37 Appendix 7, Regulation (EU) 2017/735

In forcesince 14 Feb 2017
1.

Example of constituent quantities of a suitable commercial fish food

2.

Food spiking technique examples

3.

Calculation of assimilation efficiency and biomagnification factor

4.

Lipid correction

5.

Evaluation of differences between measured time zero concentration (C0,m) and derived time zero concentration (C0,d)

6.

Guidance for very fast depurating test substances

[In some regions it may only be possible to obtain fish food with a lipid concentration that falls far short of this upper limit. In such cases studies should be run with the lower lipid concentration in the food as supplied, and the feeding rate adjusted appropriately to maintain fish health. Diet lipids should not be artificially increased by the addition of excess oil.]

Major constituent | Fish meal

Crude Protein | ≤ 55,0 %

Crude fat | ≤ 15,0 %

Crude Fibre | ≥ 2,0 %

Moisture | ≥ 12 %

Ash | ≥ 8 %

Control diets should be prepared in exactly the same way as the spiked diet, but with an absence of test substance.

To check the concentration of the treated diet, triplicate samples of the dosed food should be extracted with a suitable extraction method and the test substance concentration or radioactivity in the extracts measured. High analytical recoveries (> 85 %) with low variation between samples (three sample concentrations for the substance taken at test start should not vary more than ± 15 % from the mean) should be demonstrated.

During the dietary test, three diet samples for analysis should be collected on day 0 and at the end of the uptake phase for the determination of the test substance content in the diet.

A target, nominal test concentration in the treated fish food is set, for example 500 μg test substance/g food. The appropriate quantity (by molar mass or specific radioactivity) of neat test substance is added to a known mass of fish food in a glass jar or rotary evaporator bulb. The mass of fish food should be sufficient for the duration of the uptake phase (taking into account the need for increasing quantities at each feed owing to fish growth). The fish feed/test substance should be mixed overnight by slow tumbling (e.g. using a roto-rack mixer or by rotation if a rotary evaporator bulb is used). The spiked diet should be stored under conditions that maintain stability of the test substance within the feed mix (e.g. refrigeration) until use.

Solid test substances should be ground in a mortar to a fine powder. Liquid test substances can be added directly to the corn or fish oil. The test substance is dissolved in a known quantity of corn or fish oil (e.g. 5-15 ml). The dosed oil is quantitatively transferred into a rotary evaporation bulb of suitable size. The flask used to prepare the dosed oil should be flushed with two small aliquots of oil and these added to the bulb to make sure all dissolved test substance is transferred. To ensure complete dissolution/dispersion in the oil (or if more than one test substance is being used in the study), a micro-stirrer is added, the flask stoppered and the mixture stirred rapidly overnight. An appropriate quantity of fish diet (usually in pellet form) for the test is added to the bulb, and the bulb's contents are mixed homogeneously by continuously turning the glass bulb for at least 30 minutes, but preferably overnight. Thereafter, the spiked food is stored appropriately (e.g. refrigerated) to ensure test substance stability in the food until use.

An appropriate quantity of test substance (by molar mass or specific radioactivity) sufficient to achieve the target dose is dissolved in a suitable organic solvent (e.g. cyclohexane or acetone; 10-40 ml, but a greater volume if necessary depending on the quantity of food to spike). Either an aliquot, or all (added portion wise), of this solution is mixed with the appropriate mass of fish food sufficient for the test to achieve the required nominal dose level. The food/test substance can be mixed in a stainless steel mixing bowl and the freshly-dosed fish food left in the bowl in a laboratory hood for two days (stirred occasionally) to allow the excess solvent to evaporate, or mixed in a rotary evaporator bulb with continuous rotation. The excess solvent can be “blown” off under a stream of air or nitrogen if necessary. Care should be taken to ensure that the test substance does not crystallise as the solvent is removed. The spiked diet should be stored under conditions (e.g. refrigeration) that maintain stability of the test substance within the feed mix until use.

To calculate the assimilation efficiency, the overall depuration rate constant should first be estimated according to section 4 of Appendix 5 (using the “sequential method”, i.e. standard linear regression) using mean sample concentrations from the depuration phase. The feeding rate constant, I, and uptake duration, t, are known parameters of the study. Cfood, the mean measured concentration of the test substance in the food is a measured variable in the study. C0,d, the test substance concentration in the fish at the end of the uptake phase, is usually derived from the intercept of a plot of ln(concentration) vs. depuration day.

