Oral intake of added titanium dioxide and its nanofraction from food products, food supplements and toothpaste by the Dutch population

Abstract

Titanium dioxide (TiO₂) is commonly applied to enhance the white colour and brightness of food products. TiO₂ is also used as white pigment in other products such as toothpaste. A small fraction of the pigment is known to be present as nanoparticles (NPs). Recent studies with TiO₂ NPs indicate that these particles can have toxic effects. In this paper, we aimed to estimate the oral intake of TiO₂ and its NPs from food, food supplements and toothpaste in the Dutch population aged 2 to over 70 years by combining data on food consumption and supplement intake with concentrations of Ti and TiO₂ NPs in food products and supplements. For children aged 2–6 years, additional intake via ingestion of toothpaste was estimated. The mean long-term intake to TiO₂ ranges from 0.06 mg/kg bw/day in elderly (70+), 0.17 mg/kg bw/day for 7–69-year-old people, to 0.67 mg/kg bw/day in children (2–6 year old). The estimated mean intake of TiO₂ NPs ranges from 0.19 μg/kg bw/day in elderly, 0.55 μg/kg bw/day for 7–69-year-old people, to 2.16 μg/kg bw/day in young children. Ninety-fifth percentile (P95) values are 0.74, 1.61 and 4.16 μg/kg bw/day, respectively. The products contributing most to the TiO₂ intake are toothpaste (in young children only), candy, coffee creamer, fine bakery wares and sauces. In a separate publication, the results are used to evaluate whether the presence of TiO₂ NPs in these products can pose a human health risk.

• Food additive E 171
• long-term dietary intake
• nanomaterial
• probabilistic modelling
• TiO₂

Introduction

Titanium dioxide (TiO₂) is one of the most popular pigments in use today (TDMA, 2013). Approximately 4 million tons of this pigment are produced annually worldwide (Ortlieb, 2010), which is 70% of the total production volume of pigments (Baan, 2007). It is the brightest known white pigment and, in foods, it provides a basic white background colour that acts as an opacifier, and reflects light across most of the visible spectrum. It may also act as a barrier by physically separating other colours (Lomer et al., 2000). TiO₂ is authorised as a food colour in the European Union (EU) as E 171 (EU, 2011a) and is commonly used as a food additive (Peters et al., 2014). TiO₂ is also applied in paints, coatings, plastics, papers, inks, drugs, cosmetics (e.g. sunscreens, creams, lip balm) and toothpaste (Oomen et al., 2011; Shi et al., 2013).

TiO₂ exists in different crystal structures: anatase, rutile and brookite, or a mixture of these. The US FDA has approved the use of TiO₂ in food in 1966 by allowing levels up to 1% in food (FDA, 2015). TiO₂ in anatase form has been accepted as a food additive in the EU for decades as well at quantum satis (i.e. as much as necessary) for a selected list of products (Food additives database). Since 2004, also rutile TiO₂ is allowed (EFSA, 2004a; EU, 2009). As the anatase form has been accepted as food pigment much longer than the rutile form, it seems plausible that this form is still the most used in food products. Indeed, Yang et al. (2015), who characterised samples of food grade TiO₂ from five different vendors that ship to the US and EU, detected the anatase form in all samples and rutile in only one (Yang et al., 2015). The white colour of pigment TiO₂ is best achieved with particles of 200–300 nm, as these give an optimal diffraction of light for this colour (Braun, 1997). The Titanium Dioxide Manufacturers Association (TDMA) indicated that as the production may not result in particles of precisely this size-range, it is possible that the pigment also contains particles <100 nm (i.e. nanoparticles (NPs)) (TDMA, 2013). Indeed, the presence of NPs in E 171 and in several food products containing E 171 (such as chewing gum, coloured hard-shell candy and icing) has been reported (Peters et al., 2014; Weir et al., 2012; Yang et al., 2015). Weir et al. (2012) reported that approximately 36% of the number of particles in E 171 (from one supplier) were <100 nm in at least one dimension. Peters et al. (2014) found that 10–15% of the particles in E 171 (from seven different suppliers) were <100 nm. Recently, Yang et al. (2015) reported 17–35% and Warheit et al. (2015) reported 21%, all based on the number of particles. According to the EU recommendation 2011/696/EU (EU, 2011b), a material is called a nanomaterial when ≥50% (number based) of the particles is in the nanorange (<100 nm). With the reported fractions, E 171 is not a nanomaterial following the criteria in this recommendation. Nonetheless, it is clear that E 171 contains NPs and these are present in a series of food products that together may result in a considerable intake of TiO₂ NPs.

