4,4'-Isopropylidenediphenol, oligomeric reaction products with 1-chloro-2,3-epoxypropane, reaction products with m-phenylenebis(methylamine)

Bioaccumulation: aquatic / sediment

Link to relevant study record(s)

Reference

Endpoint:

bioaccumulation in aquatic species: fish

Type of information:

(Q)SAR

Adequacy of study:

key study

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

other: Well documented prediction using analogues for validation. Prediction within parametric domain; however, only within 50% of structural domain of QSAR prediction model.

Justification for type of information:

1. SOFTWARE
Model: OASIS CATALOGIC v. 5.11.3.
Submodel: BCF baseline model v. 01.02.

2. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
Constituent 1: c1(C(C)(C)c2ccc(OC3CO3)cc2)ccc(OCC2CO2)cc1
Constituent 2: c1(C(C)(C)c2ccc(OC3CO3)cc2)ccc(OCC(O)COc2ccc(C(C)(C)c3ccc(OC4CO4)cc3)cc2)cc1
Product of hydrolysis, HP: c1(C(C)(C)c2ccc(OC(O)CO)cc2)ccc(OCC(O)CO)cc1

3. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
- Defined endpoint: The model estimates whole fish steady state bioconcentration factor (BCF). The training set consists of 396 authentic BCF test data with Cyprinos caprio (MITI database,) and 118 BCF data estimated from dietary bioaccumulation test results with salmonids (Exxon Mobile).
Estimation BCF is based on a maximum BCF and mitigating factors that reduce the BCF. The maximum bioconcentration potential is calculated for a multi-compartment partitioning model and passive diffusion. The most relevant mitigating factor is metabolism, which is accounted for by means of a fish liver model, which has been trained to reproduce fish metabolism based on related mammalian metabolic pathways. Other mitigating factors like molecular size and ionization are also taken into account in the model (Dimitrov et al. 2005).


4. APPLICABILITY DOMAIN
i. Descriptor domain
1. Predicted Log Kow: -4.05 2. Molecular weight (MW):16.04 < MW < 1131.21: in domain
3. Predicted Water solubility (Csat): 0ii. Structural fragment domain: Structural domain is represented by the list of atom-centred fragments (accounting for the first neighbours) extracted fromtraining set chemicals which are correctly predicted. A correct prediction was assumed for those chemicals for which the residuals between predicted and observed values were less than or equal to 0.75 log units.
iii. Mechanistic domain: The model prediction is based on the assumption of a maximal uptake through passive diffusion and mitigating factors in the uptake (size, dissociation) or the elimination (metabolism).
In order to verify the relevance of the model assumption for the unknown, structural analogues were selected from the training set and processed in parallel
iv. Metabolic domain, if relevant: Covered by the structural domain.

5. ADEQUACY OF THE RESULT
The model prediction is based on the assumption of a maximal uptake through passive diffusion and mitigating factors in the uptake (size, dissociation) or the elimination (metabolism). Based on comparison of predicted BCF and experimental BCF as reported in the training set for a series of analogues, it is anticipated that the underlying assumptions of the model can be applied to the unknowns. The result is considered to adequate for risk assessment.

Principles of method if other than guideline:

Estimation of the bioconcentration factor (BCF) of BADGE-MXDA using CATABOL BCF baseline model v. 1.02

GLP compliance:

no

Remarks:

Not applicable for QSAR

Type:

BCF

Value:

4.77 dimensionless

Basis:

whole body w.w.

Calculation basis:

other: QSAR

Remarks on result:

other: Conc.in environment / dose:Not applicable

a.     Predicted value (model result):        

Table 1a: Log BCF and BCF prediction for unknowns.

Molecule	log BCF^calc	BCF
Constituent 1)	0.68± 0.11	4.77 (3.7–6.17)

Table 1b: Intermediate Results from LogBCF prediction.

	Log KOW	Log BCFmax	log BCF^calc	mitigating effect of :
				Acids	Meta­bolism	Phenols	Size3	Water solubility
1)	3.61	2.71	1.11	0	0.234	0	0.00	0.028

The developers of the QSAR present a decision rule for the interpretation of the predicted Log BCF based on the assessment of their training and validation data (See Table 1c). Based on this decision rule the BADGE MXDA adduct is not bioaccumulative with high confidence.

Table 1c: Decision rule for the interpretation of predicted Log BCF values.

