1,2-Ethanediamine, N-(2-aminoethyl)-, reaction products with glycidyl tolyl ether

Endpoint summary

Key value for chemical safety assessment

Effects on fertility

Description of key information

The substance 1,2-Ethanediamine, N-(2-aminoethyl)-, reaction products with glycidyl tolyl ether is a multiconstituent substance which predominantly consists of secondary aliphatic amines, with the exception of one CGE dimer.

Various QSAR models were run on the identified constituents of the substance. Predictions with the CAESAR model revealed some positive predictions for developmental toxicity. The DART model profiler showed positive results of the unspecific category Known precedent reproductive and developmental toxic potential; Toluene and small alkyl toluene derivatives (8a) for all structures of the substance. And in the CAT-SAR system, for Human Developmental Toxicity the CGE-DIMER revealed equivocal results for developmental toxicity.

On the basis of positive indication for developmental effects from the various QSAR models it is proposed to conduct an OECD 414 Prenatal Developmental Toxicity Study (Annex IX 8.7.2) in order to clarify specifically this concern.

Thus, an OECD 421 Reproductive and Developmental Toxicity Screening Test would provide less information and is therefore waived in accordance with column 2 of REACH Annex VIII 8.7.1.

Link to relevant study records

Reference

Reference 1

Endpoint:

screening for reproductive / developmental toxicity

Data waiving:

study scientifically not necessary / other information available

Justification for data waiving:

a screening study for reproductive/developmental toxicity does not need to be conducted because there is evidence from available information on structurally related substances, from (Q)SAR estimates or from in vitro methods that the substance may be a developmental toxicant

Reference 2

Endpoint:

reproductive toxicity, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2016

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

Gnarus systems CAT-SAR system (http://www.gnarus-systems.com/how-it-works/cat-sar-overview/)

2. MODEL (incl. version number)

Cat-SAR Human Developmental Toxicity - Mattison (QMRF 1.1)
February 2, 2016 (QMRF 2.6)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL

ME-CGE-DETA structures with same molecular formula (C14 H25 N3 O2) and molecular weight (267.37)

2-ME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCN

3-ME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCN)=CC=C1

4-ME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCN)C=C1


4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODELS

Human Teratogenicity (i.e., developmental toxicity): This model is derived from a dataset of compounds analyzed for potential human teratogenicity developed by Dr. Donald Mattison. The model contains developmental toxicity calls for 323 chemicals, 130 of which are classified as human teratogens and 193 as non-teratogens.

Ghanooni, M., D.R. Mattison, Y.P. Zhang, O.T. Macina, H.S. Rosenkranz, and G. Klopman (1997) Structural determinants associated with risk of human developmental toxicity. American Journal of Obstetrics and Gynecology, 176: 799-806.

5. APPLICABILITY DOMAIN

Generally--the applicable domain of this model is small organic molecules that would fit the general attributes of compounds contained the Human Developmental Toxicity database

Cat-SAR, unlike some other SAR expert systems, does not use default predictions. Chemicals have to be either predicted as active or inactive on account of fragments contained in their structure. If no identical fragment is found in an unknown chemical that meets rules 1-3 in section 4.4 of this QMRF, no prediction is made. In this fashion, the applicable domain is determined on a one-by-one basis.


6. ADEQUACY OF THE RESULT

Overall, the prediction of “inactive” for structural analogues of DIME-CGE-DETA were based on 17 fragments that individually related back to 263 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 739 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Guideline:

other: REACH Guidance on QSARS R.6

Principles of method if other than guideline:

Ghanooni, M., D.R. Mattison, Y.P. Zhang, O.T. Macina, H.S. Rosenkranz, and G. Klopman (1997) Structural determinants associated with risk of human developmental toxicity. American Journal of Obstetrics and Gynecology, 176: 799-806.

Specific details on test material used for the study:

CGE-DIMER
CC1=CC=C(OCC(O)COC2=CC=C(C)C=C2)C=C1

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Reproductive effects observed:

not specified

Overall, the prediction of “inactive” for structural analogues of DIME-CGE-DETA were based on 31 fragments that individually related back to 739 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 739 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).  

