Registration Dossier

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

2.1 mg/m³

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL derivation method:

ECHA REACH Guidance

Overall assessment factor (AF):

12.5

Modified dose descriptor starting point:

BMCL10

Value:

26.17 mg/m³

Explanation for the modification of the dose descriptor starting point:

ECHA guidance

AF for dose response relationship:

1

Justification:

BMDL10 derivation (no correction necessary)

AF for differences in duration of exposure:

1

Justification:

Lifetime study

AF for interspecies differences (allometric scaling):

1

Justification:

included in route to route extrapolation

AF for other interspecies differences:

2.5

Justification:

Worst case consideration (see discussion). Although humans seem to have a seems to have a more efficient detoxification pathway for hydroquinone compared to rats (justifying an AF of 1), remaining uncertainties regarding the inhalation route suggest to keep the default factor of 2.5 for interspecies differences as a precautionary approach for the inhalation route, in the absence of more specific data.

AF for intraspecies differences:

5

Justification:

Default value - ECHA guidance

AF for the quality of the whole database:

1

Justification:

Extensive dataset. Uncertainties and worst case considerations have been included in POD selection and other AF.

AF for remaining uncertainties:

1

Justification:

Uncertainties and worst case considerations have been included in POD selection and other AF

Hazard assessment conclusion:

no hazard identified

Hazard assessment conclusion:

hazard unknown (no further information necessary)

Hazard assessment conclusion:

hazard unknown (no further information necessary)

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

3.33 mg/kg bw/day

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL derivation method:

ECHA REACH Guidance

Overall assessment factor (AF):

45

Modified dose descriptor starting point:

BMDL10

Value:

150 mg/kg bw/day

Explanation for the modification of the dose descriptor starting point:

Worst case approach due to limitations identified in the repeated dermal toxicity studies

AF for dose response relationship:

1

Justification:

BMDL10 derivation (no correction necessary)

AF for differences in duration of exposure:

1

Justification:

Lifetime study

AF for interspecies differences (allometric scaling):

9

Justification:

Conservative assumption based on TK data and PBPK models (see discussion). PBPK model used to extrapolate from oral rat to human dermal dose metric predicted a greater AUC for HQ in blood, although Cmax was always lower than by the oral gavage administration. This indicates a lower external dose is need in human compared to rat to obtain the same internal AUC (based on HQ in blood).

AF for other interspecies differences:

1

Justification:

In vitro studies with human and rat hepatocytes showed a higher metabolic capacity in human hepatocytes compared to rat hepatocytes (Poet, 2004). English & Deisinger (2005) also showed similar metabolites and proportion in urine following dermal administration. Humans seem to have a seems to have a more efficient detoxification pathway for hydroquinone compared to rats supporting a AF of 1 for the toxicodynamic component of the interspecies differences.

AF for intraspecies differences:

5

Justification:

Default value for workers - ECHA guidance

AF for the quality of the whole database:

1

Justification:

Extensive dataset. Uncertainties and worst case considerations have been included in POD selection and other AF.

AF for remaining uncertainties:

1

Justification:

Extensive dataset. Uncertainties and worst case considerations have been included in POD selection and other AF.

Hazard assessment conclusion:

no hazard identified

Hazard assessment conclusion:

medium hazard (no threshold derived)

Most sensitive endpoint:

sensitisation (skin)

Hazard assessment conclusion:

medium hazard (no threshold derived)

Most sensitive endpoint:

sensitisation (skin)

Hazard assessment conclusion:

medium hazard (no threshold derived)

Rationale for DNEL point of Departure

All available studies were reviewed for potential critical effects. BMDL approach was selected to take into account the whole dose-response data, dose-response curve, and limit the impact of dose spacing used in the various studies. The lowest relevant BMDL10 value was then used to derive a DNEL for quantitative risk assessment.

Selection of tumour-response data relevant to risk assessment and suitable for modeling:

1- CNS effects, such as transient tremors or convulsions usually occurring after dosing were observed in many studies in rats and mice treated by oral gavage. Two reliable studies allowed describing a dose-response for tremor occurrence: a 2-generation study in rats (Blacker et al., 1993), and a 90-day oral study (Toppings et al., 1988). Daily observations of the animals were recorded and the sum of the observations was plotted as a function of the dose. Both datasets showed a dose-response for a BMDL derivation.