Calculated BCF	Conclusion
Log BCF^clalc≥ 3.699+0.75	Bioaccumulative – high confidence
3.699 < Log BCF^clalc< 3.699+0.75	Bioaccumulative – low confidence
3.699 - 0.75< Log BCF^clalc< 3.699	Not bioaccumulative – low confidence
Log BCF^clalc< 3.699 - 0.75	Not Bioaccumulative – high confidence

 

b.     Predicted value (comments):Metabolism and water solubility are the most relevant mitigating factor for BADGE-MXDA adduct.

c.      Input for prediction: SMILES as specified under 1.5a

d.     Descriptor values:    Prediction is based on Kow prediction based on Epiwin(US Environmental Protection Agency and Syracuse Research Corporation (SRC) 2008). Table 1b gives intermediate results used for in the calculation.

Executive summary:

The bioconcentration potential of the two major constituents in technical Bisphenol A diglycidyl ether (BADGE) and the product of hydrolysis have been assessed using a QSAR algorithm as described by Dimitrov et al (2005) and implemented in OASIS CATALOGIC. The model estimates steady state whole fish bioconcentration factor (BCF) based on a maximum BCF and mitigating factors that reduce the BCF.

Structural analogues identified from the training set have been processed parallel to the unknown. The unknowns are within the parametric domain of the model but within less than 50% of the structural domain as defined by first neighbour atom centred fragments.

To assess the relevance of the metabolism as mitigating factor for the unknowns, a comparison of the metabolic reactions proposed for the analogues and the unknowns is made. Oxidation of the epoxide to form a diol is predicted as most relevant metabolic reaction for the epoxide among the analogues and Glucuronidation on the terminal alcohol group is predicted for the analogue similar to the diol. Based on these findings the assumed metabolic reactions are considered relevant for the unknowns and the predicted corrected BCF values are considered reliable.

Assuming that the prediction falls within the range of accurate predictions as defined by the developer the following BCF are calculated for the constituents of BADGE:

 

Molecule		log BCF^calc	BCF[(mg/kg w.w.) /(mg/L)]
Constituent 1)	c1(C(C)(C)c2ccc(OC3CO3)cc2)ccc(OCC2CO2)cc1	1.11 ± 0.75	31 (5.–173)
Constituent 2)	c1(C(C)(C)c2ccc(OC3CO3)cc2)ccc(OCC(O)COc2ccc(C(C)(C)c3ccc(OC4CO4)cc3)cc2)cc1	1.49 ± 0.75	13 (2–72)
Product of hydrolysis, HP	c1(C(C)(C)c2ccc(OC(O)CO)cc2)ccc(OCC(O)CO)cc1	0.52 ± 0.75	3 (1–19)

Based on a decision rule derived by the developer of the QSAR model, Bisphenol A diglycidyl ether (BADGE) and the product of hydrolysis are not bioaccumulative with high confidence.

Description of key information

Well documented prediction using analogues for validation. Prediction within parametric domain; however, only within 50% of structural domain of QSAR prediction model.

Key value for chemical safety assessment

BCF (aquatic species):

4.77 L/kg ww

Additional information

The bioconcentration potential of the reaction mass constituents in BADGE MXDA adduct have been assessed using a QSAR algorithm as described by Dimitrov et al (2005) and implemented in OASIS CATALOGIC. The model estimates steady state whole fish bioconcentration factor (BCF) based on a maximum BCF and mitigating factors that reduce the BCF.

Structural analogues identified from the training set have been processed parallel to the unknown. The unknowns are within the parametric domain of the model but within less than 50% of the structural domain as defined by first neighbour atom centred fragments.

Assuming that the prediction falls within the range of accurate predictions as defined by the developer the following BCF are calculated for the reaction mass BAGE MXDA adduct constituents:

 

Molecule	SMILES used to describe the Material	log BCF^calc	BCF[(mg/kg w.w.) /(mg/L)]
BAGE MXDA adduct constituents	c1(C(C)(C)c2ccc(OCC(O)CNCc3cc(CN)ccc3)cc2)ccc(OCC(O)CNCc2cc(CN)ccc2)cc1	0.68 ± 0.75	4.77(3.7–6.17)

Based on a decision rule derived by the developer of the QSAR model, the reaction mass of BADGE MXDA are not bioaccumulative with high confidence.