Conclusions:

Overall, the prediction of “equivocal” for compound CGE-DIMER was based on 17 fragments that individually related back to 263 compounds in the Human Developmental Toxicity Mattison learning set with 39 of the 263 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Reference 3

Endpoint:

reproductive toxicity, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2016

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

Gnarus systems CAT-SAR system (http://www.gnarus-systems.com/how-it-works/cat-sar-overview/)

2. MODEL (incl. version number)

Cat-SAR Human Developmental Toxicity - Mattison (QMRF 1.1)
February 2, 2016 (QMRF 2.6)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL

DIME-CGE-DETA structures with same molecular formula (C24 H37 N3 O4) and molecular weight (431.58)

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1


4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODELS

Human Teratogenicity (i.e., developmental toxicity): This model is derived from a dataset of compounds analyzed for potential human teratogenicity developed by Dr. Donald Mattison. The model contains developmental toxicity calls for 323 chemicals, 130 of which are classified as human teratogens and 193 as non-teratogens.

Ghanooni, M., D.R. Mattison, Y.P. Zhang, O.T. Macina, H.S. Rosenkranz, and G. Klopman (1997) Structural determinants associated with risk of human developmental toxicity. American Journal of Obstetrics and Gynecology, 176: 799-806.

5. APPLICABILITY DOMAIN

Generally--the applicable domain of this model is small organic molecules that would fit the general attributes of compounds contained the Human Developmental Toxicity database

Cat-SAR, unlike some other SAR expert systems, does not use default predictions. Chemicals have to be either predicted as active or inactive on account of fragments contained in their structure. If no identical fragment is found in an unknown chemical that meets rules 1-3 in section 4.4 of this QMRF, no prediction is made. In this fashion, the applicable domain is determined on a one-by-one basis.


6. ADEQUACY OF THE RESULT

Overall, the prediction of “inactive” for structural analogues of DIME-CGE-DETA was based on 31 fragments that individually related back to 739 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 739 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Guideline:

other: REACH Guidance on QSARS R.6

Principles of method if other than guideline:

Ghanooni, M., D.R. Mattison, Y.P. Zhang, O.T. Macina, H.S. Rosenkranz, and G. Klopman (1997) Structural determinants associated with risk of human developmental toxicity. American Journal of Obstetrics and Gynecology, 176: 799-806.

Specific details on test material used for the study:

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Reproductive effects observed:

not specified

Overall, the prediction of “inactive” for structural analogues of DIME-CGE-DETA were based on 31 fragments that individually related back to 739 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 739 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).  

Conclusions:

Overall, the prediction of “inactive” for structural analogues of DIME-CGE-DETA were based on 31 fragments that individually related back to 739 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 739 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Reference 4

Endpoint:

reproductive toxicity, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2016

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

Gnarus systems CAT-SAR system (http://www.gnarus-systems.com/how-it-works/cat-sar-overview/)

2. MODEL (incl. version number)

Cat-SAR Human Developmental Toxicity - Mattison (QMRF 1.1)
February 2, 2016 (QMRF 2.6)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL

ME-CGE-DETA structures with same molecular formula (C14 H25 N3 O2) and molecular weight (267.37)

2-ME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCN

3-ME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCN)=CC=C1

4-ME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCN)C=C1


4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODELS

Human Teratogenicity (i.e., developmental toxicity): This model is derived from a dataset of compounds analyzed for potential human teratogenicity developed by Dr. Donald Mattison. The model contains developmental toxicity calls for 323 chemicals, 130 of which are classified as human teratogens and 193 as non-teratogens.

Ghanooni, M., D.R. Mattison, Y.P. Zhang, O.T. Macina, H.S. Rosenkranz, and G. Klopman (1997) Structural determinants associated with risk of human developmental toxicity. American Journal of Obstetrics and Gynecology, 176: 799-806.

5. APPLICABILITY DOMAIN

Generally--the applicable domain of this model is small organic molecules that would fit the general attributes of compounds contained the Human Developmental Toxicity database

Cat-SAR, unlike some other SAR expert systems, does not use default predictions. Chemicals have to be either predicted as active or inactive on account of fragments contained in their structure. If no identical fragment is found in an unknown chemical that meets rules 1-3 in section 4.4 of this QMRF, no prediction is made. In this fashion, the applicable domain is determined on a one-by-one basis.


