Bioaccumulation is the net accumulation of a contaminant in, and for some special cases on, an organism from all sources including water, air, and solid phases in the environment 1. Bioaccumulation is a key part of chemical regulatory programs such as Japan's Chemical Substances Control Law (CSCL) 2 and Registration, Evaluation, Authorization and Restriction of Chemicals registration (REACH) 3. The most widely used test guideline for measuring bioaccumulation in fish is the Organization for Economic Cooperation and Development (OECD) test guideline 305 4, which is based on a bioconcentration factor (BCF; L/kg). A BCF is preferably calculated both as the ratio of concentrations in the fish and in water at steady state (BCFss) and as a kinetic bioconcentration factor (BCFk), which is estimated as the ratio of the assumed first-order rate constants for uptake and depuration 4. Researchers and the regulatory authorities have collected BCF data for a large number of chemicals. In recent years, Japanese authorities have obtained BCF data for more than 100 chemicals every year. These data are an important resource, and they have been made available to the international regulatory community, for example, to the Stockholm Convention on Persistent Organic Pollutants (http://chm.pops.int/Home/tabid/2121/mctl/ViewDetails/EventModID/871/EventID/230/xmid/6921/Default.aspx). The collection of this knowledge has led to the development of some alternative BCF estimation methods such as quantitative structure–activity relationships 5–8 and to the development of a tiered approach 9–11 to assess the bioaccumulation of chemicals. These approaches can be used to reduce the need for in vivo testing; however, methods estimating the BCF that require fewer fish and resources when performing in vivo testing are also needed 12.
The minimized aqueous exposure test is a new approach that was proposed by Springer et al. 13, who resampled the data from the test conducted according to the OECD test guideline 305 and recalculated the BCF (BCFkm) for 25 curves (test species are not specified) to show that acceptable estimates can be obtained by a minimized study design. The test will be performed under the same experimental conditions that are specified in the standard bioconcentration test but requires sampling of the test fish at only two points in time. The OECD test guideline 305 is currently in the process of being revised, and the draft test guideline 305 was approved by the Working Group of National Coordinators of the Test Guidelines Programme 14. The minimized aqueous exposure test was incorporated into the draft guideline with some modifications to the Springer et al. 13 test design (e.g., sampling of the test fish was increased to four times during a study, not two times). One of the sampling points is to be located near the expected midpoint of the depuration curve to ensure that adequate depuration will occur by the last day of the depuration phase. However, only limited reports have evaluated the validity of BCFkm as bioaccumulation endpoints for regulatory purposes, and many regulators are still faced with the need to fully understand the capabilities and limitations of the new test designs (minimized aqueous exposure test).
In the present study, we collected the BCF data (using common carp as test species) from Japan's CSCL database (http://www.safe.nite.go.jp/jcheck/english/top.action) with a wide range of chemical substances (298 curves from 155 chemicals) and resampled the data to simulate determination of BCFkm. We used this dataset to explore the validity of BCFkm and to discuss the appropriate operation of minimized aqueous exposure tests as bioaccumulation endpoints for regulatory purposes.
MATERIALS AND METHODS
Collection of the dataset
In accord with Japan's CSCL 2, bioconcentration tests have been conducted according to the OECD test guideline 305 4, and all of the tests have been conducted in accordance with the OECD Principles on Good Laboratory Practice 15. Common carp (Cyprinus carpio) have been used as test fish and exposed to test chemicals under flow-through conditions. We collected the BCF data for 155 chemicals (including isomers) from Japan's CSCL database. Bioconcentration tests have been conducted for each chemical at one or more concentrations; therefore, we collected a total of 298 curves.