2- Renal tubular hyperplasia and adenomas were the most significant findingsin male rats treated for 2 years by oral gavage (Kari et al., 1992, NTP) or in the diet (Shibata, 1991). The tumors were observed in 2 independent experiments. Reanalysis of NTP results by Hard et al. (1997) to include consideration of chronic progressive nephropathy severity stages, showed that the renal lesions in males were located in the areas with more severe CPN, indicating that the that development of hyperplasia or adenomas had been favored by enhanced chronic progressive nephropathy. The combined hyperplasia or adenoma incidences showed a significant dose-dependent increase which was suitable for BMDL modeling.

The other findings below were not observed consistently between studies in rats and mice, or between genders, and their relevance to human cancer risk assessment has been questioned.

3- Female rats treated for 2 years did not show kidney alterations, but displayed instead dose-dependent increase in mononuclear cell leukemia, that was statistically significant at the high dose (40%), and with the incidence in both treated groups exceeding the historical means in controls, although the low dose incidence (27%) was within the historical control range (25+15%). Reanalysis by Whysner et al. (1995) concluded this findings was not related to the substance based on the historical control data and the fact that increased leukemia incidence was not observed in male rats of the same study nor in another 2-year study (Shibata et al., 1991) in either males or females, and not seen in mice of either sexes (Shibata et al., 1991). This is also one of the most frequently observed neoplasms in untreated F344 rats in NTP studies (Haseman et al., Toxicol. Pathol., 26(3):428-441, 1998). The relevance to human cancer risk of this neoplasm commonly observed in aged F344 rats is debated (Williams et al, Principles of testing for carcinogenic activity, in Principles and Methods of Toxicology, Fifth ed. A.W. Hayes, 2007, p.1265-1316; Thomas et al., Toxicol. Sci. 99:3-19, 2007). This effect was not considered relevant for a BMDL derivation.

4- Findings in mice consisted of increased incidence of hepatocellular adenomas in females in one study (Kari et al., 1992), but not in the other study conducted at a higher dose which showed adenomas in male mice instead (Shibata et al., 1991). There was no dose-dependent response for carcinomas and even a lower incidence in males of the high dose group compared to controls. The high dose group presented also a lower incidence compared to the low dose in both males and females, and this was not explained by the mortality rate which was similar in the treated groups compared to controls. B6C3F1 mice have usually a high spontaneous rate of liver tumors, and it was found also influenced by various factors including housing and diet (Haseman et al., Toxicol. Pathol., 26(3):428-441, 1998; Maronpot, J. Toxicol. Pathol. 22:11-33, 2009).

The shape of the dose-response for adenoma incidence, or for incidence of combined adenomas or carcinomas were also not appropriate for BDML modeling.

5- Additional non-neoplastic findings in mice carcinogenicity study consisted of thyroid follicular cells hyperplasia. The significance of these findings seems uncertain as these non-neoplastic effects were observed in only 2 females of the high dose group at the interim examination on week 63, and they were not observed in another 2-year mouse diet study in either males or females receiving at 1046 mg/kg/day or 1486 mg/kg/day, respectively (Shibata, 1991). No significant increase in the incidence of adenomas was observed compared to controls and incidence was within the range of historical control (NTP studies: 0 - 9%), and there was one carcinoma recorded in the high dose group. The observed increased hyperplasia could be secondary to hepatocellular lesions reported in mice and altered feedback pathways. The dose-response of hyperplasia in females was not appropriate for BMDL modeling due to the high incidence (and plateau) in both dose groups (> 80%). The females of the control group also had a noticeably high incidence (24%). The late onset of hyperplasia, combined with high levels in untreated controls, and known differences between rodents and humans regarding thyroid physiology and sensitivity to potential alterations support the lack of significance of these non-neoplastic lesions.

6- Last, in vivo genotoxicity studies were also reviewed to evaluate whether there was any reliable study that allowed dose-response analysis. Only one study reported a dose-response with micronucleated erythrocytes analysis in mice treated by intraperitoneal administration (Grawé et al., 1996). The results obtained at 4 time samplings over 24 to 72 hours post-treatment and for 3 doses were integrated by the authors to plot the AUC data as a function of dose. A 2-phase dose-response increase was obtained, suggesting a threshold of effects between 12.5 mg/kg and 25 mg/kg (ip route). However, there is no available mouse PBPK model and limited toxicokinetic data in mice to allow the use of these data for dose metric analysis and extrapolation to another more standard exposure route, and to other species. Other studies in rats (in vivo Comet assay) or transgenic mice (TGR) treated by oral gavage at doses up to 420 mg/kg/day or 200 mg/kg/day, respectively, showed no DNA effects in the organs assessed, which included liver, kidney and lung.

