reaction mass of 3-methyl-4-(2,6,6-trimethylcyclohex-2-en-1-yl)but-3-en-2-one and 1-(2,6,6-trimethylcyclohex-2-en-1-yl)pent-1-en-3-one

Bioaccumulation: aquatic / sediment

Link to relevant study record(s)

Reference

Reference 1

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a valid (Q)SAR model and falling into its applicability domain, with adequate and reliable documentation / justification

Principles of method if other than guideline:

Estimation of BCF, BAF and biotransformation rate using BCFBAF v3.01

GLP compliance:

no

Radiolabelling:

no

Test organisms (species):

other: fish

Details on estimation of bioconcentration:

BASIS INFORMATION
- Measured/calculated logPow: measured log Pow of 4,7

Type:

BCF

Value:

586.2 L/kg

Remarks on result:

other:

Remarks:

The substance is within the applicability domain of the BCFBAF submodel: Bioconcentration factor (BCF; Meylan et al., 1997/1999).

Type:

BCF

Value:

1 087 L/kg

Remarks on result:

other:

Remarks:

Upper trophic, incl. biotransformation estimates; The substance is within the applicability domain of the BCFBAF submodel: Arnot & Gobas BAF and steady-state BCF Arnot & Gobas, 2003).

Type:

BCF

Value:

4 554 L/kg

Remarks on result:

other:

Remarks:

Upper trophic, incl. biotransformation rate of zero; The substance is within the applicability domain of the BCFBAF submodel: Arnot & Gobas BAF and steady-state BCF Arnot & Gobas, 2003).

Details on kinetic parameters:

Biotransformation half-life (days): 3.374 (normalised to 10 g fish)
Biotransformation rate (kM, normalised to 10 g fish at 15 °C): 0.2054 /day

Summary Results:

Log BCF (regression-based estimate): 2.77 (BCF = 586 L/kg wet-wt)

Biotransformation Half-Life (days) : 3.37 (normalized to 10 g fish)

Log BAF (Arnot-Gobas upper trophic): 3.05 (BAF = 1.12e+003 L/kg wet-wt)

 

Log Kow (experimental): not available from database

Log Kow used by BCF estimates: 4.70 (user entered)

 

Equation Used to Make BCF estimate:

Log BCF = 0.6598 log Kow - 0.333 + Correction

 

Correction(s):                   Value

No Applicable Correction Factors

 

Estimated Log BCF = 2.768 (BCF = 586.2 L/kg wet-wt)

 

Whole Body Primary Biotransformation Rate Estimate for Fish:

Fragment Description	Coefficient value	No. compounds containing fragment in total training set	Maximum number of each fragment in any individual compound	No. of instances of each fragment for the current substance
Carbon with 4 single bonds & no hydrogens	-0.29842827	47	10	1
Ketone   [-C-C(=O)-C-]	-0.1800634	10	2	1
Methyl  [-CH3]	0.24510529	170	12	5
-CH2-  [cyclic]	0.09625069	36	12	2
-CH -  [cyclic]	0.01260466	30	17	2
-C=CH  [alkenyl hydrogen]	0.09884729	34	6	2

RESULT  |       LOG Bio Half-Life (days)           |        | 0.5282

RESULT  |           Bio Half-Life (days)           |        |  3.374

NOTE    | Bio Half-Life Normalized to 10 g fish at 15 deg C  |

 

Biotransformation Rate Constant:

kM (Rate Constant): 0.2054 /day (10 gram fish)

kM (Rate Constant): 0.1155 /day (100 gram fish)

kM (Rate Constant): 0.06496 /day (1 kg fish)

kM (Rate Constant): 0.03653 /day (10 kg fish)

 

Arnot-Gobas BCF & BAF Methods (including biotransformation rate estimates):

Estimated Log BCF (upper trophic) = 3.036 (BCF = 1087 L/kg wet-wt)

Estimated Log BAF (upper trophic) = 3.048 (BAF = 1117 L/kg wet-wt)

Estimated Log BCF (mid trophic)  = 3.085 (BCF = 1215 L/kg wet-wt)

Estimated Log BAF (mid trophic)  = 3.129 (BAF = 1347 L/kg wet-wt)

Estimated Log BCF (lower trophic) = 3.089 (BCF = 1226 L/kg wet-wt)

Estimated Log BAF (lower trophic) = 3.188 (BAF = 1541 L/kg wet-wt)

 

Arnot-Gobas BCF & BAF Methods (assuming a biotransformation rate of zero):

Estimated Log BCF (upper trophic) = 3.658 (BCF = 4554 L/kg wet-wt)

Estimated Log BAF (upper trophic) = 4.526 (BAF = 3.357e+004 L/kg wet-wt)

Reference 2

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2019

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a valid (Q)SAR model, but not (completely) falling into its applicability domain, with adequate and reliable documentation / justification

Justification for type of information:

1. SOFTWARE
OASIS Catalogic v5.12.1

2. MODEL (incl. version number)
BCF base-line model v02.09 - July 2016

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
See section 'Test Material'.

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
See attached QMRF.

5. APPLICABILITY DOMAIN
See attached QPRF.

6. ADEQUACY OF THE RESULT
- The model is scientifically valid (see attached QMRF).
- The model estimates the Bioconcentration factor (BCF) as required information point according to Regulation (EC) No 1907/2006 [REACH], Annex IX, 9.3.2 Bioaccumulation in aquatic species (preferably fish);
further related predictions: Apparent effect of mitigating factors / Maximum bioconcentration factor (BCFmax) / Maximum diameter of energetically stable conformers / Whole body primary biotransformation half-life / Metabolic biotransformation rate constant Km / Metabolites and their quantitative distribution
- See attached QPRF for reliability assessment.

