4,7-dimethyloct-6-en-3-one

Toxicity to aquatic algae and cyanobacteria

Link to relevant study record(s)

Reference

Reference 1

Endpoint:

toxicity to aquatic algae and cyanobacteria

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

Jan 2017

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
QSAR Toolbox version 3.4.0.17
2. MODEL (incl. version number)
Not applicable
3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
CCC(=O)C(C)CC=C(C)C
4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
- Defined endpoint: growth rate, EC50, 72h, Pseudokirchnerella subcapitata
- Unambiguous algorithm: EC50 = +1.56 + 0.664 * log Kow, log(1/mol/L)
- Defined domain of applicability: Aquatic toxicity classification by ECOSAR = neutral organic; Protein binding by OASIS = No alert; Acute aquatic toxicity MOA by OASIS = basesurface narcotics; Acute aquatic toxicity classification by Verhaar = Class 1 (narcosis or baseline toxicity), log Kow estimated within QSAR Toolbox of >= 2.8 and <=5.26
- Appropriate measures of goodness-of-fit and robustness and predictivity: N=10, R2=0.889, Q2=0.831
- Mechanistic interpretation: The hypothesis is that chemicals that share the same mode of action for aquatic toxicity, in this case narcosis, can be grouped together for the prediction of the daphnia short-term toxicity. The narcosis effect is based on the assumption that the substances enter the organism by absorption through passive diffusion. This mechanism is non-structure dependent and is driven by hydrophobicity. This assumption is supported by the robust trend found between EC50 and log Kow for all analogues.
5. APPLICABILITY DOMAIN
- Descriptor domain: The logKow of the target as estimated by the QSAR Toolbox v3.4.0.17 is 2.97, which falls within the applicability domain of >= 2.8 and <=5.26
- Structural and mechanistic domains: The target substance falls within the mechanistic domain, as defined by the aquatic toxicity profilers in the QSAR Toolbox (see "Defined domain of applicability" above and attached QSAR Toolbox report for full details).
- Similarity with analogues in the training set: The target and category members have been profiled using the five primary profilers relevant to acute aquatic toxicity endpoints (OECD 2013). By comparing the profiles for the target substance and the category members (see attached QSAR Toolbox report), it can be seen that the five primary profiles for all members are identical and as such it is concluded that all are expected to act via the same mode of action. Therefore the group is considered robust enough to obtain a reliable prediction. The high R2 and Q2 values for the regression equation support the reliability of the prediction.
6. ADEQUACY OF THE RESULT
- The predictions are considered reliable estimates of the algae 72 hour ErC50 as i) the model is scientifically valid and meets the requirements of the OECD principles and ii) the target falls within the applicability domain of the model.
- Thus the predicted algae 72-hr ErC50 value of 45.2 mg/L is considered adequate for regulatory purposes.

For further details see the attached QSAR Toolbox report.

Guideline:

other:

Version / remarks:

REACH Guidance: Chapter R.6: QSARs and grouping of chemicals, May 2008.

Principles of method if other than guideline:

Category formation using QSAR Toolbox and datagap filling using trend analysis.

Specific details on test material used for the study:

SMILES USED AS FOR THE MODEL
CCC(=O)C(C)CC=C(C)C

The registration substance is a monoconstituent substance with a purity of => 96%. The QSAR prediction has been performed on the structure of the single constituent.
One impurity (CAS 26902-25-0, SMILES C/C(C)=C/COC/C=C(C)/C) may be present in the registered substance at levels of between 0.1% to 4% (see endpoint record 1.2). The impurity is not relevant for classification and labelling and, given the low levels present, is not expected to significantly affect the aquatic toxicity.

