Registration Dossier

Administrative data

Key value for chemical safety assessment

Genetic toxicity in vitro

Description of key information

Titanium dioxide has been tested in bacterial reverse mutation assays, in vitro gene mutation and clastogenicity tests as well as in vivo. All tests show a negative response, thus titanium dioxide does not require classification for mutagenic properties. Endpoint Conclusion: No adverse effect observed (negative).

Endpoint conclusion
Endpoint conclusion:
no adverse effect observed (negative)

Genetic toxicity in vivo

Description of key information

Titanium dioxide has been tested in bacterial reverse mutation assays, in vitro gene mutation and clastogenicity tests as well as in vivo. All tests show a negative response, thus titanium dioxide does not require classification for mutagenic properties. Endpoint Conclusion: No adverse effect observed (negative).

Endpoint conclusion
Endpoint conclusion:
no adverse effect observed (negative)

Additional information

Germ cell mutagenicity studies with pigment-grade titanium dioxide

Pigment size titanium dioxide was tested in various in vitro and in vivo systems for its ability to induce gene, chromosome or genome mutations. The systems used include bacterial reverse mutation tests, mammalian cell mutagenicity, chromosome aberration and micronuclei test as well as in vivo experiments in mice and rats for chromosome aberration or micronuclei formation in the bone marrow or peripheral blood. The tests were conducted with two crystalline forms of titanium dioxide, namely anatase and rutile or mixtures thereof.

 

In vitro genetic toxicity tests

Gene mutation in bacteria

Myhre, A. (2011) investigated the mutagenic potential of pigment size titanium dioxide (rutile phase; 100% pure) in the bacterial reverse mutation test according OECD TG 471 and under GLP. S. typhimurium TA 1535, TA 1537, TA 98, TA 100 and E. coli WP2 uvr A were exposed using the plate incorporation method at doses of 333, 667, 1000, 3333 and 5000 µg/plate. No positive mutagenic responses were observed at any dose level or with any tester strain in either the absence or presence of S9 metabolic activation. No toxicity was observed at any dose level with any tester strain in either the absence or presence of S9. Test substance precipitation was observed at all dose levels (333 to 5000 µg/plate) with all tester strains both in the absence and presence of S9 activation. Appropriate positive controls demonstrated the sensitivity of the test system.

Callander, R.D. (1996) investigated the mutagenic potential of hydrophobically coated pigment grade titanium dioxide (rutile phase; 98.5% pure) in the bacterial reverse mutation test according OECD TG 471 and under GLP. S. typhimurium TA 1535, TA 1537, TA 98, TA 100, E. coli WP2 uvr A pKM101 and E. coli WP2 were exposed using the plate incorporation method at doses of 100, 200, 500, 1000, 2500 and 5000 µg/plate. In two separate experiments, hydrophobically coated pigment grade titanium dioxide did not induce any significant, reproducible increases of revertant colonies in any of the tester strains used, either in the presence or absence of S9 activation. Appropriate positive controls demonstrated the sensitivity of the test system.

In the DNA damage and repair test in Bacillus subtilis (recombination assay), Kanematsu, N. et al. (1980) investigated the DNA damaging potential of titanium dioxide (further details on the test item identity was not reported, pigment size only estimated by author) in B. subtilis strains H17 (rec+) and M45 (rec-). Titanium dioxide showed no genotoxic effects in the rec-assay applied in concentrations of 5-500 mM. Due to the poor reporting quality, this report is considered as supporting information only.

There was no evidence for mutagenic activity with titanium dioxide (rutile phase) in bacterial reverse mutation tests using the plate incorporation and pre-incubation method with and without metabolic activation employed at concentrations up to 5 mg/plate. The experiments were conducted with 4 different S. typhimurium and up to 2 E. coli strains. These findings are supported by a negative outcome in a DNA damage assay (rec-assay) in B. subtilis. No information on mutagenic activity in bacteria of the anatase phase is available.

 

In vitro mammalian cell gene mutation

Clay, P. (1997) investigated the mutagenic potential of hydrophobically coated pigment grade titanium dioxide (rutile phase; 98.5% pure) in the in vitro mammalian cell gene mutation test according OECD TG 476 and under GLP. Cultures of L5178Y TK+/- cells were treated in duplicate with doses of hydrophobically coated pigment grade titanium dioxide for 4 hours in the presence or absence of S9 mix at concentrations of 31, 63, 125, 250, or 500 µg/mL. In the first experiment, a concentration of 1000µg/mL resulted in the formation of heavy white precipitate which interfered with cell counting and plate scoring due to large amount of substance remaining in the medium after treatment period. The maximum concentration of 500µg/mL was considered appropriate for testing, since no interference with the scoring was observed, albeit white precipitate was still observable. No significant increases in mutant frequency, compared to the control solvent cultures, were observed in cultures with hydrophobically coated pigment grade titanium dioxide in either the presence or absence of S9 mix. Negligible cytotoxicity was observed at the highest test concentration. Appropriate positive controls demonstrated to sensitivity of the test system.

There was no evidence for mutagenic activity of hydrophobically coated pigment grade titanium dioxide (rutile phase) in the mouse lymphoma assay, up to the maximum concentration limited by heavy precipitate. No information on the mutagenic potential in mammalian cells of uncoated rutile and anatase is available.

 

In vitro mammalian chromosome aberration/micronucleus test

Glover, K.P. (2011) investigated the clastogenic potential of pigment size titanium dioxide (rutile phase; 100% pure) in the in vitro mammalian chromosome aberration test according to OECD 473 and under GLP. Chinese hamster ovary (CHO) cells were treated for 4 and 20 hours with metabolic activation and for 4 hours with S9 activation at concentrations of 25, 50, 75, 100 and 150 µg/mL in two separate experiments. The maximum concentration was limited by substantial toxicity (greater than a 50% reduction in cell growth compared with vehicle control) observed at 250 μg/mL for the 4-hour test conditions and at 100 μg/mL for the 20 hour test condition. Pigment size titanium dioxide did not induce structural or numerical chromosome aberrations in the in vitro mammalian chromosome aberration test in Chinese hamster ovary cells either with or without S9 metabolic activation. Appropriate positive controls demonstrated the sensitivity of the test system.

Fox, V. (1997) investigated the clastogenic effects of hydrophobically coated pigment grade titanium dioxide (rutile phase, 98.5% pure) in human peripheral blood lymphocytes. The study was conducted according to OECD 473 and under GLP. Duplicate cell cultures were exposed towards doses of 10 (5), 50, or 100 µg TiO2/mL for 3 hours with and without metabolic activation. The maximum concentration was limited by the reduction of the mitotic index by more than 50%. Hydrophobically coated pigment grade titanium dioxide did not induce structural or numerical chromosome aberrations in the in vitro mammalian chromosome aberration test in human lymphocytes either with or without S9 metabolic activation. Appropriate positive controls demonstrated the sensitivity of the test system.

Ivett, J.L. (1989) reported on the clastogenic potential of titanium dioxide (unknown crystalline phase, purity 98.5%) in chromosome aberration assay in CHO cells. Cells were exposed at concentrations of 0, 15, 20 or 25 µg/mL for 2 or 8 hours, with and without metabolic activation respectively. The highest dose was limited by the formation of precipitate. 100-200 Colcemid treated cells were scored analysed for aberrations. Titanium dioxide did not induce biologically relevant chromosomal aberrations in CHO cells after 2 and 8 hours treatment with and without metabolic activation.

Miller, B.M., et al.(1995) investigated the micronucleus formation in CHO cells with titanium dioxide (no further details on crystal phase or particle size were provided) with and without metabolic activation using pulse and continuous treatment. Cells were exposed to titanium dioxide concentrations of 0.025 to 10 µg/mL (without metabolic activation) and 0.25 to 10 µg/mL (with S9) were used. No increase on the MN frequency was observed up to the maximum concentration limited by precipitation of the test item.

Lu, P.J., et al. (1998) examined micronuclei formation of titanium dioxide in CHO cells using the cytokinesis block protocol with continuous treatment. The test item was described as titanium dioxide standard solution, no further details on crystal phase or particle size are provided. Cells were exposed with concentrations of 0, 1, 2, 5, 10 and 20 µM in DMSO for 18 hours. At least 1000 binucleated cells were scored per dose. The MN frequency was significantly and dose-dependently enhanced in the treated cells at the same dose range 2.5 to 3-fold. Relative cell survival was determined by a colony forming protocol. The method is insufficiently described and it is unclear how the cytotoxicity/growth inhibition was determined. The cytotoxicity tests were conducted up to 5 mM, whereas the MN experiments were conducted up to 20 mM, hence no information on the cytotoxicity for the maximum concentration is available. The maximum test item concentration of 20 mM titanium dioxide (molar mass: 79.9) corresponds to 1.6 mg/mL. As reported by other authors and depending on the cell line used, doses of 25-150 µg/mL cause significant cytotoxicity (mitotic index < 50 %), thus a ten to sixty fold higher concentrations showing no cytotoxic effects appear grossly implausible. Secondly, titanium dioxide shows a solubility of < 1µg/L in a 24 hours transformation/dissolution test (according to OECD testing series 29 and under GLP). Consequently, the solubility was exceeded in the above described experiments by more than six orders of magnitude. Although the test guidance foresees testing at concentrations with (slightly) visible precipitation, one may safely assume that any mutagenic effect observed in the experiments can be attributed to an unspecific particle effect. Based on the reasons above, the results need to be treated with great caution. Consequently the study is considered not suitable for hazard and risk assessment purposes.

Türkez et al. (2007) investigated the clastogenic potential of titanium dioxide in peripheral blood lymphocytes. The reference exhibits serious shortcomings, rendering it unsuitable for hazard assessment purposes: method is insufficiently described, cytotoxicity/growth inhibition was not measured, solubility exceeded in the experiments by more than three orders of magnitude. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide.

Riebe-Imre et al. (1994) investigated the clastogenic potential of titanium dioxide in hamster airway epithelial cell line. TiO2 induced at very low concentrations a dose-dependent increase of the MN formation in M3E3/C3 cells (nearly 2-fold at the highest concentration of 2 µg/ml). However, the control culture already shows a MN frequency of 4%, being 10-times higher compared with other in vitro test systems. This raises doubts on the suitability of the test system for the in vitro MN test - no historical control data were provided. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide.

Titanium dioxide does not induce clastogenic or aneugenic mutations in mammalian cells up to concentrations limited by toxicity or precipitation. One study shows a dose-response effect for titanium dioxide after 18 hours exposure up to a 3-fold increase in the micronucleus frequency. However, the experimental procedure raises doubt whether the positive findings were caused by the test item itself. It is highly likely that the clastogenic events were caused by an unspecific particle effect due to the excessive exposure concentrations applied in this study. Consequently, it is concluded that titanium dioxide (rutile phase and unknown phase) does not induce chromosome of genome mutations in mammalian cells. No information on clastogenic or aneugenic activity in mammalian cells of the anatase phase is available.

 

Sister chromatid exchange (induced DNA damage)

Ivett, J.L. (1989) examined the DNA damaging effect of titanium dioxide (unknown crystalline phase, purity 98.5%) in a sister chromatid exchange assay in CHO cells. Titanium dioxide was added at concentrations of 0, 2.5, 8.3 and 25 µg/ml in duplicate cultures, with and without metabolic activation for 2 and 25 hours, respectively. For both testing conditions 10 µM bromodeoxyuridine (BrdU) was added 2 hours after dosing, Colcemid was added 2-2.5 hours prior harvest. 50 and 25 cells were scored per dose in the initial and the confirmatory experiment, respectively. Titanium dioxide did not induce sister chromatid exchanges up to the highest dose, limited by solubility. The test item is therefore considered non genotoxic in the sister chromatid exchange assay.