BMDL modeling

EPA Benchmark Dose Software (BMDS 2.6.0.1) was used to determine the estimated lower bound on the benchmark dose corresponding to an extra risk of 10% (BMDL10).

As discussed above, a few critical effects were considered relevant and allowed derivation of a BMDL value.

Results of the BMDL10 for critical effects based on oral route studies

	combined incidence of renal tubule hyperplasia or adenomas	CNS effects	Tremors	Lowest NOAEL in whole dataset
	F344 rats	SD rats	SD rats	F344 rats
	2-year carcinogenicity	2-generation	90-day	2-generation
source	NTP, 1989, Kari, 1992, re-evaluated by Hard, 1997	Blacker, 1993	Toppings, 1988	Overall database
BMDS model	Multistage 2	LogProbit	Dichotomous-Hill	N/A
comments	Lowest AIC, goodness of fit	Lowest AIC, goodness of fit	Lowest AIC, best fit But model warning	N/A
BMDL10	15 mg/kg/day	63.4 mg/kg/day	61.2 mg/kg/day	NOAEL: 20 mg/kg/day

The most conservative value was obtained with the combined renal tubular hyperplasia and adenomas, with a BMDL10 of 15 mg/kg/day, which was quite consistent with the lowest NOAEL identified in the whole dataset.

Effects on the central nervous system (tremors) resulted in higher BMDL10, due to the shape of the dose-responses that were obtained in 2 studies.

The other effects in mice with dose-response allowing BMDS modeling also resulted in a BMDL10 value of 15 mg/kg/day (thyroid hyperplasia, and mice mononuclear cells leukemia although for this later one only a few models were compatible with the dataset). However, these effects were considered of questionable significance as they were not found consistently in the different studies, sexes, or species.

Rationale for route-to route extrapolation

Analysis of the available data supported a threshold assessment, and non-genotoxic mode of action, as the in vivo positive genotoxic effects were mainly observed in intraperitoneal dosing, while in one oral study a weakly positive response was observed by oral gavage. In vivo oral studies in rats and mice designed to assess DNA damages in specific organs did not show any specific gene mutation or DNA breaks in the identified target organs. The kidney tumors in F344 rats were found associated with elevated multi-glutathione conjugates metabolites and thought to originate from exacerbation of chronic progressive nephropathy which is known to occur in aging rats. The particular sensitivity of Fischer F344 was associated with metabolic capacity in kidney.

The following elements were considered:

- The genotoxic effects by the intraperitoneal route were not supported by comparable results in oral studies (TGR and in vivo Comet assay) suggesting a route-specific mode of action.

- adverse effects, in particular nephrotoxicity, could be related to a bioactivation pathway forming particular glutathionyl mono and multi-conjugates of HQ that target kidney with sensitivity due to metabolic profile differing in various rat strains. F344 Rat PBPK simulation indicate that a larger oral dose (approx. 2-fold based on lower dose range) is necessary to obtain the same internal level of glutathionyl metabolites compared to the intraperitoneal route (Corley et al., 2000).

- several toxicokinetics and PBPK modeling studies indicated that the pulmonary route and dermal application would results in a majority of glucuronic acid and sulfate conjugates formed through a detoxification pathway, in proportions similar to those obtained via the oral route, and rather low levels of the more toxic glutathionyl-HQ conjugates, contrasting with the levels seen following intraperitoneal administration.

A quantitative assessment was therefore conducted using the lowest BMDL10 identified from oral studies and route-to-route extrapolation to derive the relevant DNEL.

The most conservative rat oral BMDL10 value was considered as point of departure for the DNEL derivation, although based on available information on the mechanism of action, the relevance for human of the selected tumor effect is unlikely. Rat oral BMDL10 = 15 mg/kg bw/day.

Genotoxic effects observed in vivo in mice treated by the intraperitoneal route were considered route-specific, and associated with an elevated level of HQ-glutathione-conjugates (especially multi-glutathione conjugates) observed by this route of administration. Toxicokinetics studies following administration by the dermal route or intratracheal route as well as PBPK models showed a metabolic profile similar to that of the oral route, with a fast metabolisation and formation of a majority of glucuronide conjugates and sulfates eliminated in urine.

Despite the absence of local pulmonary metabolic activity, the very rapid metabolisation and elimination of HQ following intratracheal instillation (faster than via the intraperitoneal route) support a route-to-route extrapolation from the oral data in rats to inhalation exposure in humans. TK studies showed absorption close to 100% of HQ administered by the oral route. A similar absorption is assumed for pulmonary route in the route-to-route extrapolation (independently of whether HQ dust particles can reach the deep lung under normal conditions of use).