Principles of method if other than guideline:

Calculation using Catalogic v.5.11.19, BCF base-line model v.02.09

GLP compliance:

no

Details on estimation of bioconcentration:

BASIS FOR CALCULATION OF BCF
- Estimation software: OASIS Catalogic v5.12.1 [BCF base line model - v.02.09]

Type:

BCF

Value:

3 090 L/kg

Remarks on result:

other: considering all mitigating factors; the substance is not within the applicability domain of the model.

Type:

BCF

Value:

3 999 L/kg

Remarks on result:

other: without considering any mitigating factors; the substance is not within the applicability domain of the model.

Reference 3

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a valid (Q)SAR model and falling into its applicability domain, with adequate and reliable documentation / justification

Justification for type of information:

1. SOFTWARE
T.E.S.T. (version 4.2.1) (Toxicity Estimation Software Tool). US EPA, 2012.

2. MODEL (incl. version number)
T.E.S.T. (version 4.2.1)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
c32c(cccc3)nn(c1c(O)ccc(C(C)(C)CC(C)(C)C)c1)n2

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
See QMRF in section "Overall remarks, attachments".

5. APPLICABILITY DOMAIN
See QPRF in section "Executive summary".

6. ADEQUACY OF THE RESULT
- The model is scientifically valid (see QMRF).
- The model estimates the bioconcentration factor (BCF) as required information point under Regulation (EC) No 1907/2006 [REACH], Annex IX, 9.3.2 Bioaccumulation in aquatic species, preferably fish (see also QPRF).
- See QPRF for reliability assessment.

Principles of method if other than guideline:

T.E.S.T. is a toxicity estimation software tool. The program requires only the molecular structure of the test item, all other molecular descriptors which are required to estimate the toxicity are calculated within the tool itself. The molecular descriptors describe physical characteristics of the molecule (e.g. E-state values and E-state counts, constitutional descriptors, topological descriptors, walk and path counts, connectivity, information content, 2d autocorrelation, Burden eigenvalue, molecular property (such as the octanol-water partition coefficient), Kappa, hydrogen bond acceptor/donor counts, molecular distance edge, and molecular fragment counts). Each of the available methods uses a different set of these descriptors to estimate the toxicity. The bioaccumulation factor (BCF) was estimated using several available methods: hierarchical clustering method; FDA method, single model method; group contribution method; nearest neighbor method; consensus method. The methods were validated using statistical external validation using separate training and test data sets. The experimental data set was obtained from several different databases (Dimitrov et al., 2005; Arnot and Gobas, 2006; EURAS; Zhao, 2008). From the available data set salts, mixtures and ambiguous compounds were removed. The final data set contained 676 chemicals.

References:
- Dimitrov, S., N. Dimitrova, T. Parkerton, M. Combers, M. Bonnell, and O. Mekenyan. 2005. Base-line model for identifying the bioaccumulation potential of chemicals. SAR and QSAR in Environmental Research 16:531-554.
- Arnot, J.A., and F.A.P.C. Gobas. 2006. A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in aquatic organisms. Environ. Rev. 14:257-297.
- EURAS. Establishing a bioconcentration factor (BCF) Gold Standard Database. EURAS [cited 5/20/09]. Available from http://www.euras.be/eng/project.asp?ProjectId=92.
- Zhao, C.; Boriani, E.; Chana, A.; Roncaglioni, A.; Benfenati, E. 2008. A new hybrid system of QSAR models for predicting bioconcentration factors (BCF). Chemosphere 73:1701-1707.

GLP compliance:

no

Test organisms (species):

other: fish

Details on estimation of bioconcentration:

BASIS FOR CALCULATION OF BCF
- Estimation software: US EPA T.E.S.T. v4.2.1

Applied estimation methods:
- Hierarchical method : The toxicity for a given query compound is estimated using the weighted average of the predictions from several different cluster models.
- FDA method : The prediction for each test chemical is made using a new model that is fit to the chemicals that are most similar to the test compound. Each model is generated at runtime.
- Single model method : Predictions are made using a multilinear regression model that is fit to the training set (using molecular descriptors as independent variables).
- Group contribution method : Predictions are made using a multilinear regression model that is fit to the training set (using molecular fragment counts as independent variables).
- Nearest neighbor method : The predicted toxicity is estimated by taking an average of the 3 chemicals in the training set that are most similar to the test chemical.
- Consensus method : The predicted toxicity is estimated by taking an average of the predicted toxicities from the above QSAR methods (provided the predictions are within the respective applicability domains; recommended method by T.E.S.T. for providing the most accurate predictions).

Type:

BCF

Value:

42.74 L/kg

Remarks on result:

other: method: Consensus method. Based on the mean absolute error, the confidence in the predicted BCF values is low.

Type:

BCF

Value:

91.82 L/kg

Remarks on result:

other: method: Hierachical clustering method. Based on the mean absolute error, the confidence in the predicted BCF values is high.

Type:

BCF

Value:

74.39 L/kg

Remarks on result:

other: method: Single model method. Based on the mean absolute error, the confidence in the predicted BCF values is low.

Type:

BCF

Value:

17.76 L/kg

Remarks on result:

other: method: Group contribution method. Based on the mean absolute error, the confidence in the predicted BCF values is low.

Type:

BCF

Value:

27.51 L/kg

Remarks on result:

other: method:FDA method. Based on the mean absolute error, the confidence in the predicted BCF values is high.