Duration:

72 h

Dose descriptor:

EC50

Effect conc.:

45.2 mg/L

Basis for effect:

growth rate

Remarks on result:

other: Predicted value

Details on results:

Category definition:
The target substance was profiled using the five primary profilers relevant to acute aquatic toxicity endpoints. Since the chemical is inert (i.e. contains no structural alerts for protein binding using the mechanistic profilers), the “aquatic toxicity classification by ECOSAR” profiler was used for developing the initial category since it is the most detailed aquatic endpoint profiler in terms of the number of alerts. The target is classified as a Neutral Organic by the ECOSAR profiler. Thus an initial category was formed from chemicals in the Givaudan Aquatic Toxicity Database, that are classified as Neutral Organic (24 substances with experimental data for the endpoint tree: growth rate, EC50, 72h, Pseudokirchnerella subcapitata). Sub-categorised was performed to ensure the category only contained inert analogues that exert their toxicity via narcosis. Sub-categorisation using the ‘protein binding by OASIS V1.1’ profile, the "acute aquatic toxicity classification by Verhaar" and the "acute aquatic toxicity MOA by OASIS" resulted in a category of 16 members. Chemicals were excluded if their EC50 value was reported as a > x mg/L value (i.e. exceeds the highest concentration tested). The final category contained 10 analogues.

Prediction Approach:
Trend analysis from category members; linear approximation.
EC50 = 1.56 + 0.664 Log Kow, log(1/mol/L), N=10, r2=0.889, Q2 = 0.831

See QSAR Toolbox Prediction report for full details including the list of category members. The category members shown with respect to the descriptor log Kow are shown graphically below (section Illustration).

Validity criteria fulfilled:

yes

Remarks:

The model approach is scientifically valid and the target falls within its applicability domain (see attached QSAR Toolbox prediction report for details).

Conclusions:

The registration substance, 4,7-dimethyloct-6-en-3-one, was predicted to have a algae 72-hr ErC50 of 45.2 mg/L. The prediction result is considered reliable according to OECD principles; the prediction model approach is scientifically valid and 4,7-dimethyloct-6-en-3-one falls within its applicability domain. The predicted value is being used in a weight-of-evidence approach, along with two other reliable QSAR predictions, to fill the "toxicity to aquatic algae and cyanobacteria" endpoint.

Reference 2

Endpoint:

toxicity to aquatic algae and cyanobacteria

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

Jan 2017

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
QSAR Toolbox version 3.4.0.17
2. MODEL (incl. version number)
Not applicable
3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
CCC(=O)C(C)CC=C(C)C
4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
- Defined endpoint: growth rate, EC50, 72h, Pseudokirchnerella subcapitata
- Unambiguous algorithm: EC50 = +1.74 + 0.725 * log Kow, log(1/mol/L)
- Defined domain of applicability: Aquatic toxicity classification by ECOSAR = neutral organic; Protein binding by OASIS = No alert; Acute aquatic toxicity MOA by OASIS = basesurface narcotics; Acute aquatic toxicity classification by Verhaar = Class 1 (narcosis or baseline toxicity), log Kow estimated within QSAR Toolbox of >= -0.00488 and <=4.88
- Appropriate measures of goodness-of-fit and robustness and predictivity: N=34, R2=0.852, Q2=0.831
- Mechanistic interpretation: The hypothesis is that chemicals that share the same mode of action for aquatic toxicity, in this case narcosis, can be grouped together for the prediction of the daphnia short-term toxicity. The narcosis effect is based on the assumption that the substances enter the organism by absorption through passive diffusion. This mechanism is non-structure dependent and is driven by hydrophobicity. This assumption is supported by the robust trend found between EC50 and log Kow for all analogues.
5. APPLICABILITY DOMAIN
- Descriptor domain: The logKow of the target as estimated by the QSAR Toolbox v3.4.0.17 is 2.97, which falls within the applicability domain of >= -0.00488 and <=4.88
- Structural and mechanistic domains: The target substance falls within the mechanistic domain, as defined by the aquatic toxicity profilers in the QSAR Toolbox (see "Defined domain of applicability" above and attached QSAR Toolbox report for full details).
- Similarity with analogues in the training set: The target and category members have been profiled using the five primary profilers relevant to acute aquatic toxicity endpoints (OECD 2013). By comparing the profiles for the target substance and the category members (see attached QSAR Toolbox report), it can be seen that the five primary profiles for all members are identical and as such it is concluded that all are expected to act via the same mode of action. Therefore the group is considered robust enough to obtain a reliable prediction. The high R2 and Q2 values for the regression equation support the reliability of the prediction.
6. ADEQUACY OF THE RESULT
- The prediction is considered a reliable estimate of the algae 72-hr ErC50 as i) the model is scientifically valid and meets the requirements of the OECD principles and ii) the target falls within the applicability domain of the model.
- Using the same category of 34 analogues, a predicted was also performed using read-across based on the five nearest neigbours in log Kow space (not reported). The predicted EC50 was 37.9 mg/L, while the predicted EC50 from the trend analysis is 19.6mg/L. Thus, the trend prediction chosen is both more robust and more conservative.
- Thus the predicted algae 72-hr ErC50 value of 19.6 mg/L is considered adequate for regulatory purposes.