For 121 of the 298 curves, BCFss was not reported in the original report because the variation of the last three chemical concentrations in the fish during the uptake phases was more than ± 20%. For these curves, we used the average value of the three successive analyses at the end of the uptake phase as the BCFss. We calculated BCFk according to the OECD test guideline 305 4, because only BCFss were reported in the original Japan's CSCL test report. For 27 of the 298 curves, we could not calculate the depuration rate constant, because a decrease of the chemical concentration in the test fish was not observed during the depuration phase. For 3 of the 298 curves, the concentration of the chemical in the test fish peaked rapidly and then declined throughout the remainder of the uptake phase. Like the minimized aqueous exposure test, the kinetic approach can be applied only when the uptake and depuration of chemicals follow first-order kinetics. Therefore, we excluded the aforementioned 30 curves from the dataset and used a total of 268 curves for the subsequent analysis.
Resampling and calculation
According to the report by Springer et al. 13, BCFkm is to be calculated by using the chemical concentrations in the test fish at only two sampling points: on the last day of the 28-d uptake phase and on the last day of the 14-d depuration phase. In the present study, most of the curves we collected did not include data from either the 28th day of the uptake phase or the 14th day of the depuration phase. We used the chemical concentrations in the test fish sampled on precisely the last day of the uptake and depuration phases for the calculation (e.g., on the last day of a 60-d uptake phase and the last day of a 10-d depuration phase). We calculated the depuration rate constant (k2m; 1/d) and the uptake rate constant (k1m; 1/d) for the minimized aqueous exposure tests with the following equations 
where Cf1 is the chemical concentration in the test fish (µg/kg) sampled on the last day of the uptake phase (t1; d), Cf2 is the chemical concentration in the test fish (µg/kg) sampled on the last day of the depuration phase (t2; d), and Cw is the average chemical concentration in the test water during the uptake phase (µg/L).
The BCFkm was calculated using the following equation 
The steady-state BCF from the minimized aqueous exposure test (minimized-BCFss) was calculated using the following equation
We also estimated the time required for the chemical concentration to reach 80% of the steady-state value (t80) and the depuration half-life of the chemicals (t50) by using the following equation: t80 = 1.6/k2m and t50 = 0.693/k2m4. We then calculated the ratio of the duration of the depuration phase (td) to t50, td:t50. All calculations and statistical analysis were carried out using Microsoft Office Excel 2003 SP3.
Analysis of the BCF data in the Japan's CSCL database
The BCF in our dataset relate to compounds with a wide variety of chemical structures. Frequency distributions of the log n-octanol–water partition coefficients (log KOW) and BCFss for the chemicals analyzed in the present study are shown in Figure 1A and B. Their log KOW values ranged from −0.55 to 10.17, and the BCFss values ranged from 280 to 40,000.
The relationship between BCFss and BCFk for the 268 curves analyzed in the present study is shown in Figure 2A. A highly linear relationship was present between the logarithms of BCFss and BCFk that could be expressed by Equation 5
The average value for the BCFss:BCFk ratio was 0.98, with a range from 0.17 to 1.32, the 5th and 95th percentiles being 0.64 and 1.19, respectively (Table 1). Theoretically, the BCFss and BCFk should be similar, provided that the duration of the uptake phase was sufficient to allow the chemical concentration in the test fish to reach steady state. For 20 of the 268 curves, the duration of the uptake phase in the original test was shorter than the t80 value, an indication that the chemical concentration in the test fish had not reached its steady state. Therefore, we used BCFss for 248 curves and BCFk for 20 curves in the subsequent analysis. Hereinafter, we have designated this dataset to be BCFfull.
Table 1. Summary of the calculation results
BCFss = steady-state bioconcentration factor derived from standard bioconcentration tests (BCF); BCFk = kinetic BCF derived from standard bioconcentration tests; BCFfull = better estimation of BCF derived from standard bioconcentration tests; BCFkm = kinetic BCF derived from minimized aqueous exposure tests; minimized-BCFss = steady-state BCF derived from minimized aqueous exposure tests.