Method	Predicted value		Model statistics			MAE (in log10)
						External test set		Training set
	log BCF	BCF	r²	q²	No. of chemicals	Entire set	SC >= 0.5	Entire set	SC >= 0.5
Consensus method	1.63	42.74	-	-	-	0.51	N/A	0.42	0.49
Hierarchical clustering	1.96	91.82 (7.00-1205.16)	0.764 - 0.807	0.715 - 0.733	71 - 540 (cluster models: 3)	0.54	N/A	0.23	0.16
Single model	1.87	74.39 (5.70-971.46)	0.764	0.733	540	0.54	N/A	0.53	0.73
Group contribution	1.25	17.76 (0.50-627.44)	0.719	0.527	499	0.62	N/A	0.60	0.62
FDA	1.44	27.51 (3.04-249.15)	0.851	0.788	30	0.57	N/A	0.53	0.35
Nearest neighbor	N/A	N/A	-	-	3	0.60	N/A	0.55	N/A

Reference 4

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2019

Reliability:

2 (reliable with restrictions)

Justification for type of information:

The BCF models incorporated into the VEGA platform v.1.2.3 have been used to estimate the bioaccumulative potential of the compound in a weight-of-evidence approach.
- CAESAR v2.1.14
- Meylan v1.0.3
- KNN/Read-Across v1.1.0

Principles of method if other than guideline:

CAESAR
The BCF is estimated based on several molecular descriptors.
The applicability domain of predictions is assessed using an Applicability Domain Index (ADI) calculated by grouping several other indices, e.g. by a similarity index that consider molecule's fingerprint and structural aspects (count of atoms, rings and relevant fragments).

Meylan
The BCF is estimated based on log Kow. The applicability domain of predictions is assessed using an Applicability Domain Index (ADI) calculated by grouping several other indices, e.g. by a similarity index that consider molecule's fingerprint and structural aspects (count of atoms, rings and relevant fragments).

KNN/Read-Across
The model performs a read-across and provides a quantitative prediction of bioconcentration factor (BCF) in fish, given in log(L/kg). The read-across is based on the similarity index developed inside the VEGA platform; the index takes into account several structural aspects of the compounds, such as their fingerprint, the number of atoms, of cycles, of heteroatoms, of halogen atoms, and of particular fragments (such as nitro groups). On the basis of this structural similarity index, the three compounds from the dataset resulting most similar to the chemical to be predicted are taken into account: the estimated BCF value is calculated as the weighted average value of the experimental values of the three selected compounds, using their similarity values as weight.

Type:

BCF

Value:

75 L/kg

Remarks on result:

other: CAESAR model

Remarks:

Outside of the applicability domain but aregarded as adequate in a weight-of-evidence approach.

Type:

BCF

Value:

469 L/kg

Remarks on result:

other: Meylan model

Remarks:

Outside of the applicability domain but aregarded as adequate in a weight-of-evidence approach.

Type:

BCF

Value:

14 L/kg

Remarks on result:

other: KNN/Read-Across model

Remarks:

Outside of the applicability domain but aregarded as adequate in a weight-of-evidence approach.

CAESAR

Experimental value [log(L/kg)]: -

Predicted BCF [log(L/kg)]: 1.88

Predicted BCF [L/kg]: 75

Predicted BCF from sub-model 1 (HM) [log(L/kg)]: 1.84

Predicted BCF from sub-model 2 (GA) [log(L/kg)]: 1.92

Predicted LogP (MLogP): 3.26

Structural alerts: Carbonyl residue (SR 02); >C=O group (PG 09)

Reliability: the predicted compound is outside the Applicability Domain of the model

  

Meylan

Experimental value [log(L/kg)]: -

Predicted BCF [log(L/kg)]: 2.67

Predicted BCF [L/kg]: 469

Predicted LogP (Meylan/Kowwin): 4.55

Predicted LogP reliability: Low

MW: 205.27

Ionic compound: no

Reliability: the predicted compound is outside the Applicability Domain of the model

 

KNN/Read-Across

Experimental value [log(L/kg)]: -

Predicted BCF [log(L/kg)]: 1.16

Molecules used for prediction: 4

Reliability: the predicted compound is outside the Applicability Domain of the model

Reference 5

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a valid (Q)SAR model and falling into its applicability domain, with adequate and reliable documentation / justification

Principles of method if other than guideline:

Estimation of BCF, BAF and biotransformation rate using BCFBAF v3.01

GLP compliance:

no

Radiolabelling:

no

Test organisms (species):

other: fish

Details on estimation of bioconcentration:

BASIS INFORMATION
- Measured/calculated logPow: measured log Pow of 4,7

Type:

BCF

Value:

586 L/kg

Remarks on result:

other:

Remarks:

The substance is within the applicability domain of the BCFBAF submodel: Bioconcentration factor (BCF; Meylan et al., 1997/1999).

Type:

BCF

Value:

872 L/kg

Remarks on result:

other:

Remarks:

Upper trophic, incl. biotransformation estimates; The substance is within the applicability domain of the BCFBAF submodel: Arnot & Gobas BAF and steady-state BCF Arnot & Gobas, 2003).

Type:

BCF

Value:

4 554 L/kg

Remarks on result:

other:

Remarks:

Upper trophic, incl. biotransformation rate of zero; The substance is within the applicability domain of the BCFBAF submodel: Arnot & Gobas BAF and steady-state BCF Arnot & Gobas, 2003).