For further details see the attached QSAR Toolbox report.

Guideline:

other:

Version / remarks:

REACH Guidance: Chapter R.6: QSARs and grouping of chemicals, May 2008.

Principles of method if other than guideline:

Category formation using QSAR Toolbox and datagap filling using trend analysis.

Specific details on test material used for the study:

SMILES USED AS FOR THE MODEL
CCC(=O)C(C)CC=C(C)C

The registration substance is a monoconstituent substance with a purity of => 96%. The QSAR prediction has been performed on the structure of the single constituent.
One impurity (CAS 26902-25-0, SMILES C/C(C)=C/COC/C=C(C)/C) may be present in the registered substance at levels of between 0.1% to 4% (see endpoint record 1.2). The impurity is not relevant for classification and labelling and, given the low levels present, is not expected to significantly affect the aquatic toxicity.

Duration:

72 h

Dose descriptor:

EC50

Effect conc.:

19.6 mg/L

Basis for effect:

growth rate

Remarks on result:

other: Predicted value

Details on results:

Category definition:
The target substance was profiled using the five primary profilers relevant to acute aquatic toxicity endpoints. Since the chemical is inert (i.e. contains no structural alerts for protein binding using the mechanistic profilers), the “aquatic toxicity classification by ECOSAR” profiler was used for developing the initial category since it is the most detailed aquatic endpoint profiler in terms of the number of alerts.
The target is classified as a Neutral Organic by the ECOSAR profiler. Thus an initial category was formed from chemicals in the publicly available Toolbox aquatic toxicity databases available that are classified as Neutral Organic. A trend analysis was performed for the endpoint tree "growth rate, EC50, 72, pseudokirchnerella subcapitata" using data only according to the OECD 201 guideline. This gave an initial category of 147 chemicals. Sub-categorised was performed using the five aquatic toxicity profiles to ensure the category only contained inert analogues that exert their toxicity via narcosis (i.e. by removal of analogues with different profiles to the target). Chemicals were excluded if their EC50 value exceeded the water solubility, if the EC50 value was reported as > x mg/L or a range, and if the substance was a UVCB. The final category contained 34 analogues.

Prediction Approach:
Trend analysis from category members; linear approximation.
EC50 = 1.74 + 0.725 Log Kow, log(1/mol/L), N=34, r2=0.852, Q2 = 0.831

See QSAR Toolbox Prediction report, attached in the "full study report" section, for full details including the list of category members. The category members shown with respect to the descriptor log Kow are shown graphically below (Figure 1).

Inter-species variability considerations:
Since species variation can lead to a worse correlation and misleading results, a trend for a single species was identified. The species with the most data points was chosen. Thus the category endpoint predicted was growth rate, EC50, 72h, pseudokirchnerella subcapitata (see above). However, to explore the interspecies variability and the possible existence of a more sensitive species, the datagap filling was also performed without defining the species or guideline and setting the "data usage" as "all" as opposed to the default "average". In this way all available 72h Ec50 values were gathered for "green algae". This gave an initial category of 368 chemicals. Sub-categorisation as described above gave a final category of 106 analogues (118 values). The predicted EC50 using trend analysis was 19 mg/L. Since this is virtually identical to the prediction using the single species it confirms that no bias in the prediction was introduced by selecting only one species, pseudokirchnerella subcapitata. A plot of the data (Figure 2 below), also shows that pseudokirchnerella subcapitata is not less sensitive than any of the other species. Thus the chosen approach above is justified. The Toolbox report for the prediction using all green algae species is attached as background information.

Validity criteria fulfilled:

yes

Remarks:

The model approach is scientifically valid and the target falls within its applicability domain (see attached QSAR Toolbox prediction report for details).

