[No public or meaningful name is available]

Endpoint summary

Description of key information

Weight of Evidence: skin sensitising, 2020

1. Molecular initiating Key Event 1: negative (or minimal peptide reactivity class), DPRA, OECD TG 442C, 2019

2. Molecular initiating Key Event 2: positive (LC induction > 1.5 in 2 of 2 experiments at < 1000 µM) (n=3 replicates), KeratinoSens, OECD TG 442D, 2019

3. Molecular initiating Key Event 3: Negative (mean SMV Decision Value (DV) of -0.59, and since SMV DV<0 (n ≥ 2) not predicted to be sensitising), GARDSkin Assay, 2020

4. QSAR : DEREK NEXUS – skin sensitization : equivocal : prediction for skin sensitisation : alert: 425 Enol Ether. Predicted EC3 = 18% (weak sensitiser) (structural similarity of 12 to 19%), DEREK NEXUS v6.0.1, 2019

5. QSAR : Skin sensitization GHS v.01.02 : non-sensitiser, (ACF: 62% correct, 0% incorrect, 38% unknown), OASIS TIMES v2.30.1, 2020

6. QSAR : Skin sensitization DST (Dermal Sensitization Threshold) v.01.02 : weak/non-sensitiser, (ACF: 62% correct, 0% incorrect, 38% unknown), OASIS TIMES v2.30.1, 2020

7. QSAR : Skin sensitization with autoxidation v.24.29 : non-sensitiser, (ACF: 62% correct, 0% incorrect, 38% unknown), OASIS TIMES v2.30.1, 2020

 

Conclusion: Within the battery of in silico modelling and in vitro test assays, there are equivocal to positive predictions for sensitisation. The test item appears to be either a non-sensitiser to weak sensitiser. Using a precautionary principle conclusion through evaluation of all the available information, the substance is considered a weak sensitiser and to have a low potency (e.g. EC3 >> 2%) based on the weight of evidence. (GHS Classification: Skin Sensitisation Category 1B).

Key value for chemical safety assessment

Skin sensitisation

Link to relevant study records

Reference

Reference 1

Endpoint:

skin sensitisation: in chemico

Type of information:

experimental study

Adequacy of study:

weight of evidence

Study period:

08-08-2019 to 27-08-2019

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

guideline study with acceptable restrictions

Remarks:

Guideline study performed under GLP. All relevant validity criteria were met with acceptable restrictions. Test method considered to cover Key Event-1 under OECD 168 (2012) and OECD 255 (2016) and OECD 256 (2016).

Qualifier:

according to guideline

Guideline:

OECD Guideline 442C (In Chemico Skin Sensitisation: Direct Peptide Reactivity Assay (DPRA))

Deviations:

no

GLP compliance:

yes

Remarks:

Guideline study performed under GLP. All relevant validity criteria were met with acceptable restrictions. Test method considered to cover Key Event-1 under OECD 168 (2012) and OECD 255 (2016) and OECD 256 (2016).

Type of study:

direct peptide reactivity assay (DPRA)

Details on the study design:

Skin sensitisation (In chemico test system) - Details on study design:
- The study was conducted in accordance with the OECD TG 442C – In Chemico Skin Sensitisation: Direct Peptide Reactivity Assay (DPRA)
- The study protocol was validated with the proficiency chemicals prescribed in the OECD test guideline 442C. The results of the testing on the proficiency chemicals at the test facility is in the public domain (refer to study references provided in the full study report at the relevant test facility). All ten proficiency chemicals described in OECD TG 442C: Annex 2, were according to the test facility correctly predicted in a study conducted outside the present study. This information is in the public domain.
- Solubility of the test item in an appropriate solvent was assessed before performing the DPRA. An appropriate solvent dissolved the test item completely, i.e. by visual inspection the solution had to be not cloudy nor have noticeable precipitate. The following solvents were evaluated: acetonitrile (ACN), Milli-Q water (MQ), ACN:MQ (1:1, v/v), isopropanol (IPA), acetone:ACN (1:1, v/v), dimethylsulfoxide (DMSO):ACN (1:9, v/v), methanol and ethanol. At a concentration of 100 mM, the test item was soluble in ACN, acetone:ACN (1:1, v/v), isopropanol (IPA), acetone:ACN (1:1, v/v), dimethylsulfoxide (DMSO):ACN (1:9, v/v), methanol and ethanol ; however not soluble in MQ and ACN:MQ (1:1, v/v). Since ACN is the preferred solvent for the DPRA, this solvent was used to dissolve the test item in the study. Visual inspection of the forming of a clear solution was considered sufficient to ascertain that the test item was dissolved. The test item, positive control and peptide samples were prepared less than 4 hours before starting the incubation of the cysteine (cys) or lysine (lys) reactivity assay, respectively.
- HPLC-PDA (UV) methodology are reported in the full study report.
- Preparation of synthetic peptide solutions [Synthetic Peptide Containing Cysteine (SPCC) and Synthetic Peptide Containing Lysine (SPCL)]
1. Cysteine: A stock solution of 0.667 mM SPCC (0.501 mg SPCC/mL) was prepared by dissolving 10.1 mg of SPCC in 20.16 mL phosphate buffer pH 7.5. The mixture was stirred for 5 minutes followed by 5 minutes sonication. SPCC Reference Control solutions: Three 0.5 mM SPCC reference control (RC) solutions (RCcysA, RCcysB and RCcysC) were prepared in amber vials by mixing 750 µL of the 0.667 mM SPCC stock solution with 250 µL ACN. The SPCC was subsequently calibrated at multiple concentrations. Co-elution control, test item and positive control samples were also prepared (full details on calibration and sample preparation available in the full study report). The mean peptide concentration of Reference Controls A was 0.512 ± 0.001 mM while the mean peptide concentration of Reference Controls C was 0.510 ± 0.001 mM. The means of Reference Control samples A and C were both within the acceptance criteria of 0.50 ± 0.05 mM. This confirms the suitability of the HPLC system and indicates that the solvent (ACN) used to dissolve the test item did not impact the Percent SPCC Depletion.
2. Lysine: A stock solution of 0.667 mM SPCL (0.518 mg SPCL/mL) was prepared by dissolving 10.0 mg of SPCL in 19.31 mL of ammonium acetate buffer pH 10.2 followed by stirring for 5 minutes (and/or sonication if necessary). SPCL Reference Control solutions: Three 0.5 mM SPCL reference control (RC) solutions (RClysA, RClysB and RClysC) were prepared in amber vials by mixing 750 µL of the 0.667 mM SPCL stock solution with 250 µL ACN. The SPLC was subsequently calibrated at multiple concentrations. Co-elution control, test item and positive control samples were also prepared (full details on calibration and sample preparation available in the full study report). The mean peptide concentration of Reference Controls A was 0.504 ± 0.002 mM while the mean peptide concentration of Reference Controls C was 0.500 ± 0.004 mM. The means of Reference Control samples A and C were both within the acceptance criteria of 0.50 ± 0.05 mM. This confirms the suitability of the HPLC system and indicates that the solvent (ACN) used to dissolve the test item did not impact the Percent SPCL Depletion.
- Sample incubations: After preparation, the samples (reference controls, calibration solutions, co-elution control, positive controls and test item samples) were placed in the autosampler in the dark and incubated at 25±2.5°C. The incubation time between placement of the samples in the autosampler and analysis of the first RCcysB- or RClysB-sample was 24.1 hours. The time between the first RCcysB- or RClysB-injection and the last injection of a cysteine or lysine sequence, respectively, did not exceed 30 hours.
- Acceptability criteria:
(i) standard calibration curve(s) are to have an r2 > 0.99. (Actual: SPCC r2 = 0.999 and SPLC r2 = 0.9999)
(ii) mean Percent Peptide Depletion value of the three replicates for the positive control cinnamic aldehyde are to be between 60.8% and 100% for SPCC and between 40.2% and 69.0% for SPCL. (Actual: SPCC 70.9% ± 0.1% and SPCL 62.8% ± 0.4%)
(iii) maximum standard deviation (SD) for the positive control replicates are to be <14.9% for the Percent Cysteine Peptide Depletion and <11.6% for the Percent Lysine Peptide Depletion. (Actual SPCC PC : SD = 0.1% and SPCL PC : SD = 0.4%)
(iv) mean peptide concentration of Reference Controls A is to be 0.50 ± 0.05 mM. (Actual: Cysteine A, C reference controls: 0.512 ± 0.001 mM, and 0.510 ± 0.001 mM ; Lysine A, C reference controls: 0.504 ± 0.002 mM and 0.500 ± 0.004 mM, respectively).
(v) Coefficient of Variation (CV) of peptide areas for the nine Reference Controls B and C in ACN are to be <15.0%. (Actual: Cysteine Reference Controls B and C : CoV = 0.3% ; Lysine: Reference Controls B and C was 0.6%)
Other: Within the Cysteine Reactivity Assay: In the CC sample no peak at 220 nm and 258 nm was observed at the retention time of SPCC and SPLL. For the test item/A-cys samples, the mean SPCC A220/A258 area ratio was 37.13. For the test item/A-lys samples, the mean SPCL A220/A258 area ratio was 31.30. For each sample, a ratio in the range of 90% < mean area ratio of control samples < 110% gives a good indication that co-elution has not occurred.
All acceptance criteria were fulfilled for the PC (cinnamic aldehyde).
All relevant acceptability criteria were met.
- Synthetic peptides:
Cysteine- containing peptide: Ac-RFAACAA-COOH (MW=750.9) – full details on source provided in full study report.
Lysine-containing peptide: Ac-RFAAKAA-COOH (MW=775.9) – full details on source provided in full study report.
- Controls:
Positive control (PC): Cinnamic aldehyde (CAS 104-55-2; 99.1%) – full details on source provided in full study report.
Negative control (NC): Vehicle = Acetonitrile (ACN)
Evaluation of results: In accordance with OECD TG 442C – Table 1.
Test item reactivity was determined by mean peptide depletion and was rated as high, moderate, low, or minimal:
Mean peptide depletion [%] Reactivity
> 42.47 high reactivity (Positive)
> 22.62 < 42.47 moderate reactivity (Positive)
> 6.38 < 22.62 low reactivity (Positive)
< 6.38 minimal reactivity (Negative)

Positive control results:

- All PC acceptability criteria were met.
- PC CYS-peptide depletion (mean): 70.9% ± 0.16% (high reactivity)
- PC LYS-peptide depletion (mean): 62.8% ± 0.4% (high reactivity)

Key result

Run / experiment:

other: Mean (%)

Parameter:

other: Cys-peptide depletion

Value:

1.7 %

Vehicle controls validity:

valid

Negative controls validity:

not applicable

Positive controls validity:

valid

Remarks on result:

other: Negative : No or minimal reactivity

Remarks:

n = 3 ; See 'any other information on results incl. tables' for further information

Key result

Run / experiment:

other: Mean (%)

Parameter:

other: Lys-peptide depletion

Value:

0.1 %

Vehicle controls validity:

valid

Negative controls validity:

not applicable

Positive controls validity:

valid

Remarks on result:

other:

Remarks:

n = 3 ; Negative: No or minimal reactivity ; See 'any other information on results incl. tables' for further information

Other effects / acceptance of results:

OTHER EFFECTS:
- Visible damage on test system: None reported.
- Other: In the Cysteine reactivity assay : upon preparation, both the co-elution control (CC) as well as the test item samples were visually inspected. No precipitate or phase separation was observed in any of the samples. After incubation a phase separation was observed in the CC and test item samples. In this case, it could not be precise the amount of test item remained in the solution to react with the peptide. In the Lysine reactivity assay: upon preparation as well as after incubation a phase separation was observed in the CC and the test item samples. In this case, it could not be precise the amount of test item remained in the solution to react with the peptide..

DEMONSTRATION OF TECHNICAL PROFICIENCY: The study protocol was validated with the proficiency chemicals prescribed in the OECD test guideline 442C. The results of the testing on the proficiency chemicals at the test facility is in the public domain (refer to study references provided in the full study report at the relevant test facility). All ten proficiency chemicals described in OECD TG 442C: Annex 2, were according to the test facility correctly predicted in a study conducted outside the present study. This information is in the public domain.

ACCEPTANCE OF RESULTS:
- Acceptance criteria met for negative control: All criteria met.
- Acceptance criteria met for positive control: All criteria met.
- Acceptance criteria met for variability between replicate measurements: All criteria met.
- Range of historical values if different from the ones specified in the test guideline: Not applicable.

- Acceptability criteria:
(i) standard calibration curve(s) are to have an r2 > 0.99. (Actual: SPCC r2 = 0.999 and SPLC r2 = 0.9999)
(ii) mean Percent Peptide Depletion value of the three replicates for the positive control cinnamic aldehyde are to be between 60.8% and 100% for SPCC and between 40.2% and 69.0% for SPCL. (Actual: SPCC 70.9% ± 0.1% and SPCL 62.8% ± 0.4%)
(iii) maximum standard deviation (SD) for the positive control replicates are to be <14.9% for the Percent Cysteine Peptide Depletion and <11.6% for the Percent Lysine Peptide Depletion. (Actual SPCC PC : SD = 0.1% and SPCL PC : SD = 0.4%)
(iv) mean peptide concentration of Reference Controls A is to be 0.50 ± 0.05 mM. (Actual: Cysteine A, C reference controls: 0.512 ± 0.001 mM, and 0.510 ± 0.001 mM ; Lysine A, C reference controls: 0.504 ± 0.002 mM and 0.500 ± 0.004 mM, respectively).
(v) Coefficient of Variation (CV) of peptide areas for the nine Reference Controls B and C in ACN are to be <15.0%. (Actual: Cysteine Reference Controls B and C : CoV = 0.3% ; Lysine: Reference Controls B and C was 0.6%)
Other: Within the Cysteine Reactivity Assay: In the CC sample no peak at 220 nm and 258 nm was observed at the retention time of SPCC and SPLL. For the test item/A-cys samples, the mean SPCC A220/A258 area ratio was 37.13. For the test item/A-lys samples, the mean SPCL A220/A258 area ratio was 31.30. For each sample, a ratio in the range of 90% < mean area ratio of control samples < 110% gives a good indication that co-elution has not occurred.
All acceptance criteria were fulfilled for the PC (cinnamic aldehyde).
All relevant acceptability criteria were met.

