Registration Dossier

Toxicological information

Repeated dose toxicity: oral

Currently viewing:

Administrative data

Endpoint:
short-term repeated dose toxicity: oral
Type of information:
(Q)SAR
Adequacy of study:
supporting study
Reliability:
2 (reliable with restrictions)
Rationale for reliability incl. deficiencies:
results derived from a valid (Q)SAR model, but not (completely) falling into its applicability domain, with adequate and reliable documentation / justification
Justification for type of information:
1. SOFTWARE
https://lazar.in-silico.de/predict

2. MODEL (incl. version number)
Lazar model

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

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
- Defined endpoint: Repeat dose toxicity
- Unambiguous algorithm: Not specified
- Defined domain of applicability: The following factors affect the applicability domain of an individual prediction:
•Number of neighbours
•Similarities of neighbours
•Coherence of experimental data within neighbours
- Appropriate measures of goodness-of-fit and robustness and predictivity: An oral rat LOAEL of 68.0 mg/kg bw/day with a 95% prediction interval from 0.546 mg/kg bw/day to 8470.0 mg/kg bw/day was calculated with reference to three nearest neighbours. Regarding the relatively low similarity index of the neighbours (ca. 0.2), which does not exceed the similarity threshold of 0.5, some uncertainty remains whether β-alanine, N-(4-aminobenzoyl)- is in the applicability domain of the QSAR model.
Lazar predicted a Maximum Recommended Daily Dose (Human) of 9.68 mg/kg bw/day (95% prediction interval 0.0549 - 1710.0 mg/kg bw/day). The result was supported by experimental data on 10 nearest neighbours with similarity indices between 0.533 and 0.211. As only one neighbour showed a similarity index > 0.5 it cannot completely be excluded that the query chemical is outside the applicability domain of this model.
- Mechanistic interpretation: The following information is displayed graphically in the web interface:
•Neighbours that have been used for creating the local QSAR model, together with a graphical display of their structures, activity specific similarities, and experimental measurements
•Activating and deactivating fragments are highlighted in the query compound
•Definitions for domain specific terms can be obtained by following links in the web interface

5. APPLICABILITY DOMAIN
- Descriptor domain: Lazar identifies similar compounds in the training data (neighbours) for a given query compound, creates a local prediction model based on experimental activities of neighbours, and uses the local model to predict properties of the query compound by a regression method. The following factors affect the applicability domain of an individual prediction:
•Number of neighbours
•Similarities of neighbours
•Coherence of experimental data within neighbours
The confidence of a prediction is defined by the mean neighbour similarity.
Applicability domains are tightly integrated with the lazar framework, in that any prediction is associated with a confidence value. Cumulative plots of confidence and accuracy for the experiments for acute toxicity in fish and mutagenicity in bacteria show that the confidence value provides meaningful information, as the model accuracy decreases with decreasing confidence.
- Structural and mechanistic domains: The oral rat LOAEL was calculated with reference to three nearest neighbours. Regarding the relatively low similarity index of the neighbours (ca. 0.2), which does not exceed the similarity threshold of 0.5, some uncertainty remains whether β-alanine, N-(4-aminobenzoyl)- is in the applicability domain of the QSAR model.
prediction interval 0.0549 - 1710.0 mg/kg bw/day). The result for the Maximum Recommended Daily Dose (Human) was supported by experimental data on 10 nearest neighbours with similarity indices between 0.533 and 0.211. As only one neighbour showed a similarity index > 0.5 it cannot completely be excluded that the query chemical is outside the applicability domain of this model.
- Similarity with analogues in the training set: Not determined

6. ADEQUACY OF THE RESULT
In consideration of the basis for the prediction, the Lazar estimates of oral rat LOAEL and Maximum Recommended Daily Dose in humans are considered as reliable within the constraints of QSAR predictions.

Data source

Reference
Reference Type:
publication
Title:
Lazar: a modular predictive toxicology framework
Author:
Maunz A. et al.
Year:
2013
Bibliographic source:
Front Pharmacol, 9 April 2013. doi.org/10.3389/fphar.2013.00038
Report Date:
2018

Materials and methods

Test guideline
Qualifier:
no guideline followed
Principles of method if other than guideline:
- Software tool used including version: https://lazar.in-silico.de/predict
- Model(s) used: Lazar model
- Model description: see field 'Justification for type of information'
- Justification of QSAR prediction: see field 'Justification for type of information'
GLP compliance:
no

Test material

Reference
Name:
Unnamed
Type:
Constituent

Results and discussion

Effect levels

open allclose all
Dose descriptor:
LOAEL
Effect level:
68 other: mg/kg bw/day
Basis for effect level:
other: QSAR prediction
Dose descriptor:
other: Maximum Recommended Daily Dose (Human)
Effect level:
9.68 other: mg/kg bw/day
Basis for effect level:
other: QSAR prediction

Target system / organ toxicity

Critical effects observed:
not specified

Applicant's summary and conclusion

Conclusions:
An oral rat LOAEL of 68.0 mg/kg bw/day was calculated with reference to three nearest neighbours by the Lazar model. Lazar also predicted a Maximum Recommended Daily Dose (Human) of 9.68 mg/kg bw/day.