Recent oral studies with TiO₂ NPs indicate that these particles can have toxic effects (reviews by Iavicoli et al. (2011, 2012) and Shi et al. (2013)) and repeated oral exposure to these particles may result in tissue accumulation in the long run (Geraets et al., 2014). With a likely intake of TiO₂ NPs and recent indications of their toxic effects, it is relevant to assess whether this exposure can lead to health risks.

To assess these health risks, an estimation of the oral intake of TiO₂ NPs is necessary, and therefore we aimed to estimate the long-term oral intake of added TiO₂ (resulting from E 171) in food products in the Dutch population. Food supplements and toothpaste were also included in the intake estimation as they are also known to contain added TiO₂ and thus may contribute to the oral intake of added TiO₂. The oral intake to TiO₂ present in drugs is not taken into account, although it could contribute to the intake of specific individuals. The intake of TiO₂ NPs is estimated by combining the most recent data on the Dutch food consumption with data on concentrations of Ti or TiO₂ in food products, supplements and toothpaste. Where only measured Ti concentrations were available, TiO₂ concentrations were calculated from the measured Ti concentrations by multiplying with the TiO₂/Ti mass difference ratio of 1.67. To calculate the nanofraction of TiO₂ (<100 nm) ingested, we used a fraction of 0.31% (by mass, corresponding to 15% by number) of nano-sized particles calculated from the data of Peters et al. (2014), by multiplying the results of the TiO₂ intake estimations from food products, food supplements and toothpaste ingested (only young children) by 0.0031.

A risk assessment of the corresponding human health risk of the intake of TiO₂ NPs via these products is published as a separate publication (Heringa et al., submitted).

Methods

Consumption data of the Dutch population

The dietary intake of TiO₂ in the Dutch population was determined for three age classes (2–6 years, 7–69 years and 70+) and on the basis of three main product groups (food products, food supplements and toothpaste).

Intake calculations for children aged 2–6 were performed using food consumption data of the Dutch National Food Consumption Survey (DNFCS) – Young Children, which was conducted in 2005 and 2006 (Ocké et al., 2008). Intake calculations for the Dutch population aged 7–69 were performed using food consumption data of DNFCS 2007–2010 (van Rossum et al., 2011), and those for the (community-dwelling) population over the age of 70 (70+) using data of DNFCS-older adults, which was conducted between 2010 and 2012 (Ocké et al., 2013). In all surveys, food consumption data were collected on two non-consecutive days per individual. Food consumption data of young children were collected by means of two food records, but among the others age groups by two 24-h food recalls by using GLOBO-diet software (formal EPIC-Soft^®, IARC, Lyon, France). In total, of 1279, 3819 and 739 individuals, respectively, food consumption data were available, as well as the intake of food supplements (including name and brand of the supplements, form and number of supplements consumed per day). For a more detailed description of the three surveys, see http://www.rivm.nl/en/Topics/D/Dutch_National_Food_Consumption_Survey.

Data on toothpaste intake were only recorded in the DNFCS-Young children because it is known that young children ingest toothpaste. As children >6 years old, adults and elderly do usually not swallow toothpaste and rinse their mouth after brushing their teeth, it was assumed there was no toothpaste ingestion for these age groups. Data on toothpaste have not been recorded in DNFCS 2007–2010 and DNFCS-older adults. In the DNFCS-Young children, the number of times a child brushed its teeth was recorded. Information on the amount and the brand of the toothpaste used was not recorded. In “The Scientific Committee on Consumer Safety (SCCS) notes of guidance for the testing of cosmetic substances and their safety evaluation”, no advisory amounts or retention factor is provided for young children (SCCS, 2012). Toothpaste portions on a young child’s toothbrush were, therefore, simulated and weighed in our laboratory to make an estimation of a worst case portion of toothpaste ingested by a small child. (A worst case portion of a full bar of toothpaste (covering the whole length of a toothbrush’s head with bristles for children 2–5 years old) weighed 1.15 ± 0.14 g (average ± SD, n = 6), which is higher than the upper range level of 0.8 g given by an Scientific Committee on Consumer Products (SCCP) opinion (SCCP, 2005) (currently the SCCS). For this age group, it was assumed that all toothpaste applied on a brush is swallowed, as swallowing is found acceptable by the Dutch Society for Oral Health (2015), and even advisable to achieve a sufficient daily fluoride intake to prevent caries (SCCP, 2005; The Netherlands Nutrition Centre, 2015).