Interspecies differences – refinement using TK data, PBPK models and in vitro metabolic studies

When available, PBPK modelling can be used to adjust assessment factors in DNEL derivation (REACH guidance Annex R.8-4).

- Toxicokinetic component of interspecies differences (allometric scaling) (see appendix 1): two available PBPK models provided specific dose-metric information on rat/human differences for the oral route, as well as for the dermal route. For the oral route, HQ Cmax in blood or in liver indicate differences of 1.5 to 10.5 (Cmax), and 1.3 to 8.5 for the respective AUC, when extrapolating from rat oral external dose to human oral external dose (Gajewska, 2014); or 1.7-fold difference based on blood AUC for the critical glutathionyl-HQ metabolites (Poet, 2010). A lower oral dose is needed in humans (between 1.3 and 10-fold dose difference) to obtain the same Cmax or AUC in liver or blood, compared to rats. A value of 2.5 is used as a reasonably conservative compromise between the different predictions.

- Toxicodynamic component (residual differences):

Additional assessment factors for interspecies differences (toxicodynamic component) were not used considering the results from human in vivo data and in vitro metabolism studies reported in section 7.1 of the IUCLID (see also endpoint summary for that section). These data are indicating a very effective metabolism and detoxification of HQ in humans, more favorable than in rodents when tested in isolated hepatocytes.

In vivo data (single human volunteer): Corley et al. (1998, 2000)

In vitro data (microsomal fractions of livers from human donors and PBPK modelling): Poet et al. (2004) showed that the overall capacity for metabolism of HQ and its mono-glutathione conjugate is greater in hepatocytes from humans than in those from rats, suggesting a greater capacity for detoxification of the glutathione conjugates in humans. Metabolic rate constants were applied to an existing physiologically based pharmacokinetic model, which was used to predict total glutathione metabolites produced in the liver. The results showed that body burdens of these metabolites will be much higher in rats than in humans.

Based on these data supporting a greater sensitivity of rodent model compared to humans, no additional assessment factor is considered necessary for the toxicodynamic component of the interspecies differences for dermal and oral routes. For the inhalation route, the default value of 2.5 was applied to account for the additional uncertainties and limited available PBPK modelling data.

Overview of DNEL derivations (workers)

Oral to inhalation	values	comments
Oral rat BMDL10	15 mg/kg/day
Converted inhalation human BMCL10 (Based on ECHA guidance R8)	26.17 mg/m3	Rat SVR: 0.384 m3/kg human SVR: 6.7 m3/8 hr (rest) active worker: 10 m3/8 hr
Dose-response relationship	1	Lowest BMDL10 served as point of departure
Allometric factor (TK component)	1	Not needed (inhalation conversion)
Interspecies residual differences (TD component)	2.5	Uncertainties for the inhalation route. Although humans seem to have a more efficient detoxification pathway for hydroquinone compared to rats (justifying an AF of 1) as demonstrated by in vitro experiments in hepatocytes, uncertainties related to the inhalation route support the use of the default factor 2.5 for interspecies differences as a precautionary approach.
Intraspecies	5	Default for workers
Duration of exposure	1	2-year carcinogenicity study
Inhalation DNEL	2.1 mg/m3

oral to dermal	values	comments
Oral rat BMDL10	15 mg/kg/day	(see appendix 2)
Converted to dermal human BMDL10	150 mg/kg/day	10% dermal absorption
Dose-response relationship	1	Lowest BMDL10 served as point of departure
Allometric factor (TK component)	9	PBPK model used to extrapolate from oral rat to human dermal dose metric showed a greater AUC for HQ in blood, although Cmax was always lower than by the oral gavage administration.
Interspecies residual differences (TD component)	1	Higher metabolic capacity in human hepatocytes compared to rat hepatocytes (Poet, 2004). English & Deisinger (2005) also showed similar metabolites and proportion in urine following dermal administration.
Intraspecies	5	Default for workers
Duration of exposure	1	2-year carcinogenicity study
Dermal DNEL	3.33 mg/kg/day

Appendix 1:

PBPK model from Gajewska et al., 2014, Toxicology Letters 227:189–202

Principle: interspecies oral extrapolation based on HQ Cmax in blood, used for route-to-route extrapolation

Starting point: oral rat NOAEL = 20 mg/kg/day (lowest NOAEL for the whole database)

