Isotridecyl isononanoate

Endpoint summary

Key value for chemical safety assessment

Skin sensitisation

Link to relevant study records

Reference

Reference 1

Endpoint:

skin sensitisation, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Reliability:

3 (not reliable)

Rationale for reliability incl. deficiencies:

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

Justification for type of information:

CAESAR (VEGA)
1 Substance
1.1 CAS number 42131-27-1
1.2 EC number 255-673-5
1.3 Chemical name
IUPAC 11-Methyldodecyl 7-methyloctanoate
Other Octanoic acid, 7-methyl-, 11-methyldodecyl ester
Other Isotridecyl isononanoate
1.4 Structural formula

1.5 Structure codes
SMILES O=C(OCCCCCCCCCCC(C)C)CCCCCC(C)C
InChI InChI=1S/C22H44O2/c1-20(2)16-12-9-7-5-6-8-10-15-19-24-22(23)18-14-11-13-17-21(3)4/h20-21H,5-19H2,1-4H3
Other
Stereochemical features N/A

2 General Information
2.1 Date of QPRF 20 April 2018
2.2 Author and contact details Envigo, Shardlow Business Park London Road, Shardlow, Derbyshire, DE72 2GD

3 Prediction
3.1 Endpoint (OECD Principle 1)
Endpoint Skin Sensitisation (None vs Sensitiser)
Dependent variable Classification as sensitiser or non-sensitiser
3.2 Algorithm (OECD Principle 2)
Model or submodel name Extension of the original CAESAR model for skin sensitisation (None vs Sensitiser) within VEGA 1.1.4
Model version 2.1.6
Reference to QMRF Not available
Predicted values (model result) Sensitiser
Predicted values (comments) According to VEGA's evaluation scheme the structure is a sensitizer but the result may be not reliable.
Input for prediction Smiles
Descriptor values Not provided
3.3 Applicability domain (OECD Principle 3)
Domains i. The predicted compound is into the Applicability Domain of the model.
ii. Strongly similar compounds with known experimental value in the training set have been found.
iii. Accuracy of prediction for similar molecules found in the training set is good.
iv. Similar molecules found in the training set have experimental values that agree with the predicted
value.
v. Descriptors for this compound have values inside the descriptor range of the compounds of the
training set.
vi. All atom centered fragment of the compound have been found in the compounds of the training
set.
Structural analogues i. Isopropyl myristate
ii. 4,4-Dimethyl-2-pentadecyl-1,3-oxazol-5(4H)-one
iii. 2-heptadecyl-4,4-dimethyl-4,5-dihydro-1,3-oxazol-5-one
iv. 4,4-dimethyl-2-nonadecyl-4,5-dihydro-1,3-oxazol-5-one

Consideration on structural analogues With 89.6% the average similarity of the four most structurally similar analogues to the query structure is considered high. All four of the structures are sensitizers, thus indicating high concordance with the query structure. Three of the four structures trigger an alert from the OASIS protein binding profiler for skin sensitisation in OECD Toolbox, whereas the target molecule does not show any alerts in the profiling. Furthermore, the metabolism predictions for these molecules show different pathways and again no alerts for the query structure. It is therefore assumed to be likely that mode of action followed by the three oxazolone structures is not relatable to the target compound.

3.4 The uncertainty of the prediction (OECD principle 4)
The models own statistics indicate high reliability in the prediction, with high similarity, recognition of all compound fragments in the training set, good concordance of results with measured data etc. However it is clear to the assessor that the nearest molecules in the dataset used for the prediction have considerably different properties and functional groups to the target compound.
An assessment of these differences using the profilers in the OECD QSAR Toolbox identifies alerts for the oxazolone containing structures compared with the target structure. Furthermore assessment of the predicted metabolism of these compounds has shown different routes to breakdown products again which the target molecule does not display any alerts for.
This indicates uncertainty in the results from the model.
The Isopropyl myristate is also predicted to be sensitising.
The Data matrix with this information in has been included in the printouts section below for reference.

