Reaction mass of diethyl hydrogen 2-hydroxypropane-1,2,3-tricarboxylate and ethyl dihydrogen 2-hydroxypropane-1,2,3-tricarboxylate and triethyl citrate

Toxicity to microorganisms

Link to relevant study record(s)

Reference

Reference 1

Endpoint:

activated sludge respiration inhibition testing

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

January 25, 2018

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a (Q)SAR model, with limited documentation / justification, but validity of model and reliability of prediction considered adequate based on a generally acknowledged source

Justification for type of information:

The predictive ability of each of the QSAR methodologies was evaluated using statistical external validation.
A QSAR model has acceptable predictive power if the following conditions are satisfied:

q^2 > 0.5 (1)
R^2 > 0.6 (2)
(R^2-Ro^2)/R^2 < 0.1 and 0.85 <= k <= 1.15 (3)

where:
q^2 is the leave one out correlation coefficient for the training set
R^2 is correlation coefficient between the observed and predicted toxicities for the test set
Ro^2 is correlation coefficient between the observed and predicted toxicities for the test set with the Y-intercept set to zero (where the regression line is given by Y=kX)

The prediction accuracy will be evaluated in terms of equations (2) and (3).
In addition, the accuracy will be evaluated in terms of the RMSE (root mean square error), and the MAE (mean absolute error) for the test set.
It has been demonstrated that q^2 is not correlated with R^2 for the test set.
The prediction coverage (fraction of chemicals predicted) must be considered because the prediction accuracy (in terms of R^2 and RMSE) can sometimes be improved at the sacrifice of the prediction coverage.
For binary (active/inactive) toxicity endpoints such as developmental toxicity, the prediction accuracy is evaluated in terms of the fraction of compounds that are predicted accurately.
The prediction accuracy is evaluated in terms of three different statistics: concordance, sensitivity, and specificity.
Concordance is the fraction of all compounds that are predicted correctly (i.e. experimentally active compounds that are predicted to be active and experimentally inactive compounds that are predicted to be inactive). Sensitivity is the fraction of experimentally active compounds that are predicted to be active.
Specificity is the fraction of experimentally inactive compounds that are predicted to be inactive.

Qualifier:

equivalent or similar to guideline

Guideline:

other: ECHA Guidance on information requirements and chemical safety assessment - Chapter R.06: QSARs and grouping of chemicals

Principles of method if other than guideline:

T.E.S.T. has been developed to allow users to easily estimate toxicity using a variety of QSAR methodologies. T.E.S.T provides multiple prediction methodologies so one can have greater confidence in the predicted toxicities (assuming the predicted toxicities are similar from different methods).

(Q)SAR PREDICTION METHODOLOGIES
• Hierarchical method
The toxicity for a given query compound is estimated using the weighted average of the predictions from several different models. The different models are obtained by using Ward’s method to divide the training set into a series of structurally similar clusters. A genetic algorithm based technique is used to generate models for each cluster. The models are generated prior to runtime.

• FDA method
The prediction for each test chemical is made using a new model that is fit to the chemicals that are most similar to the test compound. Each model is generated at runtime.

• Single model method
Predictions are made using a multilinear regression model that is fit to the training set (using molecular descriptors as independent variables) using a genetic algorithm based approach. The regression model is generated prior to runtime.

• Group contribution method
Predictions are made using a multilinear regression model that is fit to the training set (using molecular fragment counts as independent variables). The regression model is generated prior to runtime.

• Nearest neighbor method
The predicted toxicity is estimated by taking an average of the three chemicals in the training set that are most similar to the test chemical.

• Consensus method
The predicted toxicity is estimated by taking an average of the predicted toxicities from the above (Q)SAR methods (provided the predictions are within the respective applicability domains).

• Random forest method
The predicted toxicity is estimated using a decision tree which bins a chemical into a certain toxicity score (i.e. positive or negative developmental toxicity) using a set of molecular descriptors as decision variables. The random forest method is currently only available for the developmental toxicity endpoint. The random forest models for the developmental toxicity endpoint were developed by researchers at Mario Negri Institute for Pharmacological Research as part of the CAESAR project (CAESAR 2009).

Analytical monitoring:

not specified

Vehicle:

not specified

Test organisms (species):

Tetrahymena pyriformis

Test type:

other: in silico estimation

Water media type:

freshwater

Total exposure duration:

48 h

Test temperature:

25 deg C

Reference substance (positive control):

not specified

Duration:

48 h

Dose descriptor:

IC50

Effect conc.:

1 742 mg/L

Nominal / measured:

estimated

Conc. based on:

test mat.