Lu, P.J., et al. (1998) examined sister chromatid exchange in CHO cells after titanium dioxide exposure. The test item was described as titanium dioxide standard solution, no further details on crystal phase or particle size are provided. Cells were exposed with concentrations of 0, 1, 2, 5 µM in DMSO for 24 hours. Colcemid was added to the cultures 2 hours prior harvest. 30 cells in metaphase were scored per dose. SCE frequencies were significantly increased in a dose-dependent manner. Relative cell survival was determined by a colony forming protocol. The method is insufficiently described and it is unclear how the cytotoxicity/growth inhibition was determined. The maximum test item concentration of 5 mM corresponds to 400 µg/mL. As reported by other authors and depending on the cell line used, doses of 25-150 µg/mL cause significant cytotoxicity (mitotic index < 50%), thus a three to sixteen fold higher concentrations showing no cytotoxic effects appear grossly implausible. Secondly, titanium dioxide shows a solubility of < 1µg/L in a 24 hours transformation/dissolution test (according to OECD testing series 29 and under GLP). Consequently, the solubility was exceeded in the above described experiments by more than five orders of magnitude. Although the test guidance foresees testing at concentrations with (slightly) visible precipitation, one may safely assume that any genotoxic effect observed in the experiments can be attributed to an unspecific particle effect. Based on the reasons above, the results need to be treated with great caution. Consequently the study is considered not suitable for hazard and risk assessment purposes.

Two studies on the in vitro SCE of titanium dioxide are available, one study showing no genotoxic effects during 2 and 25 hours exposure. The second study shows a dose-response effect for titanium dioxide after 24 hours exposure. However, the experimental procedure raises doubt whether the positive findings were caused by the test item itself. It is highly likely that the genotoxic effects were caused by an unspecific particle effect due to the excessive exposure concentrations applied in this study. Consequently, it is concluded that titanium dioxide (unknown crystalline phase) does not induce genotoxicity in the in vitro SCE assay.

 

In vivo genetic toxicity tests

 

Donner, E.M. (2011) investigated the induction of micronuclei in the bone marrow of male and female ICR mice following a single oral administration of titanium dioxide (rutile phase, pigment grade, 100% pure). The test was conducted in accordance with OECD TG 474 and under GLP. Animals (10M/10F) received doses of 500, 1000 and 2000 mg/kg bw in deionised water and were sacrificed 24 and 48 hours post exposure. At least 2000 PCEs per animal were scored for the presence of micronuclei and the proportion of PCEs among 1000 total erythrocytes (PCE/NCE ration) was determined. The test item did not induce biologically relevant increases in micronucleated polychromatic erythrocytes in the bone marrow of mice when administered up to the limit dose of 2000mg/kg bw via oral route.

In a testing programme, a series of three different pigment grade titanium dioxide samples were tested for the induction of micronuclei in the peripheral blood of rats after single oral administration (Myhre, A., 2014a, Wessels, A., 2014a and 2014b). Animals received doses of 0, 500, 1000, 2000 mg/kg bw. All tests were conducted in accordance with OECD TG 474/EC B.12 and under GLP:

 

In the study by Myhre, A. (2014a) pigment grade titanium dioxide (TiO2pg-1: anatase phase, 100% purity, d50: 120nm (primary particle size by TEM ECD), SA: 8.1 m²/g) was used as test item. The vehicle control, low, intermediate, and positive control groups contained 5 animals/sex, the high-dose group contained 7 animals/sex. Peripheral blood samples were taken via sublingual vein bleeding approx. 48 and 72 hours post exposure. A total of 20,000 RETs were analysed per blood sample via flow cytometry for the presence of micronuclei and toxicity as indicated by the frequency of immature erythrocytes among the total erythrocytes (PCE/NCE ratio). Additionally, titanium concentrations were determined in blood and liver at 48 or 72 hours. No statistically significant increases in the micronucleated RET frequency were observed in any evaluated test substance-treated group of male or female animals at either time point. There were no statistically significant decreases in the PCE/NCE ratio in any test-substance treated group. Under the conditions of this study, titanium dioxide did not induce formation of micronuclei in rat peripheral blood up to the limit dose of 2000 mg/kg bw. No discernible dose-dependent increases of titanium concentrations in the blood and liver of treated rats relative to control rats could be measured.

 

In the studies by Wessels, A. (2014a and b) two different pigment grade titanium dioxide samples were used: (i) TiO2pg-2: rutile phase, 100% purity, d50: 165nm (primary particle size by TEM ECD), SA: 7.1 m²/g (ii) TiO2pg-3: rutile phase, rutile purity, d50: 132 nm (primary particle size by TEM ECD), SA: 17.1 m²/g. All dose groups contained 5 animals/sex. Peripheral blood samples were taken via tail vein approx. 48 and 72 hours post exposure. A total of 20,000 RETs were analysed per blood sample via flow cytometry for the presence of micronuclei and toxicity as indicated by the frequency of immature erythrocytes among the total erythrocytes (PCE/NCE ratio). No statistically significant increases in the micronucleated RET frequency were observed in any evaluated test substance-treated group of male or female animals at either time point. There were no statistically significant decreases in the PCE/NCE ratio in any test-substance treated group. Under the conditions of this study, titanium dioxide did not induce formation of micronuclei in rat peripheral blood up to the limit dose of 2000 mg/kg bw.

Saber et al. (2012a) instilled intratracheally 54 µg of two different TiO2 nanoparticle types (NanoTiO2 and PhotocatTiO2) and one microscale TiO2 particle type (FineTiO2) in female mice. A comet assay was conducted with BAL fluid cells, which were extracted one day, three days, and 28 days post exposure. Comet tail lengths were calculated per treatment group. Cytotoxicity measurements (LDH and live/dead counts) were performed in a separate in vitro assay, using MutaTM Mouse lung epithelial cells and exposing them to 0 – 800 µg/ml of TiO2 particles. According to the authors, treatment with FineTiO2 and NanoTiO2 resulted in significantly increased DNA damage (1.3 fold, p < 0.05). No severe toxicity was observed in the in vitro cytotoxicity assay. The study shows major shortcomings in study design. The authors performed cytotoxicity measurements only in a separate in vitro assay and with different cells, which is not representative for the cytotoxicity in the in vivo system. Thus, any positive finding might be a cause of cytotoxicity and not direct DNA damage. Controls are partially missing or inadequate. The authors used an in vitro system (H2O2 exposed A549 cells) as positive control, which is inadequate to demonstrate proficiency. The OECD recommends usage of a group of at least three analysable animals treated with a positive control substance. Historical data are missing. BAL cell suspensions were frozen and thawed later for the comet assay analysis. However, the OECD recommends performing slide preparation as soon as possible (ideally within one hour), since freezing of tissues increases the variation between samples and is therefore not recommended for the comet assay. In addition, details on freezing procedure, storage time, and thawing are largely missing. Moreover, laboratory’s proficiency in preparing and analysing comets’ characteristics in comet assays is a prerequisite for reliable data. It was shown that there is huge variability in tail DNA intensities even within negative or positive controls (Tavares, A.M. et al. (2014); Nanogenotox Programme). One dose was testes, whereas the OECD recommends usage of at least three different doses, which does not allow a dose-response relationship analysis. Comet tail length was measured only 24 hours after treatment, exceeding the OECD recommended measurements foreseen after 2-6 h and 16-26 h after last treatment. Details on animal weights, clinical observations, housing, and test group randomization are missing. Evaluation criteria and scoring of comets are not sufficiently described. Results for two control group animals were excluded as outliers because the data points for these mice were more than two standard deviations from the rest of the data. However, the authors do not state on the experimental reason for these incidences. It is not stated whether these cases were due to the test animals per se or mistakes during the assay procedure. Further, data should not be excluded simply because being declared as outliers. The decision to remove the results from the data set on the basis of the standard deviation is arbitrarily and subjective. It would be more adequate to perform designated tests to determine objectively outliers (e.g. Grubbs test). Further, methods and evaluation criteria are insufficiently described. Statements on lysis, electrophoresis, and slide preparation are missing. The scoring system is not specified sufficiently and hedgehog quantity was not specified. The number of analysed comets is not specified. Comet tail lengths were used as endpoint for DNA damage analysis, although recommended by the OECD. Lung inflammation was concurrently observed for NanoTiO2, indicated by significantly increased neutrophil levels. Inflammation and subsequent issue damage may have potentially interfered with the comet assay and led to false positives. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Shelby, M.D., et al. (1995) investigated the clastogenic and aneugenic effect of titanium dioxide (no further details on crystal phase or particle size were provided) in male B6C3F1 mice in two parallel experiments following single intraperitoneal injection. In the micronucleus analysis, animals were dosed 0, 250, 500, 1000 mg/kg bw and 0, 500, 1000, 1500 mg/kg bw in the first and second experiment, respectively. The total dose was given in three portions over 24 hours. Animals were sacrificed 24 hours after the final injection, bone marrow and peripheral blood smears (two slides/tissue/mouse) were prepared and 2000 polychromatic erythrocytes (PCE) were scored per animal for frequency of micronucleated cells. Under the reported experimental conditions titanium dioxide did not induce significant elevated levels of micronuclei in the bone marrow cells of the mouse. Therefore titanium dioxide is considered to be non-clastogenic and non-aneugenic in this micronucleus assay. In the chromosome aberration analysis, animals were dosed 0, 625, 1250, 2500 mg/kg bw and sacrificed 17 or 36 hours post exposure. Animals received colchicine two hours prior sacrifice. Bone marrow cells were fixed, stained and 50 cells per animal were scored for presence of chromosomal aberration. Under the reported experimental conditions titanium dioxide did not induce chromosome aberrations in the bone marrow of mice. The intraperitoneal route of administration is non-physiological and therefore not considered relevant for risk assessment purposes of industrial chemicals, these test results are reported for information purposes.

 