3.5 The chemical and biological mechanisms according to the model underpinning the predicted result (OECD principle 5)
Not applicable since statistical model

4 Adequacy (Optional)
4.1 Regulatory purpose Skin sensitisation endpoint for assessing the skin sensitisation potential with in vitro and in silico methods according to ECHA Guidance, Chapter R.7a, 2016.

4.2 Approach for regulatory interpretation of the model result
Result is directly applicable since no conversion of the result is required.

4.3 Outcome The query structure is within the applicability domain of the model, and the prediction is made with good statistics, however due to concerns regarding the mode of action displayed by the similar structures in the training set, there is some uncertainty in the result.

4.4 Conclusion The prediction is not considered reliable and will not be considered with predictions from other models in a weight of evidence conclusion.

Qualifier:

no guideline followed

Principles of method if other than guideline:

- Principle of test: QSAR prediciton of Skin senistisation
- Short description of test conditions: n/a
- Parameters analysed / observed: Skin senistisation

Remarks on result:

positive indication of skin sensitisation

CAESAR (VEGA)
1	Substance
	1.1	CAS number			42131-27-1
	1.2	EC number			255-673-5
	1.3	Chemical name
			IUPAC		11-Methyldodecyl 7-methyloctanoate
			Other		Octanoic acid, 7-methyl-, 11-methyldodecyl ester
			Other		Isotridecyl isononanoate
	1.4	Structural formula
	1.5	Structure codes
			SMILES		O=C(OCCCCCCCCCCC(C)C)CCCCCC(C)C
			InChI		InChI=1S/C22H44O2/c1-20(2)16-12-9-7-5-6-8-10-15-19-24-22(23)18-14-11-13-17-21(3)4/h20-21H,5-19H2,1-4H3
			Other
			Stereochemical features		N/A
2	General Information
	2.1	Date of QPRF			20 April 2018
	2.2	Author and contact details			Envigo, Shardlow Business Park London Road, Shardlow, Derbyshire, DE72 2GD
3	Prediction
	3.1	Endpoint (OECD Principle 1)
			Endpoint		Skin Sensitisation (None vs Sensitiser)
			Dependent variable		Classification as sensitiser or non-sensitiser
	3.2	Algorithm (OECD Principle 2)
			Model or submodel name		Extension of the original CAESAR model for skin sensitisation (None vs Sensitiser) within VEGA 1.1.4
			Model version		2.1.6
			Reference to QMRF		Not available
			Predicted values (model result)		Sensitiser
			Predicted values (comments)		According to VEGA's evaluation scheme the structure is a sensitizer but the result may be not reliable.
			Input for prediction		Smiles
			Descriptor values		Not provided
	3.3	Applicability domain (OECD Principle 3)
			Domains		i.	The predicted compound is into the Applicability Domain of the model.
					ii.	Strongly similar compounds with known experimental value in the training set have been found.
					iii.	Accuracy of prediction for similar molecules found in the training set is good.
					iv.	Similar molecules found in the training set have experimental values that agree with the predicted value.
					v.	Descriptors for this compound have values inside the descriptor range of the compounds of the training set.
					vi.	All atom centered fragment of the compound have been found in the compounds of the training set.
			Structural analogues		i. Isopropyl myristate ii. 4,4-Dimethyl-2-pentadecyl-1,3-oxazol-5(4H)-one iii. 2-heptadecyl-4,4-dimethyl-4,5-dihydro-1,3-oxazol-5-one iv. 4,4-dimethyl-2-nonadecyl-4,5-dihydro-1,3-oxazol-5-one
			Consideration on structural analogues		With 89.6% the average similarity of the four most structurally similar analogues to the query structure is considered high. All four of the structures are sensitizers, thus indicating high concordance with the query structure. Three of the four structures trigger an alert from the OASIS protein binding profiler for skin sensitisation in OECD Toolbox, whereas the target molecule does not show any alerts in the profiling. Furthermore, the metabolism predictions for these molecules show different pathways and again no alerts for the query structure. It is therefore assumed to be likely that mode of action followed by the three oxazolone structures is not relatable to the target compound.
	3.4	The uncertainty of the prediction (OECD principle 4)
					The models own statistics indicate high reliability in the prediction, with high similarity, recognition of all compound fragments in the training set, good concordance of results with measured data etc. However it is clear to the assessor that the nearest molecules in the dataset used for the prediction have considerably different properties and functional groups to the target compound. An assessment of these differences using the profilers in the OECD QSAR Toolbox identifies alerts for the oxazolone containing structures compared with the target structure. Furthermore assessment of the predicted metabolism of these compounds has shown different routes to breakdown products again which the target molecule does not display any alerts for. This indicates uncertainty in the results from the model. The Isopropyl myristate is also predicted to be sensitising. The Data matrix with this information in has been included in the printouts section below for reference.
	3.5	The chemical and biological mechanisms according to the model underpinning the predicted result (OECD principle 5)
					Not applicable since statistical model
4	Adequacy (Optional)
	4.1	Regulatory purpose			Skin sensitisation endpoint for assessing the skin sensitisation potential with in vitro and in silico methods according to ECHA Guidance, Chapter R.7a, 2016.
	4.2	Approach for regulatory interpretation of the model result
					Result is directly applicable since no conversion of the result is required.
	4.3	Outcome			The query structure is within the applicability domain of the model, and the prediction is made with good statistics, however due to concerns regarding the mode of action displayed by the similar structures in the training set, there is some uncertainty in the result.
	4.4	Conclusion			The prediction is not considered reliable and will not be considered with predictions from other models in a weight of evidence conclusion.

