35 Appendix 5, Regulation (EU) 2017/735

In forcesince 14 Feb 2017
1.

Introduction

2.

Prediction of the duration of the uptake phase

3.

Prediction of the duration of the depuration phase

4.

Sequential method: determination of depuration (loss) rate constant k2

5.

Sequential method: determination of uptake rate constant k1 (aqueous exposure method only)

6.

Simultaneous method for calculation of uptake and depuration (loss) rate constants (aqueous exposure method only)

7.

Growth dilution correction for kinetic BCF and BMF

8.

Lipid normalisation to 5 % lipid content (aqueous exposure method only)

The general fish aquatic bioaccumulation model can be described in terms of uptake and loss processes, ignoring uptake with food. The differential equation (dCf/dt) describing the rate of change in fish concentration (mg·kg– 1·day– 1) is given by (1):

dC f

dt

k 1 C w k 2 k g k m k e C f | [Equation A5.1]

Where

k1First order rate constant for uptake into fish (l·kg– 1·day– 1).

k2First order rate constant for depuration from fish (day– 1).

kgFirst order rate constant for fish growth (‘growth dilution’) (day– 1)

kmFirst order rate constant for metabolic transformation (day– 1)

keFirst order rate constant for faecal egestion (day– 1)

CwConcentration in water (mg·l– 1).

CfConcentration in fish (mg·kg– 1 wet weight).

For bioaccumulating substances, it can be expected that a time-weighted average (TWA) is the most relevant exposure concentration in water (Cw) within the allowed range of fluctuation (cf. paragraph 24). It is recommended to calculate a TWA water concentration, according to the procedure in Appendix 6 of TM C.20 (2). It should be noted that the ln-transformation of the water concentration is suitable when exponential decay between renewal periods is expected, e.g. in a semi-static test design. In a flow through system, ln-transformation of exposure concentrations may not be needed. If TWA water concentrations are derived, they should be reported and used in subsequent calculations.

In a standard fish BCF test uptake and depuration can be described in terms of two first order kinetic processes.

Rate of uptake = k1 × Cw | [Equation A5.2]

Overall loss rate = (k2 + kg + km + ke) × Cf | [Equation A5.3]

At steady-state, assuming growth and metabolism are negligible (i.e. the values for kg and km cannot be distinguished from zero), the rate of uptake equals the rate of depuration, and so combining Equation A5.2 and Equation A5.3 gives the following relationship:

BCF

C f SS

C w SS

k 1

k 2

[Equation A5.4]

Where

Cf-SSConcentration in fish at steady-state (mg kg– 1 wet weight).

Cw-SSConcentration in water at steady-state (mg l– 1).

The ratio of k1/k2 is known as the kinetic BCF (BCFK) and should be equal to the steady-state BCF (BCFSS) obtained from the ratio of the steady-state concentration in fish to that in water, but deviations may occur if steady-state was uncertain or if corrections for growth have been applied to the kinetic BCF. However, as k1 and k2 are constants, steady-state does not need to be reached to derive a BCFK.

Based on these first order equations, this Appendix 5 includes the general calculations necessary for both aqueous and dietary exposure bioaccumulation methods. However, sections 5, 6 and 8 are only relevant for the aqueous exposure method but are included here as they are “general” techniques. The sequential (sections 4 and 5) and simultaneous (section 6) methods allow the calculation of uptake and depuration constants which are used to derive kinetic BCFs. The sequential method for determining k2 (section 4) is important for the dietary method as it is needed to calculate both assimilation efficiency and BMF. Appendix 7 details the calculations that are specific to the dietary method.

Before performing the test, an estimate of k2 and hence some percentage of the time needed to reach steady-state may be obtained from empirical relationships between k2 and the n-octanol/water partition coefficient (KOW) or k1 and BCF. It should be realised, however, that the equations in this section only apply when uptake and depuration follow first-order kinetics. If this is clearly not the case it is advised to seek advice from a biostatistician and/or pharmacokineticist, if predictions of the uptake phase are desirable.