6. ADEQUACY OF THE RESULT

Overall, the prediction of “inactive” for structural analogues of DIME-CGE-DETA were based on 30 fragments that individually related back to 724 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 739 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Guideline:

other: REACH Guidance on QSARS R.6

Principles of method if other than guideline:

Ghanooni, M., D.R. Mattison, Y.P. Zhang, O.T. Macina, H.S. Rosenkranz, and G. Klopman (1997) Structural determinants associated with risk of human developmental toxicity. American Journal of Obstetrics and Gynecology, 176: 799-806.

Specific details on test material used for the study:

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1

Clinical signs:

not examined

Reproductive function: oestrous cycle:

not examined

Remarks on result:

other: QSAR Predictied Value

Clinical signs:

not examined

Clinical signs:

not examined

Behaviour (functional findings):

not examined

Dose descriptor:

other:

Remarks:

QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Clinical signs:

not examined

Behaviour (functional findings):

not examined

Reproductive effects observed:

not specified

Overall, the prediction of “inactive” for structural analogues of DIME-CGE-DETA were based on 31 fragments that individually related back to 739 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 739 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).  

Conclusions:

Overall, the prediction of “inactive” for compound 2-ME-CGE-DETA was based on 29 fragments that individually related back to 709 compounds in the Human Developmental Toxicity Mattison learning set with 77 of the 709 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Overall, the prediction of “inactive” for compound 3-ME-CGE-DETA was based on 30 fragments that individually related back to 703 compounds in the Human Developmental Toxicity Mattison learning set with 79 of the 703 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Overall, the prediction of “inactive” for compound 4-ME-CGE-DETA was based on 30 fragments that individually related back to 724 compounds in the Human Developmental Toxicity Mattison learning set with 80 of the 724 compounds being classified as human developmental toxicants (noting redundancy wherein individual chemicals may relate to more than one fragment and cut-point value to separate active from inactive calls).

Reference 5

Endpoint:

reproductive toxicity, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2017-2018

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

VEGA Version 1.1.4 (Build 13/2/2017)
ww.vega-qsar.eu

2. MODEL (incl. version number)

Developmental Toxicity model (CAESAR) (version 2.1.7)


3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL

DIME-CGE-DETA structures with same molecular formula (C24 H37 N3 O4) and molecular weight (431.58)

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1


4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODELS

Developmental Toxicity model (CAESAR) (version 2.1.7)
QSAR classification model for Developmental Toxicity based on a Random Forest classification. The
model extends the original CAESAR DevTox model 1.0 developed by Istituto Mario Negri, Italy. Reference
to the original model are found on the CAESAR Project website: http://www.caesar-project.eu/ .

5. APPLICABILITY DOMAIN

The predicted structures are in the Applicability Domain of the model (VEGA reports attached)


6. ADEQUACY OF THE RESULT

Global AD Index

AD index = 0.891 - 0.897
Explanation: the predicted compounds fall into the Applicability Domain of the model.

Similarity index = 0.792 - 0.804
Explanation: only moderately similar compounds with known experimental value in the training set have been found.

Accuracy of prediction for similar molecules
Accuracy index = 1
Explanation: accuracy of prediction for similar molecules found in the training set is good.

Concordance for similar molecules
Concordance index = 1
Explanation: similar molecules found in the training set have experimental values that agree with the predicted value.

Model's descriptors range check
Descriptors range check = True
Explanation: descriptors for this compound have values inside the descriptor range of the compounds of the training set.

Atom Centered Fragments similarity check
ACF index = 1
Explanation: all atom centered fragment of the compound have been found in the compounds of the training set.

The CAESAR model predicted that the 6 molecules would display developmental toxicant effects based on read-across from experimentally-derived source data on structural analogues.

However the relevance of these results is uncertain due to structural differences with the training set molecules.

Guideline:

other: REACH Guidance on QSARS R.6

Principles of method if other than guideline:

Cassano A., Manganaro A., Martin T., Young D., Piclin N., Pintore M., Bigoni D., Benfenati E. Chemistry Central Journal 2010, 4(Suppl 1):S4 (29 July 2010)

Specific details on test material used for the study:

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Reproductive effects observed:

not specified

The CAESAR model predicted that the 6 molecules would display developmental toxicant effects based on read-across from experimentally-derived source data on structural analogues.