Details on kinetic parameters:

Biotransformation half-life (days):2.547 (normalised to 10 g fish)
Biotransformation rate (kM, normalised to 10 g fish at 15 °C): 0.2721 /day

Summary Results:

Log BCF (regression-based estimate): 2.77 (BCF = 586 L/kg wet-wt)

Biotransformation Half-Life (days) : 2.55 (normalized to 10 g fish)

Log BAF (Arnot-Gobas upper trophic): 2.95 (BAF = 886 L/kg wet-wt)

 

Log Kow (experimental): not available from database

Log Kow used by BCF estimates: 4.70 (user entered)

 

Equation Used to Make BCF estimate:

Log BCF = 0.6598 log Kow - 0.333 + Correction

 

Correction(s):                   Value

No Applicable Correction Factors

 

Estimated Log BCF = 2.768 (BCF = 586.2 L/kg wet-wt)

 

Whole Body Primary Biotransformation Rate Estimate for Fish:

TYPE	NUM	LOG BIOTRANSFORMATION FRAGMENT DESCRIPTION	COEFF	VALUE
Frag	1	Carbon with 4 single bonds & no hydrogens	-0.2984	-0.2984
Frag	1	Ketone  [-C-C(=O)-C-]	-0.1801	-0.1801
Frag	4	Methyl [-CH3]	0.2451	0.9804
Frag	1	-CH2- [linear]	0.0242	0.0242
Frag	2	-CH2- [cyclic]	0.0963	0.1925
Frag	1	-CH - [cyclic]	0.0126	0.0126
Frag	3	-C=CH [alkenyl hydrogen]	0.0988	0.2965
Frag	3	-C=CH [alkenyl hydrogen]	0.0000	0.0000
L Kow	*	Log Kow =  4.70 (user-entered  )	0.3073	1.4445
MolWt	*	Molecular Weight Parameter		-0.5291
Const	*	Equation Constant		-1.5371
RESULT	LOG Bio Half-Life (days		0.4061
RESULT	Bio Half-Life (days		2.547
NOTE	Bio Half-Life Normalized to 10 g fish at 15 deg C

 

Biotransformation Rate Constant:

kM (Rate Constant): 0.2721 /day (10 gram fish)

kM (Rate Constant): 0.153 /day (100 gram fish)

kM (Rate Constant): 0.08604 /day (1 kg fish)

kM (Rate Constant): 0.04839 /day (10 kg fish)

 

Arnot-Gobas BCF & BAF Methods (including biotransformation rate estimates):

Estimated Log BCF (upper trophic) = 2.940 (BCF = 871.6 L/kg wet-wt)

Estimated Log BAF (upper trophic) = 2.947 (BAF = 885.9 L/kg wet-wt)

Estimated Log BCF (mid trophic)  = 3.005 (BCF = 1012 L/kg wet-wt)

Estimated Log BAF (mid trophic)  = 3.040 (BAF = 1096 L/kg wet-wt)

Estimated Log BCF (lower trophic) = 3.015 (BCF = 1036 L/kg wet-wt)

Estimated Log BAF (lower trophic) = 3.104 (BAF = 1270 L/kg wet-wt)

 

Arnot-Gobas BCF & BAF Methods (assuming a biotransformation rate of zero):

Estimated Log BCF (upper trophic) = 3.658 (BCF = 4554 L/kg wet-wt)

Estimated Log BAF (upper trophic) = 4.526 (BAF = 3.357e+004 L/kg wet-wt)

Reference 6

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2019

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a valid (Q)SAR model, but not (completely) falling into its applicability domain, with adequate and reliable documentation / justification

Justification for type of information:

1. SOFTWARE
OASIS Catalogic v5.12.1

2. MODEL (incl. version number)
BCF base-line model v02.09 - July 2016

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
See section 'Test Material'.

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
See attached QMRF.

5. APPLICABILITY DOMAIN
See attached QPRF.

6. ADEQUACY OF THE RESULT
- The model is scientifically valid (see attached QMRF).
- The model estimates the Bioconcentration factor (BCF) as required information point according to Regulation (EC) No 1907/2006 [REACH], Annex IX, 9.3.2 Bioaccumulation in aquatic species (preferably fish);
further related predictions: Apparent effect of mitigating factors / Maximum bioconcentration factor (BCFmax) / Maximum diameter of energetically stable conformers / Whole body primary biotransformation half-life / Metabolic biotransformation rate constant Km / Metabolites and their quantitative distribution
- See attached QPRF for reliability assessment.

Principles of method if other than guideline:

Calculation using Catalogic v.5.11.19, BCF base-line model v.02.09

GLP compliance:

no

Details on estimation of bioconcentration:

BASIS FOR CALCULATION OF BCF
- Estimation software: OASIS Catalogic v5.12.1 [BCF base line model - v.02.09]

Type:

BCF

Value:

347 L/kg

Remarks on result:

other: considering all mitigating factors; the substance is not within the applicability domain of the model.

Type:

BCF

Value:

3 648 L/kg

Remarks on result:

other: without considering any mitigating factors; the substance is not within the applicability domain of the model.

Reference 7

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a valid (Q)SAR model and falling into its applicability domain, with adequate and reliable documentation / justification

Justification for type of information:

1. SOFTWARE
T.E.S.T. (version 4.2.1) (Toxicity Estimation Software Tool). US EPA, 2012.

2. MODEL (incl. version number)
T.E.S.T. (version 4.2.1)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
c32c(cccc3)nn(c1c(O)ccc(C(C)(C)CC(C)(C)C)c1)n2

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
See QMRF in section "Overall remarks, attachments".

5. APPLICABILITY DOMAIN
See QPRF in section "Executive summary".

6. ADEQUACY OF THE RESULT
- The model is scientifically valid (see QMRF).
- The model estimates the bioconcentration factor (BCF) as required information point under Regulation (EC) No 1907/2006 [REACH], Annex IX, 9.3.2 Bioaccumulation in aquatic species, preferably fish (see also QPRF).
- See QPRF for reliability assessment.

