2,3-dichloroaniline

Endpoint summary

Description of key information

Key_Target_Skin sensitisation_Diesing, 1990: OECD 406; guinea pig; sensitising

Danish QSAR_Skin sensitisation, 2022: Battery model; positive

Key value for chemical safety assessment

Skin sensitisation

Link to relevant study records

Reference

Reference 1

Endpoint:

skin sensitisation: in chemico

Type of information:

(Q)SAR

Adequacy of study:

supporting study

Study period:

23 SEP 2022

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
OECD QSAR Toolbox v4.5

2. MODEL (incl. version number)
Allergic Contact Dermatitis, Guinea Pig and Human - Danish QSAR DB battery model

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
Nc1cccc(Cl)c1Cl

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
The Danish QSAR group applies an overall definition of structures acceptable for QSAR processing which is applicable for all the in‐house QSAR software. According to this definition accepted structures are organic substances with an unambiguous structure, i.e. so‐called discrete organics defined as: organic compounds with a defined two dimensional (2D) structure containing at least two carbon atoms, only certain atoms (H, Li, B, C, N, O, F, Na, Mg, Si, P, S, Cl, K, Ca, Br, and I), and not mixtures with two or more ‘big components’ when analyzed for ionic bonds (for a number of small known organic ions assumed not to affect toxicity the ‘parent molecule’ is accepted). Calculation 2D structures (SMILES and/or SDF) are generated by stripping off ions (of the accepted list given above). Thus, all the training set and prediction set chemicals are used in their non‐ionized form.

5. APPLICABILITY DOMAIN
Substance was predicted by the Battery model (Danish QSAR database). Battery model draw a conclusion based on 3 different models. CASE Ultra, Leadscope and SciQSAR. 2 out of 3 these models should give a same result to draw a reliable conclusion. For this endpoint, all models showed positive results, Leadscope and SciQSAR results were in domain and CASE Ultra model result was out of domain. However, 2 out of 3 results are given positive and in domain, this prediction falls in the applicability domain of the model.

Guideline:

other: ECHA guidance R.6

Version / remarks:

May 2008

Principles of method if other than guideline:

- Software tool(s) used including version: OECD QSAR Toolbox v4.5
- Model(s) used: Danish QSAR Battery model
- Model description:

CASE Ultra model: MultiCASE CASE Ultra is an artificial intelligence (AI) based computer program with the ability to learn from existing data and is the successor to the program MultiCASE MC4PC. The system can handle large and diverse sets of chemical structures to produce so‐called global (Q)SAR models, which are in reality series of local (Q)SAR models. Upon prediction of a query structure by a given model one or more of these local models, as well as global relationships if these are identified, can be involved if relevant for the query structure. CASE Ultra is a fragment‐based statistical model system. The methodology involves breaking down the structures of the training set into all possible fragments from 2 to 10 heavy (non‐hydrogen) atoms in length. The fragment generation procedure produces simple linear chains of varying lengths and branched fragments as well as complex substructures generated by combining the simple fragments. A structural fragment is considered as a positive alert if it has a statistically significant association with chemicals in the active category. It is considered a deactivating alert if it has a statistically significant relation with the inactive category. Once final lists of positive and deactivating alerts are identified, CASE Ultra attempts to build local (Q)SARs for each alert in order to explain the variation in activity within the training set chemicals covered by that alert. The program calculates multiple molecular descriptors from the chemical structure such as molecular orbital energies and two‐dimensional distance descriptors. A stepwise regression method is used to build the local (Q)SARs based on these molecular descriptors. For each step a new descriptor (modulator) is added if the addition is statistically significant and increases the cross‐validated R2 (the internal performance) of the model. The number of descriptors in each local model is never allowed to exceed one fifth of the number of training set chemicals covered by that alert. If the final regression model for the alert does not satisfy certain criteria (R2 ≥ 0.6 and Q2 ≥ 0.5) it is rejected. Therefore, not all alerts will necessarily have a local (Q)SAR. The collection of positive and deactivating alerts with or without a local (Q)SAR constitutes a global (Q)SAR model for a particular endpoint and can be used for predicting the activity of a test chemical.