Surrogate dose metric for internal concentrations: HQ Cmax in blood for interspecies extrapolation (oral route), and HQ AUC in blood for oral-to-dermal extrapolation (for human exposure assessment)

External dose Interspecies extrapolation		Dose metric
Oral rat NOAEL	Oral extrapolated human NOAEL	Oral extrapolated human NOAEL	Rat simulated blood and liver HQ AUC and Cmax at rat oral NOAEL dose
20 mg/kg/day	1.9 mg/kg/day
	Lowest simulated value, based on blood Cmax Value used for oral to dermal extrapolation (10x difference max)	16 mg/kg/day <= 2.4 mg/kg/day <= 13 mg/kg/day <= 1.9 mg/kg/day <=	AUC liver:0.819 mg/L AUC blood: 0.081 mg/L Cmax liver: 1.037 mg/L Cmax blood: 0.056 mg/L
	↓
Oral to dermal extrapolation of internal HQ concentrations in human at oral NOAEL doses
	Oral (predicted)	Dermal (predicted)	Ratio dermal/oral (po/pc)
AUC liver	0.082 mg/L	0.026 mg/L	0.317 (3.15)
AUC blood	0.062 mg/L	0.532 mg/L	8.58 =>9-fold difference
Cmax liver	0.091 mg/L	0.0006 mg/L	0.0066 (152)
Cmax blood	0.045 mg/L	0.013 mg/L	0.29 (3.5)

For the oral route, both the HQ Cmax and AUC in liver were higher than in blood. This is consistent with the rapid blood elimination rate observed in toxicokinetics studies, and the main metabolisation occurring in liver, associated with protein binding, prior to transfer to the kidneys and elimination in urine.

Predicted Cmax in liver or in blood is higher for the oral route (peak) compared to the dermal route.

The simulation of dermal route exposure at the oral rat NOAEL resulted in a 9-fold greater AUC in blood compared to the oral route (lower blood concentration but slower elimination due to slower diffusion through skin layers). The dermal threshold based on the AUC in blood is 9-fold lower than the oral NOAEL dose applied to a skin surface of 896/9 cm2 (for a 2-h exposure).

PBPK modeling of the dermal exposure showed that Blood AUC is greater than liver AUC, and also 9-fold higher than via oral exposure. Exposure is therefore greater. A 9-fold lower dermal dose is needed to reach the same blood AU in humans.

PBPK model from Poet et al., 2010, Food and Chemical Toxicology 48(11):3085-3092

The previous model was optimized.

Principle: interspecies oral extrapolation based on total GSH conjugates AUC in blood, used for route-to-route extrapolation

Starting point: oral rat NOAEL = 20 mg/kg/day (lowest NOAEL for the whole database)

Surrogate dose metric for internal concentration: total multi-glutathione conjugates in blood.

External dose Interspecies extrapolation
Oral rat NOAEL	Oral extrapolated human NOAEL	Oral extrapolated human NOAEL	Rat simulated blood AUC at rat oral NOAEL dose
20 mg/kg/day	12 mg/kg bw
	simulated value, based on AUC for total GSH conjugates in blood Value used for oral to dermal extrapolation (1.7x difference)	12 mg/kg <=	AUC blood: 18.3 mg/L/h

The ratio external dose rat/external dose human constitutes the TK part of interspecies differences:

Based on Cmax blood: 20/1.9 = 10.5 ; Based on AUC blood: 20/2.4 = 8.3

Based on Cmax liver: 20/13 = 1.5 ; Based on AUC blood: 20/16 = 1.3

Based on blood AUC for total glutathionyl conjugates : 20/12 = 1.7

HQ Cmax was predicted to be higher in liver than in blood. This is consistent with the rapid blood elimination rate observed in toxicokinetics studies, and the main metabolisation occurring in liver, associated with protein binding, prior to transfer to the kidneys and elimination in urine.

When considering the critical metabolites glutathionyl conjugates a 1.7-fold difference was predicted between rats and human for the oral route exposure.

Overall, based on 2 PBPK models, a lower oral dose is needed in humans, between 1.3 and 10-fold dose difference to obtain the same Cmax or AUC in blood or liver, compared to rats.