Principles of method if other than guideline:

T.E.S.T. is a toxicity estimation software tool. The program requires only the molecular structure of the test item, all other molecular descriptors which are required to estimate the toxicity are calculated within the tool itself. The molecular descriptors describe physical characteristics of the molecule (e.g. E-state values and E-state counts, constitutional descriptors, topological descriptors, walk and path counts, connectivity, information content, 2d autocorrelation, Burden eigenvalue, molecular property (such as the octanol-water partition coefficient), Kappa, hydrogen bond acceptor/donor counts, molecular distance edge, and molecular fragment counts). Each of the available methods uses a different set of these descriptors to estimate the toxicity. The bioaccumulation factor (BCF) was estimated using several available methods: hierarchical clustering method; FDA method, single model method; group contribution method; nearest neighbor method; consensus method. The methods were validated using statistical external validation using separate training and test data sets. The experimental data set was obtained from several different databases (Dimitrov et al., 2005; Arnot and Gobas, 2006; EURAS; Zhao, 2008). From the available data set salts, mixtures and ambiguous compounds were removed. The final data set contained 676 chemicals.

References:
- Dimitrov, S., N. Dimitrova, T. Parkerton, M. Combers, M. Bonnell, and O. Mekenyan. 2005. Base-line model for identifying the bioaccumulation potential of chemicals. SAR and QSAR in Environmental Research 16:531-554.
- Arnot, J.A., and F.A.P.C. Gobas. 2006. A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in aquatic organisms. Environ. Rev. 14:257-297.
- EURAS. Establishing a bioconcentration factor (BCF) Gold Standard Database. EURAS [cited 5/20/09]. Available from http://www.euras.be/eng/project.asp?ProjectId=92.
- Zhao, C.; Boriani, E.; Chana, A.; Roncaglioni, A.; Benfenati, E. 2008. A new hybrid system of QSAR models for predicting bioconcentration factors (BCF). Chemosphere 73:1701-1707.

GLP compliance:

no

Test organisms (species):

other: fish

Details on estimation of bioconcentration:

BASIS FOR CALCULATION OF BCF
- Estimation software: US EPA T.E.S.T. v4.2.1

Applied estimation methods:
- Hierarchical method : The toxicity for a given query compound is estimated using the weighted average of the predictions from several different cluster models.
- FDA method : The prediction for each test chemical is made using a new model that is fit to the chemicals that are most similar to the test compound. Each model is generated at runtime.
- Single model method : Predictions are made using a multilinear regression model that is fit to the training set (using molecular descriptors as independent variables).
- Group contribution method : Predictions are made using a multilinear regression model that is fit to the training set (using molecular fragment counts as independent variables).
- Nearest neighbor method : The predicted toxicity is estimated by taking an average of the 3 chemicals in the training set that are most similar to the test chemical.
- Consensus method : The predicted toxicity is estimated by taking an average of the predicted toxicities from the above QSAR methods (provided the predictions are within the respective applicability domains; recommended method by T.E.S.T. for providing the most accurate predictions).

Type:

BCF

Value:

39.95 L/kg

Remarks on result:

other: method: Consensus method. Based on the mean absolute error, the confidence in the predicted BCF values is low.

Type:

BCF

Value:

79.46 L/kg

Remarks on result:

other: method: Hierachical clustering method. Based on the mean absolute error, the confidence in the predicted BCF values is high.

Type:

BCF

Value:

74.94 L/kg

Remarks on result:

other: method: Single model method. Based on the mean absolute error, the confidence in the predicted BCF values is low.

Type:

BCF

Value:

21.54 L/kg

Remarks on result:

other: method: Group contribution method. Based on the mean absolute error, the confidence in the predicted BCF values is low.

Type:

BCF

Value:

6.27 L/kg

Remarks on result:

other: method:FDA method. Based on the mean absolute error, the confidence in the predicted BCF values is high.

Method	Predicted value		Model statistics			MAE (in log10)
						External test set		Training set
	log BCF	BCF	r²	q²	No. of chemicals	Entire set	SC >= 0.5	Entire set	SC >= 0.5
Consensus method	1.63	42.74	-	-	-	0.51	N/A	0.42	0.49
Hierarchical clustering	1.96	91.82 (7.00-1205.16)	0.764 - 0.807	0.715 - 0.733	71 - 540 (cluster models: 3)	0.54	N/A	0.23	0.16
Single model	1.87	74.39 (5.70-971.46)	0.764	0.733	540	0.54	N/A	0.53	0.73
Group contribution	1.25	17.76 (0.50-627.44)	0.719	0.527	499	0.62	N/A	0.60	0.62
FDA	1.44	27.51 (3.04-249.15)	0.851	0.788	30	0.57	N/A	0.53	0.35
Nearest neighbor	N/A	N/A	-	-	3	0.60	N/A	0.55	N/A

Reference 8

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2019

Reliability:

2 (reliable with restrictions)

Justification for type of information:

The BCF models incorporated into the VEGA platform v.1.2.3 have been used to estimate the bioaccumulative potential of the compound in a weight-of-evidence approach.
- CAESAR v2.1.14
- Meylan v1.0.3
- KNN/Read-Across v1.1.0

Principles of method if other than guideline:

CAESAR
The BCF is estimated based on several molecular descriptors.
The applicability domain of predictions is assessed using an Applicability Domain Index (ADI) calculated by grouping several other indices, e.g. by a similarity index that consider molecule's fingerprint and structural aspects (count of atoms, rings and relevant fragments).

Meylan
The BCF is estimated based on log Kow. The applicability domain of predictions is assessed using an Applicability Domain Index (ADI) calculated by grouping several other indices, e.g. by a similarity index that consider molecule's fingerprint and structural aspects (count of atoms, rings and relevant fragments).

KNN/Read-Across
The model performs a read-across and provides a quantitative prediction of bioconcentration factor (BCF) in fish, given in log(L/kg). The read-across is based on the similarity index developed inside the VEGA platform; the index takes into account several structural aspects of the compounds, such as their fingerprint, the number of atoms, of cycles, of heteroatoms, of halogen atoms, and of particular fragments (such as nitro groups). On the basis of this structural similarity index, the three compounds from the dataset resulting most similar to the chemical to be predicted are taken into account: the estimated BCF value is calculated as the weighted average value of the experimental values of the three selected compounds, using their similarity values as weight.