Leadscope model: Leadscope Predictive Data Miner is a software program for systematic sub‐structural analysis of a chemical using predefined structural features stored in a template library, training set‐dependent generated structural features (scaffolds) and calculated molecular descriptors. The feature library contains approximately 27,000 pre‐defined structural features and the structural features chosen for the library are motivated by those typically found in small molecules: aromatics, heterocycles, spacer groups, simple substituents. Leadscope allows for the generation of training set‐dependent structural features (scaffold generation), and these features can be added to the pre‐defined structural features from the library and be included in the descriptor selection process. It is possible in Leadscope to remove redundant structural features before the descriptor selection process and only use the remaining features in the descriptor selection process. Besides the structural features Leadscope also calculates eight molecular descriptors for each training set structure: the octanol/water partition coefficient (alogP), hydrogen bond acceptors (HBA), hydrogen bond donors (HBD), Lipinski score, atom count, parent compound molecular weight, polar surface area (PSA) and rotatable bonds. These eight molecular descriptors are also included in the descriptor selection process.

SciQSAR model: For a binary classification problem SciQSAR uses discriminant analysis (DA) to make a (Q)SAR model. SciQSAR implements a broad range of discriminant analysis (DA) methods including parametric and nonparametric approaches. The classic parametric method of DA is applicable in the case of approximately normal within‐class distributions. The method generates either a linear discriminant function (the withinclass covariance matrices are assumed to be equal) or a quadratic discriminant function (the within‐class covariance matrices are assumed to be unequal). When the distribution is assumed to not follow a
particular law or is assumed to be other than the multivariate normal distribution, non‐parametric DA methods can be used to derive classification criteria. The non‐parametric DA methods available within SciQSAR include the kernel and k‐nearest‐neighbor (kNN) methods. The main types of kernels implemented in SciQSAR include uniform, normal, Epanechnikov, bi‐weight, or tri‐weight kernels, which are used to estimate the group specific density at each observation. Either Mahalanobis or Euclidean distances can be used to determine proximity between compound‐vectors in multidimensional descriptor space. When the kNN method is used, the Mahalanobis distances are based on the pooled covariance matrix. When the kernel method is used, the Mahalanobis distances are based on either the individual within‐group covariance matrices or the pooled covariance matrix. (Contrera et al. 2004)

- Justification of QSAR prediction: see field 'Justification for type of information'

Specific details on test material used for the study:

Nc1cccc(Cl)c1Cl

Group:

test chemical

Run / experiment:

mean

Remarks on result:

positive indication of skin sensitisation

Interpretation of results:

study cannot be used for classification

Remarks:

QSAR alone is not sufficient for classification

Conclusions:

The skin sensitisation QSAR calculation based on Battery model in Danish QSAR Database gives a positive result. This prediction falls in the applicability domain of the model.

Executive summary:

The skin sensitisation QSAR calculation based on Battery model in Danish QSAR Database gives a positive result. This prediction falls in the applicability domain of the model.

Reference 2

Endpoint:

skin sensitisation: in vivo (non-LLNA)

Type of information:

read-across from supporting substance (structural analogue or surrogate)

Adequacy of study:

key study

Reliability:

1 (reliable without restriction)

Justification for type of information:

In this justification, the read-across concept is applied. Please refer to a full version of Read-across statement attached in the section 13 "Assessment reports".

Reason / purpose for cross-reference:

read-across source

Reading:

rechallenge

Hours after challenge:

72

Group:

negative control

Dose level:

5 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

0

Total no. in group:

10

Clinical observations:

none

Reading:

rechallenge

Hours after challenge:

72

Group:

test chemical

Dose level:

5 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

4

Total no. in group:

20

Clinical observations:

4 animals with discrete or patchy erythema

Reading:

rechallenge

Hours after challenge:

48

Group:

negative control

Dose level:

5 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

1

Total no. in group:

10

Clinical observations:

1 animal with dicrete or patchy erythema

Reading:

rechallenge

Hours after challenge:

48

Group:

test chemical

Dose level:

5 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

14

Total no. in group:

20

Clinical observations:

14 animals with discrete or patchy erythema

Reading:

rechallenge

Hours after challenge:

72

Group:

negative control

Dose level:

25 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

5

Total no. in group:

10

Clinical observations:

5 animals with discrete or patchy erythema

Reading:

rechallenge

Hours after challenge:

72

Group:

test chemical

Dose level:

25 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

17

Total no. in group:

20

Clinical observations:

17 animals with discrete or patchy erythema

Reading:

rechallenge

Hours after challenge:

48

Group:

negative control

Dose level:

25 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

8

Total no. in group:

10

Clinical observations:

7 animals with discrete or patchy erythema, 1 animal with moderate and confluent erythema

Reading:

rechallenge

Hours after challenge:

48

Group:

test chemical

Dose level:

25 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

19

Total no. in group:

20

Clinical observations:

13 animals with discrete or patchy erythema, 6 animals with moderate and confluent erythema

Reading:

2nd reading

Hours after challenge:

72

Group:

negative control

Dose level:

50 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

7

Total no. in group:

10

Clinical observations:

7 animals with discrete or patchy erythema

Reading:

2nd reading

Hours after challenge:

72

Group:

test chemical

Dose level:

50 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

18

Total no. in group:

20

Clinical observations:

11 animals with discrete or patchy erythema, 4 animals with moderate and confluent erythema and 3 animals with intense erythema and swelling

Reading:

1st reading

Hours after challenge:

48

Group:

negative control

Dose level:

50 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

9

Total no. in group:

10

Clinical observations:

9 animals with discrete or patchy erythema

Reading:

1st reading

Hours after challenge:

48

Group:

test chemical

Dose level:

50 % 3,4-dichloroaniline in propylene glycol

No. with + reactions:

20

Total no. in group:

20

Clinical observations:

15 animals with moderate to confluent erythema; 5 animals with discrete or patchy erythema

During the range finding test for the challenge concentration acute dermal toxicity was observed after simultaneous exposure of each of five guinea pigs to 4 occlusive gauze patches treated with either 0.5 mL 6, 12, 25 or 50 % 3,4 -dichloroaniline in propylene glycol. Two animals were found dead after 24 h and the remaining three were euthanized for animal welfare reasons due to severe signs of toxicity.

 

Conclusions:

3,4 -dichloroaniline was found to be sensitising in this guinea pig maximisation test.

Executive summary:

3,4 -dichloroaniline was tested for skin sensitisation according to OECD Guideline 406 in a Magnusson and Kligman test in male guinea pigs. Although irritation was also evident in the control group after challenge or rechallenge with either 50, 25 or 5 % 3,4 -dichloroaniline in propylene glycol (w/w) the incidence and severity of irritation was always higher in the induced test group. Therefore 3,4 -dichloroaniline was concluded to be sensitising in this guinea pig maximisation test.

Endpoint conclusion

Endpoint conclusion:

adverse effect observed (sensitising)

Additional information:

Guinea pig maximisation Test - Read across to 3,4 -dichloroaniline

3,4-dichloroaniline was tested for skin sensitisation according to OECD Guideline 406 in a Magnusson and Kligman test in male guinea pigs. Although irritation was also evident in the control group after challenge or rechallenge with either 50, 25 or 5 % 3,4-dichloroaniline in propylene glycol (w/w) the incidence and severity of irritation was always higher in the induced test group. Therefore 3,4-dichloroaniline was concluded to be sensitising in this guinea pig maximisation test. Based on the read across approach, this result is also fully applicable to the registered substance 2,3-dichloroaniline.

The skin sensitisation QSAR calculation based on Battery model in Danish QSAR Database gives a positive result. This prediction falls in the applicability domain of the model.

Respiratory sensitisation

Endpoint conclusion

Endpoint conclusion:

no study available

Justification for classification or non-classification

Read-across approach data on 3,4-dichloroaniline (CAS 95-76-1, source substance) is used to fill data gaps for the structural isomer 2,3-dichloroaniline (CAS 608-27-5, target substance), in accordance with Regulation No 1907/2006 (REACH), Annex XI. Skin sensitisation study was done on male guinea pigs with 3,4-dichloroaniline, irritation was evident in the control group after challenge or rechallenge with either 50, 25 or 5 % 3,4-dichloroaniline in propylene glycol (w/w), the incidence and severity of irritation was higher in the induced test group. Therefore 3,4-dichloroaniline was concluded to be skin sensitising and classified as Skin Sens. 1 (H317). This result is fully valid for the target substance 2,3-dichloroaniline without restrictions. Also, the QSAR data supports the positive result of skin sensitisation (inside applicability domain, positive for skin sensitisation). Thus, the target substance is considered to be positive for endpoint skin sensitisation and it is classified as Skin Sens. 1 (H317).