Evaluation of the BCFkm for the bioaccumulation endpoints
The relationship between BCFfull and BCFkm for the 268 curves analyzed in the present study is shown in Figure 2B. A highly linear relationship was found between the logarithms of BCFfull and BCFkm that could be expressed by Equation 6
The average value of the BCFfull:BCFkm ratio was 1.04, with a range from 0.54 to 1.93, the 5th and 95th percentiles being 0.74 and 1.45, respectively (Table 1). The relationship between the BCFfull:BCFkm and td:t50 ratios is shown in Figure 3. For approximately 30% of the curves, the BCFfull:BCFkm ratio was outside its corresponding 5th and 95th percentile ranges when the td:t50 ratio was less than 1. Springer et al. 13 pointed out that the reliability of the BCFkm decreases when the depuration phase is shorter than the t50 value of the chemicals. These results clearly suggest that the depuration phase should be greater than the t50 values of the chemicals.
The relationship between the minimized-BCFss and BCFkm for the 268 curves analyzed in the present study is shown in Figure 2C. A highly linear relationship was seen between the logarithms of the minimized-BCFss and BCFkm that could be expressed by Equation 7
The average value of the minimized-BCFss:BCFkm ratio was 0.97, with a range from 0.29 to 1.77, the 5th and 95th percentiles being 0.74 and 1.00, respectively (Table 1). For 15 of the 268 curves, the minimized-BCFss:BCFkm ratio was less than the 5th percentile. For all of these curves, the duration of the uptake phase in the original test was shorter than the t80 value, an indication that the chemical concentration in the test fish had not reached its steady state. In such cases, the BCFkm is a more suitable bioaccumulation endpoint.
For 5 of the 268 curves, the minimized-BCFss:BCFkm ratio was greater than 1.32, which was the highest value of the BCFss:BCFk ratio. This might be an indication of excessive analytical or sampling variability; in such cases, the results of a standard bioconcentration test will be necessary to obtain robust BCF.
We collected the BCF data (298 curves from 155 chemicals, using common carp as test species) from Japan's CSCL database and resampled the data to simulate determination of BCFkm.
Combining our results and proposals of Springer et al. 13, the minimized aqueous exposure test may provide BCF estimates with an accuracy and precision comparable to those of a standard bioconcentration test when the following criteria are met: (1) uptake and depuration of the chemical in the test fish follow approximately first-order kinetics; (2) the duration of the depuration phase in the test is longer than the t50 value of the chemicals; and (3) the minimized-BCFss is not greatly higher than the BCFkm (e.g., the minimized-BCFss:BCFkm ratio < 1.32).
In the draft OECD test guideline 305 14, sampling of the test fish is to occur at four times: at the middle of the uptake phase, and at the initiation (i.e., end of uptake phase), middle, and termination of the depuration phase. One of the samplings is to be located near the expected midpoint of the depuration curve. The main purpose of sampling at the anticipated midpoint of the depuration phase is to monitor the depuration to ensure that adequate depuration will occur by the last day of the depuration phase. If there are three points during the depuration phase, k2m could be estimated using simple linear regression of the logarithms of the chemical concentration in the test fish versus time; researchers and regulatory authorities can confirm whether the uptake and the depuration of the chemical in the test fish follow approximately first-order kinetics.
In Japan's CSCL 2, the bioaccumulation criteria for a highly bioaccumulative substance is a BCF > 5,000, and in the REACH 3, the bioaccumulation criteria for bioaccumulative (B) substances is a BCF > 2,000, whereas that of very bioaccumulative (vB) substances is a BCF > 5,000. From our results (i.e., the 5th and 95th percentiles of BCFfull:BCFkm ratio being 0.74 and 1.45), BCFfull 2,000 corresponds to BCFkm 1,400 to 2,700, whereas BCFfull 5,000 corresponds to BCFkm 3,400 to 6,800. The use and interpretation of minimized aqueous exposure tests may depend on the regulatory framework; the standard bioconcentration test should be performed when the resulting BCFkm is in the region of regulatory concern. Additional data and analysis will help to ensure the appropriate operation of the minimized aqueous exposure test as bioaccumulation endpoints for regulatory purposes.
The present study was commissioned in part by the Ministry of Economy, Trade and Industry of Japan.