TiO₂ concentrations in food products

Publications in which total-Ti and/or TiO₂ concentrations in food products are reported were selected from the literature. Analytically, TiO₂ concentrations in food products can be determined by determination of total-Ti (all elemental Ti present in a sample) and/or TiO₂ particle concentrations. In total, four publications reported the presence of total-Ti and/or TiO₂ particle concentrations in food products: Sugibayashi et al. (2008) (Ti levels), Weir et al. (2012) (Ti levels), Lomer et al. (2000) (Ti levels) and Peters et al. (2014) (Ti and TiO₂ particle levels). Total-Ti and TiO₂ particle concentrations collected from these studies and used in the intake calculations in this study are listed in the Supplementary Material Table SM-1. The data of Peters et al. (2014) showed that a good correlation exists between the level of total-Ti (as measured by inductively coupled plasma mass spectrometry (ICP-MS)) and the level of TiO₂ particles (as measured by AF4-ICP-MS) in the products tested. Furthermore, Peters et al. (2014) did not detect Ti or TiO₂ NPs in seven commonly used food products where E 171 or TiO₂ was not mentioned on the label (limit of quantification (LOQ) of 10 mg Ti/kg product). Hence, we assume that all Ti in the other food products results from added TiO₂.

We included data from Weir et al. (2012), Lomer et al. (2000) and Peters et al. (2014) in our study. To this end, total-Ti concentrations were converted into TiO₂ concentrations by multiplying with the mass difference ratio of 1.67, presuming all elemental Ti originates from TiO₂. The total-Ti concentrations in vegetables and meat published by Sugibayashi et al. (2008) were not included because the addition of TiO₂ to these products seems unlikely and because no quality control information on interferences was given potentially resulting in considerable errors in total-Ti measurements.

Further data on which kind of products contain E 171 or “TiO₂” on the ingredient list were retrieved from the Innova database. The Innova database is a commercial food and beverage product database containing information on ingredients, food composition, price, portion sizes, brand and manufacturer and several other variables for foods (including pet foods), supplements and oral care products (See: http://www.innovadatabase.com/home/index.rails). One of the main purposes of the Innova database is to register new launches (in more than 70 countries). Updating the database by deleting items that are not for sale anymore or updating information if the ingredients or composition of the product have changed is of lower priority. Keeping this in mind, the Innova database provides further insight into the type of products and product categories to which TiO₂ is added. The information retrieved from the Innova database was used to identify data gaps in the food products and product categories with added TiO₂. These products and product categories were compared with the food products selected by Weir et al. (2012), Lomer et al. (2000) and Peters et al. (2014). From this comparison, it was concluded that for all relevant food categories information on the amount of TiO₂ was present in at least one of these three publications.

Initial analyses of the available data from the literature revealed that the concentrations in products that seemed to contribute most to the intake of TiO₂ were often based on only one Ti concentration level of a US product. For those product groups (milk products, fine bakery wares, confectionery, sauces and beverages), more (representative) Ti concentration levels of products specific for the Dutch market were required. Since some studies revealed background levels of Ti in raw milk (Anderson, 1992; Grebennikov et al., 1964), we determined both total-Ti concentrations and TiO₂ particles in 21 samples of Dutch dairy products and six raw (cow) milk samples (see below) to obtain insight in background levels in these products. In addition, one sample of Dutch fine bakery wares (a typical Dutch cake with icing and cream), one confectionery product (winegums), one sample of water-based ice, nine samples of beverages (including soy milk, energy, sport and soft drinks and syrups) and one sample of salad dressing and coffee creamer (powdered) were analysed for total-Ti concentrations. Samples of two food supplements (both multivitamins) were also included because measurements of Ti or TiO₂ particles in food supplements were lacking. (See Table 1 and Supplementary Material Table SM-2 for product list and overview of all samples).