Saber et al. (2012b) instilled intratracheally pure TiO2 nanoparticle suspension, pure carbon black suspension, or with suspended dusts of indoor paint with or without TiO2 nanoparticles in female mice. Instilled concentration of pure nanomaterials and dusts were 18, 54, and 162 µg and 54, 162, and 486 µg, respectively. Comet assays were conducted with BAL fluid and hepatic cells one day, three days, and 28 days post exposure. Comet tail lengths were normalized to positive controls, and calculated per treatment group. According to the authors, treatment with the highest TiO2 nanoparticle dose resulted in significantly increased DNA damage in hepatic cells one day post exposure. Carbon black induced DNA damage in BAL cells after one day at 18 and 162 µg/mouse, after three days at 162 µg/mouse, and after 28 days at all concentrations. DNA damage in hepatic cells was observed at all concentrations, but only one and 28 days post exposure. TiO2 nanoparticle-containing dust of paint induced no DNA damage. Dust of TiO2 free paint induced DNA damage only in hepatic cells at the highest concentration 28 days post exposure, which was interpretedf as chance finding. The study shows major shortcomings in study design. The authors did not perform any cytotoxicity measurement. However, analysis of cytotoxicity is important, since cytotoxic conditions interfere with the outcomes of the comet assay and lead to artefactual positive findings. Controls are partially missing or inadequate. The authors used an in vitro system (H2O2 exposed A549 cells) as positive control, which is inadequate since it is not representative for the in vivo conditions. Historical data are missing. BAL cell suspensions were frozen and thawed later for the comet assay analysis. However, the OECD recommends performing slide preparation as soon as possible (ideally within one hour), since freezing of tissues increases the variation between samples and is therefore not recommended for the comet assay. In addition, details on freezing procedure, storage time, and thawing are largely missing. The lysing procedure varied significantly in time (1-3.5 h), which hampers the comparability of the results. Therefore, the authors normalised the treated sample to a concurrent in vitro positive control, which is an inadequate normalisation. Although the variability was admitted by the authors, it has not been taken into account in the interpretation of the results. Moreover, laboratory’s proficiency in preparing and analysing comets’ characteristics in comet assays is a prerequisite for reliable data. It was shown that there is huge variability in tail DNA intensities even within negative or positive controls (Tavares, A.M. et al. (2014); Nanogenotox Programme). One dose was testes, whereas the OECD recommends usage of at least three different doses, which does not allow a dose-response relationship analysis. Comet tail length was measured only 24 hours after treatment, exceeding the OECD recommended measurements foreseen after 2-6 h and 16-26 h after last treatment. Details on animal weights, clinical observations, housing, and test group randomization are missing. Evaluation criteria and scoring of comets are not sufficiently described. Further, methods and evaluation criteria are insufficiently described. Statements on lysis details, electrophoresis, and slide preparation are missing. The scoring system is not specified sufficiently and hedgehog quantity was not specified. The number of analysed comets is not specified. Comet tail lengths were used as endpoint for DNA damage analysis, however, it is recommended by the OECD to use %tail DNA Lung inflammation was concurrently observed for NanoTiO2, indicated by significantly increased neutrophil levels. Inflammation and subsequent issue damage may have potentially interfered with the comet assay and led to false positives. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Sychewa et al. (2011) conducted a micronucleus assay in male mice after administration of titanium dioxide via gavage at doses of 40, 200, 1000 mg/kg bw/day for consecutive 7 days. The publication shows reporting and experimental deficiencies. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide. Some experimental procedure were insufficiently described, some procedures were not guideline compliant/deviate from established procedures e.g. lack of fixation, unclear staining procedure, unclear scoring for cytotoxicity. Experimental procedure for cell isolation from brain tissue and liver is missing. Unclear how the mitotic index was determined in forestomach and colon epithelial cells. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Sychewa et al. (2011) conducted a comet assay in male mice after administration of titanium dioxide via gavage at doses of 40, 200, 1000 mg/kg bw/day for consecutive 7 days. The publication shows reporting and experimental deficiencies. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide. Some experimental procedure were insufficiently described, some procedures were not guideline compliant/deviate from established procedures e.g. lack of fixation, unclear staining procedure, unclear scoring for cytotoxicity. Experimental procedure for cell isolation from brain tissue and liver is missing. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

The in vivo data for pigment grade titanium dioxide present a consistent pattern. In none of the relevant and reliable studies using physiological route of application, show an increase of chromosomal aberrations or micronuclei. The analysis of titanium concentration in liver and blood after single oral application up to the limit dose of 2000 mg/kg bw indicates a lack of absorption and systemic availability of titanium dioxide. It is therefore concluded that titanium dioxide does not cause clastogenic or aneugenic events in animals.


 

Germ cell mutagenicity studies with ultrafine titanium dioxide

A large number of references/publications dealing with the in vitro genetic toxicity potential of engineered nanomaterials were identified. All references were rated based on minimum criteria for genetic toxicity tests with engineered nanomaterials, as recommended in various recent reviews on the applicability of standard OECD test methods for poorly soluble nano particles (Greim 2010, Warheit 2010, Landsiel 2009). References not fulfilling the minimum quality criteria are reported in summary entries for information purposes, presenting some essential details only.

The comprehensive NanoGenotox testing programme up to six different titanium dioxide nano materials were assessed for their mutagenicity and genotoxicity in a large number of in vitro and in vivo assays. The outcome of this programme is discussed below in a separate chapter. The results of that testing programme are reported in individual robust study summaries for all test system and test item combinations.

 

 

In vitro genetic toxicity tests

Gene mutation in bacteria

Landsiel, R., et al. (2010) investigated the mutagenic potential of two coated ultrafine titanium dioxide samples (1: Titanium dioxide, aluminium hydroxide, dimethicone/methicone copolymer, purity of titanium dioxide core: ≥99% rutile phase, titanium dioxide -content: 79–89%; 2: Titanium dioxide, dimethoxydiphenylsilane/ triethoxycaprylylsilane crosspolymer, hydrated silica, aluminium hydroxide, purity of titanium dioxide core: ≥99% rutile phase, titanium dioxide -content: 69–73%) in the bacterial reverse mutation test according OECD 471 and under GLP. S. typhimurium TA 1535, TA 1537, TA 98, TA 100 and TA 102 were exposed with concentrations of 20, 100, 500, 1000, 2500 and 5000 µg per plate using the plate incorporation and plate incubation protocol. Both samples did not show mutagenic activity in the standard plate incorporation nor in the preincubation assay with and without exogenous metabolic activation tested up to 5,000 mg per plate.

Ford, L.S. (2006) investigated the mutagenic potential of ultrafine titanium dioxide (87% rutile, 13% anatase phase) in the bacterial reverse mutation test according OECD TG 471 and under GLP. S. typhimurium TA 1535, TA 1537, TA 98 and TA 100; E. coli WP2 uvr A were exposed using the plate incorporation and preincubation method at doses of 100, 333, 1000, 3333, and 5000 µg/plate. Under the conditions of this study, ultrafine titanium dioxide showed no evidence of mutagenicity in the bacterial reverse mutation test either in the presence or absence of S9 activation. Appropriate positive controls demonstrated to sensitivity of the test system.

There was no evidence for mutagenic activity with titanium dioxide in bacterial reverse mutation tests using the plate incorporation and pre-incubation method with and without metabolic activation employed at concentrations up to 5 mg/plate. Results were obtained using pure rutile phase or a mixed rutile/anatase phase titanium dioxide particles. The experiments were conducted with 5 different S. typhimurium and 1 E. coli strain.

 

In vitro mammalian chromosome aberration/micronucleus test

Riley, S. (1999) investigated the induction of chromosomal aberrations in CHO cells by trimethoxy caprylylsilane coated ultrafine titanium dioxide (crystal phase: 85% anatase, 15% rutile, 96.5% pure, 7.5% coating). The study was conducted according to OECD 473 and under GLP. Cells were exposed to, in experiment 1: 68.72, 209.7 and 800 µg/mL (-S9) and 167.8, 640, and 800 µg/ml (+S9); experiment 2: 167.8, 512 and 800 µg/ml (+S9). The metabolic activation in experiments 1 and 2 was derived from Aroclor 1254 and ß-naphthoflavone induced male rats, respectively. The highest concentration induced approx. 60% and 29% reduction in cell number in the absence and presence of metabolic activation, respectively. Treatment of cultures with titanium dioxide in the absence and presence of S9 resulted in frequencies of cells with structural aberrations that were similar to frequencies seen in concurrent solvent control cultures. Aberrant cell frequency in the majority of concurrent solvent control cultures and of T 805 treated cultures fell with the historical negative (normal) control range

Mixed phase ultrafine titanium dioxide (89.3% titanium dioxide (89.3% rutile, 10.7% anatase) 7.2% aluminium hydroxide 1.1% amorphous silica 1.8% trimethylolpropane 0.6% Citric acid) was tested by Donner, E.M. (2006) for the induction of structural and numerical chromosomal aberrations in CHO cells. In the main experiments the test item was applied at concentrations of:

• 125, 250, 750, 1250, and 2500 μg/mL (4h, -S9), three top doses used for CA analysis

• 62.5, 125, 250, 500, 700 μg/mL (4h, +S9), lowest three concentrations used for CA analysis

• 6.25, 12.5, 25, 50, 100 μg/mL (20h, -S9), three top doses used for CA analysis.

Cultures at 4 with and 20 hours without metabolic activation were limited by cytotoxicity. Under the conditions of this study, titanium dioxide did not induce structural or numerical chromosome aberrations in the in vitro mammalian chromosome aberration test in Chinese hamster ovary cells in either the non-activated or activated test conditions.

Landsiel, R., et al. (2010) investigated the clastogenic and aneugenic potential of coated ultrafine titanium dioxide (Titanium dioxide, aluminium hydroxide, dimethicone/methicone copolymer, purity of titanium dioxide core: ≥99% rutile phase, titanium dioxide-content: 79–89%, surface area (BET): 100 m²/g) in the in vitro mammalian cell micronucleus test according to OECD TG 487 and under GLP. V79 cells were exposed for 4 and 24 hours to concentrations of 75, 150, and 300 µg/mL and 18.8, 37.5, and 75.0 µg/mL, respectively. Quadruplicate cultures were prepared and 2,000 cells were analysed for micronuclei for each test group. Toxicity indicated by strongly reduced quality of the cells was observed from 625 µg/mL onward after 4 h treatment and from 156.3 µg/mL onward after 24 h treatment. In addition, the occurrence of precipitation on the slides strongly influenced the toxicity assessment from 625 µg/mL onward. Ultrafine titanium dioxide did not increase the number of cells containing micronuclei up to the highest evaluable concentrations of the test substance, thus is considered not to be clastogenic or aneugenic substance under the experimental conditions of this study.

Jugan, M.L. et al. (2012) investigated the clastogenic and aneugenic potential of two ultrafine titanium dioxide samples (A12: self-synthesized, 95% anatase, mean diameter: 12 nm, specific surface area: 92 m²/g; A25: Aeroxide P25, 86% anatase, mean diameter: 24 nm, specific surface area: 46 m²/g) in the in vitro mammalian cell micronucleus test according to OECD TG 487. A549 cells were exposed for 24 hours to concentrations of 50, 100 and 200 µg/mL in triplicate, followed by 24 hours cytoB treatment. Cytotoxicity, measured by MTT and colony forming ability, was measured after 48 hours exposure, showing no excessive toxicity up to 100 µg/mL. Approximately 1500 nuclei were scored per slide, showing a decrease in the portion of binucleated cells compared with the negative control. The number of clastogenic and aneugenic events did not increase for any of the concentrations tested.

Catalán, J. et al. (2012) investigated the clastogenic potential of ultrafine titanium dioxide (anatase phase, 99.7% pure, BET surface area: 222 m²/g) in human lymphocytes, in accordance with OECD TG 473. Cells were exposed at concentrations of 6.25, 12.5, 25, 50, 100, 150 and 300 µg/mL. The nanomaterials were ultrasonicated prior to exposure, to ensure maximum dispersion. Cells were exposed for 24, 48 and 72 hours without metabolic activation. The mitotic indices were not significantly affected by the nanomaterials at any exposure time. Two hundred metaphases were scored per dose and replicate (100 cells/ replicate culture), making a total of 400 cells for each experimental point. Authors conclude a dose-dependent increase at 100 and 300 μg/mL for all aberrations without gaps after 48 hours exposure. However, these isolated statistical positive finding were caused by the low aberration frequency in the control culture at that exposure time. A dose-response relationship has not been demonstrated at this experimental time, and no dose-dependent or statistically significant increase has been observed after 24 or 72 hours exposure. High background levels of %cells with aberrations in the control cultures over all experiments raises doubts on the suitability of the test system - %CA without gaps reached up to 3%, being 3fold above the expected value. It is therefore concluded that ultrafine titanium dioxide (anatase phase) does not induce chromosomal aberrations in human lymphocytes. Due to the use of an unsuitable test system, the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

The in vitro micronucleus experiment published by Kansara (2015) exhibits significant shortcomings putting into question the results of this study. The exposure concentrations between 25 and 100mg/L exceeded the solubility of titanium dioxide more than 4000 to 16,000 fold. Cytotoxicity was measured using the colourimetric MTT assay – any interference with the test item was not determined, thus underestimation of the cytotoxicity cannot be excluded. Further, the measurement of cell viability using the MTT assay is considered unsuitable when using PSLTs, CBPI is considered the more appropriate method. Cells were harvested and cleaned via centrifugation, which does not separate the test item from the cells, thus exposure duration becomes ill defined. In addition to these shortcomings, further absence of mycoplasma infection, cell culture conditions were not reported, cell proliferation/cytotoxicity (CBPI or RI from at least 500 cells) was not stated, test item insufficiently described. Historical control for the negative and positive control data are not reported.