Reference 6

Endpoint:

reproductive toxicity, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2017-2018

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

VEGA Version 1.1.4 (Build 13/2/2017)
ww.vega-qsar.eu

2. MODEL (incl. version number)

Estrogen Receptor-mediated effect (IRFMN/CERAPP) 1.0.0

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL

CGE_DIMER
CC1=CC=C(OCC(O)COC2=CC=C(C)C=C2)C=C1


4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODELS

The model providess a qualitative prediction for Estrogen Receptor mediated effect classification model
for endocrine disruptor screening. It is implemented inside the VEGA online platform, accessible at:
http://www.vega-qsar.eu/

The model has been built as a set of rules, extracted with Sarpy software from a dataset obtained from a
collection of high-quality estrogen receptor (ER) signaling data (1529 chemicals screened across 18
high-throughput screening assays integrated into a single score), from the ToxCast program (Judson RS
et al., “Integrated model of chemical perturbations of a biological pathway using 18 in vitro highthroughput screening assays for the estrogen receptor”, Toxicol Sci 148(1):137-54, 2015). The model
has been developed within the framework of the Collaborative Estrogen Receptor Activity Prediction
Project (CERAPP): Mansouri K et al., “CERAPP: Collaborative Estrogen Receptor Activity Prediction
Project”, Environ Health Perspect, 2016.

The Sarpy software has been used with a cross-validated procedure, ending with the extraction of two
sets of rules (structural alerts) related to ER-mediacted effect activity and inactivity (for a total of 61
rules). These rules have been further divided, according to their statistical significance, into a sub-set of
rules with strong statistical evidence and another one of rules with weaker evidence. These rules are
expressed SMARTS representing molecular fragments.

If at least one rule for activity is matching with the given compound, a “Active” or “Possible active”
prediction is given, depending on the statistical evidence of the rule. If no active rules are found, but at
least one rule for non-activity is matching with the given compound, a “NON-Active” or “Possible
NON-active” prediction is given, depending on the statistical evidence of the rule. If no rules are
matching at all, no prediction is provided.


5. APPLICABILITY DOMAIN

The predicted structures are in the Applicability Domain of the model (VEGA reports attached)


6. ADEQUACY OF THE RESULT

Global AD Index

AD index = 0.931
Explanation: the predicted compounds fall into the Applicability Domain of the model.

Similarity index = 0.866
Explanation: only moderately similar compounds with known experimental value in the training set have been found.

Accuracy of prediction for similar molecules
Accuracy index = 1
Explanation: accuracy of prediction for similar molecules found in the training set is good.

Concordance for similar molecules
Concordance index = 1
Explanation: similar molecules found in the training set have experimental values that agree with the predicted value.

Model's descriptors range check
Descriptors range check = True
Explanation: descriptors for this compound have values inside the descriptor range of the compounds of the training set.

Atom Centered Fragments similarity check
ACF index = 1
Explanation: all atom centered fragment of the compound have been found in the compounds of the training set.



The Prediction for all structures is NON-active, the result appears reliable.
ER non-activity alert no. 29; ER possible non-activity alert no. 3

Guideline:

other: REACH Guidance on QSARS R.6

Principles of method if other than guideline:

Mansouri K et al., “CERAPP: Collaborative Estrogen Receptor Activity Prediction Project”, Environ Health Perspect, 2016.

Specific details on test material used for the study:

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Reproductive effects observed:

not specified

Using the VEGA Estrogen Receptor-mediated effect (IRFMN/CERAPP) 1.0.0 the Prediction for all structures is NON-active, the result appears reliable.

ER non-activity alert no. 25; ER possible non-activity alert no. 3

All structures fall within the applicability domain for the model and the results are deemed reliable based on experimental data from similar compounds.