Table 1. Average measured total-Ti concentrations, and subsequently calculated TiO₂ concentrations, in selected, representative Dutch food products, raw (cow) milk, and food supplements. See Supplementary Material Table SM-2 for further details. Samples rich in calcium were analysed by ICP-HRMS, others by ICP-QMS. LOQ = 0.05 mg Ti/kg product.

	Number of samples (>LOQ)^a	Mean total-Ti (mg/kg product) (±SD)	Min total-Ti (mg/kg product)	Max total-Ti (mg/kg product)	Mean TiO2^b (mg/kg product)
Samples analysed by ICP-HRMS
Raw (cow) milk	6 (6)	0.31 (±0.23)	0.05	0.63	0.51
Regular dairy products (i.e. milk, yoghurt)	11 (10)	0.47 (±0.46)	<LOQ	1.46	0.79
Processed dairy products	10 (5)	0.12 (±0.17)	<LOQ	0.57	0.21
Soy milk	2 (2)	0.33 (±0.01)	0.32	0.34	0.55
Dutch cake with icing and cream	1 (1)	0.23	0.23	0.23	0.38
Coffee creamer (powdered)	1 (1)	1640	1640	1640	2739
Samples analysed by ICP-QMS
Energy drink (containing caffein)	1 (1)	0.07	0.07	0.07	0.11
Soft drink	2 (2)	0.06 (±0.00)	0.06	0.07	0.11
Sports drink	2 (2)	0.09 (±0.05)	0.05	0.12	0.14
Syrup	2 (2)	0.17 (±0.00)	0.17	0.17	0.28
Ice (water-based)	1 (1)	0.16	0.16	0.16	0.26
Wine gums	1 (1)	0.25	0.25	0.25	0.42
Salad dressing	1 (1)	0.43	0.43	0.43	0.72
Food supplement (multivitamin)	2 (2)	744 (±1009)	31	1458	1242

^aIn parenthesis the number of samples above the LOQ. For samples with a Ti concentration < LOQ, a value half LOQ (0.025 mg Ti/kg product) was used in the calculation of the mean value.

^b TiO₂ concentrations were calculated by multiplying the mean total-Ti levels with the mass difference factor of 1.67.

TiO₂ concentrations in food supplements

To estimate the contribution of added TiO₂ intake via food supplements, it was assumed that all food supplements as consumed in the DNFCS contain TiO₂. The TiO₂ concentration in supplements was based on the analyses of two multivitamin tablets in the present study. The average TiO₂ concentration in these tablets was extrapolated to dragées and capsules using the estimated relative TiO₂ levels mentioned in Böckmann et al. (2000). They reported estimated TiO₂ weight percentages of 0.01–1.0% (by mass) for tablets (average 0.5% (by mass)), 1.5% (by mass) for dragées and 2.0% (by mass) for capsules (Böckmann et al., 2000). Dragées and capsules were therefore assumed to contain three and four times more TiO₂ than tablets, respectively.

However, in the DNFCS, the intake of supplements is recorded as number of supplements consumed, and the weight of the specific supplements was not reported in the DNFCS. Therefore, the average weight was calculated from known weights of around 40 different food supplements available on the Dutch market. The calculated average weights per supplement type were: effervescent tablet 4.36 g; capsule 0.89 g; dragée 0.40 g and tablet 0.82 g.

TiO₂ concentrations in toothpaste

The TiO₂ concentration in toothpaste was calculated using the average and maximum total-Ti levels in three toothpaste products determined by Peters et al. (2014) by AF4-ICP-MS. These products do not consist of special toothpaste products intended to be used by young children, but as data on such children’s toothpaste are lacking, and no information is available indicating the concentration of TiO₂ could be different in regular toothpaste compared to young children’s toothpaste, we used the Ti-concentration from the investigated toothpaste in the intake estimation of young children (2–6 years). The total-Ti concentration was converted to the TiO₂ concentration by multiplying with the mass difference ratio of 1.67, resulting in an average concentration of 6.13 × 10³ mg TiO₂/kg product, and maximum concentration of 9.3 × 10³ mg TiO₂/kg product.