An estimate of k2 (day– 1) may be obtained by several methods. For example, the following empirical relationships could be used in the first instance [As with every empirical relationship, it should be verified that the test substance falls within the applicability domain of the relationship]:

log k2 = 1,47 – 0,414logKOW | (r2=0,95) [(3); Equation A5.5]

or

k 2

k 1

BCF

[Equation A5.6]

Where k1 = 520 × W– 0,32 (for substances with a log KOW > 3) | (r2=0,85)[(4); Equation A5.7]

AndBCF10 0,910logK OW1,975log6,8107K OW10,786 | (r2=0,90)[(5); Equation A5.8]

W = mean treated fish weight (grams wet weight) at the end of uptake/start of depuration [The weight of fish at the end of the uptake phase can be estimated from previous study data or knowledge of the test species' likely increase in size from a typical test starting weight over a typical uptake duration (e.g. 28 days).]

For other related relationships see (6). It may be advantageous to employ more complicated models in the estimation of k2 if, for example, it is likely that significant metabolism may occur (7) (8). However as the complexity of the model increases, greater care should be taken with the interpretation of the predictions. For example the presence of nitro groups might indicate fast metabolism, but this is not always the case. Therefore the user should weigh up the predictive method results against chemical structure and any other relevant information (for example preliminary studies) in the scheduling of a study.

The time to reach a certain percentage of steady-state may be obtained, by applying the k2-estimate, from the general kinetic equation describing uptake and depuration (first-order kinetics), assuming growth and metabolism is negligible. If substantial growth occurs during the study, the estimations described below will not be reliable. In such cases, it is better to use the growth corrected k2g as described later (see Section 7 of this Appendix):

dC f

dt

k 1C W k 2C f | [Equation A5.9]

or, if Cw is constant:

C f

k 1

k 2

C W1e k 2t | [Equation A5.10]

When steady-state is approached (t → ∞), Equation A5.10 may be reduced (cf. (9) (10)) to:

C f

k 1

k 2

C W | [Equation A5.11]

or

C f

C w

k 1

k 2

BCF | [Equation A5.12]

Then BCF × Cw is an approximation to the concentration in the fish at steady-state (Cf-SS). [Note: the same approach can be used when estimating a steady-state BMF with the dietary test. In this case, BCF is replaced with BMF and Cw with Cfood, concentration in the food, in the equations above]

Equation A5.10 may be transcribed to:

C fC f SS1e k 2t | [Equation A5.13]

or

C f

C f SS

1e k 2t | [Equation A5.14]

Applying Equation A5.14, the time to reach a certain percentage of steady-state may be predicted when k2 is pre-estimated using Equation A5.5 or Equation A5.6.

As a guideline, the statistically optimal duration of the uptake phase for the production of statistically acceptable data (BCFK) is that period which is required for the curve of the logarithm of the concentration of the test substance in fish plotted against linear time to reach at least 50 % of steady-state (i.e. 0,69/k2), but not more than 95 % of steady-state (i.e. 3,0/k2) (11). In case accumulation reaches beyond 95 % of steady-state, calculation of a BCFSS becomes feasible.

The time to reach 80 percent of steady-state is (using Equation A5.14):

0,801e k 2t | [Equation A5.15]

or

t 80ln0,20

k 2

1,6

k 2

[Equation A5.16]

Similarly the time to reach 95 percent of steady-state is:

t 95

ln 0,05

k 2

3,0

k 2

[Equation A5.17]

For example, the duration of the uptake phase (i.e. time to reach a certain percentage of steady-state, e.g. t80 or t95) for a test substance with log KOW = 4 would be (using Equation A5.5, Equation A5.16 and Equation A5.17):

logk2 = 1,47 – 0,414 · 4

k2 = 0,652 day– 1

t801,60,6522,45days59 hours

or t953,00,6524,60days110 hours

Alternatively, the expression:

teSS = 6,54 · 10 – 3 · KOW + 55,31 (hours) | [Equation A5.18]

may be used to calculate the time for effective steady-state (teSS) to be reached (12). For a test substance with log KOW = 4 this results in:

teSS = 6,54 · 10 – 3 · 104 + 55,31 = 121 hours

A prediction of the time needed to reduce the body burden to a certain percentage of the initial concentration may also be obtained from the general equation describing uptake and depuration (assuming first order kinetics, cf. Equation A5.9 (1) (13).