Interpretation of results:

GHS criteria not met

Conclusions:

The query structure is within the applicability domain of the model, and the prediction is made with good statistics, however due to concerns regarding the mode of action displayed by the similar structures in the training set, there is some uncertainty in the result.

The prediction is not considered reliable and will not be considered with predictions from other models in a weight of evidence conclusion.

Reference 2

Endpoint:

skin sensitisation, other

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 limited documentation / justification

Justification for type of information:

DEREK (skin sensitisation)
1 Substance
1.1 CAS number 42131-27-1
1.2 EC number 255-673-5
1.3 Chemical name
IUPAC 11-Methyldodecyl 7-methyloctanoate
Other Octanoic acid, 7-methyl-, 11-methyldodecyl ester
Other Isotridecyl isononanoate
1.4 Structural formula

1.5 Structure codes
SMILES O=C(OCCCCCCCCCCC(C)C)CCCCCC(C)C
InChI InChI=1S/C22H44O2/c1-20(2)16-12-9-7-5-6-8-10-15-19-24-22(23)18-14-11-13-17-21(3)4/h20-21H,5-19H2,1-4H3
Other
Stereochemical features N/A

2 General Information
2.1 Date of QPRF 20 April 2018
2.2 Author and contact details Envigo, Shardlow Business Park, London Road, Shardlow, Derbyshire, DE72 2GD

3 Prediction
3.1 Endpoint (OECD Principle 1)
Endpoint Skin Sensitisation
Dependent variable Not applicable
3.2 Algorithm (OECD Principle 2)
Model or submodel name Skin sensitisation in mammal
Model version DEREK Nexus 5.0.2, Nexus: 2.1.1.
Knowledge Base: Derek KB 2015 2.0, Version 2.0 from 28/01/2016
Reference to QMRF The QMRF with the identifier Q13-34-36-315 is available from the JRC QMRF inventory (http://qsardb.jrc.it/qmrf/).
Predicted values (model result) Skin sensitisation in mammal is NON-SENSITISER
Predicted values (comments) No misclassified or unclassified features
Input for prediction Smiles
Calculated descriptor values Descriptor Value
LogP 9.97
LogKp 2.28

3.3 Applicability domain (OECD Principle 3)
Domains No structural alerts or examples for skin sensitisation Structure does not contain any unclassified or misclassified features
Therefore no Domain is available to report
Structural analogues N/A
Consideration on structural analogues N/A.
3.4 The uncertainty of the prediction (OECD principle 4)
The software has not identified any substructures within the compound that match its database of alerts. While some confidence can be placed in the softwares prediction, this result alone should not be considered conclusive on its own. The uncertainty of the prediction cannot be directly measured/assessed as no statistics are reported by the model.
3.5 The chemical and biological mechanisms according to the model underpinning the predicted result (OECD principle 5)
The query structure does not match any structural alerts or examples for skin sensitisation in Derek. Additionally, the query structure does not contain any unclassified or misclassified features and is consequently predicted to be a non-sensitiser.