 

Browning et al. (2014) examined the potential of titanium dioxide P25 nanoparticles to induce chromosomal aberrations in primary and immortalised human skin fibroblast cells (BJ cells). The cytotoxicity and genotoxicity testing showed no statistically significant effects at any concentration of 10, 50 and 100 µg/cm² compared to the negative control. The information does not allow an independent assessment due to major methodical and reporting deficiencies and is not in accordance with any guideline. The justification for dose selection is not given by the authors and the choice of the cell lines is not justified. Moreover the chosen cell model is inappropriate for the assessment of TiO2 nanoparticles effects to the skin. Furthermore, since the criteria for the chromosomal aberration analysis is not given and the number of metaphases counted was to low (in comparison with test guideline OECD 473 In vitro mammalian chromosome aberration assay), it is not possible to draw an independent conclusion on the effects of the test substance on the treated cells. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Stoccoro et al. (2017) The aim of the study was to investigate potential chromosomal damage caused by treatment with different types of TiO2 nanoparticles and to reveal differences in the toxic properties of coated vs non-coated nanoparticles. Human alveolar type II-like epithelial A549 cells were treated with 10, 20, and 40 µg/cm² (32, 64, and 128 µg/ml) TiO2 nanoparticle suspensions for 48 h. Afterwards, a micronucleus test was performed and 1000 binucleated A549 cell were scored for the occurrence of micronuclei, nuclear buds (NBUD), and nucleoplasmic bridges (NPB). Cytotoxic and cytostatic effects were measured by analysing clone formation efficiency (CFE), cytokinesis-block proliferation index (CBPI), replication index (RI), apoptotic cell ratio, and necrotic cell ratio. The study shows serious shortcomings in study design and reporting: Statements on the test material are partially missing. Purity of the test material is not stated. No statement on osmolality of the administered TiO2 nanoparticle suspension in cell culture. Details on the micronucleus test method are missing (slide preparation, cell scoring system, and DNA counterstaining). Essential details on cell culture are unclear or largely missing. The authors do not mention whether the test was conducted with or without exogenous metabolic activation of the cells. Cell harvest times after treatment are not specified. Evaluation criteria of the micronucleus test are unclear or missing. The authors provide no information on scorable or non-scorable cells. The results section shows some deficiencies in reporting. Historical data are missing. Individual culture data are not presented and the data are not summarised in tabular form. It is stated that TiO2 nanoparticle resulted in increased cytotoxicity and necrosis rates. Cytotoxicity could have had interfered with the test leading to false positives. Further, the authors analysed 1000 binucleated cells. However, the OECD recommends the analysis of at least 2000 binucleated cells. The authors do not state on dose-response relationship of the micronucleus occurrence after TiO2 treatment. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Based on the reliable studies summarised above, ultrafine titanium dioxide does not induce clastogenic or aneugenic mutations in mammalian cells up to concentrations limited by toxicity or precipitation. Consequently, it is concluded that ultrafine titanium dioxide does not induce chromosome of genome mutations in mammalian cells.

 

DNA damage assays

Jugan, M.L. et al. (2012) investigated the DNA damaging potential of five ultrafine titanium dioxide samples via comet assay: TiO2-A12 (95% anatase, mean diameter: 12 nm, specific surface area: 92 m²/g, spherical shape), TiO2-R20 NPs (90% rutile, mean diameter: 21 nm, specific surface area: 73 m²/g, spherical shape), TiO2-A25 (AEROXIDE® P25, 86% anatase, mean diameter: 24 nm, specific surface area: 46 m²/g, spherical shape), TiO2-R68 (elongated shaped 100% rutile, mean diameter: L=68 nm D=9 nm, specific surface area: 118 m²/g), TiO2-A140 (100% anatase, mean diameter: 142 nm, specific surface area: 10 m²/g). Genotoxicity was assessed by using the alkaline comet assay after exposure for 4, 24 or 48 h to 100 µg/mL ultrafine titanium dioxide in A549 cells. Comets were scored on at least 50 cells per slide. Results are presented as median of % DNA in comet tail ± standard deviation. After 4 h of exposure, whatever the TiO2-NPs, a significant increase in the level of DNA breaks was observed. This increase in the level of breaks further increased after 24 h of exposure, still it was statistically significant only after exposure to TiO2-A12, -A25 and -R20, but not to TiO2-R68 and -A140. After 48 h of exposure, the frequency of breaks drastically decreased in exposed cells. It reached the basal level, i.e., the level of DNA breakage observed in unexposed cells, in all samples except in cells exposed to TiO2-A12, where a significant DNA fragmentation was still observed. In a parallel experiment, oxidative DNA damages by detection of 8-oxodGuo lesions were observed, showing a similar pattern as the comet assay. Authors hypothesized that the observed DNA damages were (i) caused by ROS being formed by an unknown mechanism, state however that a direct mechanism can be excluded. The DNA damages observed after 4 and 24 hours abated after 48 hours to a level comparable with the negative control. Consequently, the DNA damage observed after 4 and 24 hours is a reversible effect, which was not detectable after 48 hours via comet assay or via micronucleus test (see above). Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Petković, J. et al. (2011) investigated the DNA damaging effect of ultrafine titanium dioxide in human hepatoma cells (HepG2). Two different samples were used: TiO2-An: particle size <25 nm, 99.7% pure, specific surface area (BET): 129.3 m²/g, particle shape: spherical crystallites, XRD: Anatase type; TiO2-Ru: particle size ~10x40 nm, 99.5% pure, specific surface area (BET): 116.7 m²/g, particle shape: elongated crystallites, XRD: Rutile type. Cells were exposed at concentrations of 0, 1, 10, 100 and 250 µg/mL for 2, 4 and 24 hours. In each experiment positive controls (0.5 mM t-BOOH and 50 mM BaP) and a vehicle control (cell growth medium containing 10% PBS) were included. 50 randomly selected nuclei per experimental point were analysed. Three independent experiments were performed for each of the treatment conditions. The percent of tail DNA was used to measure the level of DNA damage. Cytotoxicity was assessed in a parallel experiment via MTT assay. The weak statistical significant findings for ultrafine anatase sample do neither show a time nor a dose dependency. The result on DNA damage for ultrafine anatase is therefore of questionable biological relevance. Ultrafine rutile did not show an increase of DNA damage.

 

The publication by Kansara (2015) exhibits significant reporting deficiencies, unclear experimental conditions - culture concentrations were given as µg/mL, which is pointless for adherent growing cells - basic information on the cell lines were not reported, i.e. stability of modal chromosome number, cell doubling time and absence of mycoplasma infection - cell seeding conditions were not reported - information on the usage of metabolic activation was missing - cell proliferation/cytotoxicity (CBPI or RI from at least 500 cells) was not stated - information on pH missing - negative control was not clearly described (vehicle, medium, untreated) - historical control data for the negative and positive control is missing - individual data was not presented. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Stoccoro et al. (2017) investigated DNA damage caused by treatment with different types of TiO2 nanoparticles and to reveal differences in the toxic properties of coated vs non-coated nanoparticles. A549 cells were treated with 10, 20, and 40 µg/cm² (32, 64, and 128 µg/ml) TiO2 NP suspensions for 48 h. Afterwards, cells remain untreated or were treated with either Fpg or Endo III to detect oxidized bases. A comet assay was performed and % tail DNA was determined. Colony forming efficiency (CFE) was determined as measure of cytotoxicity. Further, ratios of apoptotic and necrotic cells were determined. According to the authors, with TiO2 Aeroxide P25 showed the strongest capacity to induce DNA damage characterized by a dose-dependent and significant increase of % tail DNA at all concentrations tested. Pristine and silica coated TiO2 NPs showed a significantly increased DNA damage at higher concentrations. Treatment with silica-coated TiO2 NPs resulted in no significant induction of DNA damage. Concurrently, TiO2 showed a cytotoxic capacity characterized by a slight but significant decrease of the CFE at all concentrations (1.25, 2.5, 5, 10, 20, 40, and 80 µg/cm²) and exposure durations (24, 48, and 72 h), except for 1.25 and 10 µg/cm² after an exposure period of 48 h. The study has several shortcomings in study design and reporting. Statements on the test material are partially missing. Purity of the test material is not stated. No statement on osmolality and pH of the administered TiO2 nanoparticle suspension in cell culture. The methods section is described insufficiently. It was not specified whether an alkaline or neutral comet assay was performed. Harvest times after treatment are not specified, modal chromosome number, cell proliferation rate, and number of passages are not stated. Number of scored cells, hedgehog number is not specified and not historical control data were provided. The description of statistical analysis is insufficient. The authors do not specify whether the data sets were tested for normal distribution or homogeneity of variances. The authors state that TiO2 nanoparticle resulted in increased cytotoxicity and necrosis rates. Noteworthy, that cytotoxicity is a confounder of comet assay, leading to false positive findings. Number of analysed cell is too low the OECD test guideline for in vivo testing recommends scoring of at least 150 randomly selected cells. Further, the author used lesion-specific endonucleases to reveal oxidative DNA damages induced by TiO2 nanoparticles. However, it is reported that treatment with these exogenous enzymes increases the inter-sample-variability. Thus, positive findings provide only limited data for a hazard conclusion. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

El Yamani et al. (2017) tested the in vitro genotoxicity of four different nanomaterials by using a high throughput system, which should improve the output of comet assays for risk assessment purposes. A549 and TK6 cells were treated with TiO2 NM100 at 10 different concentrations for 3 or 24 h. An alkaline comet assay was performed with cells treated with or without Formamidopyrimidne-DNA Glycosylase (Fpg) enzyme, which detects oxidised bases. Afterwards, % tail DNA was determined. Cytotoxicity was determined using colony forming efficiency (CFE) assay and the colorimetric AlamarBlue assay. According to the authors, 3 h treatment with TiO2 NM100 suspensions resulted in a dose-dependent increase of DNA damage in both cell lines. The same was observable when using the lesion-specific Fpg endonuclease, indicating oxidation of purines. However, DNA damage was strongly reduced at 24 h of exposure. The DNA damaging capacity was generally higher for A549 cells than for TK6 cells. There was no significant cytotoxicity observable for A549 cells independent of the concentration and assay used. The study shows several shortcomings in study design and reporting. The test material is not characterised sufficiently. Purity and impurity elements of the test material are not stated. Information on comet assay evaluation criteria are unclear or missing. The authors do not state on scorable vs. non-scorable cells and hedgehog occurrence is not stated. Details on statistical analysis are missing. Performance of a trend test was not specified. There are no statements on the normality of homogeneity of variances of the analysed data. Historical data are missing. A vehicle control was not used in the assay. Cytotoxicity assays were only performed for the A549 cells but not for the TK6 cells. Measurement of cytotoxicity is essential to interpret the findings, since cytotoxicity is reported to be a confounder of comet assay potentially leading to false positives. Further, the author used lesion-specific endonucleases to reveal oxidative DNA damages induced by TiO2 nanoparticles. However, it is reported that treatment with these exogenous enzymes increases the inter-sample-variability. Thus, positive findings provide only limited data for a hazard conclusion. Only treatment with 140 µg/ml TiO2 NM100 resulted in a significantly increased DNA damage only in TK6 cells after 3 h of exposure. Such isolated finding at one concentration and one time point is likely to be a chance finding and should have been verified in a second experiment. The authors do not state on pH effects or effects on osmolality of the treatment suspension in cell culture. Results of CFE assay are shown as negative control-normalised values. The displayed results for TiO2 NM100 treated cells show significant cytotoxicity at the lowest concentration tested. However, the cytotoxicity decreases with increasing TiO2 NM100 concentrations, which is implausible. Number of analysed cell is too low, the OECD test guideline for in vivo testing recommends scoring of at least 150 randomly selected cells. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Haase et al. (2017) determined the genotoxic capacities of 16 nanomaterials using an in vitro 3D human bronchial tissue model. The model was treated with a TiO2 NM-105 suspension for 60 h at a concentration of 50 µg/cm². An alkaline comet assay was performed and DNA damage measured by % tail DNA was determined. Cytotoxicity was determined by analysing the LDH activity and ATP vs total protein ratio. According to the authors, they were not able to detect increased DNA damage. Cytotoxic effects were observed in the ATP, but not in the LDH assay. However, the authors stated that the ATP showed strong variability and needs to be optimized. 