For the depuration phase, Cw (or Cfood for the dietary test) is assumed to be zero. The equation may then be reduced to:

dC f

dt

k 2C f | [Equation A5.19]

or

C fC f,0 e k 2t | [Equation A5.20]

where Cf,0 is the concentration at the start of the depuration period.

50 percent depuration will then be reached at the time (t50):

Cf

Cf,0

12e– k2t50

or

t50– ln0,50

k2

0,693

k2

Similarly 95 percent depuration will be reached at:

t95– ln0,05

k2

3,0

k2

If 80 % uptake is used for the first period (1,6/k2) and 95 % loss in the depuration phase (3,0/k2), then depuration phase is approximately twice the duration of the uptake phase.

Note that the estimations are based on the assumption that uptake and depuration patterns will follow first order kinetics. If first-order kinetics is obviously not obeyed, these estimations are not valid.

Most bioconcentration data have been assumed to be ‘reasonably’ well described by a simple two-compartment/two-parameter model, as indicated by the rectilinear curve which approximates to the points for concentrations in fish (on an ln scale), during the depuration phase.

Note that deviations from a straight line may indicate a more complex depuration pattern than first order kinetics. The graphical method may be applied for resolving types of depuration deviating from first order kinetics.

To calculate k2 for multiple time (sampling) points, perform a linear regression of ln(concentration) versus time. The slope of the regression line is an estimate of the depuration rate constant k2 [In most programs that allow a linear regression, also standard errors and confidence interval (CI) of the estimates are given, e.g. in Microsoft Excel using the Data Analysis tool pack.]. From the intercept the average concentration in the fish at the start of the depuration phase (C0,d; which equals the average concentration in the fish at the end of the uptake phase) can easily be calculated (including error margins) [In most programs that allow a linear regression, also standard errors and confidence interval (CI) of the estimates are given, e.g. in Microsoft Excel using the Data Analysis tool pack.]:

C0,d = eintercept | [Equation A5.21]

To calculate k2 when only two time (sampling) points are available (as in the minimised design), substitute the two average concentrations into the following equation

k 2ln C flln C f2

t 2 t 1

[Equation A5.22]

Where ln(Cf1) and ln(Cf2) are the natural logarithms of the concentrations at times t1 and t2, respectively, and t2 and t1 are the times when the two samples were collected relative to the start of depuration [In contrast with the linear regression method, using this formula will not yield a standard error for k2.].

To find a value for k1 given a set of sequential time concentration data for the uptake phase, use a computer program to fit the following model:

C ftC wt

k 1

k 2

1e k 2t | [Equation A5.23]

Where k2 is given by the previous calculation, Cf(t) and Cw(t) are the concentrations in fish and water, respectively, at time t.

To calculate k1 when only two time (sampling) points are available (as in the minimised design), use the following formula:

k 1

C f k 2

C w1e k 2t

[Equation A5.24]

Where k2 is given by the previous calculation, Cf is the concentration in fish at the start of the depuration phase, and Cw is the average concentration in the water during the uptake phase [In contrast with a linear fitting procedure, this method will usually not yield a standard error or confidence interval for the estimated k1.].

Visual inspection of the k1 and k2 slopes when plotted against the measured sample point data can be used to assess goodness of fit. If it turns out that the sequential method has given a poor estimate for k1 then the simultaneous approach to calculate k1 and k2 should be applied (see next section 6). Again, the resulting slopes should be compared against the plotted measured data for visual inspection of goodness of fit. If the goodness of fit is still poor this may be an indication that first order kinetics do not apply and other more complex models should be employed.

Computer programs can be used to find values for k1 and k2 given a set of sequential time concentration data and the model:

C fC w

k 1

k 2

1e k 2t | 0 < t < tc | [Equation A5.25]

C fC w

k 1

k 2

e k 2t t c e k 2t | t > tc | [Equation A5.26]

where

tctime at the end of the uptake phase.