4 Adequacy (Optional)
4.1 Regulatory purpose Skin sensitisation endpoint for assessing the skin sensitisation potential with in vitro and in silico methods according to ECHA Guidance, Chapter R.7a, 2016.

4.2 Approach for regulatory interpretation of the model result
Result is directly applicable since no conversion of the result is required.

4.3 Outcome Non-sensitiser

4.4 Conclusion Non-sensitiser, no misclassified or unclassified features
The query structure does not match any structural alerts or examples for skin sensitisation in Derek. Additionally, the query structure does not contain any unclassified or misclassified features.

Qualifier:

no guideline followed

Principles of method if other than guideline:

- Principle of test: QSAR prediciton of Skin senistisation
- Short description of test conditions: n/a
- Parameters analysed / observed: Skin senistisation

Remarks on result:

no indication of skin sensitisation

DEREK (skin sensitisation)
	1	Substance
		1.1	CAS number			42131-27-1
		1.2	EC number			255-673-5
		1.3	Chemical name
				IUPAC		11-Methyldodecyl 7-methyloctanoate
				Other		Octanoic acid, 7-methyl-, 11-methyldodecyl ester
				Other		Isotridecyl isononanoate
		1.4	Structural formula
		1.5	Structure codes
				SMILES		O=C(OCCCCCCCCCCC(C)C)CCCCCC(C)C
				InChI		InChI=1S/C22H44O2/c1-20(2)16-12-9-7-5-6-8-10-15-19-24-22(23)18-14-11-13-17-21(3)4/h20-21H,5-19H2,1-4H3
				Other
				Stereochemical features		N/A
	2	General Information
		2.1	Date of QPRF			20 April 2018
		2.2	Author and contact details			Envigo, Shardlow Business Park, London Road, Shardlow, Derbyshire, DE72 2GD
	3	Prediction
		3.1	Endpoint (OECD Principle 1)
				Endpoint		Skin Sensitisation
				Dependent variable		Not applicable
		3.2	Algorithm (OECD Principle 2)
				Model or submodel name		Skin sensitisation in mammal
				Model version		DEREK Nexus 5.0.2, Nexus: 2.1.1. Knowledge Base: Derek KB 2015 2.0, Version 2.0 from 28/01/2016
				Reference to QMRF		The QMRF with the identifier Q13-34-36-315 is available from the JRC QMRF inventory (http://qsardb.jrc.it/qmrf/).
				Predicted values (model result)		Skin sensitisation in mammal is NON-SENSITISER
				Predicted values (comments)		No misclassified or unclassified features
				Input for prediction		Smiles
				Calculated descriptor values		Descriptor Value LogP 9.97 LogKp 2.28
		3.3	Applicability domain (OECD Principle 3)
				Domains		No structural alerts or examples for skin sensitisation Structure does not contain any unclassified or misclassified features Therefore no Domain is available to report
				Structural analogues		N/A
				Consideration on structural analogues		N/A.
		3.4	The uncertainty of the prediction (OECD principle 4)
						The software has not identified any substructures within the compound that match its database of alerts. While some confidence can be placed in the softwares prediction, this result alone should not be considered conclusive on its own. The uncertainty of the prediction cannot be directly measured/assessed as no statistics are reported by the model.
		3.5	The chemical and biological mechanisms according to the model underpinning the predicted result (OECD principle 5)
						The query structure does not match any structural alerts or examples for skin sensitisation in Derek. Additionally, the query structure does not contain any unclassified or misclassified features and is consequently predicted to be a non-sensitiser.
	4	Adequacy (Optional)
		4.1	Regulatory purpose			Skin sensitisation endpoint for assessing the skin sensitisation potential with in vitro and in silico methods according to ECHA Guidance, Chapter R.7a, 2016.
		4.2	Approach for regulatory interpretation of the model result
						Result is directly applicable since no conversion of the result is required.
		4.3	Outcome			Non-sensitiser
		4.4	Conclusion			Non-sensitiser, no misclassified or unclassified features The query structure does not match any structural alerts or examples for skin sensitisation in Derek. Additionally, the query structure does not contain any unclassified or misclassified features.

