Reaction mass ofDisodium [1-{[2-(hydroxy-kO)-3,5-dinitrophenyl]diazenyl-kN1}naphthalen-2-olato(2-)-kO][3-(hydroxy-kO)-4-{[2-(hydroxy-kO)naphthalen-1-yl]diazenyl-kN1}-7-nitronaphthalene-1-sulfonato(3-)]chromate(2-) andDisodium [1-{[2-(hydroxy-kO)-3,5-dinitrophenyl]diazenyl-kN2}naphthalen-2-olato(2-)-kO][3-(hydroxy-kO)-4-{[2-(hydroxy-kO)naphthalen-1-yl]diazenyl-kN1}-7-nitronaphthalene-1-sulfonato(3-)]chromate(2-)

Categories

Category name:

6.1.3 CLUSTER 10

Justifications and discussions

Category definition:

Similarity among molecules is calculated in a pair-wise manner, first using a molecular fingerprint and
then calculating the difference between the two fingerprints. All possible pairs of molecules are
combined, and the similarities obtained are stored in a matrix with dimensions n x n, where n is the
number of analyzed molecules.
There are different types of fingerprints and different similarity metrics. The fingerprints used here are
the classical Daylight fingerprints, implemented in Python and called RDKit fingerprint (it can be
found and used inside the library RDKit of Python). We found best results using the RDKit fingerprints
respect to Morgan ones or others. After the two molecules are fingerprinted, they can be compared and
the Tanimoto Coefficient (TC) returns the similarity score, evaluating the ratio between identical and
different substructures, with general formula:
TC = a / (a + b + c)
Here, given the two molecules compared, mol1 and mol2:
- a is the number of substructures found in mol1 and in mol2 (shared)
- b is the number of substructures found in mol1 but not in mol2 (not shared)
- c is the number of substructures found in mol2 but not in mol1 (not shared)
Practically, TC is equal to the number of common substructures divided by all substructures (both
shared and not shared). It ranges from 0 to 1, being maximum when the two molecules are identical,
and minimal when the two molecules do not share any substructures. In case of substances formed by
two or more components the same procedure is adopted, because it is always possible to evaluate the
number of common and different substructure. In this case the fingerprints represent the entire
substance and not the single components.

Category description:

The target molecules (the molecules to be predicted) can be clustered together based on their similarity
score. The grouping is useful because the molecules in the same cluster can be considered together and
predicted using the same procedure of reasoning

Category rationale:

REPORTING FORMAT FOR THE CATEGORY APPROACH
Type of information: read-across based on grouping of substances (category approach)
Names and identifications:
312- EC 944-038-4 - Acid Black 107 (target);
313- EC 260-906-9 - Acid Black 194-172 (target);
314- EC 275-033-9 - Acid Blue 158 (target);
375- EC 235-628-6 - Acid Blue 193 (target);
24- EC 414-290-5 - Acid Blue FC 77890 (source - supporting study);
12- EC 402-500-8 - Säure Blau 82 226 (source - key study).
Reliability: 2 (reliable with restriction)
1. CATEGORY HYPOTHESIS
The category described in this report could be justified by the presence of some similarities in the molecular structures, that are at the basis of the category construction with the targets and the similar sources. The cluster analysis included in the category also other similar chemicals, but they were discarded for the lack of reliable data on toxicity or for the presence of relevant molecular differences, as well as the presence of a lower similarity index than the other members.
All the four targets of the cluster are acid dyes that present one diazo bond in the central part of the molecule and that form metal complexes. The targets 313 and 375 are constituted by two identical components, while in the target 312 the components differ for the presence of a benzene ring instead a naphthalene ring in one of them. The target 314 is formed by a single molecular component with a diazo bond that binds two hydroxynapthalenesulphonic groups, similar to the components of the other targets. All the chemicals of the category present a chromium (III) together with one or more sodium counterions.
In detail:
• The target 312 is a disodium [1-{[2-(hydroxy-kO)-3,5-dinitrophenyl]diazenyl-kN1}naphthalen-2-olato(2-)-kO][3-(hydroxy-kO)-4-{[2-(hydroxy-kO)naphthalen-1-yl]diazenyl-kN1}-7-nitronaphthalene-1-sulphonato(3-)]chromate(2-); disodium [1-{[2-(hydroxy-kO)-3,5-dinitrophenyl]diazenyl-kN2}naphthalen-2-olato(2-)-kO][3-(hydroxy-kO)-4-{[2-(hydroxy-kO)naphthalen-1-yl]diazenyl-kN1}-7-nitronaphthalene-1-sulphonato(3-)]chromate(2-), formed by a7-nitro-3-hydroxy-1-naphtalenesulphonic acid attached by a diazo bond to a 2-naphtol, while the second component presents a 2-naphtol linked by a diazo bond to a 3-nitrophenol.
• The target 313 is a trisodium bis[3-hydroxy-4-[(2-hydroxy-1-naphthyl)azo]-7-nitronaphthalene-1-sulphonato(3-)]chromate(3-); a metal-complex diazo dye formed by a chromium trivalent and two identical molecules which have a 7-nitro-3-hydroxy-1-naphtalenesulphonic acid attached by a diazo bond to a 2-naphtol.
• The target 314 is a sodium;chromium(3+);3-oxido-4-[(1-oxido-8-sulphonatonaphthalen-2-yl)diazenyl]naphthalene-1-sulphonate; a metal-complex formed by a chromium trivalent and a 8-hydroxy-1-naphtalenesulphonic acid is attached by a diazo bond to a 3-hydroxy-1-naphtalenesulphonic acid.
• The target 375 is a disodium;chromium(3+);1-[(2-oxidonaphthalen-1-yl)diazenyl]-4-sulphonaphthalen-2-olate;3-oxido-4-[(2-oxidonaphthalen-1-yl)diazenyl]naphthalene-1-sulphonate; a metal-complex diazo dye formed by a chromium trivalent and two identical molecules which present a 3-hydroxy-1-naphtalenesulphonic acid attached by a diazo bond to a 2-naphtol.
For the targets, the available information about the composition of the dye is reported in the section 1.2 - Composition of the registration dossier.
According to the cluster analysis, conducted for every single target chemical, the source chemicals that present good similarity index and reliable experimental data for the endpoint considered are reported below in the data matrix. In particular it should be noted that the clustering analysis shows for the four targets several common source chemicals, and this should also justify the category hypothesis. The sources that present an index similarity <0.6 and with relevant differences in molecular structure were not included in this category, except the source 9 that presents a lower similarity index (anyway >0.5) only for the targets 314 and 372. It was included in this assessment to clarify eventual concerns related to the metal complex structure of this chemical which, unlike the other sources included in this category, exhibits a different toxicity behaviour.
The similar sources which derive from the cluster analysis and that show reliable data and similar fragments are:
• The source 302 that is a sodium;3-hydroxy-4-[(1-hydroxynaphthalen-2-yl)diazenyl]-7-nitronaphthalene-1-sulphonate;
• The source 24 that is a trilithium;chromium(3+);4-[[4-(diethylamino)-2-oxidophenyl]diazenyl]-3-oxidonaphthalene-1-sulphonate;
• The source 12 that is a trisodium (6-anilino-2-(5-nitro-2-oxidophenylazo)-3-sulphonato-1-naphtholato)(4-sulphonato-1,1'-azodi-2,2'-naphtholato)chromate(1-);
• The source 280 that is a sodium;4-[(2-hydroxynaphthalen-1-yl)diazenyl]benzenesulphonate
• The source 292 that is a tetrasodium;3-[[4-[[4-[(6-amino-1-hydroxy-3-sulphonatonaphthalen-2-yl)diazenyl]-7-sulphonatonaphthalen-1-yl]diazenyl]naphthalen-1-yl]diazenyl]naphthalene-1,5-disulphonate