Determination of total-Ti concentration in raw milk and selected Dutch products

The selected food products and supplements were of common brands, bought at large Dutch supermarket chains; raw cow milk samples were obtained directly from the farm (see Table 1 and Supplementary Material Table SM-2 for product list). Total-Ti concentrations of the selected Dutch food products, supplements and raw cow milk were determined by acidic sample digestion followed by ICP-MS of the digests. For non-dairy product samples, a quadrupole ICP-MS (ICP-QMS) was used to measure total-Ti in the acidic digests. The ICP-MS was operated in collision cell mode to minimise polyatomic interferences from ³⁶Ar¹²C and ³²S¹⁶O and data were acquired at m/z 47 and 48. The LOQ of the ICP-QMS method was 0.05 mg Ti/kg product. For raw milk and other dairy samples, a sector-field ICP-high resolution mass spectrometer (HRMS) was used to measure total-Ti in the acidic digests in standard mode and TiO₂ particles in aqueous suspensions in single-particle mode. The ICP-HRMS was operated in medium resolution mode and data were acquired at m/z 46.95 to avoid (polyatomic) interferences from ³⁶Ar¹²C, ³²S¹⁶O and ⁴⁸Ca. The LOQ of the ICP-HRMS was 0.05 mg Ti/kg product. A detailed description can be found elsewhere (Peters et al., 2014,2015).

Fraction of nanoparticles in TiO₂ (E 171)

Several different values for the number-based NP fraction (fraction <100 nm) of TiO₂ (E 171) have been reported: 36% (Weir et al., 2012), 10–15% Peters et al. (2014), 17–35% Yang et al. (2015) and 21% Warheit et al. (2015). We used the data reported by Peters et al. (2014) as they were derived from E 171 samples from seven different suppliers and, therefore, represent the biggest dataset. In addition, Peters et al. (2014) found that the size distribution of TiO₂ particles in E 171 is similar to the size distribution of TiO₂ particles in food products and toothpaste.

In the study by Peters et al. (2014), per TiO₂ sample, the particle distribution in eight size bins was determined by two methods: TEM and AF4-ICP-MS. With TEM, NP size were reported in size bins of 40–65 nm and 65–100 nm, and particles up to 1600 nm were detected. With AF4-ICP-MS, NP size bins of 25–40 nm, 40–65 nm and 65–100 nm were recorded, and particles up to 1000 nm were determined. Per method, the average diameter per particle size bin was used to calculate the average volume (as a sphere) and weight per particle size bin, which was multiplied with the percentage of particles in the respective particle size bin. The sum of weights resulting from these two (TEM) or three (AF4-ICP-MS) particle size bins, divided by the total sum of weights of all particle size bins, resulted in a mass-based NP fraction. The outcome from the seven different E 171 samples and both methods ranged from 0.012% (by mass) to 0.31% (by mass). We use the upper value of the range indicated by the data of Peters et al. (2014), i.e. 0.31% (by mass), to calculate the intake of NP fraction of TiO₂ (TiO₂ NPs) from food products and food supplements. Peters et al. (2014) showed that the particle size distribution of TiO₂ in toothpaste was similar to that of E 171. Thus, to calculate the intake of NP fraction of TiO₂ (TiO₂ NPs) from toothpaste ingested (only young children) we applied this fraction of 0.0031 as well.

Assigning TiO₂ levels to foods and supplements consumed and toothpaste

The analysed food products as reported by Weir et al. (2012), Lomer et al. (2000), Peters et al. (2014) and the additionally analysed food products for this study were linked to comparable foods consumed in the three DNFCSs (see Supplementary Material Table SM-1). This resulted in the following six food categories with foods containing TiO₂:

• Confectionery (chocolate products, sweets, candy bars, chewing gum).

• Fine bakery wares (cake, pies, pastries, cookies and biscuits).

• Sauces (dressings and savoury sauces).

• Milk products (cheeses (soft and hard), coffee creamer and milk powder, ices and desserts, cappuccino, dairy drinks).

• Drinks (sports and soft drinks, fruit drinks).

• Cereal products (popcorn and rice, breakfast cereals).