Reference 7

Endpoint:

reproductive toxicity, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2017-2018

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

VEGA Version 1.1.4 (Build 13/2/2017)
ww.vega-qsar.eu

2. MODEL (incl. version number)

Estrogen Receptor-mediated effect (IRFMN/CERAPP) 1.0.0

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL

DIME-CGE-DETA structures with same molecular formula (C24 H37 N3 O4) and molecular weight (431.58)

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1


4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODELS

The model provides a qualitative prediction for Estrogen Receptor mediated effect classification model
for endocrine disruptor screening. It is implemented inside the VEGA online platform, accessible at:
http://www.vega-qsar.eu/

The model has been built as a set of rules, extracted with Sarpy software from a dataset obtained from a
collection of high-quality estrogen receptor (ER) signaling data (1529 chemicals screened across 18
high-throughput screening assays integrated into a single score), from the ToxCast program (Judson RS
et al., “Integrated model of chemical perturbations of a biological pathway using 18 in vitro highthroughput screening assays for the estrogen receptor”, Toxicol Sci 148(1):137-54, 2015). The model
has been developed within the framework of the Collaborative Estrogen Receptor Activity Prediction
Project (CERAPP): Mansouri K et al., “CERAPP: Collaborative Estrogen Receptor Activity Prediction
Project”, Environ Health Perspect, 2016.

The Sarpy software has been used with a cross-validated procedure, ending with the extraction of two
sets of rules (structural alerts) related to ER-mediacted effect activity and inactivity (for a total of 61
rules). These rules have been further divided, according to their statistical significance, into a sub-set of
rules with strong statistical evidence and another one of rules with weaker evidence. These rules are
expressed SMARTS representing molecular fragments.

If at least one rule for activity is matching with the given compound, a “Active” or “Possible active”
prediction is given, depending on the statistical evidence of the rule. If no active rules are found, but at
least one rule for non-activity is matching with the given compound, a “NON-Active” or “Possible
NON-active” prediction is given, depending on the statistical evidence of the rule. If no rules are
matching at all, no prediction is provided.


5. APPLICABILITY DOMAIN

The predicted structures are in the Applicability Domain of the model (VEGA reports attached)


6. ADEQUACY OF THE RESULT

Global AD Index

AD index = 0.903 - 0.907
Explanation: the predicted compounds fall into the Applicability Domain of the model.

Similarity index = 0.816 - 0.822
Explanation: only moderately similar compounds with known experimental value in the training set have been found.

Accuracy of prediction for similar molecules
Accuracy index = 1
Explanation: accuracy of prediction for similar molecules found in the training set is good.

Concordance for similar molecules
Concordance index = 1
Explanation: similar molecules found in the training set have experimental values that agree with the predicted value.

Model's descriptors range check
Descriptors range check = True
Explanation: descriptors for this compound have values inside the descriptor range of the compounds of the training set.

Atom Centered Fragments similarity check
ACF index = 1
Explanation: all atom centered fragment of the compound have been found in the compounds of the training set.



The Prediction for all structures is NON-active, the result appears reliable.
ER non-activity alert no. 25; ER possible non-activity alert no. 3

Guideline:

other: REACH Guidance on QSARS R.6

Principles of method if other than guideline:

Mansouri K et al., “CERAPP: Collaborative Estrogen Receptor Activity Prediction Project”, Environ Health Perspect, 2016.

Specific details on test material used for the study:

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Reproductive effects observed:

not specified

Using the VEGA Estrogen Receptor-mediated effect (IRFMN/CERAPP) 1.0.0 the Prediction for all structures is NON-active, the result appears reliable.

ER non-activity alert no. 25; ER possible non-activity alert no. 3

All structures fall within the applicability domain for the model and the results are deemed reliable based on experimental data from similar compounds.

Reference 8

Endpoint:

reproductive toxicity, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2017-2018

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

VEGA Version 1.1.4 (Build 13/2/2017)
ww.vega-qsar.eu

2. MODEL (incl. version number)

Estrogen Receptor-mediated effect (IRFMN/CERAPP) 1.0.0

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL

ME-CGE-DETA structures with same molecular formula (C14 H25 N3 O2) and molecular weight (267.37)

2-ME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCN

3-ME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCN)=CC=C1

4-ME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCN)C=C1


4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODELS

The model providess a qualitative prediction for Estrogen Receptor mediated effect classification model
for endocrine disruptor screening. It is implemented inside the VEGA online platform, accessible at:
http://www.vega-qsar.eu/

The model has been built as a set of rules, extracted with Sarpy software from a dataset obtained from a
collection of high-quality estrogen receptor (ER) signaling data (1529 chemicals screened across 18
high-throughput screening assays integrated into a single score), from the ToxCast program (Judson RS
et al., “Integrated model of chemical perturbations of a biological pathway using 18 in vitro highthroughput screening assays for the estrogen receptor”, Toxicol Sci 148(1):137-54, 2015). The model
has been developed within the framework of the Collaborative Estrogen Receptor Activity Prediction
Project (CERAPP): Mansouri K et al., “CERAPP: Collaborative Estrogen Receptor Activity Prediction
Project”, Environ Health Perspect, 2016.