Interpretation of results:

GHS criteria not met

Conclusions:

The software has not identified any substructures within the compound that match its database of alerts. While some confidence can be placed in the softwares prediction, this result alone should not be considered conclusive on its own. The uncertainty of the prediction cannot be directly measured/assessed as no statistics are reported by the model.
Non-sensitiser, no misclassified or unclassified features
The query structure does not match any structural alerts or examples for skin sensitisation in Derek. Additionally, the query structure does not contain any unclassified or misclassified features.

Reference 3

Endpoint:

skin sensitisation, other

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 limited documentation / justification

Justification for type of information:

1. SOFTWARE
QSAR Toolbox 4.1

2. MODEL (incl. version number)
QSAR Toolbox 4.1
Database version: 4.1
TPRF v4.1

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
CC(C)CCCCCCCCCCOC(=O)CCCCCC(C)C

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
[Explain how the model fulfils the OECD principles for (Q)SAR model validation. Consider attaching the QMRF or providing a link]
see attached justificaiton

5. APPLICABILITY DOMAIN
[Explain how the substance falls within the applicability domain of the model]
see attached ustification

6. ADEQUACY OF THE RESULT
[Explain how the prediction fits the purpose of classification and labelling and/or risk assessment]
Prediction is considered to be reliable since the nearest neighbours are of moderate (>50%) similarity to the target, and while the mechanism is ill defined, there is concordance within the category substances which provides some confidence in the prediction.

Qualifier:

no guideline followed

Principles of method if other than guideline:

- Principle of test: QSAR prediciton of Skin senistisation
- Short description of test conditions: n/a
- Parameters analysed / observed: Skin senistisation

Remarks on result:

no indication of skin sensitisation

Interpretation of results:

GHS criteria not met

Conclusions:

Prediction with OECD Toolbox was performed first via the automatic workflow, then manually amended after the workflow failed to conclude a result for skin sensitisation. Isotridecyl isononanoate did not trigger an alert by any of the skin sensitisation profilers and so the automatic workflow selected the ECOSAR profiler as a starting point, this was then subcategorised using maximal data by substance type, Protein binding alerts for skin sensitization by oasis, Protein binding alerts for skin sensitization by oasis accounting autoxidation simulator, Protein binding alerts for skin sensitization by oasis accounting skin metabolism simulator, but the automated work flow could not conclude on a suitable grouping. At this point the prediction was manually further subcategorised by structural similarity to 50%. The resulting category are a group of similar substances to the target molecule which show the same predicted autoxidation and skin metabolism pathway and all carry negative study data, which can be seen from the resultant data matrix. There is high similarity of the substances to the query structure, high concordance of results within the category. As such the result is considered to be reliable.

Reference 4

Endpoint:

skin sensitisation, other

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:

TOPKAT
1 Substance
1.1 CAS number 42131-27-1
1.2 EC number 255-673-5
1.3 Chemical name
IUPAC 11-Methyldodecyl 7-methyloctanoate
Other Octanoic acid, 7-methyl-, 11-methyldodecyl ester
Other Isotridecyl isononanoate
1.4 Structural formula

1.5 Structure codes
SMILES O=C(OCCCCCCCCCCC(C)C)CCCCCC(C)C
InChI InChI=1S/C22H44O2/c1-20(2)16-12-9-7-5-6-8-10-15-19-24-22(23)18-14-11-13-17-21(3)4/h20-21H,5-19H2,1-4H3
Other
Stereochemical features N/A