Table 1.0 – Acceptability of the DPRA

	Cysteine reactivity assay		Lysine reactivity assay
	Acceptability criteria	Results for SPCC	Acceptability criteria	Results for SPCL
Correlation coefficient (r2) standard calibration curve	>0.99	0.9999	>0.99	0.9999
Mean peptide concentration RC-A samples (mM)	0.50 ± 0.05	0.512 ± 0.001	0.50 ± 0.05	0.504 ± 0.002
Mean peptide concentration RC-C samples (mM)	0.50 ± 0.05	0.510 ± 0.002	0.50 ± 0.05	0.500 ± 0.004
CV (%) for RC samples B and C	<15.0	0.3	<15.0	0.6
Mean peptide depletion PC (cinnamic aldehyde) (%)	60.8-100	70.9	40.2-69.0	62.8
SD of peptide depletion PC (cinnamic aldehyde) (%)	<14.9	0.1	<11.6	0.4
SD of peptide depletion for the test item (%)	<14.9	0.6	<11.6	0.2

Where: RC = Reference Control; CV = Coefficient of Variation; SD = Standard Deviation

 

Table 2.0 – Results of the DPRA with the test item

	SPCC depletion (CYSTEINE)		SPCL depletion (LYSINE)		Mean of SPCC and SPCL depletion
	Mean	± SD	Mean	± SD
Test item	1.7%	±0.6%	0.1%	±0.2%	0.9%

Interpretation of results:

other: The test item gave a negative in DPRA and was classified in the “Negative: No or minimal reactivity class” using the Cysteine 1:10 / Lysine 1:50 prediction model. The result will be considered within a weight of evidence assessment for C&L purposes

Conclusions:

The test item gave a negative in DPRA and was classified in the “Negative: No or minimal reactivity class” using the Cysteine 1:10 / Lysine 1:50 prediction model. The result will be considered within a weight of evidence assessment for Classification and Labelling purposes

Executive summary:

The study was performed to the OECD TG 442C in chemico Direct peptide reactivity Assay (DPRA) guideline under GLP. The test item was assessed for reactivity to model synthetic peptides containing either cysteine (SPCC) or lysine (SPCL). After incubation of the test item with either SPCC or SPCL, the relative peptide concentration was determined by High-Performance Liquid Chromatography (HPLC) with gradient elution and photodiode array (PDA) detection at 220 nm and 258 nm. SPCC and SPCL Percent Depletion Values were calculated and used in a prediction model which allows assigning the test item to one of four reactivity classes used to support the discrimination between sensitizers and non-sensitizers. Acetonitrile (ACN) was found to be an appropriate solvent to dissolve the test item.Upon preparation of the SPCC test item samples, no precipitate or phase separation was observed in any of the samples. After incubation of the SPCC test item samples, a phase separation was observed. Upon preparation as well as after incubation of the SPCL test item samples, a phase separation was observed.Since phase separation was observed after the incubation period for SPCC and SPCL, it cannot be precise as to the amount of test item which remained in the solution to react with the peptide. In the cysteine reactivity assay the test item showed 1.7% SPCC depletion while in the lysine reactivity assay the test item showed 0.1% SPCL depletion. The mean of the SPCC and SPCL depletion was 0.9% and as a result the test item was considered to be negative in the DPRA and classified in the “negative or no or minimal reactivity class” when using the Cysteine 1:10 / Lysine 1:50 prediction model. All relevant test acceptability criteria were met with acceptable restrictions.

Reference 2

Endpoint:

skin sensitisation: in vitro

Type of information:

experimental study

Adequacy of study:

weight of evidence

Study period:

08-07-2019 to 26-08-2019

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

guideline study with acceptable restrictions

Remarks:

Guideline study performed under GLP. All relevant validity criteria were met. Test method considered to cover Key Event-2 under OECD 168 (2012) and OECD 255 (2016) and OECD 256 (2016).

Qualifier:

according to guideline

Guideline:

OECD Guideline 442D (In Vitro Skin Sensitisation: ARE-Nrf2 Luciferase Test Method)

Deviations:

no

GLP compliance:

yes

Remarks:

Guideline study performed under GLP. All relevant validity criteria were met. Test method considered to cover Key Event-2 under OECD 168 (2012) and OECD 255 (2016) and OECD 256 (2016).

Type of study:

activation of keratinocytes

Details on the study design:

Skin sensitisation (In vitro test system) - Details on study design:
- The study was conducted in accordance with the OECD TG 442D – In Vitro Skin Sensitisation: ARE-Nrf2 Luciferase Test Method and EURL ECVAM DB-ALM Protocol no. 155: KeratinoSens™, (Adopted March, 2018).
- The study protocol was validated with the proficiency chemicals prescribed in the OECD test guideline 442D. The results of the testing on the proficiency chemicals at the test facility is in the public domain (refer to study references provided in the full study report at the relevant test facility). All ten proficiency chemicals described in OECD TG 442C: Annex 1, were according to the test facility correctly predicted in a study conducted outside the present study. This information is in the public domain.
- A solubility test was performed. The test item was suspended in DMSO to a final concentration of 200 mM (glassy suspension). The stock was sonicated (Time: 18 min; Temperature: 18.0 – 27.0°C). The 100-fold dilution of the 200 mM DMSO stock formed a homogeneous solution (slight precipitation). The 100-fold dilution of the 100 mM DMSO stock in DMEM glutamax formed also a homogeneous solution (no precipitation). The 2000 μM (200 mM stock) concentration was selected as highest concentration for the main assay (highest dose required in the current guideline).
- Test Item preparation and concentrations: In the main experiments the test item was suspended in dimethyl sulfoxide (DMSO) at 200 mM (glassy) and dissolved in DMSO at 50 mM (colourless) in experiment 1 and 2, respectively. The stock was sonicated in experiment 1 only (Time: 13 min; Temperature: 25.0 – 31.0°C). From this stock 11 spike solutions in DMSO were prepared (2-fold and 1.5-fold dilution series in experiment 1 and 2, respectively). The stock and spike solution were diluted 25-fold with exposure medium. These solutions were diluted 4-fold with exposure medium in the assay resulting in final test concentrations of 2000, 1000, 500, 250, 125, 63, 31, 16, 7.8, 3.9, 2.0 and 0.98 μM and 5.8, 8.7, 13, 20, 29, 44, 66, 148, 222, 333 and 500 μM in experiment 1 and 2 respectively (final concentration DMSO of 1%). All concentrations of the test item were tested in triplicate. No precipitation was observed at the start and end of the incubation period in the 96-well plates.
- Positive Control preparation and concentrations: The positive control was Ethylene dimethacrylate glycol (EDMG), for which a 2-fold dilution series ranging from 0.78 to 25 mM were prepared in DMSO and diluted to the final concentration ranges of 7.8 to 250 µM (final concentration DMSO of 1%). All concentrations of the positive control were tested in triplicate. The PC formulation was used in other GLP studies concurrently to the present KERATINOSENS study.
- Acceptability criteria:
All acceptability criteria were met.
(i) The luciferase activity induction obtained with the positive control, Ethylene dimethacrylate glycol, should be above the threshold of 1.5 in at least one of the tested concentrations (from 7.8 to 250 µM). Actual PC: Experiment 1: 102 µM; Experiment 2: 33 µM.
(ii) The EC1.5 should be within two standard deviations of the historical mean. Moreover, the induction for Ethylene dimethacrylate glycol at 250 μM should be higher than 2-fold. Or if not achieved should give a satisfactory dose-response. The EC1.5 was 102 µM and 33 µM in experiment 1 and 2, respectively and the dose response in both experiments was greater than 2-fold (2.31-fold and 3.03-fold in experiment 1 and 2, respectively).
(iii) average coefficient of variation of the luminescence reading for the negative (solvent) control DMSO should be below 20% in each repetition: the % variability in solvent controls: Actual: Experiment 1: 4.5% ; Experiment 2: 9.0%.

Cell line used:
The KeratinoSens™ cell line is derived from the human keratinocyte culture HaCaT. It contains a stable insertion of a Luciferase gene under the control of the ARE-element of the gene. The cell line was developed by supplier (full details in the full study report). Upon receipt, cells are propagated (e.g. 2 to 4 passages) and stored frozen as a homogeneous stock. Cells from this original stock can be propagated up to a maximum passage number from the frozen stock (i.e. 25) and are employed for routine testing using the appropriate maintenance medium.

Cell culture and exposure:
Cells are grown for 24 h in 96-well plates. The maintainece medium was Dulbecco’s minimal (DMEM glutamax) supplemented with 9.1% (v/v) heat-inactivated (56°C; 30 min) fetal calf serum and geneticin (500 µg/mL). Cells were subcultured upon reaching 80-90% confluency. To maintain the integrity of the response, the cells were grown for more than one passage from the frozen stock and were not cultured for more than 25 passages from the frozen stock (P+25). One day prior to testing cells were harvested and distributed into 96-well plates (10,000 cells/well) in basic medium (Dulbecco’s minimal (DMEM glutamax) supplemented with 9.1% (v/v) heat-inactivated (56°C; 30 min) fetal calf serum). For each repetition, three replicates were used for the luciferase activity measurements, and one parallel replicate used for the MTT cell viability assay. The cells were incubated overnight in the incubator. The passage number used was P+13 in experiment 1 and P+3 in experiment 2. The medium was removed and replaced with fresh culture medium (150 μL culture medium containing serum but without Geneticin) to which 50 μL of the 25-fold diluted test item and control items were added. The exposure medium was Dulbecco’s minimal (DMEM glutamax) supplemented with 1% (v/v) heat-inactivated (56°C; 30 min) fetal calf serum. Three wells per plate were left empty (no cells and no treatment) to assess background values.

The following parameters are calculated in the KeratinoSens test method:
(i) The maximal average fold induction of luciferase activity (Imax) value observed at any concentration of the tested chemical and positive control
(ii) The EC1.5 value representing the concentration for which induction of luciferase activity is above the 1.5 fold threshold (i.e. 50% enhanced luciferase activity) was obtained
(iii) The IC50 and IC30 concentration values for 50% and 30% reduction of cellular viability.

- Cell viability assay MTT:
Test item IC50: IC50 value as the concentration in μM reducing the viability by 50%
Experiment 1: mean (n=3) 101 μM ; and Experiment 2: mean (n=3) 87 μM
Test item IC30: IC30 value as the concentration in μM reducing the viability by 30%
Experiment 1: mean (n=3) 74 μM ; and Experiment 2: mean (n=3) 61 μM

- Luciferase assay
Imax indicating maximum fold-induction up to concentration 2000 μM
Experiment 1: mean (n=3) 2.18 and Experiment 2: mean (n=3) 2.13

- Determinations:
EC1.5
Experiment 1: A dose related luminescence activity induction was observed after treatment with test item : EC1.5 = 8.8 μM
Experiment 2: A dose related luminescence activity induction was observed after treatment with test item : EC1.5 = 7.8 μM

Evaluation criteria
Test item considered ‘negative’ where
1. The Imax is higher than (>) 1.5 fold and statistically significantly different as compared to the vehicle (negative) control (as determined by a two-tailed, unpaired Student’s t-test)
2. The cellular viability is higher than (>) 70% at the lowest concentration with induction of luciferase activity above 1.5 fold (i.e. at the EC1.5 determining concentration)
3. The EC1.5 value is less than (<) 1000 μM (or < 200 µg/mL for test chemicals with no defined MW)
4. There is an apparent overall dose-response for luciferase induction
Negative results obtained with concentrations <1000 µM or 200 µg/mL and which do not reach cytotoxicity (< 70% viability) at the maximal tested concentration should be considered as inconclusive.

Results:
2 out of 2 positive experiments (each in triplicate). The test item showed toxicity with IC30 values of 74 μM and 61 μM and IC50 values of 101 μM and 87 μM in experiment 1 and 2, respectively. A biologically relevant, dose-related induction of the luciferase activity (EC1.5 values of 8.8 μM and 7.8 μM in experiment 1 and 2, respectively) was measured in both experiments. The maximum luciferase activity induction (Imax) was 2.18-fold and 2.13-fold in experiment 1 and 2 respectively. The cells were in these experiments incubated with the test item in a concentration range of 0.98 – 2000 µM (2-fold dilution steps) for 48 hours ± 1 h. The activation of the ARE-dependent pathway was assessed by measuring the luminescence induction compared to the vehicle control. In addition, the viability was assessed with an MTT assay. The test item is classified as positive in the KeratinoSens assay since positive results (>1.5-fold induction) were observed at test concentrations < 1000 µM (and/or with a cell viability of > 70% compared to the vehicle control).