Basis for effect:

growth inhibition

Remarks on result:

other: Consensus method

Reported statistics and error estimates:

Statistical external validation
The prediction results for the IGC50 test set were as follows:

Method R2 (R^2-R0^2)/R^2 k RMSE MAE Coverage
Hierarchical 0.719 0.023 0.978 0.539 0.358 0.933
FDA 0.747 0.056 0.988 0.489 0.337 0.978
Group contribution 0.682 0.065 0.994 0.575 0.411 0.955
Nearest neighbor 0.600 0.170 0.976 0.638 0.451 0.986
Consensus 0.764 0.065 0.983 0.475 0.332 0.983

The Consensus method achieved the best results. The R2 value for the consensus method in version 4.1 of TEST was slightly lower than the value for version 4.0. This is due to the fact that the data set has been expanded to include a wider variety of chemical classes.

Predicted T. pyriformis IGC50 (48 hr) for 1,2-DEC from Consensus method:

Prediction results
Endpoint	Experimental value	Predicted value
T. pyriformis IGC50(48 hr) -Log10(mol/L)	N/A	2.23
T. pyriformis IGC50(48 hr) mg/L	N/A	1741.88

Individual Predictions
Method	Predicted value -Log10 (mol/L)
Hierarchical clustering	N/A
Group contribution	2.01
FDA	1.82
Nearest neighbor	2.85

Validity criteria fulfilled:

yes

Conclusions:

The IC50, on the basis of growth inhibition effect (48h) of 1,2-diethyl citrate on Tetrahymena pyriformis, was estimated to be 1742 mg/L.

Executive summary:

The IC50, on the basis of growth inhibition effect (48h) of 1,2-diethyl citrate on Tetrahymena pyriformis, was estimated to be 1742 mg/L.

Reference 2

Endpoint:

activated sludge respiration inhibition testing

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

January 25, 2018

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a (Q)SAR model, with limited documentation / justification, but validity of model and reliability of prediction considered adequate based on a generally acknowledged source

Justification for type of information:

The predictive ability of each of the QSAR methodologies was evaluated using statistical external validation.
A QSAR model has acceptable predictive power if the following conditions are satisfied:

q^2 > 0.5 (1)
R^2 > 0.6 (2)
(R^2-Ro^2)/R^2 < 0.1 and 0.85 <= k <= 1.15 (3)

where:
q^2 is the leave one out correlation coefficient for the training set
R^2 is correlation coefficient between the observed and predicted toxicities for the test set
Ro^2 is correlation coefficient between the observed and predicted toxicities for the test set with the Y-intercept set to zero (where the regression line is given by Y=kX)

The prediction accuracy will be evaluated in terms of equations (2) and (3).
In addition, the accuracy will be evaluated in terms of the RMSE (root mean square error), and the MAE (mean absolute error) for the test set.
It has been demonstrated that q^2 is not correlated with R^2 for the test set.
The prediction coverage (fraction of chemicals predicted) must be considered because the prediction accuracy (in terms of R^2 and RMSE) can sometimes be improved at the sacrifice of the prediction coverage.
For binary (active/inactive) toxicity endpoints such as developmental toxicity, the prediction accuracy is evaluated in terms of the fraction of compounds that are predicted accurately.
The prediction accuracy is evaluated in terms of three different statistics: concordance, sensitivity, and specificity.
Concordance is the fraction of all compounds that are predicted correctly (i.e. experimentally active compounds that are predicted to be active and experimentally inactive compounds that are predicted to be inactive). Sensitivity is the fraction of experimentally active compounds that are predicted to be active.
Specificity is the fraction of experimentally inactive compounds that are predicted to be inactive.

Qualifier:

equivalent or similar to guideline

Guideline:

other: ECHA Guidance on information requirements and chemical safety assessment - Chapter R.06: QSARs and grouping of chemicals

Principles of method if other than guideline:

T.E.S.T. has been developed to allow users to easily estimate toxicity using a variety of QSAR methodologies. T.E.S.T provides multiple prediction methodologies so one can have greater confidence in the predicted toxicities (assuming the predicted toxicities are similar from different methods).

(Q)SAR PREDICTION METHODOLOGIES
• Hierarchical method
The toxicity for a given query compound is estimated using the weighted average of the predictions from several different models. The different models are obtained by using Ward’s method to divide the training set into a series of structurally similar clusters. A genetic algorithm based technique is used to generate models for each cluster. The models are generated prior to runtime.

• FDA method
The prediction for each test chemical is made using a new model that is fit to the chemicals that are most similar to the test compound. Each model is generated at runtime.

• Single model method
Predictions are made using a multilinear regression model that is fit to the training set (using molecular descriptors as independent variables) using a genetic algorithm based approach. The regression model is generated prior to runtime.

• Group contribution method
Predictions are made using a multilinear regression model that is fit to the training set (using molecular fragment counts as independent variables). The regression model is generated prior to runtime.