 

Merely one reliable reference is available reporting on DNA damage in mammalian cells after exposure towards ultrafine titanium dioxide. The work by Petkovic shows an increase of DNA damage in mammalian cells after treatment with ultrafine anatase with no dose- or time-response relationship. The presence of oxidative DNA lesions indicate an indirect mode of action. Whether this was cause by ROS formed on the surface of the particles by light/UV-irradiation remains unclear. The lack of dose or time dependency also remains unexplained. Based on the given information it is concluded that ultrafine titanium dioxide does not cause significant and persistent DNA damage to an extent which is considered relevant for the human health hazard assessment.

 

In vivo genetic toxicity tests

In vivo chromosome aberration/micronucleus tests

In a testing programme, a series of three different ultrafine titanium dioxide samples were tested for the induction of micronuclei in the peripheral blood of rats after single oral administration (Myhre, A., 2014b and 2014c, Wessels, A., 2014c). Animals received doses of 0, 500, 1000, 2000 mg/kg bw. All tests were conducted in accordance with OECD TG 474/EC B.12 and under GLP:

In the study by Myhre, A. (2014b and 2014c) two ultrafine titanium dioxide samples (TiO2uf-1: 89.3% anatase phase, 10.7% rutile phase, D50(TEM ECD)=19 nm, SA: 50.4 m²/g; TiO2uf-3: rutile phase, D50(TEM ECD)=22 nm, SA: 58.8 m²/g) were used as test item. The vehicle control, low, intermediate, and positive control groups contained 5 animals/sex, the high-dose group contained 7 animals/sex. Peripheral blood samples were taken via sublingual vein bleeding approx. 48 and 72 hours post exposure. A total of 20,000 RETs were analysed per blood sample via flow cytometry for the presence of micronuclei and toxicity as indicated by the frequency of immature erythrocytes among the total erythrocytes (PCE/NCE ratio). Additionally, titanium concentrations were determined in blood and liver at 48 or 72 hours. One mid-dose animal was accidentally killed on TD3 prior to blood collection and was therefore not considered in the MN analysis (Myhre 2014c, TiO2uf-3). This has no impact on the outcome or validity of this study since it was a mid-dose animal and there is no indication of any test substance related effects in this or any other test substance treated group. No statistically significant increases in the micronucleated RET frequency were observed in any evaluated test substance-treated group of male or female animals at either time point. There were no statistically significant decreases in the PCE/NCE ratio in any test-substance treated group. Under the conditions of this study, titanium dioxide did not induce formation of micronuclei in rat peripheral blood up to the limit dose of 2000 mg/kg bw. No discernible dose-dependent increases of titanium concentrations in the blood and liver of treated rats relative to control rats could be measured.

In the studies by Wessels, A. (2014c) ultrafine titanium dioxide (TiO2uf-2) was used: anatase phase, D50(TEM ECD)=19 nm, SA: 82 m²/g. All dose groups contained 5 animals/sex. Peripheral blood samples were taken via tail vein approx. 48 and 72 hours post exposure. A total of 20,000 RETs were analysed per blood sample via flow cytometry for the presence of micronuclei and toxicity as indicated by the frequency of immature erythrocytes among the total erythrocytes (PCE/NCE ratio). No statistically significant and dose-dependent increases in the micronucleated RET frequency were observed in any evaluated test substance-treated group of male or female animals at either time point. Two isolated statistical significant increased were seen for males at 2000 mg/kg bw and for females at 500 mg/kg bw. Since these findings (i) do not show a dose response relationship (ii) are within the range of the historical negative control data, they are not considered biologically relevant. There were no statistically significant decreases in the PCE/NCE ratio in any test-substance treated group. Under the conditions of this study, titanium dioxide did not induce formation of micronuclei in rat peripheral blood up to the limit dose of 2000 mg/kg bw.

Sadiq, R. et al. (2012) investigated the induction of micronuclei in the peripheral blood of male B6C3F1 mice. Five animals per dose group were exposed via intravenous injection of three doses (0.5, 5, 50mg/kg/day) ultrafine titanium dioxide (anatase, primary particle size: 12 nm, aggregate size in medium approx. 170 nm) over three consecutive days, the maximum dose was limited by the feasibility to administer higher doses by that method. Concurrent positive and negative control groups were used. The frequency of micronucleated reticulocytes was determined by the acquisition of approximately 20,000 C71-positive reticulocytes for each animal. Bone marrow exposure was demonstrated by toxicity and elevated titanium levels in the bone marrow tissue, analysed via ICP-MS (fold increase over control ranged from 12.1 to 14.2). Titanium dioxide nanoparticles were not clastogenic or aneugenic in the micronucleus assay in mice at the dose levels showing target organ exposure.

 

Sychewa et al. (2011) conducted a micronucleus assay in male mice after administration of titanium dioxide via gavage at doses of 40, 200, 1000 mg/kg bw/day for consecutive 7 days. The publication shows reporting and experimental deficiencies. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide. Some experimental procedure were insufficiently described, some procedures were not guideline compliant/deviate from established procedures e.g. lack of fixation, unclear staining procedure, unclear scoring for cytotoxicity. Experimental procedure for cell isolation from brain tissue and liver is missing. Unclear how the mitotic index was determined in forestomach and colon epithelial cells. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

The micronucleus study by Dobrzynska (2014) exhibits significant methodological deficiencies, intravenous administration is an unphysiological route of application and bears only very limited value for risk assessment purposes. The number of PCE analysed is too low, thus statistical robustness of the results is impaired. No positive control was used, thus sensitivity of the system was not demonstrated. Historical control data are not reported. No information on tissue toxicity was reported, thus it remains questionable whether positive findings were artefactual due to overt toxicity or a true genotoxic response. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

In the study by Relier et al. (2017) male Sprague-Dawley rats were intratracheally instilled three times, over a period of eight days, with a TiO2 nanoparticle suspension with concentrations of 0.17, 0.83, and 3.33 mg/mL which resulted in doses of 0.5, 2.5, and 10 mg/kg. A micronucleus assay was conducted with extractions from bone marrow before treatment, 2 hours and 35 days after treatment. 2000 PCE were scored for micronuclei frequency. Cytotoxicity was assessed by analysing the relative number of immature erythrocytes. According to the authors, the number of micronuclei was significantly increased after 35 days at all three doses, in a dose-independent manner. However, it was also stated that the observed changes were slight and might therefore be regarded as biologically irrelevant. The study shows inconsistencies in reporting, study design shortcomings and results, which are not biologically meaningful. The authors state that the rats were instilled three times with 100 µl TiO2 nanoparticle suspensions with a concentration of 0.17, 0.83, and 3.33 mg/mL allegedly resulting in total doses of 0.5, 0.25, and 10 mg/kg. Recalculating the (not reported) bodyweight of the animals, based on the given parameters (i) instillation volume, (ii) concentration of the test item in the vehicle and (iii) the resulting dose, would result in a total body weight of 102, 99.6 and 99.9g for the low, mid and high-dose animals. A typical weight of a 10 weeks old, male Sprague-Dawley rat (gender, species, and age used in this study) is about 300 g. It is therefore unclear whether the authors applied wrong doses due to a calculation error or whether the animals were of a different age (for 100g male Sprague-Dawley rats one would assume 4-5 week old animals). The authors state that they triggered hyperventilation for a better distribution of the test substance in the lungs. However, they make no statements on the induction process per se. They do not state on potential side-effects of this treatment on the results. This treatment further increases the artificiality of the exposure route and renders the study of little relevance for risk assessment purposes. Authors make no statements on the animal weights before or during the study. Furthermore, data on housing, clinical observations, environmental conditions, test group randomization, euthanasia, and food consumption are missing. The preparation of the bone marrow is insufficiently described. Evaluation criteria for the micronuclei assay are not described. Historical data are missing, the historical data for rat vehicle controls of other reputable laboratories show a range of 0.03-0.22% in micronucleation frequency. The MN frequencies measured over all doses and time points range between approx. 0.05 and 0.15% and therefore within the historical negative control range. Due to a lack of a dose response relationship and a MN frequency, which is within the historical control range, the significantly increased MN frequencies should have been considered as chance finding with no biological relevance. Based on the above-mentioned severe shortcomings, the reference is considered not reliable and was disregarded in the hazard assessment.

 

Troullier et al (2009) treated male C57Bl/6Jpun/pun mice for five days with P25 TiO2 nanoparticles suspended in drinking water at concentrations of 60, 120, 300, and 600 µg/ml (total dose after five days: 50, 100, 250, and 500 mg/kg). A micronuclei assay was conducted with peripheral blood taken before and after treatment. Approximately 2000 erythrocytes were scored for micronuclei occurrence. According to the authors, micronuclei frequency was significantly increased only at the highest TiO2 nanoparticle dose (500 mg/kg), when compared to untreated controls. The study shows serious reporting and experimental shortcomings. The actual dose received was calculated on an average 30g body weight and an average of 5 mL water intake. This is an unjustified simplification, since the bodyweight for 12-20 week old animals differs significantly between e.g. male and female animals, also the average bodyweight of a 12-20 week old female mouse ranges between 21 and 24g. Such dose calculation. The author state that four to five months old mice were used. The current OECD test guideline recommends using young animals at an age of six to ten weeks. The authors do not justify selected species, strains, and dosing. The transgenic C57Bl/6Jpun/pun mouse is an uncommon strain for this assay, with no historical data available. Due to the very low solubility and the difference in density of a factor of 4, one have to assume that titanium dioxide settles on the bottom of the flask, leading to a higher total dose. The authors verified neither the actual dose received nor the homogeneity of the suspension before during and after the dosing of the animals using an appropriate analytical method. Authors provide test results a bar chart in which the MN results of the four dose groups were presented parallel to the control groups. From the different height of the control bars one has to assume that also four different control groups (with n=5) were used. This is a disproportionate use of animals and is not in-line with the ethical principles in toxicological testing practice. The MN frequency for untreated mice was given in the text as 4.3 ±0.93/2000 cells, whereas the bar chart reports approx. 4.3/1000 cells. In case the latter value it the true measured frequency, it would be above reported control data for mice, which is 2.5 ±0.46/1000 PCE (Zeiger et al. 2009). Cytotoxicity measurements were not performed, which does not allow a conclusive statement, whether a true test item specific effect was observed or an artefactual response due to increased cytotoxicity. Clinical observations were not reported. Time points of blood sampling and analysis, the proportion of immature erythrocytes to total erythrocytes is not specified. The authors conducted a student’s t-test to test for significant changes before and after treatment. However, the student’s t-test is not a suitable for paired data. Individual animal values and results from the lower doses (50, 100, 250 mg/kg) are not specified. Further, the authors state not on a potential dose-response relationship. Moreover, the author state that the highest TiO2 nanoparticle dose leads to an inflammatory response, measured via qRT-PCR of pro-inflammatory marker gene expression. Inflammatory responses could have interfered with the results and could have led to false positive findings in the genotoxicity assays. Based on the above-mentioned shortcomings, the reference is rated not reliable and disregarded for the hazard assessment.