Conclusions:

The registration substance, 4,7-dimethyloct-6-en-3-one, was predicted to have a algae 72-hr ErC50 of 19.6 mg/L. The prediction result is considered reliable according to OECD principles; the prediction model approach is scientifically valid and 4,7-dimethyloct-6-en-3-one falls within its applicability domain. The predicted value is being used in a weight-of-evidence approach, along with two other reliable QSAR predictions, to fill the "toxicity to aquatic algae and cyanobacteria" endpoint.

Reference 3

Endpoint:

toxicity to aquatic algae and cyanobacteria

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

Jan 2017

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
ECOSAR v1.11

2. MODEL (incl. version number)
Neutral Organic SAR

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

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
The ECOSAR Class Program and underlying methodology have been developed over a period of more than 25 years by EPA/OPPT. The current ECOSAR version (v1.11, May 2012) is programmed to identify
111 chemical classes and allows access to 704 QSARs for numerous endpoints and organisms. The neutral organic 96-hr Green Algae EC50 QSAR is one of these QSARs. The model is valid according to the
following five OECD principles.
1) Defined endpoint: Green Algae 96-hr EC50
2) Unambiguous algorithm: Log 96-h EC50 (mmol/L) = -0.6922 (log Kow) + 0.9253
3) Defined domain of applicability: The Neutral Organic SAR is applicable to non-ionizable and non-reactive chemicals with a Log Kow of < 5 and a molecular weight of < 1000
4) Appropriate measures of goodness-of-fit and robustness and predictivity: a) internal performance; n = 41 (training set), r2 = 0.6782, b) External validation; A number of supporting validation exercises have been performed in conjunction with EPA and other stakeholders on the ECOSAR models and are listed in the "Methodology Document for the Ecological Structure-Activity Relationship Model (ECOSAR) Class Program v1.11"
5) Mechanistic interpretation: The mechanistic basis for the model is nonpolar narcosis. The narcosis effect is based on the assumption that the substances enter the body of the organism by absorption through passive diffusion. This mechanism is non-structure dependent and is driven by hydrophobicity. Thus, it is possible to observe a trend between EC50 and the log Kow of the narcotic chemicals, regardless of structural similarity.

5. APPLICABILITY DOMAIN
- Descriptor domain: The logKow of 4,7-dimethyloct-6-en-3-one, as automatically estimated within the ECOSAR programme, is 2.97 which falls within the applicability domain of < 5.0.
Note: All ECOSAR QSARs, including the Neutral Organic SAR, have been derived using predicted log Kow values (KOWWIN version 1.67) for the training set chemicals to minimize potential measurement
variability which may arise from inconsistent laboratory test conditions. In ECOSAR version 1.11, the log Kow values are predicted using the KOWWIN v1.68 program from U.S. EPA’s EPISuite model.
- Structural and mechanistic domains: The molecular weight of the target is substance 154.25, which falls within the applicability domain of < 1000. Based on the chemical structure, ECOSAR v1.1 did not attribute 4,7-dimethyloct-6-en-3-one to a particular chemical class. Therefore, it is considered that the Neutral Organic SAR is the most appropriate ECOSAR model for 4,7-Dimethyloct-6-en-3-one.
- Similarity with analogues in the training set: The Neutral Organic SAR is applicable to narcotic chemicals, regardless of structural similarity. To this end, structural similarity to analogues in the training set is not relevant.
- Other considerations: Potential structural analogues were identified using the EPA Analog Identification methodology (AIM) and the Research Institute of Fragrance materials (RIFM) database. Two analogues were found with reliable experimental data. The 96 hour EC50 was predicted for both analogues using the ECOSAR neutral organic SAR. The predicted values were lower than the experimental values (see other information on results incl tables) indicating that the Neutral Organic SAR provides a reliable conservative estimate of the aquatic toxicity of the structural analogues, which supports the applicability of this model to 4,7-dimethyloct-6-en-3-one.

6. ADEQUACY OF THE RESULT
- The prediction is considered a reliable estimate of the green algae 72 hour EC50 as i) the model is scientifically valid and meets the requirements of the OECD principles and ii) the target falls within the
applicability domain of the model.
- Thus the predicted green algae 72-hr EC50 value of 11.4 mg/L is considered adequate for regulatory
purposes.
- Furthermore, based on the comparison of the predicted and experimental data for two structural analogues, the predicted value is considered conservative and worst-case.