Positive control results:

All acceptability criteria were met.
(i) The luciferase activity induction obtained with the positive control, Ethylene dimethacrylate glycol, should be above the threshold of 1.5 in at least one of the tested concentrations (from 7.8 to 250 µM). Actual PC: Experiment 1: 102 µM; Experiment 2: 33 µM.
(ii) The EC1.5 should be within two standard deviations of the historical mean. Moreover, the induction for Ethylene dimethacrylate glycol at 250 μM should be higher than 2-fold. Or if not achieved should give a satisfactory dose-response. The EC1.5 was 102 µM and 33 µM in experiment 1 and 2, respectively and the dose response in both experiments was greater than 2-fold in experiment 2 (2.31-fold and 3.03-fold in experiment 1 and 2, respectively).
(iii) average coefficient of variation of the luminescence reading for the negative (solvent) control DMSO should be below 20% in each repetition: the % variability in solvent controls: Actual: Experiment 1: 4.5% ; Experiment 2: 9.0%.

Historical results for luciferase induction by the positive control in the test laboratory: average and standard deviations from 341 valid runs are presented in the full study report.

Key result

Run / experiment:

other: mean Experiment 1 (n = 3)

Parameter:

other: EC1.5

Remarks:

/ µM

Value:

8.8

Vehicle controls validity:

valid

Negative controls validity:

not applicable

Positive controls validity:

valid

Remarks on result:

positive indication of skin sensitisation

Key result

Run / experiment:

other: mean Experiment 2 (n=3)

Parameter:

other: EC1.5

Remarks:

/ µM

Value:

7.8

Vehicle controls validity:

valid

Negative controls validity:

not applicable

Positive controls validity:

valid

Remarks on result:

positive indication of skin sensitisation

Other effects / acceptance of results:

OTHER EFFECTS:
- Visible damage on test system: None reported.

DEMONSTRATION OF TECHNICAL PROFICIENCY: The study protocol was validated with the proficiency chemicals prescribed in the OECD test guideline 442D. The results of the testing on the proficiency chemicals at the test facility is in the public domain (refer to study references provided in the full study report at the relevant test facility). All ten proficiency chemicals described in OECD TG 442C: Annex 1, were according to the test facility correctly predicted in a study conducted outside the present study. This information is in the public domain.

ACCEPTANCE OF RESULTS:
- Acceptance criteria met for negative control: All criteria met.
- Acceptance criteria met for positive control: All criteria met.
- Acceptance criteria met for variability between replicate measurements: All criteria met.
- Range of historical values if different from the ones specified in the test guideline: Not applicable.

- Acceptability criteria:
All acceptability criteria were met.
(i) The luciferase activity induction obtained with the positive control, Ethylene dimethacrylate glycol, should be above the threshold of 1.5 in at least one of the tested concentrations (from 7.8 to 250 µM). Actual PC: Experiment 1: 102 µM; Experiment 2: 33 µM.
(ii) The EC1.5 should be within two standard deviations of the historical mean. Moreover, the induction for Ethylene dimethacrylate glycol at 250 μM should be higher than 2-fold. Or if not achieved should give a satisfactory dose-response. The EC1.5 was 102 µM and 33 µM in experiment 1 and 2, respectively and the dose response in both experiments was greater than 2-fold in experiment 2 (2.31-fold and 3.03-fold in experiment 1 and 2, respectively).
(iii) average coefficient of variation of the luminescence reading for the negative (solvent) control DMSO should be below 20% in each repetition: the % variability in solvent controls: Actual: Experiment 1: 4.5% ; Experiment 2: 9.0%.

Historical results for luciferase induction by the positive control in the test laboratory: average and standard deviations from 341 valid runs are presented in the full study report.

Interpretation of results:

other: The test item gave 2 out of 2 positive experiments (each in triplicate). The result will be considered within a weight of evidence assessment for C&L purposes

Conclusions:

Under the condition of this study, the test item is considered to be sensitising to the skin. The test item gave 2 out of 2 positive experiments (each in triplicate) with >1.5-fold induction observed at test concentrations < 1000 µM (and/or with a cell viability of >70% compared to the vehicle control).

Executive summary:

The study was performed to the OECD TG 442D in vitro Skin Sensitisation guideline: ARE-Nrf2 Luciferase Test Method under GLP. The objective of this study was to evaluate the ability of the test item, to activate the antioxidant/electrophile responsive element (ARE)-dependent pathway in the KeratinoSens assay in two independent experiments. The test item was dissolved in dimethyl sulfoxide at 200 mM. From this stock 11 spike solutions in DMSO were prepared. The stock and spike solutions were diluted 100-fold in the assay resulting in test concentrations of 0.98 – 2000 μM (2-fold dilution series). The highest test concentration was the highest dose required in the current guideline. In experiment 2, a narrower dose-response analysis was performed using a lower dilution factor of 1.5-fold. This was to investigate further the induction observed at 16 μM in experiment 1.No precipitate was observed at any dose level tested. The luciferase activity induction obtained with the positive control, Ethylene dimethacrylate glycol, was statistically significant above the threshold of 1.5-fold in at least one concentration. The EC1.5 of the positive control was between two standard deviations of the historical mean (102 µM and 33 µM in experiment 1 and 2, respectively). A dose response was observed in both experiments and the induction at 250 µM was higher than 2-fold in experiment 2 (2.31-fold and 3.03-fold in experiment 1 and 2, respectively). The average coefficient of variation of the luminescence reading for the vehicle (negative) control DMSO was below 20% (4.5% and 9.0% in experiment 1 and 2, respectively). The test item showed toxicity with IC30 values of 74 μM and 61 μM and IC50 values of 101 μM and 87 μM in experiment 1 and 2, respectively.A biologically relevant, dose-related induction of the luciferase activity (EC1.5 values of 8.8 μM and 7.8 μM in experiment 1 and 2, respectively) was measured in both experiments. The maximum luciferase activity induction (Imax) was 2.18-fold and 2.13-fold in experiment 1 and 2 respectively. The cells were in these experiments incubated with the test item in a concentration range of 0.98 – 2000 µM (2-fold dilution steps) for 48 hours ± 1 h. The activation of the ARE-dependent pathway was assessed by measuring the luminescence induction compared to the vehicle control. In addition, the viability was assessed with an MTT assay. The test item is classified as positive in the KeratinoSens assay since positive results (>1.5-fold induction) were observed at test concentrations < 1000 µM and/or with a cell viability of > 70% compared to the vehicle control. All relevant test acceptability criteria were met.

Reference 3

Endpoint:

skin sensitisation: in vitro

Type of information:

experimental study

Adequacy of study:

weight of evidence

Study period:

21-01-2020 to 04-02-2020

Reliability:

2 (reliable with restrictions)

Rationale for reliability incl. deficiencies:

study well documented, meets generally accepted scientific principles, acceptable for assessment

Remarks:

Well reported non-GLP study using a test method validated by the EURL ECVAM [test method: TM2011-09 (EU)] and included in the OECD Test Guideline Program (TGP no. 4.106). All relevant validity criteria were met. Test method considered to cover Key Event-3 under OECD 168 (2012) and OECD 255 (2016) and OECD 256 (2016).

Qualifier:

according to guideline

Guideline:

other: OECD Test Guideline Program (TGP no. 4.106)

Version / remarks:

Test protocol in accordance with the test method included in OECD Test Guideline Program (TGP no. 4.106).

Deviations:

yes

Remarks:

minor deviation not considered to impact the validity of the study ; see "Principles of method if other than guideline" for further information

Qualifier:

according to guideline

Guideline:

other: EURL ECVAM test method: TM2011-09 (EU)

Version / remarks:

Test protocol in accordance with the test method included in EURL ECVAM validation

Deviations:

yes

Remarks:

minor deviation not considered to impact the validity of the study ; see "Principles of method if other than guideline" for further information

Qualifier:

according to guideline

Guideline:

other: H. Johansson et al. (2019), Validation of the GARD™skin Assay for Assessment of Chemical Skin Sensitisers: Ring Trial Results of Predictive Performance and Reproducibility, Toxicol. Sci. 170, 374-381

Version / remarks:

Test method as described in the aforementioned publication and related literature citations

Deviations:

yes

Remarks:

minor deviation not considered to impact the validity of the study ; see "Principles of method if other than guideline" for further information

Qualifier:

according to guideline

Guideline:

other: H. Johansson et al. (2013), The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27, 1163–1169

Version / remarks:

Test method as described in the aforementioned publication and related literature citations

Deviations:

no

Qualifier:

according to guideline

Guideline:

other: A. Forreryd et. al. (2016), From genome-wide arrays to tailor-made biomarker readout – Progress towards routine analysis of skin sensitizing chemicals with GARD, Toxicology in Vitro 37, 178–188

Version / remarks:

Test method as described in the aforementioned publication and related literature citations

Deviations:

yes

Remarks:

minor deviation not considered to impact the validity of the study ; see "Principles of method if other than guideline" for further information

Principles of method if other than guideline:

In accordance with REACH Regulation (EC) 1907/2006: Annex XI: section 1.1.2 and/or Annex XI: section 1.4: adequate and reliable (with restrictions) study information has been provided (not in accordance with GLP) but which can be considered equivalent to the relevant test method validated by the EURL ECVAM [test method: TM2011-09 (EU)] and included in the OECD Test Guideline Program (TGP no. 4.106). Study well documented. All relevant validity criteria were met. Test method considered to cover Key Event-3 under OECD 168 (2012) and OECD 255 (2016) and OECD 256 (2016).

A minor deviation from the Standard GARD Assay Protocol was made within the study. Instead of a 2-step dilution as described in the Standard GARD Assay Protocol, the DMF diluted test Items were directly diluted in the cell medium in order to assess the highest (potential) GARD assay "in-well concentration". The cell medium diluted test item was further transferred to the SenzaCells for stimulation according to the Standard GARD Assay Protocol. Another compatible solvent: DMF was utilised rather than DMSO or water standard solvents (further information on the justification behind this deviation is given below).

Prior to the test assay: the test item was assessed for solubility and cytotoxic effect in order to establish the GARD input concentration (concentration inducing 90% relative viability). The test Item showed solubility limitations in cell medium when using the Standard GARD Assay Protocol, therefore a 1-step dilution in DMF was used to increase the in-well concentration to the upper limit of the titration range. No cellular cytotoxicity was observed, therefore the GARD input concentration for the test item was determined to 500 µM. This minor deviation was not considered to impact the validity of the study. Applicant assessment indicates: this deviation was used to enhance the sensitivity of the test assay by maximising concentration/exposure of the test item to the SenzaCell : human myeloid leukemia-derived cell line by utilising the highest soluble concentration and/or 500 µM concentration, where possible in the absence of cytotoxicity and a determinable Rv90 (relative viability 90%). References: (1) H. Johansson et al., The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27 (2013) 1163–1169 ; (2) Test system manufacturer: [Senzagen AB, Sweden] published research (2019) indicates: DMSO (0.1%), water and/or Acetone, DMF, DMF/Glycerol, Ethanol, Glycerol, Isopropanol have been determined all compatible solvents with the test system, the latter of which are used increase its applicability domain (by increasing test item solubility) when testing difficult to dissolve test items: Applicant assessment indicates these solvent options are likely to be incorporated in future OECD TG publication, and/or have been demonstrated by the manufacturer to be compatible with the test system.

GLP compliance:

no

Remarks:

See "Principles of method if other than guideline" for further information ; study well documented, meets generally accepted scientific principles, acceptable for assessment ; test method EURL ECVAM validated

Type of study:

activation of dendritic cells

Details on the study design:

Skin sensitisation (In vitro test system) - Details on study design:
- The study was conducted in accordance with the GARDSkin assay protocol ; non-exhaustive references include: OECD Test Guideline Program (TGP no. 4.106), and which has been EURL ECVAM validated according to test method: TM2011-09 (EU), and with reference to the non-exhaustive related publications: (1) H. Johansson et al. (2019), Validation of the GARD™skin Assay for Assessment of Chemical Skin Sensitisers: Ring Trial Results of Predictive Performance and Reproducibility, Toxicol. Sci. 170, 374-381 ; and/or : (2) H. Johansson et al. (2013), The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27, 1163–1169. The test assay is pending full OECD acceptance. The laboratory where the study was conducted (details given in the full study report) achieved GLP certification: 26 May 2020 (inspected 06 to 15 April 2020). Information in the public domain.
- The study protocol was previously validated (in-house laboratory and/or external validations) with reference chemicals described in the published literature. The results of the testing on the reference chemicals at the test facility is in the public domain (refer to study references provided in the full study report at the relevant test facility). Additional public information on the in-house validation can be found in: H. Johansson et al. (2014), Genomic allergen rapid detection in-house validation--a proof of concept, Toxicol. Sci. ; 139 (2) : 362-70 and/or H. Johansson et al. (2013), The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27, 1163–1169 ; and/or for external validation within: H. Johansson et al., (2019), Validation of the GARD™skin Assay for Assessment of Chemical Skin Sensitisers: Ring Trial Results of Predictive Performance and Reproducibility, Toxicol. Sci. 170, 374-381. Additional information is in the public domain on the EURL ECVAM and OECD websites.
- Details on and maintainece of the test cell line: The human myeloid leukemia-derived cell line SenzaCell (available through ATCC), acting as an in vitro model of human Dendritic Cell (DC), is maintained in a-MEM supplemented with 20% (volume/volume) fetal calf serum and 40 ng/ml recombinant human Granulocyte Macrophage Colony Stimulating Factor (rhGM-CSF). A media change during expansion is performed every 3-4 days. Working stocks of cultures are grown for a maximum of 16 passages or two months after thawing. For chemical stimulation of cells, exposed cells are incubated for 24 h at 37°C, 5% C02 and 95% humidity.
- A solubility test was performed. Prior to the test assay: the test item was assessed for solubility and cytotoxic effect in order to establish the GARD input concentration (concentration inducing 90% relative viability). The test Item showed solubility limitations in cell medium when using the Standard GARD Assay Protocol, therefore a 1-step dilution in DMF was used to increase the in-well concentration to the upper limit of the titration range. No cellular cytotoxicity was observed, therefore the “GARD input concentration” for the test item was determined to 500 µM. This minor deviation was not considered to impact the validity of the study. Applicant assessment indicates: this deviation was used to enhance the sensitivity of the test assay by maximising concentration/exposure of the test item to the SenzaCell : human myeloid leukemia-derived cell line by utilising the highest soluble concentration and/or 500 µM concentration, where possible in the absence of cytotoxicity and a determinable Rv90 (relative viability 90%). Reference: The GARD assay for assessment of chemical skin sensitizers, H. Johansson et al., Toxicology in Vitro 27 (2013) 1163–1169. 500 μM concentration is the highest concentration for the main assay (highest dose required in the protocol and literature).
- Positive Control preparation and concentrations: The positive control p-phenvlendiamine (PPD) [CAS 106-50-3] of purity ≥ 99% (source and batch information in the full study report), for which an in-well concentration of 75 µM in DMSO was utilised. Positive controls were concurrently tested in a manner similar to test item described below.
- Negative Control preparation and concentrations: The negative control Dimethylformamide (DMF) [CAS 68-12-2] of purity 99.9% (source and batch information in the full study report), for which an in-well concentration of 0.10% was utilised. Negative controls were concurrently tested in a manner similar to test item described below.
- Test Item preparation and concentrations: test item was dissolved in DMF. When test items are poorly dissolved in cell medium the maximum soluble
concentration is assessed. The assayed test item was titrated to concentrations ranging from 1 µM to the maximum soluble concentration in cell media. For freely soluble test item, 500 µM is set as the upper limit of the titration range. For test item dissolved in DMF, the in-well concentration of DMF was 0.1%. Cytotoxic effects of test item were monitored. After incubation for 24 h at 37°C, 5% C02 and 95% humidity, harvested cells were stained with the viability marker Propidium Iodide (PI) and analysed by flow cytometry. PI-negative cells were defined as viable, and the relative viability of cells stimulated with each concentration in the titration range and the Rv90 (relative viability 90%) calculated. For toxic test items, the concentration yielding 90% relative viability (Rv90) is used in the GARDSkin assay, the reason being that this concentration demonstrates bioavailability of the test item used for stimulation, while not impairing immunological responses. For non-toxic test Items, a concentration of 500 µM is used where possible. For insoluble test Items at 500 µM in cell media, the highest soluble concentration is used. Whichever is higher. This is termed the “GARD input concentration”. References: (1) H. Johansson et al., The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27 (2013) 1163–1169 ; (2) Test system manufacturer: [Senzagen AB, Sweden] published research (2019) indicates: DMSO (0.1%), water and/or Acetone, DMF, DMF/Glycerol, Ethanol, Glycerol, Isopropanol have been determined all compatible solvents with the test system, the latter of which are used increase its applicability domain (by increasing test item solubility) when testing difficult to dissolve test items: Applicant assessment indicates these solvent options are likely to be incorporated in future OECD TG publication, and/or have been demonstrated by the manufacturer to be compatible with the test system.
- Main Stimulation: In then main experiment(s) the cells are stimulated (as previously described) using the GARD input concentration. In addition to the assayed test Item a set of concurrent positive and negative controls are performed as reference and quality controls. Test item and controls were assayed in biological replicates (n=3), performed at different time-points and using different cell cultures. After incubation for 24 h at 37°C, 5% C02 and 95% humidity, cell culture was lysed in TRizol reagent and stored at -20°C until RNA was extracted. In parallel, stimulated cells were PI stained and analysed by flow cytometry to verify the expected relative viability. RNA isolation from lysed cells was performed using commercially available kits (referenced in the full study report). Total RNA was quantified, and quality controlled using BioAnalyzer equipment.
- Endpoint measurement:
(i) Gene expression analysis using Nanostring nCounter System : A total of 100 ng of RNA was used as sample input in a hybridization assay with the GPS (GARD Prediction Signature) specific Reporter CodeSet. The hybridized sample was prepared on chip using nCounter Preparation Station and individual transcripts of the GPS were quantified using Digital Analyzer. Data was imported into relevant software environment for statistical computation analysis (documented in the full study report). Raw data was normalised using a gene specific Counts Per Total Counts (CPTC) algorithm.
(ii) Prediction model and evaluation: For assessment of skin sensitisation, the Support Vector Machine (SVM) was modelled on a training data set corresponding to samples used for assay development. For a comprehensive overview of the training data set and methods, this is referenced in the full study report. Batch variations between the training data set and the test data set were eliminated using the Batch Adjustment by Reference Alignment (BARA) method, using unstimulated cells as a reference control.
Evaluation criteria
Each sample in the test set were assigned a Support Vector Machine (SVM) Decision Value (DV), based on its transcriptional levels of the GPS biomarker signature.
Test item considered ‘negative’ where: mean DV < 0 (n ≥ 2) which results in no classification as a skin sensitiser (or ‘non-sensitising’).
Test item considered ‘positive’ where: mean DV ≥ 0 (n ≥ 2) which results in classification as a skin sensitiser.
- Other: No precipitation was reported observed at any point during the study including the start and end of the incubation period in solutions and wells.
- Acceptability criteria:
All applicable acceptability criteria were met.
(1) Positive Control: p-phenvlendiamine (PPD) [CAS 106-50-3] gave a positive prediction for sensitisation (passed GARDSkin Assay Quality Control)
(2) Negative Control: Dimethylformamide (DMF) [CAS 68-12-2 gave a negative prediction for sensitisation (passed GARDSkin Assay Quality Control)

Results:
The mean SMV DV was -0.59, the GARDSkin Prediction was non-sensitising. All relevant concurrent positive and negative controls passed GARDSkin Assay Quality Control criteria. All relevant validity criteria were met.

Positive control results:

Positive Control: p-phenvlendiamine (PPD) [CAS 106-50-3] gave a positive prediction for sensitisation (PC SMV Decision Value > 0) and passed GARDSkin Assay Quality Control

Key result

Run / experiment:

other: mean (n=3)

Parameter:

other: GARDSkin SMV Decision Value

Value:

-0.59

Vehicle controls validity:

valid

Negative controls validity:

valid

Remarks:

See vehicle controls (served as negative control)

Positive controls validity:

valid

Remarks on result:

no indication of skin sensitisation

Other effects / acceptance of results:

OTHER EFFECTS:
- Visible damage on test system: None reported. No (cyto)toxicity was observed during testing with the test item to the cell line.

DEMONSTRATION OF TECHNICAL PROFICIENCY: The study was conducted in accordance with the GARDSkin assay protocol ; non-exhaustive references include: OECD Test Guideline Program (TGP no. 4.106), and which has been EURL ECVAM validated according to test method: TM2011-09 (EU), and with reference to the non-exhaustive related publications: (1) H. Johansson et al. (2019), Validation of the GARD™skin Assay for Assessment of Chemical Skin Sensitisers: Ring Trial Results of Predictive Performance and Reproducibility, Toxicol. Sci. 170, 374-381 ; and/or : (2) H. Johansson et al. (2013), The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27, 1163–1169. The test assay is pending full OECD acceptance. The laboratory where the study was conducted (details given in the full study report) achieved GLP certification: 26 May 2020 (inspected 06 to 15 April 2020). Information in the public domain. The study protocol was previously validated (in-house laboratory and/or external validations) with reference chemicals described in the published literature. The results of the testing on the reference chemicals at the test facility is in the public domain (refer to study references provided in the full study report at the relevant test facility). Additional public information on the in-house validation can be found in: H. Johansson et al. (2014), Genomic allergen rapid detection in-house validation--a proof of concept, Toxicol. Sci. ; 139 (2) : 362-70 and/or H. Johansson et al. (2013), The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27, 1163–1169 ; and/or for external validation within: H. Johansson et al., (2019), Validation of the GARD™skin Assay for Assessment of Chemical Skin Sensitisers: Ring Trial Results of Predictive Performance and Reproducibility, Toxicol. Sci. 170, 374-381. Additional information is in the public domain on the EURL ECVAM and OECD websites.

ACCEPTANCE OF RESULTS:
- Acceptance criteria met for negative control: All criteria met.
- Acceptance criteria met for positive control: All criteria met.
- Acceptance criteria met for variability between replicate measurements: No issues were reported in the full study report.
- Range of historical values if different from the ones specified in the test guideline: Not applicable.

- Acceptability criteria:
All applicable acceptability criteria were met.
(1) Positive Control: p-phenvlendiamine (PPD) [CAS 106-50-3] gave a positive prediction for sensitisation (passed GARDSkin Assay Quality Control)
(2) Negative Control: Dimethylformamide (DMF) [CAS 68-12-2 gave a negative prediction for sensitisation (passed GARDSkin Assay Quality Control)
No precipitation was reported observed at any point during the study including the start and end of the incubation period in solutions and wells.

Interpretation of results:

other: The test item gave a negative result based on the mean SMV DC < 0 (test conducted in triplicate). The result will be considered within a weight of evidence assessment for C&L purposes.

Conclusions:

Under the condition of this study, the test item is considered to be non-sensitising to the skin. The test item gave a mean Support Vector Machine (SVM) Decision Value (DV) of -0.59. The GARDSkin Prediction was non-sensitising since the mean SMV DV < 0 (n ≥ 2). All relevant concurrent positive and negative controls passed GARDSkin Assay Quality Control criteria.

Executive summary:

The study was performed to the GARDSkin (Genomic Allergen Rapid Detection) test method which has been EURL ECVAM validated [test method: TM2011-09 (EU)] and is included in the OECD Test Guideline Program (TGP no. 4.106), pending full OECD TG adoption.The test assay is detailed in the following non-exhaustive related publications: (1) H. Johansson et al. (2019), Validation of the GARD™skin Assay for Assessment of Chemical Skin Sensitisers: Ring Trial Results of Predictive Performance and Reproducibility, Toxicol. Sci. 170, 374-381 ; and/or : (2) H. Johansson et al. (2013), The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27, 1163–1169 ; and/or (3) A. Forreryd et. al. (2016), From genome-wide arrays to tailor-made biomarker readout – Progress towards routine analysis of skin sensitizing chemicals with GARD, Toxicology in Vitro 37, 178–188. The GARD platform is a versatile testing strategy that relies on gene expression analysis of cell cultures exposed to test items in vitro. The high-dimensional readout allows for gene expression analysis of customized biomarker signatures, each specific for various biological end points. In addition, the final prediction gives rise to the classification of test items based on a statistical prediction model that is in turn dependent on an end point specific training data set as to if a substance is skin sensitising or non-skin sensitising.The human myeloid leukemia-derived cell line SenzaCell (available through ATCC), acting as an in vitro model of human Dendritic Cell (DC), is maintained in a-MEM supplemented with 20% (volume/volume) fetal calf serum and 40 ng/ml recombinant human Granulocyte Macrophage Colony Stimulating Factor (rhGM-CSF). A media change during expansion is performed every 3-4 days. Working stocks of cultures are grown for a maximum of 16 passages or two months after thawing. For chemical stimulation of cells, exposed cells are incubated for 24 h at 37°C, 5% CO2 and 95% humidity.Prior to the assay: the test item was assessed for solubility and cytotoxic effect in order to establish the GARD input concentration (concentration inducing 90% relative viability).When test items are poorly dissolved in cell medium the maximum soluble concentration is assessed.The test Item showed solubility limitations in cell medium when using the Standard GARD Assay Protocol, therefore a 1-step dilution in DMF was used to increase the in-well concentration to the upper limit of the titration range.The assayed test item was titrated to concentrations ranging from 1µM to the maximum soluble concentration in cell media and/or 500µM set as the upper limit of the titration range. For test item dissolved in DMF, the in-well concentration of DMF was 0.1%. Cytotoxic effects of test item were monitored. After incubation for 24 h at 37°C, 5% CO2 and 95% humidity, harvested cells were stained with the viability marker Propidium Iodide (PI) and analysed by flow cytometry. PI-negative cells were defined as viable, and the relative viability of cells stimulated with each concentration in the titration range and the Rv90 (relative viability 90%) calculated. The test item indicated no cellular cytotoxicity, therefore the GARD input concentration was determined to 500 µM for the test item.The concurrent positive control was p-phenvlendiamine (PPD) [CAS 106-50-3], for which an in-well concentration of 75µM in DMSO was utilised.The concurrent negative control was dimethylformamide (DMF) [CAS 68-12-2], for which an in-well concentration of 0.10%was utilised.Under the condition of this study, the test item is considered to be non-sensitising to the skin. The test item gave a meanSupport Vector Machine (SVM)Decision Value (DV) of -0.59. The GARDSkin Prediction was non-sensitising since themean SMV DV<0 (n ≥ 2). All relevant concurrent positive and negative controls passed GARDSkin Assay Quality Control criteria.All relevant validity criteria were met.