• Nearest neighbor method
The predicted toxicity is estimated by taking an average of the three chemicals in the training set that are most similar to the test chemical.

• Consensus method
The predicted toxicity is estimated by taking an average of the predicted toxicities from the above (Q)SAR methods (provided the predictions are within the respective applicability domains).

• Random forest method
The predicted toxicity is estimated using a decision tree which bins a chemical into a certain toxicity score (i.e. positive or negative developmental toxicity) using a set of molecular descriptors as decision variables. The random forest method is currently only available for the developmental toxicity endpoint. The random forest models for the developmental toxicity endpoint were developed by researchers at Mario Negri Institute for Pharmacological Research as part of the CAESAR project (CAESAR 2009).

Analytical monitoring:

not specified

Vehicle:

not specified

Test organisms (species):

Tetrahymena pyriformis

Test type:

other: in silico estimation

Water media type:

freshwater

Total exposure duration:

48 h

Test temperature:

25 deg C

Reference substance (positive control):

not specified

Duration:

48 h

Dose descriptor:

IC50

Effect conc.:

2 858 mg/L

Nominal / measured:

estimated

Conc. based on:

test mat.

Basis for effect:

growth inhibition

Remarks on result:

other: Consensus method

Reported statistics and error estimates:

Statistical external validation
The prediction results for the IGC50 test set were as follows:

Method R2 (R^2-R0^2)/R^2 k RMSE MAE Coverage
Hierarchical 0.719 0.023 0.978 0.539 0.358 0.933
FDA 0.747 0.056 0.988 0.489 0.337 0.978
Group contribution 0.682 0.065 0.994 0.575 0.411 0.955
Nearest neighbor 0.600 0.170 0.976 0.638 0.451 0.986
Consensus 0.764 0.065 0.983 0.475 0.332 0.983

The Consensus method achieved the best results. The R2 value for the consensus method in version 4.1 of TEST was slightly lower than the value for version 4.0. This is due to the fact that the data set has been expanded to include a wider variety of chemical classes.

Predicted T. pyriformis IGC50 (48 hr) for 1-MEC citrate from Consensus method:

Prediction results
Endpoint	Experimental value	Predicted value
T. pyriformis IGC50(48 hr) -Log10(mol/L)	N/A	1.89
T. pyriformis IGC50(48 hr) mg/L	N/A	2858

Individual Predictions
Method	Predicted value -Log10 (mol/L)
Hierarchical clustering	N/A
Group contribution	1.48
FDA	1.60
Nearest neighbor	2.58

Validity criteria fulfilled:

yes

Conclusions:

The IC50, on the basis of growth inhibition effect (48h) of 1-ethyl citrate on Tetrahymena pyriformis, was estimated to be 2858 mg/L.

Executive summary:

The IC50, on the basis of growth inhibition effect (48h) of 1-ethyl citrate on Tetrahymena pyriformis, was estimated to be 2858 mg/L.

Reference 3

Endpoint:

activated sludge respiration inhibition testing

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

January 25, 2018

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

results derived from a (Q)SAR model, with limited documentation / justification, but validity of model and reliability of prediction considered adequate based on a generally acknowledged source

Justification for type of information:

The predictive ability of each of the QSAR methodologies was evaluated using statistical external validation.
A QSAR model has acceptable predictive power if the following conditions are satisfied:

q^2 > 0.5 (1)
R^2 > 0.6 (2)
(R^2-Ro^2)/R^2 < 0.1 and 0.85 <= k <= 1.15 (3)

where:
q^2 is the leave one out correlation coefficient for the training set
R^2 is correlation coefficient between the observed and predicted toxicities for the test set
Ro^2 is correlation coefficient between the observed and predicted toxicities for the test set with the Y-intercept set to zero (where the regression line is given by Y=kX)

The prediction accuracy will be evaluated in terms of equations (2) and (3).
In addition, the accuracy will be evaluated in terms of the RMSE (root mean square error), and the MAE (mean absolute error) for the test set.
It has been demonstrated that q^2 is not correlated with R^2 for the test set.
The prediction coverage (fraction of chemicals predicted) must be considered because the prediction accuracy (in terms of R^2 and RMSE) can sometimes be improved at the sacrifice of the prediction coverage.
For binary (active/inactive) toxicity endpoints such as developmental toxicity, the prediction accuracy is evaluated in terms of the fraction of compounds that are predicted accurately.
The prediction accuracy is evaluated in terms of three different statistics: concordance, sensitivity, and specificity.
Concordance is the fraction of all compounds that are predicted correctly (i.e. experimentally active compounds that are predicted to be active and experimentally inactive compounds that are predicted to be inactive). Sensitivity is the fraction of experimentally active compounds that are predicted to be active.
Specificity is the fraction of experimentally inactive compounds that are predicted to be inactive.