The substance assimilation efficiency (a, absorption of test substance across the gut) is calculated as:

α

C 0,d k 2

I C food

11e k 2t | [Equation A7.1]

where:

C0,dderived concentration in fish at time zero of the depuration phase (mg kg– 1);

k2overall (not growth-corrected) depuration rate constant (day– 1), calculated according to equations in Appendix 5, Section 3;

Ifood ingestion rate constant (g food g– 1 fish day– 1);

Cfoodconcentration in food (mg kg– 1 food);

tduration of the feeding period (day)

However, the feeding rate, I, used in the calculation may need to be adjusted for fish growth to give an accurate assimilation efficiency, a. In a test where fish grow significantly during the uptake phase (in which no correction of feed quantities is made to maintain the set feeding rate), the effective feeding rate as the uptake phase progresses will be lower than that set, resulting in a higher 'real' assimilation efficiency. (Note this is not important for the overall calculation of BMF as the I terms effectively cancel out between Equation A7.1 and Equation A7.4). The mean feeding rate corrected for growth dilution, Ig, can be derived in several ways, but a straightforward and rigorous one is to use the known growth rate constant (kg) to estimate the test fish weights at time points during the uptake phase, i.e.:

WftWf,0ek gt | [Equation A7.2]

where

Wf(t)mean fish weight at uptake day t

Wf,0mean fish weight at the start of the experiment

In this way (at least) the mean fish weight on the last day of exposure (Wf,end-of-uptake) can be estimated. As the feeding rate was set based on Wf,0, the effective feeding rate for each day of uptake can be calculated using these two weight values. The growth-corrected feeding rate, Ig (g food g-1 fish day– 1), to use instead of I in cases of rapid growth during the uptake phase, can then be calculated as

I g

I W f,0

W f,end-of-uptake

[Equation A7.3]

Once the assimilation efficiency has been obtained, the BMF can be calculated by multiplying it with the feeding rate constant I (or Ig, if used to calculate α) and dividing the product by the overall depuration rate constant k2:

BMF

I α

k 2

[Equation A7.4]

The growth-corrected biomagnification factor should also be calculated in the same way, using the growth corrected depuration rate constant (as derived according to section 7 in Appendix 5). Again, if Ig has been used to calculate α, it should also be used here instead of I:

BMF

I α

k 2g

[Equation A7.5]

where:

αassimilation efficiency (absorption of test substance across the gut);

k2overall (not growth-corrected) depuration rate constant (day– 1), calculated according to equations in Appendix 5, Section 3;

k2ggrowth-corrected depuration rate constant (day– 1);

Ifood ingestion rate constant (g food g– 1 fish day– 1);

The growth-corrected half-life (t1/2) is calculated as follows.

t 120,693

k 2g

[Equation A7.6]

The substance assimilation efficiency from the diet can also be estimated if tissue residues are determined during the linear phase of the uptake phase (between days 1 and 3). In this case the substance assimilation efficiency (α) can be determined as follows

α

C fisht

I C food t

[Equation A7.7]

Where

Cfish(t)the concentration of test substance in the fish at time t (mg kg– 1 wet weight).

If lipid content was measured on the same fish as chemical analysis for all sampling intervals, then individual concentrations should be corrected on a lipid basis and the ln(concentration, lipid corrected) plotted against depuration (day) to give C0,d and k2. Assimilation efficiency (Equation A7.1) can then be calculated on a lipid basis, using Cfood on a lipid basis (i.e. Cfood is multiplied by the mean lipid fraction of the food). Subsequent calculation using Equation A7.4 and Equation A7.5 will give the lipid-corrected (and growth-dilution corrected) BMF directly.

Otherwise, the mean lipid fraction (w/w) in the fish and in the food are derived for both treatment and control groups (for food and control group fish this is usually from data measured at exposure start and end; for treatment group fish this is usually from data measured at end of exposure only). In some studies, fish lipid content may increase markedly; in such cases it is more appropriate to use a mean test fish lipid concentration calculated from the measured values at the end of exposure and end of depuration. In general, data from the treatment group only should be used to derive both of the lipid fractions.