Appendix 2

Available data for the dermal route were compared to identify the most appropriate point of departure for the dermal DNEL derivation

	Lowest oral BMDL10	14-day dermal	90-day dermal
	F344 rats	F344 rats	F344 rats
	2-year carcinogenicity	14-day	90-day
source	NTP, 1989, Kari, 1992, re-evaluated by Hard, 1997	NTP, 1989	David et al., 1998
Comments	Oral-to-dermal extrapolation	No effects. Short exposure period	No effects. Cream formulation at 0.5% HQ, highest dose is low
POD	150 mg/kg/day (oral 15 mg/kg/day converted to dermal external dose)	3840 mg/kg/day	74 mg/kg/day
Dose-response	1 (BMDL)	1 (no systemic effects)	1 (no systemic effects)
Exposure duration	1	10	3
Allometric scaling (TK)	9	9	9
Residual differences (TD)	1	1	1
Intraspecies	5 (worker) 10 (general public)	5 (worker) 10 (general public)	5 (worker) 10 (general public)
Quality of the database	1	1	1
Remaining uncertainties	1	2 (uncertainties linked to short test period)	1 (low dose tested, highly conservative POD)
DNEL	3.33 mg/kg/day (workers) 1.66 mg/kg/day (gen. pop)	4.3 mg/kg/day (workers) 2.13 mg/kg/day (gen. pop)	0.55 mg/kg/day (workers) 0.27 mg/kg/day (gen. pop)

Due to study limitations in the dermal repeated dose toxicity studies, a route-to-route extrapolation is considered appropriate, considering the conservative factors applied and relying on information from toxicokinetic studies and PBPK models which showed rapid metabolisation and elimination of HQ through metabolites similar to the oral route.

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

1.05 mg/m³

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL derivation method:

ECHA REACH Guidance

Overall assessment factor (AF):

25

Modified dose descriptor starting point:

BMCL10

Value:

26.17 mg/m³

Explanation for the modification of the dose descriptor starting point:

ECHA guidance.

AF for dose response relationship:

1

Justification:

BMDL10 derivation (no correction necessary)

AF for differences in duration of exposure:

1

Justification:

Lifetime study

AF for interspecies differences (allometric scaling):

1

Justification:

Included in route to route extrapolation

AF for other interspecies differences:

2.5

Justification:

Worst case consideration (see discussion). Although humans seem to have a seems to have a more efficient detoxification pathway for hydroquinone compared to rats (justifying an AF of 1), other considerations provided by PBPK models suggest to keep the default factor of 2.5 for interspecies differences as a precautionary approach for the inhalation route, in the absence of more specific data.

AF for intraspecies differences:

10

Justification:

Default value - ECHA guidance

AF for the quality of the whole database:

1

Justification:

Extensive dataset. Uncertainties and worst case considerations have been included in POD selection and other AF.

AF for remaining uncertainties:

1

Justification:

Uncertainties and worst case considerations have been included in POD selection and other AF

Hazard assessment conclusion:

no hazard identified

Hazard assessment conclusion:

hazard unknown (no further information necessary)

Hazard assessment conclusion:

hazard unknown (no further information necessary)

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

1.66 mg/kg bw/day

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL derivation method:

ECHA REACH Guidance

Overall assessment factor (AF):

90

Modified dose descriptor starting point:

BMDL10

Value:

150 mg/kg bw/day

Explanation for the modification of the dose descriptor starting point:

Worst case approach due to limitations identified in the repeated dermal toxicity studies

AF for dose response relationship:

1

Justification:

BMDL10 derivation (no correction necessary)

AF for differences in duration of exposure:

1

Justification:

Lifetime study

AF for interspecies differences (allometric scaling):

9

Justification:

Conservative assumption based on TK data and PBPK models (see discussion). PBPK model used to extrapolate from oral rat to human dermal dose metric predicted a greater AUC for HQ in blood, although Cmax was always lower than by the oral gavage administration. This indicates a lower external dose is need in human compared to rat to obtain the same internal AUC (based on HQ in blood).

AF for other interspecies differences:

1

Justification:

In vitro studies with human and rat hepatocytes showed a higher metabolic capacity in human hepatocytes compared to rat hepatocytes (Poet, 2004). English & Deisinger (2005) also showed similar metabolites and proportion in urine following dermal administration. Humans seem to have a seems to have a more efficient detoxification pathway for hydroquinone compared to rats supporting a AF of 1 for the toxicodynamic component of the interspecies differences.

AF for intraspecies differences:

10

Justification:

Default for general population

AF for the quality of the whole database:

1

Justification:

Extensive dataset. Uncertainties and worst case considerations have been included in POD selection and other AF.

AF for remaining uncertainties:

1

Justification:

Uncertainties and worst case considerations have been included in POD selection and other AF.