Reference 4

Endpoint:

skin sensitisation, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2019

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

Remarks:

)SAR predictions from DEREK NEXUS v6.0.1 [DEREK NEXUS – skin sensitization]. QMRF and/or QPRF provided.

Justification for type of information:

1. SOFTWARE
DEREK NEXUS – skin sensitization; within DEREK NEXUS v6.0.1
Contact : LHASA Limited,
Granary Warf House,
2 Canal Wharf,
Leeds,
LS11 5PS,
United Kingdom
URL: www.lhasalimited.org
More information available at:
https://www.lhasalimited.org/products/derek-nexus.htm
and
https://www.lhasalimited.org/products/skin-sensitisation-assessment-using-derek-nexus.htm

2. MODEL (incl. version number)
DEREK NEXUS – skin sensitization; within DEREK NEXUS v6.0.1
26 July 2010 (model development and/or QMRF publication) ; 01 December 2017 (current model version)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
See attached: ‘DEREK Prediction on Skin Sensitization of [test item name]’ (report number); dated 08 August 2019
See attached QPRF: ‘DEREK NEXUS – skin sensitization; within DEREK NEXUS v6.0.1 : Prediction on Skin Sensitisation of [test item name]’ version 1.0; dated 18 July 2019.
QPRF is in appendix to report 08 August 2019

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
Full details of the method are provided in the attached QMRF named ‘DEREK NEXUS – skin sensitization’ for DEREK NEXUS v6.0.1 model ; version 1.9; 01 December 2017

5. APPLICABILITY DOMAIN
See ‘any other information on results incl. tables’.
See attached: ‘DEREK Prediction on Skin Sensitization of [test item name]’ (report number); dated 08 August 2019
See attached QPRF: ‘DEREK NEXUS – skin sensitization; within DEREK NEXUS v6.0.1 : Prediction on Skin Sensitisation of [test item name]’ version 1.0; dated 18 July 2019.
QPRF is in appendix to report 08 August 2019
- Descriptor/Parameter domain: All constituents are in domain. The scope of the structure-activity relationships describing the skin sensitisation endpoint are defined by model developer to be the applicability domain for the model. Therefore, if a chemical activates an alert describing a structure-activity for skin sensitisation it can considered to be within the applicability domain of the model. The applicability of potency predictions may be judged, and modified, by the user based on the displayed data for nearest neighbours (in the training set). If a chemical does not activate an “alert” or “reasoning rule” then DEREK makes a negative prediction. The applicability of the negative prediction to the query chemicals can be determined by an expert, if required ,by investigating the presence (or absence) of misclassified and/or unclassified features of the query chemical.
- Structural and mechanistic domains: The substructure of the substance falls within the structural fragment domain for alert 425 Enol Ether. The structural fragment of the substance falls within the mechanistic domain of the molecules in the training set on which the alert is based.
- Conclusion: The potential mechanism for this alert is thought to occur via metabolism or oxidation of the molecule resulting in the formation of an electrophilic species capable of Schiff base reactions with nucleophilic groups. All similar chemicals contain an enol ether. These chemicals fall within the scope of alert 425. The query chemical contains an enol ether of which the double bond is substituted with other carbons and one oxygen. The closest structurally related analogues only have a structural similarity of 12 – 19%. The 3 structural analogues used for calculating the EC3 are 2 non-sensitisers and 1 weak sensitiser. DEREK NEXUS predicted an LLNA EC3 of 18% (weak sensitiser) for the query chemical based on LLNA data from three structurally related analogues (structural similarity of 12 to 19%). Considering the overall structural resemblance is low the accuracy of the EC3 prediction is uncertain i.e. EQUIVOCAL (there is an equal weight of evidence for and against the proposition).

6. ADEQUACY OF THE RESULT
1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Guideline:

other: REACH Guidance on QSARs R.6, May/July 2008

Principles of method if other than guideline:

Full details of the method are provided in the attached QMRF named ‘DEREK NEXUS – skin sensitization’ for software package: DEREK NEXUS v6.0.1 ; version 1.9; 01 December 2017
- The model applies the following methodology to generate predictions:
(i) Structural alerts
(ii) Logic of argumentation
(iii) Nearest neighbours (with same alert as query chemical) based on Tanimoto similarity
(iv) Feature-based database search
The following descriptors exist in the model: (1) Markush structures encoding activating and deactivating features (known as “patterns” in the DEREK NEXUS knowledge-base and/or (2) 2D structural fingerprints
There is an a priori assumption that patterns, physico-chemical properties and associated reasoning will or can be used to model toxicity within DEREK NEXUS. Further relevant toxicology experts identified that (1) predictions of potency (LLNA, EC3) could be made by using nearest neighbours within/to the same structural alert ; (2) misclassified and unclassified features were useful descriptors for determining the reliability of the predictions for non-alerting chemicals.
The experimental values in the training set and validation set were measured using one or more methods equivalent or similar to the following guidelines:
- Guinea pig data, such as the Buehler and Maximisation tests eq. or similar to OECD TG 406
- Local Lymph Node Assay (LLNA) eq. or similar to OECD TG 429
- Human Repeat Insult Patch Test (HRIPT).
Secondary data sources of toxicity data such as: (i) BgVV categories and (ii) R43 DSD classification. Additionally. alert authors consider both mechanistic evidence and chemical properties (such as reactivity).
- The model and the training sets are collated and published by LHASA Limited, Granary Warf House, 2 Canal Wharf, Leeds, LS11 5PS, United Kingdom
The training set of the model currently consists of more than 650 chemicals (proprietary)
Alert authors design the patterns to describe the activating and deactivating features found during expert assessment of the alert training set. Structural fingerprints are generated from the. 469 compounds in the local lymph node data set. Misclassified and unclassified features generated by processing a large Skin Sensitisation reference set {comprising 1346 sensitisers and 1420 non-sensitisers) against DEREK NEXUS (v6.0) fragments.
A full list of experimental reference citations is provided in the DEREK NEXUS software with additional reference citations in the QMRF attached.
DEREK NEXUS is a proprietary, rule-based expert system for the prediction of toxicity. The model knowledge base is composed of alerts, examples and reasoning-rules which may each contribute to the predictions made by the software system. Each alert in DEREK NEXUS describes a chemical substructure believed to be responsible for inducing a specific toxicological outcome (often referred to as a toxicophore). Alerts are derived by experts, using toxicological data and information regarding the biological mechanism of action . Where relevant, metabolism data may be incorporated into all alert, enabling the prediction of chemicals which are not directly toxic but are metabolised to an active species. The derivation of each alert is described in the alert comments along with supporting references and example chemicals where possible. By reporting this information to the user, DEREK NEXUS provides highly transparent predictions. The use of structural alerts for the prediction of toxicity is both widely understood and the subject of many publications.
Details could be found in:
(1) D.M. Sanderson & C.G Earnshaw CG (1991), Computer prediction of possible toxic action from chemical structure, The DEREK system. Human and Experimental Toxicology 10, 261-273.
(2) P.N. Judson, C.A. Marchant & J.D. Vessey (2003), Using argumentation for absolute reasoning about the potential toxicity of chemicals. Journal of Chemical Information
and Computer Sciences 43, 1364-1310.
(3) C.A. Marchant, K.A. Briggs & A. Long (2003), In silico tools for sharing data and knowledge on toxicity and metabolism: DEREK Windows, Meteor , and Vitic. Toxicology Mechanisms and Methods 18, 177-187.
(4) P.N. Judson, S.A. Stalford & J. Vessey (2013), Assessing confidence in predictions made by knowledge-based systems. Toxicology Research 2, 70-79.
- Justification of QSAR prediction: The result should be considered in relation to corresponding information presented and in accordance with the tonnage driven information requirements of REACH Regulation (EC) 1907/2006 in a weight of evidence. For classification purposes, Regulation (EC) No 1272/2008 allows the use of non-standard approaches as mentioned in 4.1.1.2.2. Annex XI, 1.3 allows the use of a (Q)SAR instead of testing when certain conditions are met (use of valid (Q)SAR, adequate and reliable information is provided etc). The present predictions are intended to be used in a weight of evidence approach as mentioned in Annex IX, 1.2: using the present predictions to support experimental data (GLP and/or non-GLP ; completed to an applicable guidelines and/or using methods which are scientifically validated and/or from reports that are sufficiently documented).

Specific details on test material used for the study:

Detailed information on the 'test material identity' is provided in the attached QSAR Prediction Reporting Format (QPRF) document including information on individual constituents.

Key result

Parameter:

EC3

Remarks:

%

Value:

18

Test group / Remarks:

falls within the structural fragment domain for alert 425 Enol Ether ; result considered EQUIVOCAL

Remarks on result:

positive indication of skin sensitisation based on QSAR/QSPR prediction

Remarks:

DEREK NEXUS predicted an LLNA EC3 of 18% (weak sensitiser) for the query chemical based on LLNA data from three structurally related analogues (structural similarity of 12 to 19%). Considering the overall structural resemblance is low the accuracy of the EC3 prediction is uncertain i.e. EQUIVOCAL (there is an equal weight of evidence for and against the proposition).

1. Defined Endpoint:

QMRF 4. Human health effects

QMRF 4.6 Skin sensitisation

 

Reference to type of model used and description of results:

DEREK NEXUS – skin sensitization, 01 December 2017 (current model version)

Platform version: DEREK NEXUS v6.0.1

 

2. Description of results and assessment of reliability of the prediction:

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: SENSITISER – Alert: 425 Enol Ether

PARENT SMILES.2: SENSITISER – Alert: 425 Enol Ether

DEREK EC3 model 1.2.0 – Predicted EC3 = 18% (weak sensitiser) (n=3 structurally related analogues (structural similarity of 12 to 19%). Considering the overall structural resemblance is low the accuracy of the EC3 prediction is uncertain i.e. EQUIVOCAL (there is an equal weight of evidence for and against the proposition).

 

Full details of the constituents are provided in the attached : ‘DEREK Prediction on Skin Sensitization of [test item name]’ (report number); dated 08 August 2019

See attached QPRF: ‘DEREK NEXUS – skin sensitization; within DEREK NEXUS v6.0.1 : Prediction on Skin Sensitisation of [test item name]’ version 1.0; dated 18 July 2019.

QPRF is in appendix to report 08 August 2019

Inputs Used for Test Item: All modelling was based on input of SMILES notation (or SMILES that is representative for each constituent) of the test item

Assessment of the substance within the applicability domain as documented within the corresponding QMRF named ‘DEREK NEXUS – skin sensitization’ for DEREK NEXUS v6.0.1 model ; version 1.9; 01 December 2017 – section 5; indicates:

That for all constituents:

- Descriptor/Parameter domain: All constituents are in domain. The scope of the structure-activity relationships describing the skin sensitisation endpoint are defined by model developer to be the applicability domain for the model. Therefore, if a chemical activates an alert describing a structure-activity for skin sensitisation it can considered to be within the applicability domain of the model. The applicability of potency predictions may be judged, and modified, by the user based on the displayed data for nearest neighbours (in the training set). If a chemical does not activate an “alert” or “reasoning rule” then DEREK makes a negative prediction. The applicability of the negative prediction to the query chemicals can be determined by an expert, if required, by investigating the presence (or absence) of

misclassified and/or unclassified features of the query chemical.

- Structural and mechanistic domains: The substructure of the substance falls within the structural fragment domain for alert 425 Enol Ether. The structural fragment of the substance falls within the mechanistic domain of the molecules in the training set on which the alert is based.

- Conclusion: The potential mechanism for this alert is thought to occur via metabolism or oxidation of the molecule resulting in the formation of an electrophilic species capable of Schiff base reactions with nucleophilic groups. All similar chemicals contain an enol ether. These chemicals fall within the scope of alert 425. The query chemical contains an enol ether of which the double bond is substituted with other carbons and one oxygen. The closest structurally related analogues only have a structural similarity of 12 – 19%. The 3 structural analogues used for calculating the EC3 are 2 non-sensitisers and 1 weak sensitiser. DEREK NEXUS predicted an LLNA EC3 of 18% (weak sensitiser) for the query chemical based on LLNA data from three structurally related analogues (structural similarity of 12 to 19%). Considering the overall structural resemblance is low the accuracy of the EC3 prediction is uncertain i.e. EQUIVOCAL (there is an equal weight of evidence for and against the proposition). The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

 

3. Uncertainty of the prediction and mechanistic domain:

It is noted that the software model itself determines if the test item falls within the general domain requirements ; then the appropriate structural and mechanistic/metabolic domains. Full details are provided of the methodology in the corresponding QMRF for model ‘DEREK NEXUS – skin sensitization’ for DEREK NEXUS v6.0.1 model ; version 1.9; 01 December 2017. Interested parties are invited to contact the model author.