 

The reliable in vivo data for ultrafine titanium dioxide present a consistent pattern. In none of the studies administering titanium dioxide via physiological route in rats or mice an increase of micronuclei was observed. The maximum doses/concentrations were limited by technical feasibility or the limit as foreseen in the guideline. It is therefore concluded that titanium dioxide does not cause local or systemic clastogenic or aneugenic events in animals.

 

DNA damage assays

Two comet assays investigated the DNA damage effect of nano-sized titanium dioxide following intratracheal instillation (Naya et al., 2012) or inhalation (Landsiel et al., 2010).

In the study by Naya et al. young male Crl:CD(SD) rats were instilled with a single dose of 1 and 5mg/kg or with a repeated dose of 0.2 and 1 mg/kg once a week for 5 weeks with ultrafine titanium dioxide (primary particle size: 19 nm, anatase phase, 99.99% pure, surface area (BET): 316 m²/g). The histopathological findings 3 and 24 hours after single instillation showed a slight infiltration of alveolar macrophages and/or neutrophils at 5 and 1 and 5 mg/kg dose group, respectively. For the repeated treatment, a similar inflammatory response was observed 3 hours after final dosing only in the high-dose group animals. No DNA damage has been observed via alkaline comet assay in the lungs epithelial cells, after scoring 100 cells per rat. A positive control substance induced significant DNA damage after 3 hours, demonstrating the sensitivity of the test system. Consequently, ultrafine titanium dioxide does not induce DNA damage after single or repeated intratracheal instillation in rats. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide.

In the study by Landsiel et al. young male Crl:Wi Han rats were exposed by inhalation to 10 mg/m³ of coated ultrafine titanium dioxide (Titanium dioxide, aluminium hydroxide, dimethicone/methicone copolymer, purity of titanium dioxide core: ≥99% rutile phase, titanium dioxide-content: 79–89%, surface area (BET): 100 m²/g). The alkaline comet assay was performed in alveolar lavage cells three weeks after exposure. Compared with the negative control no increase in the tail intensity, tail length or tail moment, nor in the hedgehogs formed upon inhalation of ultrafine titanium dioxide was observed. Consequently ultrafine titanium dioxide does not induce DNA damage after inhalation exposure in rats.

The work by Lindberg et al. investigated the induction of local or systemic genotoxic events in C57BL/6J mice after inhalation of self-synthesised ultrafine titanium dioxide (74% anatase, 26% brookite, agglomerate size: 10-60 nm, average primary particles size: 21 nm, surface area (BET): 61 m²/g) at 0.8, 7.2, and (the highest concentration allowing stable aerosol production) 28.5 mg/m³. DNA damage was assessed by the alkaline comet assay in lung epithelial alveolar type II and Clara cells sampled immediately following the exposure. The percentage of DNA in the comet tail from 100 cells per animal (two replicates, 50 cells each) was used as a measure of the amount of DNA damage. MN were analysed by acridine orange staining in blood PCEs collected 48 h after the last exposure. Blood samples were collected from the tails of the mice and whole blood smears were prepared. The frequency of micronucleated erythrocytes in 2000 PCEs and in 2000 normochromatic erythrocytes per mouse was scored. Also the percentage of PCEs was assessed (scored until the number of PCEs or NCEs reached 2000) to indicate possible bone marrow toxicity. The experiments were conducted using self-synthesised material by thermal decomposition of titanium tetraisopropoxide. The reaction mixture was directly attached to the inhalation chamber with no purification step in between. One has therefore to assume that animals were exposed towards a mixture of reaction products and not a pure titanium dioxide sample. A dose-dependent deposition of ultrafine titanium dioxide in lung tissue was seen. Although the highest exposure level produced a clear increase in neutrophils in BAL fluid, indicating an inflammatory effect, no significant effect on the level of DNA damage in lung epithelial cells or micronuclei in PCEs was observed, suggesting no genotoxic effects by the 5-day inhalation exposure to ultrafine titanium dioxide (anatase phase). Consequently, ultrafine titanium dioxide does not induce DNA damage after inhalation exposure in rats. Based on the unclear exposure conditions and the test item being a reaction mixture rather than a pure substance, the study is considered not relevant and was disregarded for hazard assessment.

 

For the evaluation of the references by Saber et al. (2012a and b), please refer to the discussion of the in vivo studies with pigment-grade titanium dioxide presented above.

 

Sychewa et al. (2011) conducted a comet assay in male mice after administration of titanium dioxide via gavage at doses of 40, 200, 1000 mg/kg bw/day for consecutive 7 days. The publication shows reporting and experimental deficiencies. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide. Some experimental procedure were insufficiently described, some procedures were not guideline compliant/deviate from established procedures e.g. lack of fixation, unclear staining procedure, unclear scoring for cytotoxicity. Experimental procedure for cell isolation from brain tissue and liver is missing. Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

The micronucleus study by Dobrzynska (2014) exhibits significant methodological deficiencies, intravenous administration is an unphysiological route of application and bears only very limited value for risk assessment purposes. Samples were collected 24 hours, 1 week and 4 weeks after the single injection (OECD foresees at both 2 - 6 and 16 - 26 hours after a single administration). A signle dose was tested, which precludes a dose-response relationship analysis. No positive control was used, thus sensitivity of the system was not demonstrated. Historical control data are not reported. No information on tissue toxicity was reported, thus it remains questionable whether positive findings were artefactual due to overt toxicity or a true genotoxic response. The number of analysed cells was too low, individual results were not reported and acceptability and evaluation criteria not reported Based on the above-mentioned shortcomings the reference is considered not reliable and therefore disregarded for the hazard assessment.

 

Meena et al. (2014) report the induction of adverse effect in testes after i.v. injection of TiO2 NP. The results of this study cannot be evaluated, due to severe reporting deficiencies. The methods section describes the dissection and preparation of the brain, whereas the results section presented findings in the testes. Furthermore, the experimental design exhibits shortcomings, such as inphysiological route of application, testes cell type not characterised, gross necropsy not reported or performed, clinical signs not reported, positive control missing, number of cells analysed too low, individual results were missing. Consequently, the reference is rated not reliable and therefore disregarded for the hazard assessment.

 

In the study by Relier et al. (2017) male Sprague-Dawley rats were intratracheally instilled three times, over a period of eight days, with a TiO2 nanoparticle suspension with concentrations of 0.17, 0.83, and 3.33 mg/mL which resulted in doses of 0.5, 2.5, and 10 mg/kg. Comet assays were conducted with lung, peripheral blood, and liver cells extracted two hours post exposure, and 35 days post exposure. The median of 150 tail DNA percentages was calculated per rat. Six rats per dose group were analysed. Hedgehog percentages were analysed as indicators for cytotoxicity. According to the authors, treatment with two highest doses of TiO2 nanoparticles induced significant DNA damages in liver cells at both post-exposure time points. In lung cells, a biologically meaningful and significant increase was only observed after 35 days in both doses. In blood cells, TiO2 nanoparticles did not induce DNA damage at biological relevant levels, at any dose and point in time. Exposure with TiO2 did not result in increased hedgehog levels. The study shows inconsistencies in reporting, study design shortcomings and results which are not biologically meaningful. The authors state that the rats were instilled three times with 100 µl TiO2 nanoparticle suspensions with a concentration of 0.17, 0.83, and 3.33 mg/mL allegedly resulting in total doses of 0.5, 0.25, and 10 mg/kg. Recalculating the (not reported) bodyweight of the animals, based on the given parameters (i) instillation volume, (ii) concentration of the test item in the vehicle and (iii) the resulting dose, would result in a total body weight of 102, 99.6 and 99.9g for the low, mid and high-dose animals. A typical weight of a 10 weeks old, male Sprague-Dawley rat (gender, species, and age used in this study) is about 300 g. It is therefore unclear whether the authors applied wrong doses due to a calculation error or whether the animals were of a different age (for 100g male Sprague-Dawley rats one would assume 4-5 week old animals). The authors state that they triggered hyperventilation for a better distribution of the test substance in the lungs. However, they make no statements on the induction process per se. They do not state on potential side-effects of this treatment on the results. This treatment further increases the artificiality of the exposure route and renders the study of little relevance for risk assessment purposes. Authors make no statements on the animal weights before or during the study. Furthermore, data on housing, clinical observations, environmental conditions, test group randomization, euthanasia, and food consumption are missing. The authors report a significant increase of DNA damage in lung cells, after 35 days, at the highest doses. However, the reported significantly increased mean tail intensities are still in the acceptable range of vehicle controls (OECD Series on Testing and Assessment No. 196). A lack of historical data precludes evaluation of the reported results for the other organs. Labs proficiency in preparing and analysing tail DNA intensities in comet assay is a prerequisite for reliable data. It was shown that there is huge variability in tail intensities even within negative or positive controls (Tavares, A.M. et al. (2014); Nanogenotox Programme). This variability in combination with lack of controls and historical data renders positive findings of this assay of little relevance for risk assessment purposes. Analysis of hedgehog percentages as measurement for cytotoxicity is questionable, since their etiology and relation to cytotoxicity is not clarified so far. Only for the lung a second indicator, LDH level in BAL fluid, was used as measurement for cytotoxicity. The analysis of cytotoxicity is an important parameter, since cytotoxicity is known to be a potent confounder . Further, lung inflammation, indicated by significantly increased neutrophil levels, was observed concurrently at the highest dose. Inflammation could have potentially interfered with the comet assay and led to false positives in the lung test. Although the authors cite that Inflammatory cells have been shown to induce mutagenic effects after TiO2 inhalation exposure (Driscoll et al., 1997b), it was not taken into account in the overall conclusion.

 

Asare et al. (2015) administered male, wildtype and Ogg1(-/-) KO, mice via single intravenous TiO2 nanoparticle suspension injections at a dose of 2 mg/ml (final dose: 5 mg/kg). Extracted cells remained either untreated or were treated with Fpg enzyme. Comet assays were conducted with lung, liver, and testis cells extracted one and seven days post exposure. The median of 100 tail DNA percentages was calculated per mouse. Five to six mice per treatment were analysed. According to the authors, TiO2 nanoparticle-exclusive treatment does not result in a statistically significant increase of DNA damage after both post exposure periods. However, treatment of extracted cells with Fpg enzyme leads to a significant increase of DNA damage only in testis and after seven days post exposure. The study shows some study design shortcomings. The authors administered the test material intravenously into the tail vein. However, intravenous injection is an unphyisiological route of exposure and bears only very limited value for hazard assessment purposes. Furthermore, the study does not fulfil several requirements of the OECD test guideline 489, as follows: The authors analysed 100 comets, although a minimum of 150 are recommended. Only one does was used in the experiments, which precludes any conclusion on a dose-response relationship. The OECD guideline recommends a sampling period of two to six hours after the last treatment. Particularly in case of an intravenous administration, shorter sampling intervals should have been chosen due to a rapid systemic distribution. Animals were analysed one and seven days after the last treatment. Transient DNA lesion could have been lost during this period. Moreover, test material’s cytotoxicity and clinical observations of the test animals are not specified. Hedgehog occurrence is not specified. Furthermore, historical data is missing and the results of the positive control are not reported. Laboratory’s proficiency in preparing and analysing tail DNA intensities in comet assays is a prerequisite for reliable data. A huge variability in tail DNA intensities was shown even within negative or positive controls (Tavares, A.M. et al. (2014); Nanogenotox Programme). This variability in combination with lack of controls and historical data renders positive findings of this assay of little relevance for hazard assessment purposes. The test material was insufficiently characterised. Due to an absence of a clear designation, purity or impurity information, it remains unclear whether the test item was in fact of industrial origin and therefore of relevance for the hazard and risk assessment of titanium dioxide. Description of the methods is not sufficient, i.e. details on tissue preparation, and electrophoresis conditions are largely missing. Body weight, age of the wildtype mice, caging, acclimation and food consumption were not reported. Recalculating the animal weights on the basis of the given concentration of the suspension, total administered dose and the range of the administered volume, it is concluded that the body weight was ranging between 20-28 g (difference > 20%) per mice. However, the OECD recommends that the weight variation at study start should not exceed 20%. The authors show a significant increase of comet tail intensity for testis cells being treated with TiO2 nanoparticles and crude Fpg enzyme extraction. However, combined treatment bears only very limited value for risk assessment purposes. Based on the above-mentioned shortcomings the reference is rated not reliable and therefore disregarded for the hazard assessment.