2 General Information
2.1 Date of QPRF 20 April 2018
2.2 Author and contact details Envigo, Shardlow Business Park, London Road, Shardlow, Derbyshire, DE72 2GD

3 Prediction
3.1 Endpoint (OECD Principle 1)
Endpoint Skin Sensitisation (None vs Sensitiser)
Dependent variable Classification as sensitizer or non-sensitizer
3.2 Algorithm (OECD Principle 2)
Model or submodel name Toxicity Prediction (Extensible)Toxicity Prediction (Extensible) Skin sensitisation (None vs Sensitizer)
Model version 4.5
Reference to QMRF The corresponding QMRF with the identifier Q50-54-55-509 is available at http://qsardb.jrc.it/qmrf/index.jsp. The original data set was extended with 42 additional compounds from the Envigo database.
Predicted values (model result) Non-Sensitizer
Predicted values (comments) A Bayesian score of -18.33 being well below the best split of -1.075 and a probability of 0.01 suggest confidence in the prediction.
Input for prediction Smiles
Caclulated descriptor values Descriptor Value
LogP 8.45
Molecular weight (g/mol) 340.584
Number of hydrogen bond donors 0
Number of hydrogen bond acceptors 2
Number of rotatable bonds in the molecule 18
The fraction of polar surface area over the total molecular surface area 0.06
FCFP_12: Unit functional class extended-connectivity atom type fingerprint with a maximum length of 12 bonds Not applicable



3.3 Applicability domain (OECD Principle 3)
Domains i. All properties and OPS components are within expected ranges.
ii. Considerations on the mechanism domain are not applicable since the contributing FCFP_12 features of the model are selected purely on their Bayesian score (statistical model)
Structural analogues i. Isostearyl neopentanoate
ii. n-Octylpyrrolidone
iii. Ethyl hexyl palmitate
iv. Octadecanoyl chloride

Consideration on structural analogues With 0.53 the average distance of the four analogues to the query structure is considered poor. 3 out of the four structures are non-sensitizers, thus indicating high concordance with the predicted result of query structure. Accuracy between predicted and actual result is high as all four analogues were predicted correctly.
3.4 The uncertainty of the prediction (OECD principle 4)
Uncertainty is indicated due to poor similarity.
3.5 The chemical and biological mechanisms according to the model underpinning the predicted result (OECD principle 5)
Not applicable since statistical model

4 Adequacy (Optional)
4.1 Regulatory purpose Skin sensitisation endpoint for assessing the skin sensitisation potential with in vitro and in silico methods according to ECHA Guidance, Chapter R.7a, 2016.

4.2 Approach for regulatory interpretation of the model result
Result is directly applicable since no conversion of the result is required.

4.3 Outcome Due to poor similarity, some uncertainty can be concluded, however as all other prediction statistics are good, there is moderate confidence in the prediction.

4.4 Conclusion The prediction is considered to be of moderate reliability and will be used together with predictions from other models in a weight of evidence conclusion.

Qualifier:

no guideline followed

Principles of method if other than guideline:

- Principle of test: QSAR prediciton of Skin senistisation
- Short description of test conditions: n/a
- Parameters analysed / observed: Skin senistisation

Remarks on result:

no indication of skin sensitisation

TOPKAT
1	Substance
	1.1	CAS number			42131-27-1
	1.2	EC number			255-673-5
	1.3	Chemical name
			IUPAC		11-Methyldodecyl 7-methyloctanoate
			Other		Octanoic acid, 7-methyl-, 11-methyldodecyl ester
			Other		Isotridecyl isononanoate
	1.4	Structural formula
	1.5	Structure codes
			SMILES		O=C(OCCCCCCCCCCC(C)C)CCCCCC(C)C
			InChI		InChI=1S/C22H44O2/c1-20(2)16-12-9-7-5-6-8-10-15-19-24-22(23)18-14-11-13-17-21(3)4/h20-21H,5-19H2,1-4H3
			Other
			Stereochemical features		N/A
2	General Information
	2.1	Date of QPRF			20 April 2018
	2.2	Author and contact details			Envigo, Shardlow Business Park, London Road, Shardlow, Derbyshire, DE72 2GD
3	Prediction
	3.1	Endpoint (OECD Principle 1)
			Endpoint		Skin Sensitisation (None vs Sensitiser)
			Dependent variable		Classification as sensitizer or non-sensitizer
	3.2	Algorithm (OECD Principle 2)
			Model or submodel name		Toxicity Prediction (Extensible)Toxicity Prediction (Extensible) Skin sensitisation (None vs Sensitizer)
			Model version		4.5
			Reference to QMRF		The corresponding QMRF with the identifier Q50-54-55-509 is available athttp://qsardb.jrc.it/qmrf/index.jsp. The original data set was extended with 42 additional compounds from the Envigo database.
			Predicted values (model result)		Non-Sensitizer
			Predicted values (comments)		A Bayesian score of -18.33 being well below the best split of -1.075 and a probability of 0.01 suggest confidence in the prediction.
			Input for prediction		Smiles
			Caclulated descriptor values		Descriptor Value LogP 8.45 Molecular weight (g/mol) 340.584 Number of hydrogen bond donors 0 Number of hydrogen bond acceptors 2 Number of rotatable bonds in the molecule 18 The fraction of polar surface area over the total molecular surface area 0.06 FCFP_12: Unit functional class extended-connectivity atom type fingerprint with a maximum length of 12 bonds Not applicable
	3.3	Applicability domain (OECD Principle 3)
			Domains		i.	All properties and OPS components are within expected ranges.
					ii.	Considerations on the mechanism domain are not applicable since the contributing FCFP_12 features of the model are selected purely on their Bayesian score (statistical model)
			Structural analogues		i. Isostearyl neopentanoate   ii. n-Octylpyrrolidone   iii. Ethyl hexyl palmitate   iv. Octadecanoyl chloride
			Consideration on structural analogues		With 0.53 the average distance of the four analogues to the query structure is considered poor. 3 out of the four structures are non-sensitizers, thus indicating high concordance with the predicted result of query structure. Accuracy between predicted and actual result is high as all four analogues were predicted correctly.
	3.4	The uncertainty of the prediction (OECD principle 4)
					Uncertainty is indicated due to poor similarity.
	3.5	The chemical and biological mechanisms according to the model underpinning the predicted result (OECD principle 5)
					Not applicable since statistical model
4	Adequacy (Optional)
	4.1	Regulatory purpose			Skin sensitisation endpoint for assessing the skin sensitisation potential with in vitro and in silico methods according to ECHA Guidance, Chapter R.7a, 2016.
	4.2	Approach for regulatory interpretation of the model result
					Result is directly applicable since no conversion of the result is required.
	4.3	Outcome			Due to poor similarity, some uncertainty can be concluded, however as all other prediction statistics are good, there is moderate confidence in the prediction.
	4.4	Conclusion			The prediction is considered to be of moderate reliability and will be used together with predictions from other models in a weight of evidence conclusion.

Interpretation of results:

GHS criteria not met

Conclusions:

Due to poor similarity, some uncertainty can be concluded, however as all other prediction statistics are good, there is moderate confidence in the prediction.
The prediction is considered to be of moderate reliability and will be used together with predictions from other models in a weight of evidence conclusion.

Endpoint conclusion

Endpoint conclusion:

no adverse effect observed (not sensitising)

Justification for classification or non-classification

TOPKAT, DEREK, and OECD Toolbox predicted Isotridecyl isononanoateto be non-sensitising, the statistics related to these predictions suggest these results are reliable. The results obtained from the Danish QSAR Database were conclusive, with all results proposed to be negative for skin sensitisation, and also all within the applicability domain of the respective models. With the exception of the disregarded VEGA prediction, which is discussed futher below, all other predictions are negative and the lack of mechanism highlighted by DEREK and the OECD toolbox is supported by the similar structures identified in the TOPKAT which upon assessment also show similar profiles. Therefore, the overall battery of results is considered to be of good reliability.