Type:

BCF

Value:

43 L/kg

Remarks on result:

other: CAESAR model

Remarks:

Outside of the applicability domain but aregarded as adequate in a weight-of-evidence approach.

Type:

BCF

Value:

430 L/kg

Remarks on result:

other: Meylan model

Remarks:

Outside of the applicability domain but aregarded as adequate in a weight-of-evidence approach.

Type:

BCF

Value:

14 L/kg

Remarks on result:

other: KNN/Read-Across model

Remarks:

Outside of the applicability domain but aregarded as adequate in a weight-of-evidence approach.

CAESAR

Experimental value [log(L/kg)]: -

Predicted BCF [log(L/kg)]: 1.63

Predicted BCF [L/kg]: 43

Predicted BCF from sub-model 1 (HM) [log(L/kg)]: 1.6

Predicted BCF from sub-model 2 (GA) [log(L/kg)]: 1.92

Predicted LogP (MLogP): 3.26

Structural alerts: Carbonyl residue (SR 02); >C=O group (PG 09)

Reliability: the predicted compound is outside the Applicability Domain of the model

  

Meylan

Experimental value [log(L/kg)]: -

Predicted BCF [log(L/kg)]: 2.63

Predicted BCF [L/kg]: 430

Predicted LogP (Meylan/Kowwin): 4.5

Predicted LogP reliability: Moderate

MW: 205.27

Ionic compound: no

Reliability: the predicted compound is outside the Applicability Domain of the model

 

KNN/Read-Across

Experimental value [log(L/kg)]: -

Predicted BCF [log(L/kg)]: 1.15

Molecules used for prediction: 4

Reliability: the predicted compound is outside the Applicability Domain of the model

Description of key information

There is a high probability that the reaction mass does not accumulate in organisms.

Additional information

The bioaccumulative potential of the mixture of isomers was estimated in a weight of evidence approach using different QSAR models (Table 1) and molecular parameters (i.e. logKow, molecular weight and molecular dimensions). Therefore the two isomers with the biggest contingent (65mol% and 22 mol%) in the reaction mass were evaluated (Table 1).

Table 1: Used isomers reaction mass.

Component	Amount in mol% in reaction mass	Smilescode	Chemical Name
01	65	CC(=O)\C(=C\C1C(=CCCC1(C)C)C)\C	3-methyl-4-(2,6,6-trimethylcyclohex-2-en-1-yl)but-3-en-2-one
02	22	CCC(=O)\C=C\C1C(=CCCC1(C)C)C	1-(2,6,6-trimethylcyclohex-2-en-1-yl)pent-1-en-3-one

 

As supporting information several QSAR models were used to calculate the BCF (Table 2). The results and further information on the reliability (e.g. applicability domain) can be found in Table 3.

 

Table 2: QSAR models used in the weight of evidence approach.

Software	Version	Model	Version	Sub-model
EPISuite	v4.11	BCFBAF model	v3.01	Regression-based estimate
				Arnot-Gobas BCF & BAF methods
VEGA	v1.1.3	CAESAR	v2.1.14
		Meylan	v1.0.3
		KNN/Read-Across	v1.1.0
T.E.S.T	v4.2.1	Hierarchical clustering
		FDA method
		Single model
		Group contribution
		Nearest neighbor
		Consensus
OASIS Catalogic	v5.13.1	BCF base-line model	v03.10

 

 

 

 

Table 3: Results and further details of the supporting QSAR estimations.

Compound		01		02
SMILES code		CC(=O)\C(=C\C1C(=CCCC1(C)C)C)\C		CCC(=O)\C=C\C1C(=CCCC1(C)C)C
Chemical Name		3-methyl-4-(2,6,6-trimethylcyclohex-2-en-1-yl)but-3-en-2-one		1-(2,6,6-trimethylcyclohex-2-en-1-yl)pent-1-en-3-one
Software	Model	Results [L/kg]	Remarks	Results [L/kg]	Remarks
EPISuite	BCFBAF model Regression-based estimate	586	within the applicability domain	586	within the applicability domain
	BCFBAF model Arnot-Gobas BCF & BAF methods incl. biotransformation rate estimates	1087	within the applicability domain	872	within the applicability domain
	BCFBAF model Arnot-Gobas BCF & BAF methods incl. biotransformation rate of zero	4554	within the applicability domain	4554	within the applicability domain
VEGA	CAESAR	75	outside the applicability domain	43	outside the applicability domain
	Meylan	469	outside the applicability domain	430	outside the applicability domain
	KNN/Read-Across	14	outside the applicability domain	14	outside the applicability domain
T.E.S.T	Hierarchical clustering	91.82	- within applicability domain - Confidence in the estimated BCF: high	79.46	- within applicability domain - Confidence in the estimated BCF: high
	Single model	74.39	- within applicability domain - Confidence in the estimated BCF: low	74.94	- within applicability domain - Confidence in the estimated BCF: low
	Group contribution	17.76	- within applicability domain - Confidence in the estimated BCF: low	21.54	- within applicability domain - Confidence in the estimated BCF: low
	FDA method	27.51	- within applicability domain - Confidence in the estimated BCF: high	6.27	- within applicability domain - Confidence in the estimated BCF: high
	Nearest neighbor	n/a	n/a	n/a	n/a
	Consensus	42.74	- within applicability domain - Confidence in the estimated BCF: low	39.95	- within applicability domain - Confidence in the estimated BCF: low
OASIS Catalogic	BCF base-line model	3090 [BCFmax = 3999]	- all mitigating factors applied - within parameter and mechanistic domain - outside structural fragment domain	347 [BCFmax = 3648]	- all mitigating factors applied - within parameter and mechanistic domain - outside structural fragment domain