The training set is embedded in the software of the model; refer to the QMRF which is available in ‘DEREK NEXUS – skin sensitization’ for DEREK NEXUS v6.0.1. The training set data is proprietary and not made publicly available by the model developers. DEREK NEXUS predictive performance against a combined human dataset had an accuracy of 76% (cited in the attached QMRF & QPRF). External validation is carried out on each knowledge base release. The data sets used are in the public domain, however curated versions used by the model developer are proprietary and not available. Three published datasets have been used for alert validation: (1) M.T. Cronin & D.A. Basketter et al. (1994), Multivariate QSAR analysis of a Skin Sensitization database. SAR and QSAR in Environmental Research 2, 159-179. (2) G.F. Gerberick et al. (2005), Compilation of historical local lymph node data for evaluation of skin sensitization : alternative methods. Dermatitis 16, 157-202 ; and/or related: (3) P.S. Kern et al. (2010), Local lymph node data for the evaluation of skin sensitization alternatives : a second compilation. Dermatitis 21, 8-10. (4) a collection of local node assay data for 137 chemicals published in Contact Dermatitis which have been extracted from Vitic Nexus (date: 13 September 2012). Secondary to this a further assessment of the relationship between likelihood-levels and prediction accuracy has been assessed (Judson et al. 2013) and several external evaluations have been published (Rorije et al., 2013 and/or Nukada et al., 2013).

 

Uncertainty in the prediction relates to:

Any available statistical characteristics are provided (i) for the prediction: in the attached QPRF and (ii) for the model: in the attached QMRF, respectively. Noting that the mode structural alerts are knowledge-based rather than statistically based. All similar chemicals contain an enol ether. These compounds fall within the scope of alert 425. The 3 structural analogues used for calculating the EC3 are 2 non-sensitizers and 1 weak sensitizer. The query chemical contains an enol ether of which the double bond is substituted with other carbons and one oxygen. DEREK NEXUS predicted an LLNA EC3 of 18% (weak sensitiser) for the query chemical based on LLNA data from three structurally related analogues (structural similarity of 12 to 19%). Considering the overall structural resemblance is low the accuracy of the EC3 prediction is uncertain i.e. EQUIVOCAL (there is an equal weight of evidence for and against the proposition). No further comments on the uncertainty of the prediction are provided by the author of the present QPRF. A summary of provided information is presented by the applicant in attachment with relevant citations. It is noted that the (Q)SAR prediction will not be used as a standalone prediction. The prediction is fit for regulatory purpose when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes

Interpretation of results:

other: The test item gave an EQUIVOCAL prediction for skin sensitisation : alert: 425 Enol Ether. Predicted EC3 = 18% (weak sensitiser) (structural similarity of 12 to 19%). The result will be considered within a weight of evidence assessment for C&L purposes.

Conclusions:

The results are adequate for the regulatory purpose.

Executive summary:

DEREK NEXUS v6.0.1, DEREK NEXUS – skin sensitization

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: SENSITISER – Alert: 425 Enol Ether

PARENT SMILES.2: SENSITISER – Alert: 425 Enol Ether

DEREK EC3 model 1.2.0 – Predicted EC3 = 18% (weak sensitiser) (n=3 structurally related analogues (structural similarity of 12 to 19%). Considering the overall structural resemblance is low the accuracy of the EC3 prediction is uncertain i.e. EQUIVOCAL (there is an equal weight of evidence for and against the proposition).

 

Adequacy of the QSAR:

1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Reference 5

Endpoint:

skin sensitisation, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2020

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

Remarks:

(Q)SAR predictions from OASIS TIMES v2.30.1.11 using the submodels: (i) Skin sensitization GHS v.01.02 ; (ii) Skin sensitization DST (Dermal Sensitization Threshold) v.01.02 and/or (iii) Skin sensitization with autoxidation v.24.29. QMRF and/or QPRF for each submodel are provided individually.

Justification for type of information:

1. SOFTWARE
Skin sensitization GHS v.01.02; within OASIS TIMES v2.30.1.11
Contact LMC University:
Prof. As. Zlatarov,
LMC University,
Laboratory of Mathematical Chemistry,
Bourgas,
Bulgaria
URL: www.oasis-lmc.org
More information available at:
http://oasis-lmc.org/products/models/human-health-endpoints/skin-sensitization.aspx

2. MODEL (incl. version number)
OASIS TIMES v2.30.1 model v.01.02, Skin sensitization GHS
March 2017 (model development and/or QMRF publication) ; January 2020 (current model version)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
See attached: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (GHS v.01.02)’ version 1.0; dated 09 September 2020.

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
Full details of the method are provided in the attached QMRF named ‘Skin sensitization GHS model’ for OASIS TIMES v2.30.1 model v.01.02, Skin sensitization GHS ; version 1.4; 01 January 2020

5. APPLICABILITY DOMAIN
See ‘any other information on results incl. tables’.
See attached: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (GHS v.01.02) ’ version 1.0; dated 09 September 2020.
- Descriptor/Parameter domain: All constituents are in domain.
- Structural and mechanistic domains: Correct fragments = 62%, Incorrect fragments = 0%, Unknown fragments = 38%, Total Structural domain : (Out of domain - 2 chemicals). All constituents are within the mechanistic domain.

6. ADEQUACY OF THE RESULT
1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Guideline:

other: REACH Guidance on QSARs R.6, May/July 2008

Principles of method if other than guideline:

Full details of the method are provided in the attached QMRF named ‘Skin sensitization GHS model’ - model: Skin sensitization GHS v.01.02 - software package: OASIS TIMES v2.30.1, version 1.4; date: 01 March 2017; updated 01 January 2020.
- The model applies the following methodology to generate predictions:
The LMC stepwise approach was used to define the applicability domain. It contains three layers:
(i) General properties requirements (log KOW, MW, log WS)
(ii) Structural domain (Atom Cantered Fragments (ACFs))
(iii) Metabolic domain
Then assessment of interpolation space - estimates the population density of the parametric space defined by the explanatory variables of the QSAR models by making use the training set chemicals
Details could be found in:
Dimitrov S, Dimitrova G., Pavlov T., Dimitrova D., Patlevisz G., Niemela J., Mekenyan O., A stepwise approach for defining the applicability domain of SAR and QSAR models, J. Chem. Inf. Model., 45, 839-849 (2005)
- The model and the training sets are collated and published by Prof. As. Zlatarov, LMC University, (Bulgaria).
The training set of the model currently consists of 517 chemicals (proprietary)
The experimental values in the training set and validation set were measured using one or more methods equivalent or similar to the following guidelines:
- Local Lymph Node Assay (LLNA) eq. or similar to OECD TG 429
- Human Repeat Insult Patch Test (HRIPT).
- Other appropriately documented data, (collated by the model authors).
All available skin sensitization data (EC3,% and NOEL values in μg/cm2) from both tests were stored in the training set. In case of multiple data the worst case scenario (or expert judgement) is applied. Most potent data (EC3, NOEL) are selected to represent the individual training set chemical(s).
A full list of experimental reference citations is provided in the OASIS TIMES software with additional reference citations in the QMRF attached.
- Justification of QSAR prediction: The result should be considered in relation to corresponding information presented and in accordance with the tonnage driven information requirements of REACH Regulation (EC) 1907/2006 in a weight of evidence. For classification purposes, Regulation (EC) No 1272/2008 allows the use of non-standard approaches as mentioned in 4.1.1.2.2. Annex XI, 1.3 allows the use of a (Q)SAR instead of testing when certain conditions are met (use of valid (Q)SAR, adequate and reliable information is provided etc). The present predictions are intended to be used in a weight of evidence approach as mentioned in Annex IX, 1.2: using the present predictions to support experimental data (GLP and/or non-GLP ; completed to an applicable guidelines and/or using methods which are scientifically validated and/or from reports that are sufficiently documented).

Specific details on test material used for the study:

Detailed information on the 'test material identity' is provided in the attached QSAR Prediction Reporting Format (QPRF) document including information on individual constituents.

Key result

Remarks on result:

no indication of skin sensitisation based on QSAR/QSPR prediction

1. Defined Endpoint:

QMRF 4. Human health effects

QMRF 4.6 Skin sensitisation

 

Reference to type of model used and description of results:

Skin sensitization GHS v.01.02, January 2020 (current model version)

Platform version: OASIS TIMES v2.30.1.11

 

2. Description of results and assessment of reliability of the prediction:

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: NON-SENSITISER

PARENT SMILES.2: NON-SENSITISER

Full details of the constituents are provided in the attached QPRF: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (GHS v.01.02)’ version 1.0; dated 09 September 2020.

Inputs Used for Test Item: All modelling was based on input of SMILES notation for each constituent of the test item

Assessment of the substance within the applicability domain as documented within the corresponding QMRF named ‘Skin sensitization GHS model’ for OASIS TIMES v2.30.1 model v.01.02, Skin sensitization GHS ; version 1.4; 01 January 2020 – section 5; indicates:

That for all constituents:

Descriptor/Parameter domain (e.g. Log Kow and MW) : All constituents are in domain.

Structural and mechanistic domains: (e,g, ACF) Correct fragments = 62%, Incorrect fragments = 0%, Unknown fragments = 38%, Total Structural domain : (Out of domain - 2 chemicals). All constituents are within the mechanistic domain by assessment of the interpolation space – estimation of the population density of the parametric space defined by the explanatory variables of the QSAR models by making use the training set chemicals

Details could be found in:

Dimitrov S, Dimitrova G., Pavlov T., Dimitrova D., Patlevisz G., Niemela J., Mekenyan O., A stepwise approach for defining the applicability domain of SAR and QSAR models, J. Chem. Inf. Model., 45, 839-849 (2005)

The substance constituents are predominantly within the model structural domain. The relatively high percentage of correctly predicted atom centred fragments (ca. 60%) allows that the conclusion that the predictions are relevant. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

 

3. Uncertainty of the prediction and mechanistic domain:

The training set is embedded in the software of the model; refer to the QMRF which is available in OASIS TIMES v.2.30.1.11. The training set data is proprietary and not made publicly available by the model developers. It is noted that the software model itself determines if the test item falls within the general properties requirements [lipophilicity (log KOW), molecular weight (MW) and/or water solubility (WS)] domain; then the appropriate structural and mechanistic/metabolic domains. Full details are provided of the methodology in the corresponding QMRF for model OASIS TIMES v.2.30.1.11, Skin sensitization GHS v.01.02, January 2020. Interested parties are invited to contact the model author.

Uncertainty in the prediction relates to:

Statistical characteristics are provided (i) for the prediction: in the attached QPRF and (ii) for the model: in the attached QMRF, respectively. No further comments on the uncertainty of the prediction are provided by the author of the present QPRF. A summary of provided information is presented by the applicant in attachment with relevant citations. It is noted that the (Q)SAR prediction will not be used as a standalone prediction. The prediction is fit for regulatory purpose when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Interpretation of results:

other: The test item gave a negative prediction for skin sensitisation. The result will be considered within a weight of evidence assessment for C&L purposes

Conclusions:

The results are adequate for the regulatory purpose.

Executive summary:

OASIS TIMES v2.30.1, Skin sensitization GHS v.01.02

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: NON-SENSITISER

PARENT SMILES.2: NON-SENSITISER

Conclusion: The substance is not expected to be a skin sensitiser.

 

Adequacy of the QSAR:

1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Reference 6

Endpoint:

skin sensitisation, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2020

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

Remarks:

(Q)SAR predictions from OASIS TIMES v2.30.1.11 using the submodels: (i) Skin sensitization GHS v.01.02 ; (ii) Skin sensitization DST (Dermal Sensitization Threshold) v.01.02 and/or (iii) Skin sensitization with autoxidation v.24.29. QMRF and/or QPRF for each submodel are provided individually. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Justification for type of information:

1. SOFTWARE
Skin sensitization with autoxidation v.24.29; within OASIS TIMES v2.30.1.11
Contact LMC University:
Prof. As. Zlatarov,
LMC University,
Laboratory of Mathematical Chemistry,
Bourgas,
Bulgaria
URL: www.oasis-lmc.org
More information available at:
http://oasis-lmc.org/products/models/human-health-endpoints/skin-sensitization.aspx

2. MODEL (incl. version number)
OASIS TIMES v2.30.1 model v.24.29, Skin sensitization with autoxidation
November 2013 (model development and/or QMRF publication) ; January 2020 (current model version)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
See attached: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (Autoxidation v.24.29)’ version 1.0; dated 09 September 2020.

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
Full details of the method are provided in the attached QMRF named ‘TIMES (TIssue MEtabolism Simulator) model for Skin sensitization (TIMES-SS model)’ for OASIS TIMES v2.30.1 model v.24.29, Skin sensitization with autoxidation ; version 1.9; 01 January 2020

5. APPLICABILITY DOMAIN
See ‘any other information on results incl. tables’.
See attached: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (Autoxidation v.24.29)’ version 1.0; dated 09 September 2020.
- Descriptor/Parameter domain: All constituents are in domain.
- Structural and mechanistic domains: Correct fragments = 62%, Incorrect fragments = 0%, Unknown fragments = 38%, Total Structural domain : (Out of domain - 2 chemicals). All constituents are within the mechanistic domain.