• The source 9 that is a trisodium;6-[(5-chloro-2-hydroxy-4-nitrophenyl)diazenyl]-5-hydroxynaphthalene-1-sulphonic acid;chromium.
For the sources 302, 280 and 292, the study of Croce et al. (2017) indicates a dye contents higher than 70% for each tested formulation. However, information on the identity of impurities is not available. No other information on impurities is available for the other source compounds.
As described above, the four targets present some similarities; therefore, a category approach (many to many), based on the analysis of the similarities of the source molecules (see the data matrix below) could be justified for all the members of the cluster. Based on the similarities, the data quality, and the reliability of the sources, it is possible to include in the category the chemicals 24 and 12. The other sources present some relevant differences and do not add any additional information to characterize this category approach; therefore, they were discarded in consideration of their lower similarity index. Moreover, some of them do not form metal complexes or, even if present, they present additional molecular differences (as the halogens in the source 9) and different counterions (as the Cu instead of Cr in the source 9). The data on water solubility are also comparable for the majority of the members of this category.
2. CATEGORY JUSTIFICATION
The RAAF Scenario is the number 6 (Different compounds have the same type of effect(s)).
• Justify why all compounds - targets and sources - fall in the same category
The category approach can be explained considering that all the category members are in majority diazo dyes which form a metal complex, and which have structural similarities: the four targets and the source chemicals are characterized by the presence of one naphthalene sulphonic acid further substituted with nitro and/or hydroxy functional groups and linked by a diazo bond to another aromatic fragment. They also present similar molecular weight, except for the target 314 that has a single component, instead of the two similar substructures that form the metal complexes; therefore, the molecular weight is about the half compared with other members of the category. All of them are highly soluble in water and their water solubility values are comparable, except for the target 313 and the source 24 where the values are an order of magnitude higher.
As described above they present several structural similarities; therefore, the category construction is justified from a molecular point of view.
The similar sources included in the category present also several similarities with the targets. The source 24 is a metal-complex diazo dye formed by a chromium trivalent and two identical components: a 3-hydroxy-1-naphtalenesulphonic acid attached by a diazo bond to a 3-diethyl-phenol. Lithium counter ions are present instead of sodium counter ions, aside from the Cr3+.
The source 12 is a metal-complex diazo dye showing a trivalent chromium and sodium counterions. One of the sub-structures has a 3-hydroxy-1-naphtalenesulphonic acid linked by a diazo bond to a 2-naphtol, while the latter is a 4-nitrophenol linked by a diazo bond to a 4-hydroxy-6-phenylamino-2-naphtalenesulphonic acid.
Based on the molecular similarities these two sources were selected respectively as supporting study and key study for the category justification, due also to the high reliability of their experimental data, as assessed in the light of their accordance with a Good Laboratory Practises (GLP) standard (i.e., K1 = reliable without restrictions).
The experimental data of these sources show a general toxicity for the endpoint considered (see the data matrix below). Indeed, both are in the range 10-100 mg/L indicating a moderate toxicity. This justifies the category construction with the targets.
The only difference that needs to be mentioned is that the source 24 presents a diethylaminophenol, which is absent in all the targets, but no alerts are known for this substructure in literature for the endpoint of interest nor for the presence of particular metabolic pathways in the aquatic organisms.
The sources 302, 292 and 280 have been discarded in principle because they do not form the metal complex, but also for the reliability of their experimental data: they present some uncertainties and therefore were classified as reliable with some restrictions (K2), for the reasons explained in the paragraph below.
The source 9, despite it forms the metal complex, is not included in the category not so much for the presence of a halogen, as mainly for the difference in the metal (Cu instead of Cr), and the reliability of the experimental toxicity for the endpoint, as detailed below.
Justify for the selection of the experimental source data
The use of sources 12 and 24, instead the other potential sources, is justified by the reliability of the data for the endpoint and the similarity index. The values of the sources 12 and 24 are based on experimental analysis and was classified with high reliability (K1). Both follow a standard GLP protocol (EU Method C.2), and no particular deviations are reported. The analysis results also present NOEC values (50 and 7.81 mg/l after 48h for the source 24 and 12, respectively) that agrees with the EC50 available data.