The Sarpy software has been used with a cross-validated procedure, ending with the extraction of two
sets of rules (structural alerts) related to ER-mediacted effect activity and inactivity (for a total of 61
rules). These rules have been further divided, according to their statistical significance, into a sub-set of
rules with strong statistical evidence and another one of rules with weaker evidence. These rules are
expressed SMARTS representing molecular fragments.

If at least one rule for activity is matching with the given compound, a “Active” or “Possible active”
prediction is given, depending on the statistical evidence of the rule. If no active rules are found, but at
least one rule for non-activity is matching with the given compound, a “NON-Active” or “Possible
NON-active” prediction is given, depending on the statistical evidence of the rule. If no rules are
matching at all, no prediction is provided.


5. APPLICABILITY DOMAIN

The predicted structures are in the Applicability Domain of the model (VEGA reports attached)


6. ADEQUACY OF THE RESULT

Global AD Index

AD index = 0.903 - 0.907
Explanation: the predicted compounds fall into the Applicability Domain of the model.

Similarity index = 0.816 - 0.822
Explanation: only moderately similar compounds with known experimental value in the training set have been found.

Accuracy of prediction for similar molecules
Accuracy index = 1
Explanation: accuracy of prediction for similar molecules found in the training set is good.

Concordance for similar molecules
Concordance index = 1
Explanation: similar molecules found in the training set have experimental values that agree with the predicted value.

Model's descriptors range check
Descriptors range check = True
Explanation: descriptors for this compound have values inside the descriptor range of the compounds of the training set.

Atom Centered Fragments similarity check
ACF index = 1
Explanation: all atom centered fragment of the compound have been found in the compounds of the training set.



The Prediction for all structures is NON-active, the result appears reliable.
ER non-activity alert no. 25; ER possible non-activity alert no. 3

Guideline:

other: REACH Guidance on QSARS R.6

Principles of method if other than guideline:

Mansouri K et al., “CERAPP: Collaborative Estrogen Receptor Activity Prediction Project”, Environ Health Perspect, 2016.

Specific details on test material used for the study:

2-2’-DIME-CGE-DETA
CC1=CC=CC=C1OCC(O)CNCCNCCNCC(O)COC1=C(C)C=CC=C1

2-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)=CC=C1

2-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=C(C)C=CC=C2)C=C1

3-3’-DIME-CGE-DETA
CC1=CC(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)=CC=C1

3-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC(C)=CC=C2)C=C1

4-4’-DIME-CGE-DETA
CC1=CC=C(OCC(O)CNCCNCCNCC(O)COC2=CC=C(C)C=C2)C=C1

Remarks on result:

other: QSAR Predictied Value

Remarks on result:

other: QSAR Predictied Value

Reproductive effects observed:

not specified

Using the VEGA Estrogen Receptor-mediated effect (IRFMN/CERAPP) 1.0.0 the Prediction for all structures is NON-active, the result appears reliable.

ER non-activity alert no. 25; ER possible non-activity alert no. 3

All structures fall within the applicability domain for the model and the results are deemed reliable based on experimental data from similar compounds.

Effect on fertility: via oral route

Endpoint conclusion:

no study available

Effect on fertility: via inhalation route

Endpoint conclusion:

no study available

Effect on fertility: via dermal route

Endpoint conclusion:

no study available

Effects on developmental toxicity

Description of key information

The substance 1,2-Ethanediamine, N-(2-aminoethyl)-, reaction products with glycidyl tolyl ether is a multiconstituent substance which predominantly consists of secondary aliphatic amines, with the exception of one CGE dimer.