 

The information (follow-up study of Heinrich et al. 1995), contained in the reference by Gallagher et al. (1994) represents in vivo DNA damage experiments in mammalian cells (32P-Postlabeling) with very limited value for risk assessment purposes. The study design described in the reference for detecting genotoxicity is not in accordance with any harmonised protocol (e.g. only one concentration tested) and is consequently inappropriate for risk and hazard assessment purposes under REACH (ECHA guidance R4 in conjunction with regulation (EC) 1907/2006, Annexes VII-X). The information contained therein were included for information purposes only.

 

Boisen et al. (2012) exposed pregnant C57BL/6 mice by whole-body inhalation to TiO2 UV-Titan L181 nanoparticles (~42.4 mg UV-Titan/m³) or filtered clean air on gestation days (GD) 8–18. Female C57BL/6 F1 offspring were raised to maturity and mated with unexposed CBA males. The F2 descendants were collected and ESTR germline mutation rates in this generation were estimated from full pedigrees of F1 female mice (UV-Titan-exposed F2 offspring). ESTR mutation rates were examined. ESTR mutation rates of 0.029 (maternal allele) and 0.047 (paternal allele) in UV-Titan-exposed F2 offspring were not statistically different from those of F2 controls: 0.037 (maternal allele) and 0.061 (paternal allele). The generated information is not required for building an expert judgement and further assessment of the test substance under REACH (ECHA guidance R4 in conjunction with regulation (EC) 1907/2006, Annexes VII-X) for the following reasons:

(i) Titanium dioxide nanoparticles are not able to pass the blood stream via the blood-air barrier after inhalation. The major amount of particles will be eliminated through muco-ciliary escalation and subsequent swallowing. It is questionable if any nanoparticles could ever reach the target tissue, particularly because the placental barrier existed as a second effective barrier in exposed maternal animals.

(ii) The experimental conditions / housing conditions of animals are insufficiently described, important data e.g. age of animals at study initiation and housing conditions are not specified.

(iii) The study design is not in accordance with any guideline and not performed under GLP directive. Testing only one concentration is uncommon and an analytical verification of the concentration is recommended.

(iv) small effective sample size and the very large maternal dose used in the study hamper interpretation

(v) the study does not investigate a direct mutagenic effect on exposed animals but rather epigenetic effects transmitted from the parent to the descendants. No plausible mechanism has been presented by the authors for the alleged epigenetic effects.

 

Jackson et al. (2013) time-mated female mice (C57BL/6Bom-Tac) were exposed to filtered clean air or approx. 42 mg/m³ UV-Titan on gestation day (GD) 8–18 for 1 h/day by a whole-body exposure. DNA strand breaks were examined in bronchoalveolar lavage (BAL) and liver cells of the time-mated mice (5 and 26–27 days after inhalation exposure), and in livers of the offspring (post-natal days (PND) 2 and 22) using the alkaline Comet assay. Furthermore, hepatic gene expression in newborns using DNA microarrays was investigated. No statistically significant genotoxic effect was measured in maternal exposed mice and offspring compared to the control (filtered air). The newborn offspring, irrespective of nanoparticle exposure, had higher levels of DNA strand breaks compared with the offspring at weaning (p-value = 0.001). Changes in gene expression in the offspring were observed and suggested to be a secondary effect to maternal inflammation after exposure. The generated information concerning genotoxicity and toxicogenomics is not required for building an expert judgement and further assessment of the test substance under REACH (ECHA guidance R4 in conjunction with regulation (EC) 1907/2006, Annexes VII-X) for the following reasons:

(i) The study design is not in accordance with any guideline. The acceptability criteria of the OECD test guideline 489 (in vivo mammalian alkaline comet assay)were not fulfilled since there is only one concentration tested, adequate number of cells (> 150 cells per slide) have not been analysed, a positive control was not specified in the results and the sampling time is not in accordance with the recommendation (2 - 6 hours after last treatment).

(ii) A significant dose-response related effect could not be observed (due to lack of concentrations tested).

 

Troullier et al (2009) treated male C57Bl/6Jpun/pun mice for five days with P25 TiO2 nanoparticles suspended in drinking water at concentrations of 60, 120, 300, and 600 µg/ml (total dose after five days: 50, 100, 250, and 500 mg/kg). An alkaline comet assay was conducted with peripheral blood taken before and after treatment. 150 to 200 comets per sample were randomly captured and scored for change in tail moment. According to the authors, tail moments of the comets were significantly increased after TiO2 nanoparticle treatment at the highest dose (500 mg/kg), when compared to untreated controls. Thus, the authors concluded that the test item is able to induce DNA strand breaks. The study shows serious reporting and experimental shortcomings. The actual dose received was calculated on an average 30g body weight and an average of 5 mL water intake. This is an unjustified simplification, since the bodyweight for 12-20 week old animals differs significantly between e.g. male and female animals, also the average bodyweight of a 12-20 week old female mouse ranges between 21 and 24g. Such dose calculation. The author state that four to five months old mice were used. The current OECD test guideline recommends using young animals at an age of six to ten weeks. The authors do not justify selected species, strains, and dosing. Laboratory’s proficiency in preparing and analyzing tail DNA intensities in comet assays is a prerequisite for reliable data. It was shown that there is huge variability in tail DNA intensities even within negative or positive controls (Tavares, A.M. et al. (2014); Nanogenotox Programme). This variability in combination with lack of controls and historical data renders positive findings of this assay of little relevance for risk assessment purposes. The transgenic C57Bl/6Jpun/pun mouse is an uncommon strain for this assay, with no historical data available. Due to the very low solubility and the difference in density of a factor of 4, one have to assume that titanium dioxide settles on the bottom of the flask, leading to a higher total dose. The authors verified neither the actual dose received nor the homogeneity of the suspension before during and after the dosing of the animals using an appropriate analytical method. Historical data and positive controls are not specified. Cytotoxicity measurements were not performed, which does not allow a conclusive statement, whether a true test item specific effect was observed or an artefactual response due to increased cytotoxicity. Clinical observations were not reported. Description of methods is insufficient. Details on tissue preparation, electrophoreses, and slide preparation are missing. Moreover, evaluation criteria are not sufficiently described. Moreover, the author state that the highest TiO2 nanoparticle dose leads to an inflammatory response, measured via qRT-PCR of pro-inflammatory marker gene expression. Inflammatory responses could have interfered with the results and could have led to false positive findings in the genotoxicity assays. The authors conducted a student’s t-test to test for significant changes before and after treatment. However, the student’s t-test is not a suitable for paired data. Based on the above mentioned shortcomings the reference is rated not reliable and thus considered unsuitable for risk assessment purposes.

 

Merely on in vivo DNA damage assay was considered reliable for hazard assessment purposes. The analysis of bronchi-alveolar lavage cells via alkaline comet assay showed no significant effect on the level of DNA damage after inhalation exposure towards 10 mg/m³ for five consecutive days. The maximum concentrations was reaching the maximum tolerated dose. It is therefore concluded that titanium dioxide does not cause local or systemic clastogenic or aneugenic events in animals.

 

 

Nanogenotox programme

In a comprehensive testing programme, a total of six different titanium dioxide nanomaterials were tested in several in vitro and in vivo test systems for their ability to induce (i) unspecific DNA damage, (ii) chromosome or genome mutation and (iii) gene mutation. An overview of the physicochemical parameters of the tested titanium dioxide samples is presented in the table below.

Table: Overview of the titanium dioxide samples used in the testing programme

Sample

Phase

Average XRD crystalline size

Average TEM particle size

Average BET & SAXS SSA

NM-100

anatase

56.7 - >100 nm

110 nm

9 m²/g

NM-101

anatase

7 nm

6 nm

316 m²/g

NM-102

anatase

21 nm

22 nm

78 m²/g

NM-103

rutile

23 nm

25 nm

51 m²/g

NM-104

rutile

23 nm

25 nm

56 m²/g

NM-105

anatase

rutile

23 nm

60 nm

24 nm

15 nm

46 m²/g

The testing programme was coordinated by the French Agency for Food, Environmental and Occupational Health & Safety (ANSES), involving 16 associated partners from 11 member states and 13 collaborating partners. The programme started in March 2010 for a period of 3 years, the final report was published in March 2013 (http://www.nanogenotox.eu/). An overview of all genetox tests executed within this programme is given in the table below.

Table: Overview of the in vitro genotoxicity assays conducted with titanium dioxide nano particles, samples NM-102 – NM-105.

Cell line

Assay

BEAS 2B

MN

Comet 3h

Comet 24h

16 HBE

MN

Comet 3h

Comet 24h

A549

MN

Comet 3h

Comet 24h

CACO-2

MN

Comet 3h

Comet 24h

Lymphocytes

MN

L5178Y

MLA

NHEK

MN

Comet 3h

Comet 24h

Skin 3D

Comet 3h

Comet 24h

 

Table: Overview of the in vitro genotoxicity assays conducted with titanium dioxide nano particles.

Species

Route

Assay

NM-101

NM-102

NM-103

NM-104

NM-105

Rat

Intratracheal

Comet

5 tissues

 

 

 

 

 

MN

1 tissue

 

 

 

 

 

Gavage

Comet

5 tissues

 

 

 

 

 

FpG Comet

2 tissues

 

 

 

 

 

MN

2 tissues

 

 

 

 

 

Intravenous

MN

1 tissue

 

 

 

 

 

Mice (LacZ)

Intravenous

gene mutation

1 tissue

 

 

 

 

 

Comet

2 tissues

 

 

 

 

 

MN

1 tissue

 

 

 

 

 

IUCLID study records were generated for each test item and test system combination, in order to report the test results transparently. This resulted in 54 study records for the in vitro systems and 23 study records for in vivo systems.

Due to the large number of test results (some of the in vitro experiments were conducted as round robin test over a range of up to 12 different laboratories), the overall results are discussed here only in summary.

For the reliability rating according to Klimisch, all experiments were checked against the relevant OECD technical guidelines relevant at the time of study conduct. The in vivo comet assay study design was checked against the OECD test guideline 489 (26 September 2014), which is based on internationally agreed study designs suggested by various authors in the public domain (e.g. Kirkland 2008, Brendler-Schwaab 2005, Burlinson 2007 and 2012, Smith 2008, Hartmann 2003, McKelvey-Martin 1993, Tice 2000, Singh 1998, Rothfuss 2010). The comparison with OECD 489 is therefore appropriate, since the majority of these protocols existed at the time of study planning and conduct.