The other sources have been discarded, despite the good reliability of the experimental data, because of their lower similarity (<0.7), for the evident differences in molecular structures and complexity (the source 280 and 292 form no metal complexes), or for the presence of different chemical groups and metals (i.e., the halogen and the Cu3+ in the chemical source 9). It should also be noted that the experimental value of toxicity for the source 9 presents a K4 reliability because it is unknown the GLP reference standard, the test duration, nor other details as water hardness. For precaution it will be assumed to be referred to a 24 h period.
The sources 302, 292 and 280 present good reliable data, despite they were classified with a K2 score because the short-term toxicity test is not referenced as in GLP, even if it was conducted on the basis of the OECD 202, as reported in the study of Croce et al. (2017) and in the protocol details supplied by the LIFE PROSIL project (http://www.life-prosil.eu/; see 6.1.3 – Cluster 10 – PROSIL Test Protocol.pdf). The test duration was of 48h and the result was referred to a limit test. Thus, the reliability of the data was classified as good (K2, reliable with restrictions) although a complete test on effect concentration for response has not been conducted.
The data on water solubility are also of good quality (K2) and are comparable for some target and the selected source belonging to the category.
• Highlights any inconsistency or underlying uncertainty
As mentioned above the data quality is very good for the sources 24 and 12 of this category. Only the experimental data for 302 was extrapolated from the bibliography, but the test protocol for the endpoint investigated is similar to the OECD 202 and the details are provided by the LIFE PROSIL project (see 6.1.3 – Cluster 10 – PROSIL Test Protocol.pdf). It is possible to assimilate this test as conducted in GLP. In any case, this source was discarded for the reasons explained above.
The only potential sources that present a different behaviour for the endpoint are the sources 9 and 292. After a first assessment, they were not included in this category.
The first one presents a moderate similarity only with target 312 (0.673), whilst for the others is near or below 0.6. Moreover, it presents a different counterion (Cu3+ instead of Cr3+) together with different moieties (the hydroxynitrobenzene ring with a halogen). Since the very high EC50 value (of 549 mg/L) leave few uncertainties about the low toxicity of the substance, the source 9 should be considered as non- toxic, despite the uncertainties on the data. This source was investigated for the sake of clarity in order to show any possible concern related to the toxic behaviour of the category. In any case, this source was not deemed to belong to the same category of the targets for the structural reasons explained above.
The second one, despite presents a similar skeleton of naphthalenesulphonic acid, it is a more complex structure which do not form a metal complex and where some moieties are lacking. Even the source 280 presents the same inconsistencies of the source 292, but in this case, it is less complex and its toxic activity, which in this case is in accordance with the targets, could be probably explained with the lower dimension of this chemical.
The little differences in water solubility data are probably related to experimental conditions (T and pH) which could affect the form of the metal complex, as well as the little differences of the molecular structure (i.e., the presence of additional sulphonates).
In light of this assessment, the category construction performed in this study could be considered reliable with few restrictions.
3. APPLICABILITY DOMAIN OF THE CATEGORY
The choice of the target and the potential source substances that compose the category was based on a clustering method described in “6.1.3 – Cluster 10 - ClusteringMethod.pdf”. This method allows to calculate the similarity of each potential source with the targets and the similarity among the targets. The comparison between two molecules is based on the Tanimoto Coefficient that compares the Daylight fingerprints (as calculated in Python by RDKit) of the two molecules.
In this case, among the pool of potential sources (the ones with a reliable value for the endpoint of interest) the sources 302, 24, 12, 9, 292 and 280 were selected based on their similarity with the targets (the data matrix reports the similarity indexes for each couple of target-source) and the quality of the available data. Data on water solubility are also considered.
The target and the source substances can be considered in the applicability domain of the category due to their high structural similarity (higher than 0.6, with some exception, but no lower than 0.55).