6. ADEQUACY OF THE RESULT
1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Guideline:

other: REACH Guidance on QSARs R.6, May/July 2008

Principles of method if other than guideline:

Full details of the method are provided in the attached QMRF named ‘TIMES (TIssue MEtabolism Simulator) model for Skin sensitization (TIMES-SS model)’ - model: Skin sensitization with autoxidation v.24.29 - software package: OASIS TIMES v2.30.1, version 1.9; date: 01 November 2013 ; updated 01 January 2020.
- The model applies the following methodology to generate predictions:
The LMC stepwise approach was used to define the applicability domain. It contains three layers:
(i) General properties requirements (log KOW, MW, log WS)
(ii) Structural domain (Atom Cantered Fragments (ACFs))
(iii) Metabolic domain
Then assessment of interpolation space - estimates the population density of the parametric space defined by the explanatory variables of the QSAR models by making use the training set chemicals
Details could be found in:
Dimitrov S, Dimitrova G., Pavlov T., Dimitrova D., Patlevisz G., Niemela J., Mekenyan O., A stepwise approach for defining the applicability domain of SAR and QSAR models, J. Chem. Inf. Model., 45, 839-849 (2005)
- The model and the training sets are collated and published by Prof. As. Zlatarov, LMC University, (Bulgaria).
The training set of the model currently consists of 886 chemicals (proprietary)
The experimental values in the training set and validation set were measured using one or more methods equivalent or similar to the following guidelines:
- Local Lymph Node Assay (LLNA) eq. or similar to OECD TG 429
- GPMT/Buehler tests eq. or similar to OECD TG 406.
- Other appropriately documented data, (collated by the model authors) to the above.
A unifying scale was derived evaluating the correlation and concordance of those chemicals that exist in two datasets to the above methods. The TIMES Skin sensitization model integrates a simulator of skin metabolism together with a list of alerts for protein binding built based on the training set of the 886 chemicals. Mechanistic rational of each alert is provided by experts based on significant reference support from the literature. A full list of experimental reference citations is provided in the OASIS TIMES software with additional reference citations in the QMRF attached.
- Justification of QSAR prediction: The result should be considered in relation to corresponding information presented and in accordance with the tonnage driven information requirements of REACH Regulation (EC) 1907/2006 in a weight of evidence. For classification purposes, Regulation (EC) No 1272/2008 allows the use of non-standard approaches as mentioned in 4.1.1.2.2. Annex XI, 1.3 allows the use of a (Q)SAR instead of testing when certain conditions are met (use of valid (Q)SAR, adequate and reliable information is provided etc). The present predictions are intended to be used in a weight of evidence approach as mentioned in Annex IX, 1.2: using the present predictions to support experimental data (GLP and/or non-GLP ; completed to an applicable guidelines and/or using methods which are scientifically validated and/or from reports that are sufficiently documented).

Specific details on test material used for the study:

Detailed information on the 'test material identity' is provided in the attached QSAR Prediction Reporting Format (QPRF) document including information on individual constituents.

Key result

Remarks on result:

no indication of skin sensitisation based on QSAR/QSPR prediction

1. Defined Endpoint:

QMRF 4. Human health effects

QMRF 4.6 Skin sensitisation

 

Reference to type of model used and description of results:

Skin sensitization with autoxidation v.24.29, January 2020 (current model version)

Platform version: OASIS TIMES v2.30.1.11

 

2. Description of results and assessment of reliability of the prediction:

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: NON-SENSITISER

PARENT SMILES.2: NON-SENSITISER

Full details of the constituents are provided in the attached QPRF: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (Autoxidation v.24.29)’ version 1.0; dated 09 September 2020.

Inputs Used for Test Item: All modelling was based on input of SMILES notation for each constituent of the test item

Assessment of the substance within the applicability domain as documented within the corresponding QMRF named ‘TIMES (TIssue MEtabolism Simulator) model for Skin sensitization (TIMES-SS model)’ for OASIS TIMES v2.30.1 model v.24.29, Skin sensitization with autoxidation ; version 1.9 ; 01 January 2020 – section 5; indicates:

That for all constituents:

Descriptor/Parameter domain (e.g. Log Kow and MW) : All constituents are in domain.

Structural and mechanistic domains: (e,g, ACF) Correct fragments = 62%, Incorrect fragments = 0%, Unknown fragments = 38%, Total Structural domain : (Out of domain - 2 chemicals). All constituents are within the mechanistic domain by assessment of the interpolation space – estimation of the population density of the parametric space defined by the explanatory variables of the QSAR models by making use the training set chemicals

Details could be found in:

Dimitrov S, Dimitrova G., Pavlov T., Dimitrova D., Patlevisz G., Niemela J., Mekenyan O., A stepwise approach for defining the applicability domain of SAR and QSAR models, J. Chem. Inf. Model., 45, 839-849 (2005)

The substance constituents are predominantly within the model structural domain. The relatively high percentage of correctly predicted atom centred fragments (ca. 60%) allows that the conclusion that the predictions are relevant. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

 

3. Uncertainty of the prediction and mechanistic domain:

An external validation set is available and publicly available. Reference: Patlewicz, G. et al. (2007), TIMES-SS – A promising tool for the assessment of skin sensitization hazard. A characterization with respect to the OECD validation principles for (Q)SARs and an external evaluation for predictivity”, Regulatory Toxicology and Pharmacology 48 : 225-239. Interested parties are invited to contact the model authors. Full details on the training set, internal and external validations, as applicable are included in the attached QMRF named ‘TIMES (TIssue MEtabolism Simulator) model for Skin sensitization (TIMES-SS model) for OASIS TIMES v2.30.1 model v.24.29, Skin sensitization with autoxidation; version 1.9; 01 January 2020. The training set is embedded in the software of the model; refer to the QMRF which is available in OASIS TIMES v.2.30.1.11. The training set data is proprietary and not made publicly available by the model developers. It is noted that the software model itself determines if the test item falls within the general properties requirements [lipophilicity (log KOW), molecular weight (MW) and/or water solubility (WS)] domain; then the appropriate structural and mechanistic/metabolic domains. Full details are provided of the methodology in the corresponding QMRF for model OASIS TIMES v.2.30.1.11, Skin sensitization with autoxidation; version 1.9; 01 January 2020.

Uncertainty in the prediction relates to:

Statistical characteristics are provided (i) for the prediction: in the attached QPRF and (ii) for the model: in the attached QMRF, respectively. A summary of provided information is presented by the applicant in attachment with relevant citations. It is noted that the (Q)SAR prediction will not be used as a standalone prediction. The prediction is fit for regulatory purpose when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Interpretation of results:

other: The test item gave a negative prediction for skin sensitisation. The result will be considered within a weight of evidence assessment for C&L purposes

Conclusions:

The results are adequate for the regulatory purpose.

Executive summary:

OASIS TIMES v2.30.1, Skin sensitization with autoxidation v.24.29

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: NON-SENSITISER

PARENT SMILES.2: NON-SENSITISER

Conclusion: The substance is not expected to be a skin sensitiser.

 

Adequacy of the QSAR:

1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Reference 7

Endpoint:

skin sensitisation, other

Type of information:

(Q)SAR

Adequacy of study:

weight of evidence

Study period:

2020

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

Remarks:

(Q)SAR predictions from OASIS TIMES v2.30.1.11 using the submodels: (i) Skin sensitization GHS v.01.02 ; (ii) Skin sensitization DST (Dermal Sensitization Threshold) v.01.02 and/or (iii) Skin sensitization with autoxidation v.24.29. QMRF and/or QPRF for each submodel are provided individually. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Justification for type of information:

1. SOFTWARE
Skin sensitization DST (Dermal Sensitization Threshold) v.01.02; within OASIS TIMES v2.30.1.11
Contact LMC University:
Prof. As. Zlatarov,
LMC University,
Laboratory of Mathematical Chemistry,
Bourgas,
Bulgaria
URL: www.oasis-lmc.org
More information available at:
http://oasis-lmc.org/products/models/human-health-endpoints/skin-sensitization.aspx

2. MODEL (incl. version number)
OASIS TIMES v2.30.1 model v.01.02, Skin sensitization DST (Dermal Sensitization Threshold)
September 2016 (model development and/or QMRF publication) ; January 2020 (current model version)

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL
See attached: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (DST v.01.02)’ version 1.0; dated 09 September 2020.

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
Full details of the method are provided in the attached QMRF named ‘Skin sensitization DST (Dermal Sensitization Threshold) model’ for OASIS TIMES v2.30.1 model v.01.02, Skin sensitization DST ; version 1.4; 01 January 2020

5. APPLICABILITY DOMAIN
See ‘any other information on results incl. tables’.
See attached: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (DST v.01.02)’ version 1.0; dated 09 September 2020.
- Descriptor/Parameter domain: All constituents are in domain.
- Structural and mechanistic domains: Correct fragments = 62%, Incorrect fragments = 0%, Unknown fragments = 38%, Total Structural domain : (Out of domain - 2 chemicals). All constituents are within the mechanistic domain.

6. ADEQUACY OF THE RESULT
1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Guideline:

other: REACH Guidance on QSARs R.6, May/July 2008

Principles of method if other than guideline:

Full details of the method are provided in the attached QMRF named ‘Skin sensitization DST (Dermal Sensitization Threshold) model’ - model: Skin sensitization DST (Dermal Sensitization Threshold) v.01.02 - software package: OASIS TIMES v2.30.1, version 1.4; date: 01 September 2016 ; updated 01 January 2020.
- The model applies the following methodology to generate predictions:
The LMC stepwise approach was used to define the applicability domain. It contains three layers:
(i) General properties requirements (log KOW, MW, log WS)
(ii) Structural domain (Atom Cantered Fragments (ACFs))
(iii) Metabolic domain
Then assessment of interpolation space - estimates the population density of the parametric space defined by the explanatory variables of the QSAR models by making use the training set chemicals
Details could be found in:
Dimitrov S, Dimitrova G., Pavlov T., Dimitrova D., Patlevisz G., Niemela J., Mekenyan O., A stepwise approach for defining the applicability domain of SAR and QSAR models, J. Chem. Inf. Model., 45, 839-849 (2005)
- The model and the training sets are collated and published by Prof. As. Zlatarov, LMC University, (Bulgaria).
The training set of the model currently consists of 557 chemicals (proprietary)
The experimental values in the training set and validation set were measured using one or more methods equivalent or similar to the following guidelines:
- Local Lymph Node Assay (LLNA) eq. or similar to OECD TG 429
- Human Repeat Insult Patch Test (HRIPT).
- Other appropriately documented data, (collated by the model authors).
All available skin sensitization data (EC3,% and NOEL values in μg/cm2) from both tests were stored in the training set. In case of multiple data the worst case scenario (or expert judgement) is applied. Most potent data (EC3, NOEL) are selected to represent the individual training set chemical(s).
A full list of experimental reference citations is provided in the OASIS TIMES software with additional reference citations in the QMRF attached.
The model can be used to discriminate HPC (High Potency Category) chemicals from non-HPC chemicals with respect to dermal sensitization threshold accounting for (a)biotic activation of chemicals.
- Justification of QSAR prediction: The result should be considered in relation to corresponding information presented and in accordance with the tonnage driven information requirements of REACH Regulation (EC) 1907/2006 in a weight of evidence. For classification purposes, Regulation (EC) No 1272/2008 allows the use of non-standard approaches as mentioned in 4.1.1.2.2. Annex XI, 1.3 allows the use of a (Q)SAR instead of testing when certain conditions are met (use of valid (Q)SAR, adequate and reliable information is provided etc). The present predictions are intended to be used in a weight of evidence approach as mentioned in Annex IX, 1.2: using the present predictions to support experimental data (GLP and/or non-GLP ; completed to an applicable guidelines and/or using methods which are scientifically validated and/or from reports that are sufficiently documented).

Specific details on test material used for the study:

Detailed information on the 'test material identity' is provided in the attached QSAR Prediction Reporting Format (QPRF) document including information on individual constituents.

Key result

Remarks on result:

positive indication of skin sensitisation based on QSAR/QSPR prediction

Remarks:

The test item gave a "Weak/Non sensitiser" prediction for skin sensitisation. The result will be considered within a weight of evidence assessment for C&L purposes

1. Defined Endpoint:

QMRF 4. Human health effects

QMRF 4.6 Skin sensitisation

 

Reference to type of model used and description of results:

Skin sensitization DST (Dermal Sensitization Threshold) v.01.02, January 2020 (current model version)

Platform version: OASIS TIMES v2.30.1.11

 

2. Description of results and assessment of reliability of the prediction:

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: WEAK/NON SENSITISER

PARENT SMILES.2: WEAK/NON SENSITISER

Conclusion: The substance has the potential to be a non-sensitiser OR a weak skin sensitiser.

Precautionary conclusion based on weight of evidence. POSITIVE : weak skin sensitiser.

 

Full details of the constituents are provided in the attached QPRF: ‘OASIS TIMES v.2.30.1.11 Prediction on Skin Sensitisation of [test item name] (DST v.01.02)’ version 1.0; dated 09 September 2020.

Inputs Used for Test Item: All modelling was based on input of SMILES notation for each constituent of the test item

Assessment of the substance within the applicability domain as documented within the corresponding QMRF named Skin sensitization DST (Dermal Sensitization Threshold) model’ for OASIS TIMES v2.30.1 model v.01.02, Skin sensitization DST ; version 1.4; 01 January 2020 – section 5; indicates:

 

That for all constituents:

Descriptor/Parameter domain (e.g. Log Kow and MW) : All constituents are in domain.

Structural and mechanistic domains: (e,g, ACF) Correct fragments = 62%, Incorrect fragments = 0%, Unknown fragments = 38%, Total Structural domain : (Out of domain - 2 chemicals). All constituents are within the mechanistic domain by assessment of the interpolation space – estimation of the population density of the parametric space defined by the explanatory variables of the QSAR models by making use the training set chemicals

Details could be found in:

Dimitrov S, Dimitrova G., Pavlov T., Dimitrova D., Patlevisz G., Niemela J., Mekenyan O., A stepwise approach for defining the applicability domain of SAR and QSAR models, J. Chem. Inf. Model., 45, 839-849 (2005)

The substance constituents are predominantly within the model structural domain. The relatively high percentage of correctly predicted atom centred fragments (ca. 60%) allows that the conclusion that the predictions are relevant. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

 

3. Uncertainty of the prediction and mechanistic domain:

The training set is embedded in the software of the model; refer to the QMRF which is available in OASIS TIMES v.2.30.1.11. The training set data is proprietary and not made publicly available by the model developers. It is noted that the software model itself determines if the test item falls within the general properties requirements [lipophilicity (log KOW), molecular weight (MW) and/or water solubility (WS)] domain; then the appropriate structural and mechanistic/metabolic domains. Full details are provided of the methodology in the corresponding QMRF for model OASIS TIMES v.2.30.1.11, Skin sensitization DST (Dermal Sensitization Threshold) v.01.02, January 2020. Interested parties are invited to contact the model author.