The reports exhibited a number of shortcomings in the description of the experimental procedures and in the presentation of the test results. Further, the overall reporting does not appears to be final, since e.g. the report on the in vivo micronucleus assay in colon of rats following gavage administration is still flagged as “to be amended”. The major shortcomings of the reporting and experimental design are given below:

General

        i.           experiments were not conducted under GLP

      ii.           acceptability criteria were not included

     iii.           evaluation criteria are unclear/were not described in sufficient detail and not according to international accepted criteria (IWGT recommendations)

    iv.           some experiments were lacking internationally agreed protocols and guidelines (e.g. in vivo micronucleus assay in spleen)

      v.           the laboratories were lacking historical control data (e.g. in vivo comet assay in spleen)

In vitro experiments

        i.           basic information on the cell lines were not reported, i.e. source, sub-type, cell doubling time, absence of mycoplasma infection

      ii.           basic information on cell culture conditions were not reported, i.e. medium, culture dishes, temperature, humidity, seeding density, transfer times

     iii.           basic information on the experimental conditions of the genetox experiments were not reported, i.e. test item application including identification of the vehicle, CytoB concentration, information on pH effects, negative control (vehicle, medium, untreated), slide preparation, number of cells analysed, cell proliferation/cytotoxicity (CBPI or RI),

    iv.           no short term (pulse) treatment was performed

In vivo experiments

        i.           in some experiments only single dose regime was used

      ii.           number of cells analysed per animal was too low or was not stated

     iii.           information on (cyto)toxicity not clearly stated

    iv.           in some experiments the experimental procedure of cell staining and slide preparation was not reported

      v.           individual animal data was not reported

    vi.           in some experiments a non-physiological route of administration was chosen, which is not relevant for the hazard assessment of industrial chemicals (for which oral, dermal and inhalation are the only anticipated routes of human exposure)

vii.         for some experiments, the number of animals per dose was not given

Such shortcomings in the reporting and the test design would normally result in a reliability rating of 3 or 4, i.e. rendering those studies not suitable for hazard assessment purposes of industrial chemicals. However, since it can be assumed that the experiments were conducted by professional labs with documented competence for the execution of those tests and that the two journal articles (published after project finalisation: Tavares, A.M. et al. (2014); Louro, H. (2014)) demonstrated that more details are available, than were provided in the Nanogenotox reports.

Consequently, all individual experiments were rated with a reliability rating of 2, with the restriction that the individual studies cannot be used as key studies, but only as a part in a weight of evidence approach. The results of the individual experiments are presented and discussed below in detail.

Results

In vitro

Four titanium dioxide samples were tested in a micronucleus assay in six different cell lines. Authors concluded that the micronucleus assay was positive for four tested TiO2 samples (NM-102 – NM-105) in NHEK cells and also in the micronucleus assay in primary lymphocytes for NM-102, NM-103 and NM-104. However, the micronucleus assay was negative for all TiO2 samples in the other four remaining cell lines.

The comet assay was positive for all TiO2 in Caco-2 cells after the 24-hour treatment except for NM-104 (negative). NM- 100 showed a positive result in the 16HBE cell line as well as NM-102 (3 hour treatment) and NM- 105 (24 hour treatment). Four TiO2 samples (NM-102 – NM-105) showed a positive result at the 3 and 24 hours treatment time in the NHEK cell line. In addition, NM-102 showed a positive result in the BEAS 2B cell line at both treatment times. Lastly, NM-102 and NM-105 caused a positive result in the A549 cell line at the 3 hour treatment time.

In 3D human reconstructed full thickness skin models, all TiO2 nanomaterials (NM-102, NM-103, NM-104, and NM-105) investigated for DNA damage were negative in the comet assay. In contrast, the chemical positive control methylmethanesulfonate (MMS) consistently generated a significant increase in DNA damage. The highest dose studied by this protocol was 246 μg/cm² skin surface which showed no interference during the analysis. Transmission electron microscopic (TEM) analysis did not identify penetration of the stratum corneum of reconstructed human full thickness skin models even after 72 hour exposure with TiO2.

The L5178Y mouse lymphoma gene mutation assay was negative for all forms of TiO2 tested (NM-102 – NM-105).

 

In vivo

After intratracheal instillation in male Spague-Dawley rats, none of the TiO2nanomaterials (NM-101 – NM-105) induced DNA damage in broncho alveolar lavage (BAL) cells in a comet assay. Two TiO2samples (NM-102 and -103) showed an insignificant increase with no dose dependency in the liver. None of the TiO2nanomaterials studied showed genotoxic effects in lung, spleen, and kidney.

Furthermore, intratracheal instillation with NM-101 – NM-105 in male Sprague-Dawley rats induced no increase in micronucleus frequency in a bone marrow micronucleus assay.

Following oral administration by gavage in male Wistar rats, some genotoxic effects were observed with the comet assay with TiO2in spleen, jejunum (NM-103), colon (NM-102 and -105), and bone marrow (NM-104). However, all TiO2 nanomaterials studied showed no genotoxic effects in liver, kidney, and lymphocytes samples. No mutagenicity in bone marrow was observed in a bone marrow micronucleus assay using the same type of rats and the same route of administration.

Micronucleus assays in rats were included using single and repeated (5 times) intravenous administrations with NM-103 and NM-104 in order to increase the micronucleus potency to reach systemic organs. No increase in micronuclei could be detected in bone marrow.

In a further series of testing C57Bl/6 pUR288 transgenic male mice were used. Negative results for bone marrow micronucleus assay after repeated intravenous (i.v.;2 times) exposure to NM-102 were obtained. The genotoxicity effect of NM-102 was also tested in a gene mutation assay (Mutation Lac Z assay) and a comet assay using the transgenic male mice and i.v. administration, which showed a negative effect in the liver and spleen.

 

Conclusion

 

The overall quality of the summary report and the individual deliverable reports of the Nanogenotox programme are insufficient to follow the experimental design and findings. In some cases the results section does not appear to match the findings reported in the results section. The material and methods section merely provides a brief description of the essential methods, which hinders an assessment of the whole study design by a peer-reviewer. A small sub-set of experiments were published after completion of the programme (Louro, H. et al. (2014); Tavares, A.M. et al. (2014)), which provides further information on the overall study design compared with the final report of the Nanogenotox Programme.

 

In vitro

The in vitro experiments returned predominantly negative results from the micronucleus tests. Merely one cell line (NHEK, which is also not commonly used in in vitro genotoxicity testing experiments) returned a dose-response relationship over all test items. However, the MNBN frequencies were within the historical data range of other cell lines and primary human lymphocytes seen by other labs (0-15 MNBN per 1000 BN cells in vehicle/negative control), which puts the biological relevance into question whether this dose-response relationship is a true substance- induced finding. Whether this particular cell line is more susceptible to a physical insult by PSPs leading to an increased frequency of MN formation, was not further discussed by the authors. Based on the overwhelmingly negative findings with the five remaining cell lines and primary human lymphocytes, the dose-response relationship finding with NHEK cells is considered as incidental finding with no biological relevance for the human health hazard assessment. Consequently, titanium dioxide is considered non-clastogenic and non-aneugenic in a large range of in vitro mammalian cell systems. The mouse lymphoma assay in L5178Y cells returned a clear negative result over a range of four titanium dioxide samples, indicating an absence of potential to induce gene mutations by nano-sized titanium dioxide.

 

The comet assays returned largely positive findings over a wide range of different cell systems (pulmonary, intestinal, and epidermal). It is unclear whether the differences among cell systems were caused by intrinsic differences of their susceptibility, or by experimental variations. A major concern is that a dose dependency of the comet findings could not be established in any of the experiments, which raises doubts on a substance intrinsic ability to induce DNA damage. Most disturbingly, a very large inter-laboratory difference among the comet results for selected nano materials was observed, highlighting the difficulty to obtain conclusive results with this assay. Based on the outcome of the in vitro comet assays with titanium dioxide, it is concluded that there is inconclusive evidence on a DNA damaging potential of titanium dioxide. This is based on an overall lack of a dose-response relationship and the large inter-laboratory differences in results.

In summary, there is no conclusive evidence of a genotoxic potential of five different titanium dioxide samples in a number of different in vitro tests systems, covering general DNA damage, gene, chromosome or genome mutations,

 

In vivo

Inconclusive positive in vivo findings in the come assay are restricted to isolated organs or tissues and are also route dependent. After oral administration by gavage genotoxic effects were observed in spleen and jejunum (MN-103), colon (NM-102 and NM-105) and bone marrow (NM-104) in a comet assay with rats. However, those findings were largely lacking a dose-response relationship or were even not statistically significant. Further, the authors highlighted that the involved laboratories showed pronounced differences in the evaluation of the comet assays, as expressed in the figure below.

Figure: Variability of comet assay responses in liver for negative (vehicle) and positive (MMS) controls among the different laboratories involved in the WP. Each point is the median of one animal.

This highlights the lack of repeatability of these experiments and the lack of reliability of this test. In addition, some labs lacked historical control data for some tissues (e.g. spleen). Such lack of historical control data aggravates the interpretation of positive findings, since the variability of the comet findings are not known. It is therefore unclear whether a significant increase of DNA damage may have been within a normal biological variation and thus caused by an incidentally low negative control. These shortcomings for the in vivo comet assay highlight that the criteria for acceptability of that test system (especially in organ systems or tissues not commonly investigated) are not yet established.

For the other assays in rats (comet assay (intratracheal administration), bone marrow micronucleus assay (intratracheal, intravenous, and oral administration), and transgenic mice (bone marrow micronucleus assay, comet assay, gene mutation assay), an overall negative outcome was investigated over a range of up to five different titanium dioxide samples (NM-101 – NM-105).

In summary, there is no conclusive evidence of a genotoxic potential of five different titanium dioxide samples in a number of different in vivo tests systems, covering general DNA damage, gene, chromosome or genome mutations.

 

 

Discussion

 

In conclusion, the Nanogenotox testing programme almost exclusively returned negative findings in genotoxicity tests with a long-lasting testing history and for which well-established guidelines and procedures are available. In cases where inconclusive positive findings were received, these were mostly isolated findings in a single cell line and could not be repeated in other cell lines or primary human lymphocytes, thus should be regarded as not biologically relevant. The overall negative findings in the in vivo systems corroborates the mostly negative in vitro test results. For test systems investigating the induction of unspecific DNA damage (Comets assay), the in vitro tests largely returned a positive outcome, which were however not biologically relevant or not statistically significant. Similarly, the positive in vivo findings of the comet assay showed serious shortcomings in an inter-laboratory comparison. Considering the overwhelmingly negative tests results in this testing programme, it is concluded that nano-sized titanium dioxide is non-mutagenic, non-clastogenic or non-aneugenic over a wide range of different test systems.

 

 

 

Overall conclusion

The genetic toxicity of pigment grade and ultrafine titanium dioxide was assessed in various in vitro and in vivo experiments, using different crystalline forms (anatase and rutile) as well as surface modified titanium dioxide. The surface modifications applied to titanium dioxide particles is aimed to increase the dispersibility in lipid suspensions. Consequently, negative results obtained with hydrophobic coated titanium dioxide are considered as worst case exposure conditions. In summary, there is a consistent pattern showing a complete lack of genotoxic effects over a wide range of test systems for all commercially relevant crystalline forms of titanium dioxide.

Justification for classification or non-classification

Genetic toxicity, in vivo

The in vivo data for pigment grade titanium dioxide present a consistent pattern. In rigorous GLP and guideline compliant studies as discussed above, combining chromosome aberration and micronucleus endpoints up to limit doses in mice, titanium dioxide did not show test item induced genetic toxicity.

 

Genetic toxicity, in vitro

None of the in vitro genotoxicity studies rated as reliable showed any effect in bacterial reverse mutation assays, in mammalian cell gene mutation tests (TK assay) or in mammalian cell chromosome aberration or micronucleus tests, thus supporting the negative findings in the in vivo tests as cited above.

The classification criteria acc. to regulation (EC) 1272/2008 as germ cell mutagen are not met, thus no classification applicable.