 

 

 

EPISuite includes the BCFBAF model which encompasses two BCF estimation methodologies. The regression-based model is based on the work from Meylan et al. (1997 and 1999). The second model is based on the works from Arnot and Gobas (2003) and includes estimates on the biotransformation rate in fish. The regression-based model derived a BCF value of 586 L/kg for both compounds based on an experimental determined logKow of 4.7. The model did not apply any correction factors and derived the BCF by applying common statistical regression methodology (n = 396; r2= 0.792). As the substance is within the applicability domain of the model (molecular weight and logKow ranges) the result is regarded as reliable information in a weight of evidence approach. Additionally both models showed similar results for the BCF determination for both compounds. Therefore the result of the BCFBAF model is regarded as a reliable information for an estimation of bioaccumulative properties of the reaction mass.

The model based on the works of Arnot and Gobas (2003) takes the biotransformation rate of the compound into account. The model derived a BCF value of 1087 L/kg (Compound 01) resp. 872 (Compound 02). As the substance is within the applicability domain of the model (molecular weight and logKow ranges) the result is regarded as reliable information in a weight of evidence approach. Additionally both models showed a quite similar result for the BCF determination for both compounds. Therefore the result of the model based on the works of Arnot and Gobas is regarded as a reliable information for an estimation of bioaccumulative properties of the reaction mass.

 

The current VEGA package comprises three different models. (1) CAESAR, (2) Meylan and (3) KNN/Read-Across. The applicability domain of the predictions is assessed using an Applicability Domain Index (ADI) that has values from 0 (worst case) to 1 (best case). The ADI is calculated by grouping several other indices, each one taking into account a particular issue of the applicability domain. Most of the indices are based on the calculation of the most similar compounds found in the training and test set of the model, calculated by a similarity index that consider molecule's fingerprint and structural aspects (count of atoms, rings and relevant fragments).

 In regards to the CAESAR model the following indices are checked: (1) Similar molecules with known experimental value, (2) accuracy (average error) of prediction for similar molecules, (3) concordance with similar molecules (average difference between target compound prediction and experimental values of similar molecules), (4) maximum error of prediction among similar molecules, (5) atom centered fragments similarity check, (6) model descriptors range check and (7) global AD index which takes into account all the previous indices in order to give a general global assessment on the applicability domain. The model predicted BCF values of 75 L/kg (Compound 01) resp. 43 L/kg (Compound 02) but the substance was out of the applicability domain. Therefore, the results are not regarded as reliable and the model was not used for determination of the bioaccumulative potential of the reaction mass.

The Meylan model is a reconstruction of the regression-based model integrated in EPISuite (Meylan et al. 1997, 1999). The original dataset from EPISuite has been processed and cleared from duplicates and compounds provided with structures that had problems. The final dataset has 662 compounds. Similar to the CAESAR model, the applicability domain is assessed with several indices. (1) Similar molecules with known experimental value, (2) accuracy (average error) of prediction for similar molecules, (3) concordance with similar molecules (average difference between target compound prediction and experimental values of similar molecules), (4) maximum error of prediction among similar molecules, (5) logP reliability, (6) model descriptors range check and (7) global AD index. The model predicted BCF values of 469 L/kg (Compound 01) resp. 430 L/kg (Compound 02) but the substance was out of the applicability domain. Therefore, the results are not regarded as reliable and the model was not used for determination of the bioaccumulative potential of the reaction mass.

The KNN/Read-Across model performs a read-across on a dataset of 860 chemicals. The applicability domain takes the following indices into account: (1) Similar molecules with known experimental value, (2) accuracy (average error) of prediction for similar molecules, (3) concordance with similar molecules (average difference between target compound prediction and experimental values of similar molecules), (4) maximum error of prediction among similar molecules, (5) atom centered fragments similarity check, (6) global AD index. The model predicted BCF values of 14 L/kg (Compound 01) resp. 14 L/kg (Compound 02) but the substance was out of the applicability domain. Therefore, the results are not regarded as reliable and the model was not used for determination of the bioaccumulative potential of the reaction mass.

 

The T.E.S.T. package encompasses five separate methods to estimate the BCF. The sixth method is the consensus method which simply averages the results of the prediction from the other QSAR methodologies (taking into account the applicability domain of each method). This method typically provides the highest prediction accuracy since errant predictions are dampened by the predictions from the other methods. In addition, this method provides the highest prediction coverage because several methods with slightly different applicability domains are used to make a prediction. For the five separate methodologies, the results range from 17.76 to 91.82 (Compound 01) resp. 6.27 – 79.46 (Compound 02). The averaged consensus result is 42.74 (Compound01) resp. 39.95 (Compound 02). The models only make predictions if the substance is within the respective applicability domains. For the present case, the substance is within the applicability domains of four of the five models. However, the mean absolute errors regarding the similarity coefficients for both the external test sets and the training sets are above the respective thresholds. Therefore, the confidence in the estimated BCF values is low. Nevertheless, the result of the consensus method clearly showed that the estimated BCFs are way below the regulatory threshold. Therefore, the model is regarded as reliable for the estimation of the bioaccumulative potential of the reaction mass.