Hazard assessment conclusion:

no hazard identified

Hazard assessment conclusion:

medium hazard (no threshold derived)

Most sensitive endpoint:

sensitisation (skin)

Hazard assessment conclusion:

medium hazard (no threshold derived)

Most sensitive endpoint:

sensitisation (skin)

Hazard assessment conclusion:

DNEL (Derived No Effect Level)

Value:

0.6 mg/kg bw/day

Most sensitive endpoint:

carcinogenicity

Route of original study:

Oral

DNEL derivation method:

ECHA REACH Guidance

Overall assessment factor (AF):

25

Modified dose descriptor starting point:

BMDL10

Value:

15 mg/kg bw/day

Explanation for the modification of the dose descriptor starting point:

not applicable

AF for dose response relationship:

1

Justification:

BMDL10 derivation (no correction necessary)

AF for differences in duration of exposure:

1

Justification:

Lifetime study

AF for interspecies differences (allometric scaling):

2.5

Justification:

PBPK models were used for dose-metric oral rat to oral human extrapolation: HQ Cmax in blood or in liver indicate differences of 1.5 to 10.5 (Cmax), and 1.3 to 8.5 for the respective AUC, when extrapolating from rat oral external dose to human oral external dose (Gajewska, 2014); or 1.7-fold difference based on blood AUC for the critical glutathionyl-HQ metabolites (Poet, 2010). The default value of 2.5 is reasonably conservative to account for uncertainties.

AF for other interspecies differences:

1

Justification:

In vitro studies with human and rat hepatocytes showed a higher metabolic capacity in human hepatocytes compared to rat hepatocytes (Poet, 2004). Corley et al. (2000) also showed similar metabolites and proportion in urine following oral ingestion by humans or rats. Humans seem to have a seems to have a more efficient detoxification pathway for hydroquinone compared to rats supporting a AF of 1 for the toxicodynamic component of the interspecies differences. HQ is rapidly metabolized and excreted in urine after oral uptake, i.e. there is no indication for bioaccumulation of HQ in the organism.

AF for intraspecies differences:

10

Justification:

default assessment factor for general population

AF for the quality of the whole database:

1

Justification:

The dataset is considered of good quality

AF for remaining uncertainties:

1

Justification:

No remaining uncertainties considered for that exposure route

Hazard assessment conclusion:

low hazard (no threshold derived)

Hazard assessment conclusion:

medium hazard (no threshold derived)

Rationale for DNEL point of Departure

All available studies were reviewed for potential critical effects. BMDL approach was selected to take into account the whole dose-response data, dose-response curve, and limit the impact of dose spacing used in the various studies. The lowest relevant BMDL10 value was then used to derive a DNEL for quantitative risk assessment.

Selection of tumour-response data relevant to risk assessment and suitable for modeling:

1- CNS effects, such as transient tremors or convulsions usuallyoccurring after dosing were observed in many studies in rats and mice treated by oral gavage. Two reliable studies allowed describing a dose-response for tremor occurrence: a 2-generation study in rats (Blacker et al., 1993), and a 90-day oral study (Toppings et al., 1988). Daily observations of the animals were recorded and the sum of the observations was plotted as a function of the dose. Both datasets showed a dose-response for a BMDL derivation.

2- Renal tubular hyperplasia and adenomas were the most significant findingsin male rats treated for 2 years by oral gavage (Kari et al., 1992, NTP) or in the diet (Shibata, 1991). The tumors were observed in 2 independent experiments. Reanalysis of NTP results by Hard et al. (1997) to include consideration of chronic progressive nephropathy severity stages, showed that the renal lesions in males were located in the areas with more severe CPN, indicating that the that development of hyperplasia or adenomas had been favored by enhanced chronic progressive nephropathy. The combined hyperplasia or adenoma incidences showed a significant dose-dependent increase which was suitable for BMDL modeling.

The other findings below were not observed consistently between studies in rats and mice, or between genders, and their relevance to human cancer risk assessment has been questioned.

3- Female rats treated for 2 years did not show kidney alterations, but displayed instead dose-dependent increase in mononuclear cell leukemia, that was statistically significant at the high dose (40%), and with the incidence in both treated groups exceeding the historical means in controls, although the low dose incidence (27%) was within the historical control range (25+15%). Reanalysis by Whysner et al. (1995) concluded this findings was not related to the substance based on the historical control data and the fact that increased leukemia incidence was not observed in male rats of the same study nor in another 2-year study (Shibata et al., 1991) in either males or females, and not seen in mice of either sexes (Shibata et al., 1991). This is also one of the most frequently observed neoplasms in untreated F344 rats in NTP studies (Haseman et al., Toxicol. Pathol., 26(3):428-441, 1998). The relevance to human cancer risk of this neoplasm commonly observed in aged F344 rats is debated (Williams et al, Principles of testing for carcinogenic activity, in Principles and Methods of Toxicology, Fifth ed. A.W. Hayes, 2007, p.1265-1316; Thomas et al., Toxicol. Sci. 99:3-19, 2007). This effect was not considered relevant for a BMDL derivation.