This approach directly provides standard errors for the estimates of k1 and k2. When k1/k2 is substituted by BCF (cf. Equation A5.4) in Equation A5.25 and Equation A5.26, the standard error and 95 % CI of the BCF can be estimated as well. This is especially useful when comparing different estimates due to data transformation. The dependent variable (fish concentration) can be fitted with or without ln transformation, and the resulting BCF uncertainty can be evaluated.

As a strong correlation exists between the two parameters k1 and k2 if estimated simultaneously, it may be advisable first to calculate k2 from the depuration data only (see above); k2 in most cases can be estimated from the depuration curve with relatively high precision. k1 can be subsequently calculated from the uptake data using non-linear regression [It should be realised that the uncertainty in the k2 estimate is not used properly in the bioaccumulation model when this is essentially regarded as constant when fitting k1 in the sequential fit method. The resulting BCF uncertainty will therefore be different between the simultaneous and sequential fitting methods.]. It is advised to use the same data transformation when fitting sequentially.

Visual inspection of the resulting slopes when plotted against the measured sample point data can be used to assess goodness of fit. If it turns out that this method has given a poor estimate for k1 then the simultaneous approach to calculate k1 and k2 can be applied. Again, the fitted model should be compared against the plotted measured data for visual inspection of goodness of fit and the resulting parameter estimates for k1, k2 and resulting BCF and their standard errors and/or confidence intervals should be compared between different types of fit.

If the goodness of fit is poor this may be an indication that first order kinetics does not apply and other more complex models should be employed. One of the most common complications is fish growth during the test.

This section describes a standard method for correction due to fish growth during the test (so called ‘growth dilution’) which is only valid when first order kinetics applies. In case there are indications that first order kinetics do not apply, it is advised to seek advice from a biostatistician for a proper correction of growth dilution or to use the mass based approach described below.

In some cases this method for correcting growth dilution is subject to a lack of precision or sometimes does not work (for example for very slowly depurating substances tested in fast growing fish the derived depuration rate constant corrected for growth dilution, k2g, may be very small and so the error in the two rate constants used to derive it become critical, and in some cases kg estimates may be larger than k2). In such cases an alternative approach (i.e. mass approach), which also works when first order growth kinetics have not been obeyed, can be used which avoids the need for the correction. This approach is outlined at the end of this section.

For the standard method all individual weight and length data are converted to natural logarithms and ln(weight) or ln(1/weight) is plotted vs. time (day), separately for treatment and control groups. The same process is carried out for the data from the uptake and depuration phases separately. Generally for growth dilution correction it is more appropriate to use the weight data from the whole study to derive the growth rate constant (kg), but statistically significant differences between the growth rate constants derived for the uptake phase and depuration phase may indicate that the depuration phase rate constant should be used. Overall growth rates from aqueous studies for test and control groups can be used to check for any treatment related effects.

A linear least squares correlation is calculated for the ln(fish weight) vs. day (and for ln(1/weight) vs. day) for each group (test(s) and control groups, individual data, not daily mean values) for the whole study, uptake and depuration phases using standard statistical procedures. The variances in the slopes of the lines are calculated and used to evaluate the statistical significance (p = 0,05) of the difference in the slopes (growth rate constants) using the student t-test (or ANOVA if more than one concentration is tested). Weight data are generally preferred for growth correction purposes. Length data, treated in the same way, may be useful to compare control and test groups for treatment related effects. If there is no statistically significant difference in the weight data analysis, the test and control data may be pooled and an overall fish growth rate constant for the study (kg) calculated as the overall slope of the linear correlation. If statistically significant differences are observed, growth rate constants for each fish group, and/or study phase, are reported separately. The rate constant from each treated group should then be used for growth dilution correction purposes of that group. If statistical differences between the uptake and depuration phase rate constants were noted, depuration phase derived rate constants should be used.

The calculated growth rate constant (kg expressed as day-1) can be subtracted from the overall depuration rate constant (k2) to give the depuration rate constant, k2g.

k2g = k2 – kg | [Equation A5.27]

The uptake rate constant is divided by the growth-corrected depuration rate constant to give the growth-corrected kinetic BCF, denoted BCFKg (or BMFKg).