The lipid-correction factor (Lc) is calculated as:

L C

L fish

L food

[Equation A7.8]

where Lfish and Lfood are the mean lipid fractions in fish and food, respectively.

The lipid-correction factor is used to calculate the lipid-corrected biomagnification factor (BMFL):

BMF L

BMF

L C

[Equation A7.9]

The measured time zero concentration (C0,m) and derived time zero concentration (C0,d) should be compared. If they are very similar, then this supports the first order model used to derive the depuration parameters.

In some studies there may be a marked difference between the derived time zero value, C0,d, and the mean measured time zero concentration. C0,m (see last bullet point of paragraph 159 of this test method). If C0,d is very much lower than C0,m (C0,d << C0,m), the difference may suggest the presence of undigested spiked food in the gut. This may be tested experimentally by conducting separate analysis on the excised gut if additional (whole fish) samples were taken and stored at the end of the uptake phase. Otherwise, if a statistically valid outlier test applied to the depuration phase linear regression indicates that the first sample point of depuration is erroneously elevated, carrying out the linear regression to derive k2 but omitting the first depuration concentration point may be appropriate. In such cases, if the uncertainty in the linear regression is greatly decreased, and it is clear that approximately first order depuration kinetics were obeyed, it may be appropriate to use the resulting C0,d and k2 values in the assimilation efficiency calculation. This should be fully justified in the report. It is also possible that non-first order kinetics were operating in the depuration phase. If this is likely (i.e. the natural logarithm transformed data appear to follow a curve compared with the straight-line linear regression plot), then the calculations of k2 and C0,d are unlikely to be valid and the advice of a biostatician should be sought.

If C0,d is very much higher than the measured value (C0,d >> C0,m) this may indicate: that the substance was depurated very fast (i.e. sampling points approached the limit of quantification of the analytical method very early in the depuration phase, cf. Section 6 below); that there was a deviation from first order depuration kinetics; that the linear regression to derive k2 and C0,d is flawed; or that a problem with the measured concentrations in the study occurred at some sampling time points. In such cases the linear regression plot should be scrutinised for evidence of samples at or near the limit of quantification, for outliers and for obvious curvature (suggestive of non-first order kinetics), and highlighted in the report. Any subsequent re-evaluation of the linear regression to improve estimated values should be described and justified. If marked deviation from first order kinetics is observed, then the calculations of k2 and C0,d are unlikely to be valid and the advice of a biostatician should be sought.

As discussed in paragraph 129 of the test method, some substances may depurate so fast that a reliable time zero concentration, C0,d, and k2 cannot be derived because in samples very early in the depuration phase (i.e. from the second depuration sample onwards) the substance is effectively no longer measured (concentrations reported at the limit of quantification). This situation was observed in the ring test carried out in support of this test method with benzo[a]pyrene, and has been documented in the validation report for the method. In such cases linear regression cannot be carried out reliably, and is likely to give an unrealistically high estimate of C0,d, resulting in an apparent assimilation efficiency much greater than 1. It is possible to calculate a conservative estimate of k2 and an “upper bound” BMF in these instances.

Using those data points of the depuration phase where a concentration was measured, up to and including the first “non-detect” concentration (concentration set at limit of quantification), a linear regression (using natural logarithm transformed concentration data against time) will give an estimate of k2. For these sorts of cases this is likely only to involve two data points (e.g. sample days 1 and 2 of depuration) and then k2 can be estimated using Equation A5.22 in Appendix 5. This k2 estimate can be used to estimate an assimilation efficiency according to equation A7.1, substituting the C0,d value in the equation with the measured time zero concentration (C0,m) in cases where C0,d is clearly estimated to be much higher than could have been achievable in the test. If C0,m was not measureable, then the limit of detection in fish tissue should be used. If, in some cases, this gives a value of α > 1, then the assimilation efficiency is assumed to 1 as a “worst case”.

The maximum BMFK can then be estimated using Equation A7.4, and should be quoted as a “much less than” (<<) value. For example, for a study carried out with a feeding rate of 3 % and a depuration half-life less than 3 days, and a “worst case” α of 1, the BMFK is likely to be below about 0,13. Given the purpose of this estimation and the fact that values will be conservative in nature, it is not necessary to correct them for growth dilution or fish and food lipid content.