4- Findings in mice consisted of increased incidence of hepatocellular adenomas in females in one study (Kari et al., 1992), but not in the other study conducted at a higher dose which showed adenomas in male mice instead (Shibata et al., 1991). There was no dose-dependent response for carcinomas and even a lower incidence in males of the high dose group compared to controls. The high dose group presented also a lower incidence compared to the low dose in both males and females, and this was not explained by the mortality rate which was similar in the treated groups compared to controls. B6C3F1 mice have usually a high spontaneous rate of liver tumors, and it was found also influenced by various factors including housing and diet (Haseman et al., Toxicol. Pathol., 26(3):428-441, 1998; Maronpot, J. Toxicol. Pathol. 22:11-33, 2009).

The shape of the dose-response for adenoma incidence, or for incidence of combined adenomas or carcinomas were also not appropriate for BDML modeling.

5- Additional non-neoplastic findings in mice carcinogenicity study consisted of thyroid follicular cells hyperplasia. The significance of these findings seems uncertain as these non-neoplastic effects were observed in only 2 females of the high dose group at the interim examination on week 63, and they were not observed in another 2-year mouse diet study in either males or females receiving at 1046 mg/kg/day or 1486 mg/kg/day, respectively (Shibata, 1991). No significant increase in the incidence of adenomas was observed compared to controls and incidence was within the range of historical control (NTP studies: 0 - 9%), and there was one carcinoma recorded in the high dose group. The observed increased hyperplasia could be secondary to hepatocellular lesions reported in mice and altered feedback pathways. The dose-response of hyperplasia in females was not appropriate for BMDL modeling due to the high incidence (and plateau) in both dose groups (> 80%). The females of the control group also had a noticeably high incidence (24%). The late onset of hyperplasia, combined with high levels in untreated controls, and known differences between rodents and humans regarding thyroid physiology and sensitivity to potential alterations support the lack of significance of these non-neoplastic lesions.

6- Last, in vivo genotoxicity studies were also reviewed to evaluate whether there was any reliable study that allowed dose-response analysis. Only one study reported a dose-response with micronucleated erythrocytes analysis in mice treated by intraperitoneal administration (Grawé et al., 1996). The results obtained at 4 time samplings over 24 to 72 hours post-treatment and for 3 doses were integrated by the authors to plot the AUC data as a function of dose. A 2-phase dose-response increase was obtained, suggesting a threshold of effects between 12.5 mg/kg and 25 mg/kg (ip route). However, there is no available mouse PBPK model and limited toxicokinetic data in mice to allow the use of these data for dose metric analysis and extrapolation to another more standard exposure route, and to other species. Other studies in rats (in vivo Comet assay) or transgenic mice (TGR) treated by oral gavage at doses up to 420 mg/kg/day or 200 mg/kg/day, respectively, showed no DNA effects in the organs assessed, which included liver, kidney and lung.

BMDL modeling

EPA Benchmark Dose Software (BMDS 2.6.0.1) was used to determine the estimated lower bound on the benchmark dose corresponding to an extra risk of 10% (BMDL10).

As discussed above, a few critical effects were considered both relevant and allowing derivation of a BMDL value.

Results of the BMDL10 for critical effects based on oral route studies

	combined incidence of renal tubule hyperplasia or adenomas	CNS effects	Tremors	Lowest NOAEL in whole dataset
	F344 rats	SD rats	SD rats	F344 rats
	2-year carcinogenicity	2-generation	90-day	2-generation
source	NTP, 1989, Kari, 1992, re-evaluated by Hard, 1997	Blacker, 1993	Toppings, 1988	Overall database
BMDS model	Multistage 2	LogProbit	Dichotomous-Hill	N/A
comments	Lowest AIC, goodness of fit	Lowest AIC, goodness of fit	Lowest AIC, best fit But model warning	N/A
BMDL10	15 mg/kg/day	63.4 mg/kg/day	61.2 mg/kg/day	NOAEL: 20 mg/kg/day