Qualifier:

equivalent or similar to guideline

Guideline:

other: ECHA Guidance on information requirements and chemical safety assessment - Chapter R.06: QSARs and grouping of chemicals

Principles of method if other than guideline:

T.E.S.T. has been developed to allow users to easily estimate toxicity using a variety of QSAR methodologies. T.E.S.T provides multiple prediction methodologies so one can have greater confidence in the predicted toxicities (assuming the predicted toxicities are similar from different methods).

(Q)SAR PREDICTION METHODOLOGIES
• Hierarchical method
The toxicity for a given query compound is estimated using the weighted average of the predictions from several different models. The different models are obtained by using Ward’s method to divide the training set into a series of structurally similar clusters. A genetic algorithm based technique is used to generate models for each cluster. The models are generated prior to runtime.

• FDA method
The prediction for each test chemical is made using a new model that is fit to the chemicals that are most similar to the test compound. Each model is generated at runtime.

• Single model method
Predictions are made using a multilinear regression model that is fit to the training set (using molecular descriptors as independent variables) using a genetic algorithm based approach. The regression model is generated prior to runtime.

• Group contribution method
Predictions are made using a multilinear regression model that is fit to the training set (using molecular fragment counts as independent variables). The regression model is generated prior to runtime.

• Nearest neighbor method
The predicted toxicity is estimated by taking an average of the three chemicals in the training set that are most similar to the test chemical.

• Consensus method
The predicted toxicity is estimated by taking an average of the predicted toxicities from the above (Q)SAR methods (provided the predictions are within the respective applicability domains).

• Random forest method
The predicted toxicity is estimated using a decision tree which bins a chemical into a certain toxicity score (i.e. positive or negative developmental toxicity) using a set of molecular descriptors as decision variables. The random forest method is currently only available for the developmental toxicity endpoint. The random forest models for the developmental toxicity endpoint were developed by researchers at Mario Negri Institute for Pharmacological Research as part of the CAESAR project (CAESAR 2009).

Analytical monitoring:

not specified

Vehicle:

not specified

Test organisms (species):

Tetrahymena pyriformis

Test type:

other: in silico estimation

Water media type:

freshwater

Total exposure duration:

48 h

Test temperature:

25 deg C

Reference substance (positive control):

not specified

Duration:

48 h

Dose descriptor:

IC50

Effect conc.:

642 mg/L

Nominal / measured:

estimated

Conc. based on:

test mat.

Basis for effect:

growth inhibition

Remarks on result:

other: Consensus method

Reported statistics and error estimates:

Statistical external validation
The prediction results for the IGC50 test set were as follows:

Method R2 (R^2-R0^2)/R^2 k RMSE MAE Coverage
Hierarchical 0.719 0.023 0.978 0.539 0.358 0.933
FDA 0.747 0.056 0.988 0.489 0.337 0.978
Group contribution 0.682 0.065 0.994 0.575 0.411 0.955
Nearest neighbor 0.600 0.170 0.976 0.638 0.451 0.986
Consensus 0.764 0.065 0.983 0.475 0.332 0.983

The Consensus method achieved the best results. The R2 value for the consensus method in version 4.1 of TEST was slightly lower than the value for version 4.0. This is due to the fact that the data set has been expanded to include a wider variety of chemical classes.

Predicted T. pyriformis IGC50 (48 hr) for diethyl citrate (TEC) from Consensus method:

Prediction results
Endpoint	Experimental value	Predicted value
T. pyriformis IGC50(48 hr) -Log10(mol/L)	N/A	2.63
T. pyriformis IGC50(48 hr) mg/L	N/A	641.80

Individual Predictions
Method	Predicted value -Log10 (mol/L)
Hierarchical clustering	N/A
Group contribution	2.54
FDA	2.13
Nearest neighbor	3.23

Validity criteria fulfilled:

yes

Conclusions:

The IC50, on the basis of growth inhibition effect (48h) of triethyl citrate on Tetrahymena pyriformis, was estimated to be 642 mg/L.

Executive summary:

The IC50, on the basis of growth inhibition effect (48h) of triethyl citrate on Tetrahymena pyriformis, was estimated to be 642 mg/L.

Description of key information

Inhibition growth concentration (IGC50):  642 ÷ 1742 mg/L

Key value for chemical safety assessment

EC50 for microorganisms:

642 mg/L

Additional information

The inhibition growth concentration (IGC50) value of diethyl citrate technical in microorganisms (tetrahymena pyriformis) in a 48 h study, on the basis of growth inhibition effect, was estimated to be in range 642 ÷ 1742 mg/L.

For the assessment was conservatively selected the lowest value.