenia fetida data from Matscheko et al., 2002, ▲ — Lumbricus rubellus data from the NoMiracle project, and — model (3) from Table 4. This work was partially supported by the international NoMiracle contract no. 003956 under the Sixth Framework Programme (FP6). References The performance of the model that was developed based on available literature data was investigated by assessing predictive power against an independent data set that was not used in model development. Within the framework of the NoMiracle project the elimination of 15 PAHs was studied in L. rubellus that had been previously exposed (Table 2). Four of the new tested chemicals (acenaphthene; fluorene; naphthalene and acenaphthylene) were outside the model domain mainly due to the incomplete metabolic information implemented in the current simulator (lack of molecular transformation to implement all biotransformations needed for the elimination of the chemicals). For the remaining 11 chemicals experimental data were already found in literature and used in the model building. This provided an opportunity to evaluate data variation. Direct comparison of data was impossible because the concentrations reported in the literature for E. fetida and these obtained for L. rubellus within the project were measured at different time intervals from the beginning of the elimination phase. However, graphical comparison of kinetic data showed that in some cases discrepancy could be greater than 50%. As an illustration of the interlaboratory variation of data and model predictions in Fig. 2 observed and predicted concentration for Chrysene are shown. Except for two outliers good interlaboratory and model compatibility is demonstrated for the whole duration of the studied elimination period. Similar or worse results were found for the remaining chemicals tested in the project frame. Because the model predictions are within the variation of experimental data it could be assumed that the model adequately predicts the elimination concentration of chemicals falling in its applicability domain. 4. Conclusion A kinetic model for the elimination of polycyclic aromatic chemicals from earthworms (E. fetida) was developed in this study. The model includes a simulator of metabolism containing 43 Phase I and Phase II principal metabolic transformations. Model building was based on the assumption of first order kinetics for each transformation. The mathematical formalism allows other properties, such as molecular size, water solubility and hydrophobicity, to be also accounted for in the model. Experimental kinetic data for the elimination of 28 PACs collected from the literature were used to determine the model parameters. Results showed that metabolism is the most important View publication stats Amorim MJ, Sousa JP, Nogueira AJA, Soares AMVM. Bioavailability and toxicokinetics of 14C-Lindane (-HCH) in the enchytraeid Enchytraeus albidus in two soil types: the aging effect. Arch Environ Contam Toxicol 2002;43:221–8. Belfroid A, Sikkenk M, Seinen W, van Gestel K, Hermens J. The toxicokinetic behaviour of chlorbenzenes in earthworms (Eisenia andrei) experiments in soil. Environ Toxicol Chem 1994;13:93–9. Belfroid A, van den Berg M, Seinen W, Hermens J, van Gestel K. Uptake, bioavailability and elimination of hydrophobic compounds in earthworms (Eisenia andrei) in field-contaminated soil. Environ Toxicol Chem 1995;14:605–12. Brown PJ, Long SM, Spurgeon DJ, Svendsen C, Hankard PK. Toxicological and biochemical responses of the earthworm Lumbricus rubellus to pyrene, a noncarcinogenic polycyclic aromatic hydrocarbon. Chemosphere 2004;57:1675–81. Dimitrov S, Dimitrova N, Parkerton T, Comber M, Bonnell M, Mekenyan O. Base line model for identifying the bioaccumulation potential of chemicals. SAR QSAR Environ Res 2005a;16:531–54. Dimitrov S, Dimitrova G, Pavlov T, Dimitrova N, Patlewicz G, Niemela J, et al. A stepwise approach for defining applicability domain of SAR and QSAR models. J Chem Inf Model 2005b;45:839–49. Dimitrov S, Pavlov T, Nedelcheva D, Reuschenbach P, Silvani M, Bias R, et al. A kinetic model for predicting biodegradation. SAR QSAR Environ Res 2007;18:443–57. Giessing AMB, Mayer LM, Forbes TL. 1-Hydroxypyrene glucuronide as the major aqueous pyrene metabolite in tissue and gut fluid from the marine deposit-feeding polychaete Nereis diversicolor. Environ Toxicol Chem 2003;22:1107–14. Hankard PK, Svendsen C, Wright J, Wienberg C, Fishwick SK, Spurgeon DJ. Weeks JM. Biological assessment of contaminated land using earthworm biomarkers in support of chemical analysis. Sci Total Environ 2004;330:9-20. Jager T, Sànchez FAA, Muijs B, van der Velde EG, Posthuma L. Toxicokinetics of polycyclic aromatic hydrocarbons in Eisenia andrei (Oligochaeta) using spiked soil. Environ Toxicol Chem 2000;19:953–61. Jager T, Fleuren RHLJ, Hogendoorn EA, de Korte G. Elucidating the routes of exposure for organic chemicals in the earthworm, Eisenia andrei (Oligochaeta). Environ Sci Technol 2003;37:3399–404. Jager T, van der Wal L, Fleuren RHLJ, Barendregt A, Hermens JLM. Bioaccumulation of organic chemicals in contaminated soils: evaluation of bioassays with earthworms. Environ Sci Technol 2005;39:293–8. Jaworska J, Dimitrov S, Nikolova N, Mekenyan O. Probabilistic assessment of biodegradability based on metabolic pathways: CATABOL system. SAR QSAR Environ Res 2002;13:307–23. Marquardt DW. An algorithm for least-squares estimation of nonlinear parameters. J Soc Ind Appl Math 1963;11:431–41. Matscheko N, Lundstedt S, Svensson L, Harju M, Tysklind M. Accumulation and elimination of 16 polycyclic aromatic compounds in the earthworm (Eisenia fetida). Environ Toxicol Chem 2002;21:1724–9. Meharg AA, Wright J, Dyke H, Osborn D. Polycyclic aromatic hydrocarbon (PAH) dispersion and deposition to vegetation and soil following a large scale chemical fire. Environ Pollut 1998;99:29–36. Mekenyan O, Dimitrov S, Pavlov T, Veith G. A systematic approach to simulating metabolism in computational toxicology I. The TIMES heuristic modeling framework. Curr Pharm Design 2004;10:1-13. Stroomberg GJ, Zappey H, Steen RJCA, van Gestel CAM, Ariese F, Velthorst NH, et al. PAH biotransformation in terrestrial invertebrates — a new phase II metabolite in isopods and springtails. Part CComp Biochem Physiol 2004;138:129–37.