BCF Kg

k 1

k 2g

[Equation A5.28]

The growth rate constant derived for a dietary study is used in Equation A7.5 to calculate the growth corrected BMFKg (cf. Appendix 7).

An alternative to the above “growth rate constant subtraction method” that avoids the need to correct for growth can be used as follows. The principle is to use depuration data on a mass basis per whole fish rather than on a concentration basis.

Convert depuration phase tissue concentrations (mass of test substance/unit mass of fish) into mass of test substance/fish: match concentrations and individual fish weights in tabular form (e.g. using a computer spreadsheet) and multiply each concentration by the total fish weight for that measurement to give a set of mass test substance/fish for all depuration phase samples.

Plot the resulting natural logarithm of substance mass data against time for the experiment (depuration phase) as would be done normally.

For the aqueous exposure method, derive the uptake rate constant routinely (see sections 4 and 6) note that the “normal” k2 value should be used in the curve fitting equations for k1) and derive the depuration rate constant from the above data. Because the resulting value for the depuration rate constant is independent of growth as it has been derived on a mass basis per whole fish, it should be denoted as k2g and not k2.

BCF results (kinetic and steady-state) from aqueous exposure tests should also be reported relative to a default fish lipid content of 5 % wet weight, unless it can be argued that the test substance does not primarily accumulate in lipid (e.g. some perfluorinated substances may bind to proteins). Fish concentration data, or the BCF, need to be converted to a 5 % lipid content wet weight basis. If the same fish were used for measuring substance concentrations and lipid contents at all sampling points, this requires each individual measured concentration in the fish to be corrected for that fish's lipid content.

C f,L0,05

L

C f | [Equation A5.29]

where

Cf,Llipid-normalised concentration in fish (mg kg– 1 wet weight)

Llipid fraction (based on wet weight)

Cfconcentration of test substance in fish (mg kg– 1 wet weight)

If lipid analysis was not conducted on all sampled fish, a mean lipid value is used to normalise the BCF. For the steady-state BCF, the mean value recorded at the end of the uptake phase in the treatment group should be used. For the normalisation of a kinetic BCF there may be some cases where a different approach is warranted, for example if the lipid content changed markedly during the uptake or depuration phase. However a feeding rate that minimises dramatic changes in lipid content should be used anyway routinely.

BCF KL0.05

L n

BCF K | [Equation A5.30]

where

BCFKLlipid-normalised kinetic BCF (L kg– 1)

Lnmean lipid fraction (based on wet weight)

BCFKkinetic BCF (L kg– 1)

(1)

Arnot J.A. and Gobas F.A.P.C. (2004), A food web bioaccumulation model for organic chemicals in aquatic ecosystems, Environ. Toxicol. Chem. 23: 2343–2355.

(2)

Chapter C.20 of this Annex, Daphnia magna Reproduction Test.

(3)

Spacie A. and Hamelink J.L. (1982), Alternative models for describing the bioconcentration of organics in fish. Environ. Toxicol. Chem. 1: 309-320.

(4)

Sijm D.T.H.M., Verberne M.E., de Jonge W.J., Pärt P. and Opperhuizen A. (1995), Allometry in the uptake of hydrophobic chemicals determined in vivo and in isolated perfused gills. Toxicol. Appl. Pharmacol. 131: 130-135.

(5)

Bintein S., Devillers J. and Karcher W. (1993), Nonlinear dependence of fish bioconcentration on n-octanol/water partition coefficient. SAR QSAR Environ. Res. 1: 29-39.

(6)

Kristensen P. (1991), Bioconcentration in fish: comparison of BCF's derived from OECD and ASTM testing methods; influence of particulate matter to the bioavailability of chemicals. Danish Water Quality Institute, Hørsholm, Denmark.

(7)

Arnot J.A., Meylan W., Tunkel J., Howard P.H., Mackay D., Bonnell M. and Boethling R.S. (2009), A quantitative structure-activity relationship for predicting metabolic biotransformation rates for organic chemicals in fish. Environ. Toxicol. Chem. 28: 1168-1177.

This article is long: the rest is at the official publication.