Uncertainty in the prediction relates to:

Statistical characteristics are provided (i) for the prediction: in the attached QPRF and (ii) for the model: in the attached QMRF, respectively. No further comments on the uncertainty of the prediction are provided by the author of the present QPRF. A summary of provided information is presented by the applicant in attachment with relevant citations. It is noted that the (Q)SAR prediction will not be used as a standalone prediction. The prediction is fit for regulatory purpose when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Interpretation of results:

other: The test item gave a "Weak/Non sensitiser" prediction for skin sensitisation. The result will be considered within a weight of evidence assessment for C&L purposes

Conclusions:

The results are adequate for the regulatory purpose.

Executive summary:

OASIS TIMES v2.30.1, Skin sensitization DST (Dermal Sensitization Threshold) v.01.02

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: WEAK/NON SENSITISER

PARENT SMILES.2: WEAK/NON SENSITISER

Conclusion: The substance has the potential to be a non-sensitiser OR a weak skin sensitiser. Precautionary conclusion based on weight of evidence. POSITIVE : weak skin sensitiser.

 

Adequacy of the QSAR:

1) QSAR model is scientifically valid. 2) The substance falls within the (general) applicability domain of the QSAR model and/or with additional interpretation by expert judgement. 3) The prediction is fit for regulatory purpose. Specifically when combined with further information such as physico-chemistry and in chemico and in vitro testing and further modelling. The result will be considered within a weight of evidence assessment for C&L and risk assessment purposes.

Endpoint conclusion

Endpoint conclusion:

adverse effect observed (sensitising)

Additional information:

Skin Sensitisation:

1. Key Study – Molecular initiating Key Event 1: DPRA, OECD TG 442C, 2019 : The study was performed to the OECD TG 442C in chemico Direct peptide reactivity Assay (DPRA) guideline under GLP. The test item was assessed for reactivity to model synthetic peptides containing either cysteine (SPCC) or lysine (SPCL). After incubation of the test item with either SPCC or SPCL, the relative peptide concentration was determined by High-Performance Liquid Chromatography (HPLC) with gradient elution and photodiode array (PDA) detection at 220 nm and 258 nm. SPCC and SPCL Percent Depletion Values were calculated and used in a prediction model which allows assigning the test item to one of four reactivity classes used to support the discrimination between sensitizers and non-sensitizers. Acetonitrile (ACN) was found to be an appropriate solvent to dissolve the test item. Upon preparation of the SPCC test item samples, no precipitate or phase separation was observed in any of the samples. After incubation of the SPCC test item samples, a phase separation was observed. Upon preparation as well as after incubation of the SPCL test item samples, a phase separation was observed. Since phase separation was observed after the incubation period for SPCC and SPCL, it cannot be precise as to the amount of test item which remained in the solution to react with the peptide. In the cysteine reactivity assay the test item showed 1.7% SPCC depletion while in the lysine reactivity assay the test item showed 0.1% SPCL depletion. The mean of the SPCC and SPCL depletion was 0.9% and as a result the test item was considered to be negative in the DPRA and classified in the “negative or no or minimal reactivity class” when using the Cysteine 1:10 / Lysine 1:50 prediction model. All relevant test acceptability criteria were met with acceptable restrictions.

 

2. Key Study – Molecular initiating Key Event 2: KeratinoSens, OECD TG 442D, 2019: The study was performed to the OECD TG 442D in vitro Skin Sensitisation guideline: ARE-Nrf2 Luciferase Test Method under GLP. The objective of this study was to evaluate the ability of the test item, to activate the antioxidant/electrophile responsive element (ARE)-dependent pathway in the KeratinoSens assay in two independent experiments. The test item was dissolved in dimethyl sulfoxide at 200 mM. From this stock 11 spike solutions in DMSO were prepared. The stock and spike solutions were diluted 100-fold in the assay resulting in test concentrations of 0.98 – 2000 μM (2-fold dilution series). The highest test concentration was the highest dose required in the current guideline. In experiment 2, a narrower dose-response analysis was performed using a lower dilution factor of 1.5-fold. This was to investigate further the induction observed at 16 μM in experiment 1.No precipitate was observed at any dose level tested. The luciferase activity induction obtained with the positive control, Ethylene dimethacrylate glycol, was statistically significant above the threshold of 1.5-fold in at least one concentration. The EC1.5 of the positive control was between two standard deviations of the historical mean (102 µM and 33 µM in experiment 1 and 2, respectively). A dose response was observed in both experiments and the induction at 250 µM was higher than 2-fold in experiment 2 (2.31-fold and 3.03-fold in experiment 1 and 2, respectively). The average coefficient of variation of the luminescence reading for the vehicle (negative) control DMSO was below 20% (4.5% and 9.0% in experiment 1 and 2, respectively). The test item showed toxicity with IC30 values of 74 μM and 61 μM and IC50 values of 101 μM and 87 μM in experiment 1 and 2, respectively. A biologically relevant, dose-related induction of the luciferase activity (EC1.5 values of 8.8 μM and 7.8 μM in experiment 1 and 2, respectively) was measured in both experiments. The maximum luciferase activity induction (Imax) was 2.18-fold and 2.13-fold in experiment 1 and 2 respectively. The cells were in these experiments incubated with the test item in a concentration range of 0.98 – 2000 µM (2-fold dilution steps) for 48 hours ± 1 h. The activation of the ARE-dependent pathway was assessed by measuring the luminescence induction compared to the vehicle control. In addition, the viability was assessed with an MTT assay. The test item is classified as positive in the KeratinoSens assay since positive results (>1.5-fold induction) were observed at test concentrations < 1000 µM and/or with a cell viability of > 70% compared to the vehicle control. All relevant test acceptability criteria were met.

 

3. Key study – Molecular initiating Key Event 3: GARDSkin Assay, 2020: The study was performed to the GARDSkin (Genomic Allergen Rapid Detection) test method which has been EURL ECVAM validated[test method: TM2011-09 (EU)] and is included in the OECD Test Guideline Program (TGP no. 4.106), pending full OECD TG adoption.The test assay is detailed in the following non-exhaustive related publications: (1) H. Johansson et al. (2019), Validation of the GARD™skin Assay for Assessment of Chemical Skin Sensitisers: Ring Trial Results of Predictive Performance and Reproducibility, Toxicol. Sci. 170, 374-381 ; and/or : (2) H. Johansson et al. (2013), The GARD assay for assessment of chemical skin sensitizers, Toxicology in Vitro 27, 1163–1169 ; and/or (3) A. Forreryd et. al. (2016), From genome-wide arrays to tailor-made biomarker readout – Progress towards routine analysis of skin sensitizing chemicals with GARD, Toxicology in Vitro 37, 178–188. The GARD platform is a versatile testing strategy that relies on gene expression analysis of cell cultures exposed to test items in vitro. The high-dimensional readout allows for gene expression analysis of customized biomarker signatures, each specific for various biological end points. In addition, the final prediction gives rise to the classification of test items based on a statistical prediction model that is in turn dependent on an end point specific training data set as to if a substance is skin sensitising or non-skin sensitising. The human myeloid leukemia-derived cell line SenzaCell (available through ATCC), acting as an in vitro model of human Dendritic Cell (DC), is maintained in a-MEM supplemented with 20% (volume/volume) fetal calf serum and 40 ng/ml recombinant human Granulocyte Macrophage Colony Stimulating Factor (rhGM-CSF). A media change during expansion is performed every 3-4 days. Working stocks of cultures are grown for a maximum of 16 passages or two months after thawing. For chemical stimulation of cells, exposed cells are incubated for 24 h at 37°C, 5% CO2 and 95% humidity. Prior to the assay: the test item was assessed for solubility and cytotoxic effect in order to establish the GARD input concentration (concentration inducing 90% relative viability).When test items are poorly dissolved in cell medium the maximum soluble concentration is assessed. The test Item showed solubility limitations in cell medium when using the Standard GARD Assay Protocol, therefore a 1-step dilution in DMF was used to increase the in-well concentration to the upper limit of the titration range.The assayed test item was titrated to concentrations ranging from 1µM to the maximum soluble concentration in cell media and/or 500µM set as the upper limit of the titration range. For test item dissolved in DMF, the in-well concentration of DMF was 0.1%. Cytotoxic effects of test item were monitored. After incubation for 24 h at 37°C, 5% CO2 and 95% humidity, harvested cells were stained with the viability marker Propidium Iodide (PI) and analysed by flow cytometry. PI-negative cells were defined as viable, and the relative viability of cells stimulated with each concentration in the titration range and the Rv90 (relative viability 90%) calculated. The test item indicated no cellular cytotoxicity, therefore the GARD input concentration was determined to 500 µM for the test item. The concurrent positive control was p-phenvlendiamine (PPD) [CAS 106-50-3], for which an in-well concentration of 75µM in DMSO was utilised. The concurrent negative control was dimethylformamide (DMF) [CAS 68-12-2], for which an in-well concentration of 0.10%was utilised. Under the condition of this study, the test item is considered to be non-sensitising to the skin. The test item gave a mean Support Vector Machine (SVM)Decision Value (DV) of -0.59. The GARDSkin Prediction was non-sensitising since the mean SMV DV<0 (n ≥ 2). All relevant concurrent positive and negative controls passed GARDSkin Assay Quality Control criteria. All relevant validity criteria were met.

 

4. QSAR : DEREK NEXUS – skin sensitization, DEREK NEXUS v6.0.1, 2019 :

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: SENSITISER – Alert: 425 Enol Ether

PARENT SMILES.2: SENSITISER – Alert: 425 Enol Ether

DEREK EC3 model 1.2.0 – Predicted EC3 = 18% (weak sensitiser) (n=3 structurally related analogues (structural similarity of 12 to 19%). Considering the overall structural resemblance is low the accuracy of the EC3 prediction is uncertain i.e. EQUIVOCAL (there is an equal weight of evidence for and against the proposition).

 

5. QSAR : Skin sensitization GHS v.01.02, OASIS TIMES v2.30.1, 2020 :

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: NON-SENSITISER

PARENT SMILES.2: NON-SENSITISER

Conclusion: The substance is not expected to be a skin sensitiser.

(ACF: 62% correct, 0% incorrect, 38% unknown)

 

6. QSAR : Skin sensitization DST (Dermal Sensitization Threshold) v.01.02, OASIS TIMES v2.30.1, 2020 :

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: WEAK/NON SENSITISER

PARENT SMILES.2: WEAK/NON SENSITISER

Conclusion: The substance has the potential to be a non-sensitiser OR a weak skin sensitiser.

Precautionary conclusion based on weight of evidence. POSITIVE : weak skin sensitiser.

(ACF: 62% correct, 0% incorrect, 38% unknown)

 

7. QSAR : Skin sensitization with autoxidation v.24.29, OASIS TIMES v2.30.1, 2020 :

Results:

The parent chemical is a mixture or multi-constituent substance

PARENT SMILES.1: NON-SENSITISER

PARENT SMILES.2: NON-SENSITISER

Conclusion: The substance is not expected to be a skin sensitiser.

(ACF: 62% correct, 0% incorrect, 38% unknown)

 

Weight of Evidence Conclusion:

The applicant assesses by expert judgement the available information. Within the battery of in silico modelling and in vitro test assays, there are equivocal to positive predictions for sensitisation. The test item appears to be either a non-sensitiser to weak sensitiser. Using a precautionary principle conclusion through evaluation of all the available information, the substance is considered a weak sensitiser and to have a low potency (e.g. EC3 >> 2%) based on the weight of evidence. The substance is assigned to GHS Classification: Skin Sensitisation Category 1B. Further information concerning the weight of evidence conclusion is attached by the applicant to this endpoint.

Respiratory sensitisation

Endpoint conclusion

Endpoint conclusion:

no study available

Justification for classification or non-classification

The substance meets classification criteria under Regulation (EC) No 1272/2008 for skin sensitisation: category 1B.

 

Within the battery of in silico modelling and in vitro test assays, there is are equivocal to positive predictions for sensitisation. The test item appears to be either a non-sensitiser to weak sensitiser. Using a precautionary principle conclusion through evaluation of all the available information, the substance is considered a weak sensitiser and to have a low potency (e.g. EC3 >> 2%) based on the weight of evidence. The substance is assigned to GHS Classification: Skin Sensitisation Category 1B.

 

Further information concerning the weight of evidence conclusion is attached by the applicant to this endpoint.

References:

(1) ECHA Guidance on Information Requirements and Chemical Safety Assessment (Chapter R.7a: Endpoint Specific Guidance, July 2017)

(2) OECD (2017) Guidance Document on the Reporting of Defined Approaches and Individual Information Sources to be Used within Integrated Approaches to Testing and Assessment (IATA) for Skin Sensitisation. OECD Series on Testing and Assessment No. 256