 

The BCF base-line model integrated in OASIS Catalogic reflects the current understanding of the process by which lipophilic organic chemicals are bioaccumulated in fish through the respiratory organs only. Chemicals bioaccumulating by other mechanisms (e.g., binding to proteins) are considered out of the mechanistic domain of the model. The model consists of two major components: a model for predicting the maximum potential for bioaccumulation based solely on chemicals’ lipophilicity (i.e., BCFmax model), and a set of mitigating factors that account for the reduction of the bioaccumulation potential of chemicals based on chemical (e.g., molecular size and ionization) and organism-dependent factors (e.g., metabolism). BCFmax model is a theoretical model based on the assumption that the only driving force of bioconcentration is lipophilicity and the effect of any other factors are insignificant. It mathematical formalism is derived considering multi-compartment diffusion. The bioconcentration predicted by BCFmax model could be limited by variety of mitigating factors that account for the reduction of the bioaccumulation potential of chemicals based on chemical and organism-dependent factors. The effect of mitigating factors mathematically is quantified by probabilities: to penetrate through the cell membrane, to be ionized, to be metabolised, etc. In the BCF base-line model the tissue metabolism simulator is used to account for the effect of metabolism. It consists of a consequence of spontaneous abiotic and enzyme controlled steps. Probabilities of these molecular transformations are assessed by fitting the training set data. The CATALOGIC platform utilizes a multi-stage applicability domain that has been described by Dimitrov et al. (2005). The applicability domain of the BCF base-line model contains three layers: (1) General properties requirements. These requirements specify in the domain only those chemicals that fall in the range of variation of physicochemical properties that may affect significantly the quality of the measured endpoint. For the BCF base-line model attention is focused on lipophilicity (log KOW), molecular weight (MW) and water solubility (WS). Only correctly predicted chemicals from the training set are used to determine the range of variation of these properties. (2) The structural domain. It determines the maximum structural similarity between the target chemical and chemicals from the training set. The structural neighborhood of atom-centered fragments (ACF) accounting for 1st neighbors, atom type, hybridization and attached hydrogen atoms are used to determine this similarity. The target chemical could contain the following types of ACF:

 

-      Fragments present in correctly predicted training chemicals only (i.e. correct fragments)

-      Fragments found both in correctly and non-correctly predicted training chemicals (i.e. fuzzy fragments). These fragments are treated as correct fragments

-      Fragments present in non-correctly predicted training chemicals only (i.e. incorrect fragments),

-      Fragments not present in the training chemicals (i.e. unknown fragments).

 

(3) The mechanistic domain.It discriminates between modes of bioaccumulation - passive (partitioning in lipid phase) or active (based on protein binding). Only chemicals with expected passive diffusion driven bioaccumulation are considered to be in the mechanistic domain of the model.

 

In the present case, both isomers fulfill the general properties requirements, i.e. its logKow, molecular weight and water solubility are within the ranges of the model. Furthermore, both isomers are within the mechanistic domain, i.e. it is expected to be taken up by passive diffusion only. However, both isomers are not in the structural domain. For component 01 and component 02 only 46.67% resp. 53.33% of its fragments could be found in correctly predicted training set chemicals. The remaining 53.33% resp. 46.67% are not present in the training set chemicals. If a chemical is out of at least one of the specified layers mentioned above, it will be classified as out of the applicability domain. This classification means that the prediction falls in the extrapolation space but the prediction still could be reliable. The estimated BCF (all mitigating factors applied) were determined to be 3090 L/kg (Compound 01) resp. 347 L/kg (Compound 02). For Compound 01 no mitigating effect of Metabolism was taken into account, resulting in similar values for the calculated BCF (3090 L/kg) and the calculated BCF max (3999 L/kg). As it is assumable, that the molecule will be metabolized and as the reliability of the model calculation for Compound 01 is low, this value is considered as not reliable.

The results of the calculation using VEGA and OASIS Catalogic were regarded as not reliable and therefore not used in the bioaccumulative assessment. According to the results of EPISuite and the T.E.S.T. package based on the two major compounds, the reaction mass may regarded as being not bioaccumulative.

 

Conclusion

The target of this analysis was a reaction mass consisting of 4 identified compounds of which two compounds (Component 01 and Component 02) were considered in an analysis regarding the bioaccumulative properties. According the chemical identification report provided in chapter 1.4 Analytical Information, these two components represent the majority of the reaction mass. Additionally the two other components found in the reaction mass show a very similar structure to the two chosen main components. Therefore, it can be assumed, that the analysis of the two compounds represents the whole reaction mass.

Regarding the outcome of most of the used models for the two main components, a bioaccmulative potential regarding the BCF value was not shown. Although one model for one of the compounds calculated a BCF value higher than 2000, the result of the model is regarded as not reliable. Taking into account only the results of models were a suitable reliability was given, there is a high probability that the reaction mass is not bioaccumulative.

 

 

References

Meylan, W.M., Howard, P.H, Aronson, D., Printup, H. and S. Gouchie.  1997.  "Improved Method for Estimating Bioconcentration Factor (BCF) from Octanol-Water Partition Coefficient", SRC TR-97-006 (2nd Update), July 22, 1997; prepared for: Robert S. Boethling, EPA-OPPT, Washington, DC; Contract No. 68-D5-0012; prepared by: ; Syracuse Research Corp., Environmental Science Center, 6225 Running Ridge Road, North Syracuse, NY 13212.

 Meylan,WM, Howard,PH, Boethling,RS et al. 1999.  Improved Method for Estimating Bioconcentration / Bioaccumulation Factor from Octanol/Water Partition Coefficient. Environ. Toxicol. Chem. 18(4): 664-672 (1999).

Arnot JA, Gobas FAPC. 2003. A generic QSAR for assessing the bioaccumulation potential of organic chemicals in aquatic food webs.QSAR and Combinatorial Science 22: 337-345.

Dimitrov S, Dimitrova N, Parkerton T, Comber M, Bonnell M and Mekenyan O. “Base-line model for identifying the bioaccumulation potential of chemicals”, SAR and QSAR in Environmental Research 16(6), 2005