Registration Dossier

Toxicological information

Genetic toxicity: in vitro

Currently viewing:

Administrative data

Endpoint:
in vitro gene mutation study in bacteria
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: Leadscope Model Applier, version 2.1.2.

2. MODEL: Leadscope QSAR Genetic Toxicity - Salmonella, v3

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL:
Structural formula: C18H27N3O9S3
Structural codes:
a. SMILES: O=C(OCCn1c(=O)n(c(=O)n(CCOC(=O)CCS)c1=O)CCOC(=O)CCS)CCS
b. InChI: InChI=1S/C18H27N3O9S3/c22-13(1-10-31)28-7-4-19-16(25)20(5-8-29-14(23)2-11-32)18(27)21(17(19)26)6-9-30-15(24)3-12-33/h31-33H,1-12H2
c. Other structural representation: mol file used and included in the test material information.

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
- Defined endpoint: Mutagenicity - microbial in vitro Salmonella
- Unambiguous algorithm: The predictive algorithm is based on a Partial Logistic Regression (PLS), which uses structural features and eight calculated properties (i.e., MW, LogP, polar surface area, H bond acceptors, H bond donors, no. rotational bonds and Lipinski score (rule violation)) as molecular descriptors.
- Defined domain of applicability: Leadscope uses two parameters to guide the applicability of model domain: 1) having at least one structural feature defined in the model in addition to all the property descriptors; 2) having at least one chemical in a training neighbourhood with at least 30% global similarity to the test structure.
- Appropriate measures of goodness-of-fit and robustness and predictivity: Please see attached QMRF.
- Mechanistic interpretation: not applicable.

5. APPLICABILITY DOMAIN
The target TEMPIC is within the applicability domain, since 6 structural features were found and 33 analogues with a similarity >30%were identified in the model training set.
- Descriptor domain: property descriptors for the target compound have values inside the descriptor range of the compounds of the training set.
- Structural fragment domain: 6 model structural features were identified in the target (majority of negative features, i.e. features mainly represented in negative training compounds).
- Similarity with analogues in the training set: 33 analogues with a similarity >30% were identified in the training set. However, the identified analogues exhibited little similarity toward the target (similarity indices lower than 0.5), thus preventing a robust assessment of prediction reliability.
Please see attached QPRF for structural analogues and further details.

6. ADEQUACY OF THE RESULT
The target TEMPIC was predicted negative for Salmonella in vitro mutagenicity (Ames test) and the prediction was assessed as borderline reliable. This QSAR prediction indicates that the target TEMPIC does not have the potential to induce gene mutation and could be used to assess the mutagenic potential of the substance (e.g., to support the conclusion for no classification for mutagenicity).
This negative Salmonella in vitro mutagenicity QSAR prediction is assessed as adequate for regulatory purposes.

Data source

Reference
Reference Type:
other: Software
Title:
Unnamed
Year:
2013

Materials and methods

Test guideline
Qualifier:
according to
Guideline:
other: REACH Guidance on QSARs R.6 (2008)
Principles of method if other than guideline:
- Software tool used including version: Leadscope Model Applier, version 2.1.2
- Model used:Leadscope QSAR Genetic Toxicity - Salmonella v3
- Model description: see fields 'Justification for type of information' and 'Attached justification'
- Justification of QSAR prediction: see fields 'Justification for type of information' and 'Attached justification'
Type of assay:
bacterial reverse mutation assay

Test material

Reference
Name:
Unnamed
Type:
Constituent
Test material form:
liquid: viscous
Specific details on test material used for the study:
SMILES: O=C(OCCn1c(=O)n(c(=O)n(CCOC(=O)CCS)c1=O)CCOC(=O)CCS)CCS
InChI: InChI=1S/C18H27N3O9S3/c22-13(1-10-31)28-7-4-19-16(25)20(5-8-29-14(23)2-11-32)18(27)21(17(19)26)6-9-30-15(24)3-12-33/h31-33H,1-12H2

Results and discussion

Test results
Species / strain:
other: S. typhimurium
Genotoxicity:
negative
Additional information on results:
Moderate uncertainty was associated with the negative prediction generated for the target TEMPIC since:
- the identified training set analogues did not show a sufficiently high structural similarity (structural similarity > 0.30, but < 0.50).
Overall, the negative Salmonella in vitro mutagenicity QSAR prediction was assessed as borderline reliable.
Remarks on result:
no mutagenic potential (based on QSAR/QSPR prediction)

Applicant's summary and conclusion

Conclusions:
The target TEMPIC is predicted negative for Salmonella in vitro mutagenicity (Ames test). The prediction is assessed as borderline reliable and adequate for regulatory purposes.
Executive summary:

This study was designed to generate in silico (non-testing) genotoxicity data as Salmonella in vitro mutagenicity for TEMPIC. A reliability score of 2 was assigned, since results were derived from a valid QSAR model with adequate and reliable documentation/justification.

The Leadscope QSAR Genetic Toxicity - Salmonella v3 model, implemented in Leadscope Model Applier (version 2.1.2), was employed. This QSAR model estimates the probability that a compound will result positive in the Ames test. Leadscope results include a mutagenicity prediction (positive, negative or not in domain), a positive prediction probability and two parameters which assess model applicability domain, i.e. Model Features Count and 30% Similarity Training Neighbours Count.

Leadscope predicted the target as negative for Salmonella in vitro mutagenicity, based on a positive prediction probability of 0.16. The target compound was included in the applicability domain of the model since 6 model features and 33 analogues with a similarity >30% were identified in the training set. However,

the identified analogues exhibited a little similarity toward the target (similarity indices lower than 0.5), thus preventing a robust assessment of prediction.

The prediction was assessed as borderline reliable, and adequate for regulatory purposes.