The CAESAR (VEGA) model identified the substance as sensitising, this was however discounted as the conclusion is based on compounds which were dissimilar to the target chemical, specifically in terms of functional groups. The three compounds were oxazolones which would possess different properties to the target. This was confirmed by running profilers through OECD QSAR Toolbox on the compounds to identify any alerts related not only to the compounds, but their potential metabolites (in case of pre/pro haptens). It was identified that these values were different to those of the target molecule and thus as the hypothesis was supported by this evidence, the result of the CAESAR model was discounted.

The negative prediction by TOPKAT is characterised by some uncertainties with regard to the most similar structurers in the training set, thus indicating only moderate confidence in the prediction. Nevertheless, the prediction statistics (probability, enrichment, baysian score/best split difference, concordance of measured data and accuracy of predictions) were all high indicating some amount of confidence in the prediction. Furthermore, the four compounds were included into a category with the target compound in OECD QSAR Toolbox and their similarity compared using the Dice atom paired fragments method commonly used in this software. The result indicated an average similarity of 58.6%, therefore the similarity could be considered moderate as opposed to poor, affording further confidence in the prediction. It was concluded that the model would be considered as part of the overall weight of evidence.

 

The DEREK skin sensitisation module indicated the query structure does not match any structural alerts or examples for skin sensitisation in Derek. Additionally, the query structure does not contain any unclassified or misclassified features and is consequently predicted to be a non-sensitiser.by the software. The uncertainty of the prediction cannot be directly measured/assessed as no statistics are reported by the model. The result will be used as part of the weight of evidence.

 

Prediction with OECD Toolbox was performed first via the automatic workflow, then manually amended after the workflow failed to conclude a result for skin sensitisation. Isotridecyl isononanoate did not trigger an alert by any of the skin sensitisation profilers and so the automatic workflow selected the ECOSAR profiler as a starting point, this was then subcategorised using maximal data by substance type, Protein binding alerts for skin sensitization by oasis, Protein binding alerts for skin sensitization by oasis accounting autoxidation simulator, Protein binding alerts for skin sensitization by oasis accounting skin metabolism simulator, but the automated work flow could not conclude on a suitable grouping. At this point the prediction was manually further subcategorised by structural similarity to 50%. The resulting category are a group of similar substances to the target molecule which show the same predicted autoxidation and skin metabolism pathway and all carry negative study data, which can be seen from the resultant data matrix. There is high similarity of the substances to the query structure, high concordance of results within the category. As such the result is considered to be reliable.

 

Table1    Prediction results

Model	Prediction result	Reliability
TOPKAT extended	Non-sensitising	Moderate
TOPKAT non extended	Non-sensitising	Moderate
DEREK	Non-sensitising	Moderate
OECD Toolbox	Non-sensitising	Moderate
CAESAR (VEGA)	Sensitising	Discounted
Toxtree	No alerts identified	Not applicable
Danish (Q)SAR Database	Non-Sensitising	Not applicable

 

Conclusion

There is evidence that Isotridecyl isononanoate is not classified as a skin sensitizer.

From the predictions with OECD Toolbox, DEREK, and TOPKAT, the models were in concordance, identifying no alerts on the structure or its metabolites. In addition, the Danish QSAR database and Toxtree are also in agreement with these results.

The CAESAR model in Vega drew a conclusion from its assessment that the target substance was nearest to several oxazolone compounds which were classified as sensitizers. As discussed above and below, these compounds are dissimilar to the target in terms of functional groups, and the identified alerts for these compounds are different to that of the target substance when assessed in the QSAR Toolbox. Furthermore, the identified metabolism products from these compounds differ entirely to the target substances metabolites and pathways, it was determined that this prediction was not appropriate and it was discounted from the conclusion.

Therefore the identified conclusions from OECD QSAR Toolbox, US EPA TEST, and DEREK NEXUS were considered valid, and in concordance, providing a weight of evidence proposing that the substance should be considered not classified for skin sensitisation.