Modeling the Bioconcentration of Organic Compounds by Fish: A Novel Approach HENK J. M. VERHAAR,* JOOST DE JONGH, AND JOOP L. M. HERMENS Research Institute of Toxicology, Utrecht University, P.O. Box 80.176, NL-3508 TD Utrecht, The Netherlands Bioconcentration is generally assumed to be similar to the partitioning of compounds between an aqueous and lipoid phase. In the case of bioconcentration lipoid phase is thought to consist of the lipid and lipoid tissues of an organism, e.g. a fish. However, it is frequently observed that for highly lipophilic compounds, defined as compounds with a high octanol -water partition coefficient, or log K ow , that simple view seems to break down. At low and intermediate log K ow there seems to be a linear relationship between bioconcentration, expressed as log BCF, and log K ow . At high log K ow deviations are seen, generally indicating that for these compounds BCF increases less than proportionally with increasing log K ow . Many different explanations and accompanying correlation models have been proposed to date. We present here a model that is capable of explaining the different shapes of the BCF/ K ow relationships empirically observed, based on the effect of the presence of tiny amounts of different types of organic material in the aqueous phase. The actual shape of the nonlinear curve simply depends on the amount of organic material in the water, the difference in “ affinity” for hydrophobic compounds of the organic material in water and lipid in fish, and to a lesser extent the amount of lipid in fish. The model presented also fits a large dataset of experimental BCF values with fitted parameter values that are consistent with the biology and physical chemistry of the described system. Introduction Bioconcentration of xenobiotic organic chemicals, such as pollutants, from water into fish, or other aquatic biota, is a veryimportant process in the chain ofevents leadingto toxic effects of many such compounds in the environment. Furthermore, bioconcentration kinetics is one of the pro- cesses that govern the function of LC50-vs-time behavior (1-4). Numerous authors have in the past observed that bio- concentration factors (log BCF) show a positive linear correlation with the log Kow of numerous compounds (see e.g. refs 5 or 6).Ofcourse,as a first order approximation,this is only to be expected, since the assumption is that the more hydrophobic a compound is, the more it will eventually end up in lipoid tissues in aquatic organisms. It has also been observed frequently that at the high log Kow end ofthe range (see e.g. ref 7) as well as at the very low end (8), the linear correlation ofBCFwith log Kow seemstobreakdown,resulting in a constant BCF at the hydrophilic end and in a leveling off or even decreasing BCF at the hydrophobic end of the range. A similar effect has been observed in numerous quantitative structure -activity relationships (QSAR) for several pharmacological endpoints, in which cases authors suggested using bilinear or even parabolic functions of log Kow to model endpoints. It has to be noted here that in pharmacology the fact that effects are generally recorded at a single restricted timepoint after administering a single, or sometimes multiple discrete, dose(s) is for a large part responsible for these observations (9). In studies ofthe correlation ofBCFwith log Kow anumber of explanations have been suggested to explain the above mentioned nonlinear relationships (5, 10-15).Explanations range from simple considerations of nonequilibrium for highly lipophilic compounds to size limitations (13), sug- gesting that an upper limit exists on a molecule’s effective diameter,abovewhich nomoleculeisabletopassmembrane pores, and to increased (induced) biodegradation or biotrans- formation. Some authors have indicated that the decreased bioavailabilityofhighlylipophiliccompoundsmightexplain this behavior (15, 16). For example Gobas et al. (15) have shown that a reduction in bioavailabilitymayresult in a strong underestimation of BCF. Here we report on a more general mathematical model that iscapable ofexplainingthisBCF/log Kow behaviorbased on the assumption of a reduced bioavailability. Background and Theory Recently De Jongh and co-workers (17) described a model that estimates the tissue-to-blood partition coefficient for organic chemicals, for use in PB-PKmodeling, from log Kow, depending largely on the water and lipid content or the respective compartments. Following the success of that particular model, we realized that the same description that was used in that study to describe the tissue and blood compartment sviz. two compartments consisting of water and comparable lipids in nonequal ratios and separated by a membrane that is permeable to organic compounds scan also be used to more accurately describe the water/fish system. The primary difference between our present description of the fish/water system and the system that is implicit in most approaches to estimating BCF from compound hy- drophobicity is the explicit inclusion of a “lipidlike” sub- compartment in the aqueous compartment. It is of course unthinkable that any real experimental setup in which fish are introduced in an aqueous environment could be de- scribed as consistingoffish in a pure aqueous compartment. Even if the water were pure to start with, the introduction of the fish would “instantaneously” provide an amount of dissolved organic matter to the aqueous phase, e.g. origi- nating from their mucous layer or from feces. The approach that we will be presenting is capable of explaining the observed BCF/log K ow behavior even with minute amounts of organic matter in the aqueous compartment. The basic assumptions here is that in any BCF system under consid- eration there is a “dissolved”organic carbon or “lipid”fraction in the aqueousphase,however small,and that given itsnature and especially amount it is experimentally impossible to determine the truly dissolved amount of compound using state-of-the-art procedures.Thissmallamount willbe shown subsequently, to make a huge difference in the observed BCF. *Correspondingauthorfax: +31-70-4260001;e-mail: verhaar.h@ oag.nl.Currentaddress: OpdenKamp Adviesgroep,Koninginnegracht 23, 2514 AB The Hague, The Netherlands. Environ. Sci. Technol. 1999, 33, 4069-4072 10.1021/es980709u CCC: $18.00  1999 American Chemical Society VOL. 33, NO. 22, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4069 Published on Web 10/07/1999