More detailed information on the APPLICABILITY DOMAIN and ADEQUACY OF THE RESULT is provided in the attached QPRF report.

Guideline:

other:

Version / remarks:

REACH Guidance: Chapter R.6: QSARs and grouping of chemicals, May 2008.

Principles of method if other than guideline:

ECOSAR v.1.11; Neutral Organics, 96-hr Green Algae EC50 regression QSAR model.

Specific details on test material used for the study:

SMILES USED AS FOR THE MODEL
CCC(=O)C(C)CC=C(C)C

The registration substance is a monoconstituent substance with a purity of => 96%. The QSAR prediction has been performed on the structure of the single constituent.
One impurity (CAS 26902-25-0, SMILES C/C(C)=C/COC/C=C(C)/C) may be present in the registered substance at levels of between 0.1% to 4% (see endpoint record 1.2). The impurity is not relevant for classification and labelling and, given the low levels present, is not expected to significantly affect the aquatic toxicity.

Duration:

96 h

Dose descriptor:

EC50

Effect conc.:

11.4 mg/L

Remarks on result:

other: Predicted value

Details on results:

- The ECOSAR v1.11 results file is attached.
- The predicted value is considered a reliable estimate because the model is scientifically valid and 4,7-dimethyloct-6-en-3-one falls within its applicability domain.
- Furthermore, based on experimental and predicted data for two structural analogues (see table below and QMRF report for details), the prediction is considered conservative and worst case.

Table 1:Predicted and experimental algal inhibition EC50 values for analogues

CAS	Log Kow (EPI)	Experimental 72-hr EC50 mg/L	Predicted (ECOSAR v1.11 Neutral Organics) Green algae 96-hr EC50 mg/L
110-93-0	2.062	ca. 116 (growth rate)	39.742
74338-72-0	3.915	7.2 (biomass) 13.3 (growth rate)	2.993

 

The results indicate that the Neutral Organic ECOSAR provides a reliable and conservative estimate of the aquatic toxicity of the structural analogues. This supports the reliability of the predicted value for 4,7-dimethyloct-6-en-3-one.

Validity criteria fulfilled:

yes

Remarks:

The model approach is scientifically valid and the target falls within its applicability domain (see attached QPRF report for details).

Conclusions:

The registration substance, 4,7-dimethyloct-6-en-3-one, was predicted to have a algae 96-hr EC50 of 11.4 mg/L using the ECOSAR (v1.11) neutral organics SAR. The prediction result is considered reliable according to OECD principles; the QSAR model is scientifically valid and 4,7-dimethyloct-6-en-3-one falls within its applicability domain. Based on experimental and predicted data for two structural analogues, the prediction is considered conservative and worst case. The predicted value is being used in a weight-of-evidence approach, along with two other reliable QSAR predictions, to fill the "toxicity to aquatic algae and cyanobacteria" endpoint.

Description of key information

In the absence of experimental data, the use of predicted values in a weight of evidence approach has been used to assess toxicity to aquatic algae. Two well-established software tools have been used, OECD QSAR Toolbox v3.4.0.17 and ECOSAR v1.11. Two predictions were performed using the QSAR Toolbox; one using a category formed from Givaudans' proprietary database (2017a) and one using a category formed from the aquatic toxicity databases publicly available in the Toolbox (2017b). ECOSAR v1.1 did not attribute 4,7-dimethyloct-6-en-3-one to a particular chemical class and therefore the Neutral Organic SAR was used (2017c). The predicted algae EC50 values were 45.2, 19.6 and 11.4 mg/L respectively. The prediction results are considered reliable according to OECD principles. The prediction models are scientifically valid and 4,7-dimethyloct-6-en-3-one falls within their applicability domain.

The predicted values are within the same acute aquatic toxicity classification band of >10 and <=100mg/L and as such are considered adequate for regulatory purposes. Furthermore, the models are independent from each other (e.g. based on different training sets). Thus, the agreement among the predictions increases the confidence in the reliability of the predictions.