Various QSAR models were run on the identified constituents of the substance. Predictions with the CAESAR model revealed some positive predictions for developmental toxicity. The DART model profiler showed positive results of the unspecific category Known precedent reproductive and developmental toxic potential; Toluene and small alkyl toluene derivatives (8a) for all structures of the substance. And in the CAT-SAR system, for Human Developmental Toxicity the CGE-DIMER revealed equivocal results for developmental toxicity.

On the basis of positive indication for developmental effects from the various QSAR models it is proposed to conduct an OECD 414 Prenatal Developmental Toxicity Study (Annex IX 8.7.2) in order to clarify specifically this concern.

Thus, an OECD 421 Reproductive and Developmental Toxicity Screening Test would provide less information and is therefore waived in accordance with column 2 of REACH Annex VIII 8.7.1.

Alternative approaches are excluded based on the following considerations:

• Available GLP studies              There is no experimental GLP- study regarding toxicity on reproduction or teratogenicity available

• Available non-GLP studies       There is no experimental study regarding toxicity on reproduction or teratogenicity available

• Grouping and read-across        Due to the concern from the QSAR calculations and the complexity of an multiconstituent read across is not further considered

• Historical human data               Not available

• (Q)SAR                                   Data provided, see above, in attached document and endpoint entries

• In vitro methods                      Due to the concern from the QSAR calculations and the complexity of an multiconstituent in in vitro systems, in vitro testing is not further considered

• Weight of evidence                    Data generated using QSAR raise a concern for developmental toxicity, which should be clarified by the proposed OECD 414 study, there is no further data available for a weight of evidence approach

• Substance-tailored exposure  

driven testing [if applicable]           Not applicable. In order to clarify the concern, the OECD study 414 is considered appropriate

Link to relevant study records

Reference

Endpoint:

developmental toxicity

Type of information:

experimental study planned

Justification for type of information:

The attached document provides evidence generated for a testing proposal for developmental toxicity (OECD 414) using weight of evidence from structurally similar substances, identified using validated QSAR models.

The substance 1,2-Ethanediamine, N-(2-aminoethyl)-, reaction products with glycidyl tolyl ether is a multiconstituent substance which predominantly consists of secondary aliphatic amines, with the exception of one CGE dimer.
Various QSAR models were run on the identified constituents of the substance. Predictions with the CAESAR model revealed some positive predictions for developmental toxicity. The DART model profiler showed positive results of the unspecific category Known precedent reproductive and developmental toxic potential; Toluene and small alkyl toluene derivatives (8a) for all structures of the substance. And in the CAT-SAR system, for Human Developmental Toxicity the CGE-DIMER revealed equivocal results for developmental toxicity.
On the basis of positive indication for developmental effects from the various QSAR models it is proposed to conduct an OECD 414 Prenatal Developmental Toxicity Study (Annex IX 8.7.2) in order to clarify specifically this concern.
Thus, an OECD 421 Reproductive and Developmental Toxicity Screening Test would provide less information and is therefore waived in accordance with column 2 of REACH Annex VIII 8.7.1.

Alternative approaches are excluded based on the following considerations:
• Available GLP studies There is no experimental GLP- study regarding toxicity on reproduction or teratogenicity available
• Available non-GLP studies There is no experimental study regarding toxicity on reproduction or teratogenicity available
• Grouping and read-across Due to the concern from the QSAR calculations and the complexity of an multiconstituent read across is not further considered
• Historical human data Not available
• (Q)SAR Data provided, see above, in attached document and endpoint entries
• In vitro methods Due to the concern from the QSAR calculations and the complexity of an multiconstituent in in vitro systems, in vitro testing is not further considered
• Weight of evidence Data generated using QSAR raise a concern for developmental toxicity, which should be clarified by the proposed OECD 414 study, there is no further data available for a weight of evidence approach
• Substance-tailored exposure
driven testing [if applicable] Not applicable. In order to clarify the concern, the OECD study 414 is considered appropriate

Qualifier:

according to guideline

Guideline:

OECD Guideline 414 (Prenatal Developmental Toxicity Study)

Version / remarks:

rats

Species:

rat

Effect on developmental toxicity: via oral route

Endpoint conclusion:

no study available (further information necessary)

Effect on developmental toxicity: via inhalation route

Endpoint conclusion:

no study available

Effect on developmental toxicity: via dermal route

Endpoint conclusion:

no study available