For food products for which more concentration data were available the average TiO₂ concentration was calculated and used for the intake estimation (see Supplementary Material Table SM-1). However, when Dutch data (i.e. data of Peters et al. (2014) and the concentration data in this study) of a specific product were available, this were given priority and only this calculated TiO₂ level was included (as for chewing gum). When no concentration data were available for an individual food item, the best possible match was made based on expert judgement. For example, the TiO₂ concentration of chocolate muffins was based on that of chocolate cookies.

For some products, the determined TiO₂ concentrations had to be adjusted to fit the form of the product as consumed and as reported in the DNFCS. TiO₂ concentrations in fluids (e.g. dressing) analysed by Weir et al. (2012) were reported in μg/ml, while the consumption in the DNFCS is given in mass (i.e. grams) of the product. These data were expressed in μg/mg (like in other products) assuming a density of the fluids the same as of water (1 g/ml). For cappuccino prepared from powder, the levels from the coffee and tea creamers were used, but diluted 50-fold, as cappuccino is known to consist of 2% coffee creamer.

Lomer et al. reported for several products values below the detection limit (LOD, which was 0.01 μg Ti/mg). In these cases, a value of half LOD was taken for further calculations. However, in the case of soft cheeses, this value of half the LOD would have become the maximum TiO₂ level reported, as all soft cheeses in Lomer et al. (2000) gave < LOD values, while Weir et al. (2012) reported lower, but measured values (2.45−9.22*10⁻⁰⁴μg/mg). As we found it unreasonable to let a value below LOD dominate over measured values, we omitted the data of Lomer et al. (2000) in the calculation of average levels in soft cheeses.

Intake calculations

Consistent with the aim to estimate the long-term exposure, we were interested in an “average” scenario, assuming that the Dutch population varies in the brands consumed and thus be exposed to the average TiO₂ concentration of the brands available for each food product, or at least not consume the worst case brand for each type of food products during its entire life. Furthermore, to determine long-term intake, ideally, statistical models should be used that correct the variation in long-term intake between individuals for the within individual (between days) variation (Hoffmann et al., 2002; Nusser et al., 1996; Slob, 1993). An important prerequisite for this is that the logarithmically transformed daily intake distribution is normally distributed (de Boer et al., 2009). Since the intake data were not normally distributed for TiO₂ (not shown), the observed individual means (OIM) method was used. The OIM method calculates the intake per day per subject and averages the intake of the two recall days per subject. The Monte Carlo Risk Assessment programme, Release 8.1 (Biometris - Wageningen University and Research Centre, Wageningen, The Netherlands) was used for the intake assessment (de Boer & van der Voet, 2015). All intake estimates were weighted for deviances in socio-demographic factors (including age, gender, educational level of the head of the household, region and urbanisation) and season (to correct for a higher representation of consumption data in winter and autumn than spring and summer), and additionally for day of the week, to make the results representative of the relevant Dutch population and for all days of the week and all seasons (Ocké et al., 2008,2013; van Rossum et al., 2011). By using the bootstrap approach, the uncertainty in the dietary intake assessment due to the sampling size of the food consumption data was quantified. The uncertainty is reported as the 95% confidence interval around the percentiles of intake.

We used a fraction (0.31% by mass) of nano-sized particles calculated from the data of Peters et al. (2014) to calculate the nanofraction of TiO₂ (<100 nm) ingested, by multiplying the results of the TiO₂ intake estimations from food products, food supplements and toothpaste ingested (only young children) by 0.0031.

Results

Total-Ti concentrations measured in raw milk, selected Dutch food products and food supplements

Raw cow milk samples contained Ti, with an average concentration of 0.31 mg Ti/kg (Table 1). The average Ti concentration of “regular” milk products, including fresh milk (skimmed and semi-skimmed), ultra-heat treated milk (semi-skimmed and unskimmed) and plain yoghurt, was in the same range: 0.47 mg Ti/kg product (Table 1). In addition, results of specific TiO₂ particles analysis by ICP-HRMS and confirmation by electron microscopy (data not shown) demonstrated that TiO₂ particles were present both in samples of raw milk as well as in regular milk products. Therefore, we concluded that the Ti-levels in these regular milk products are likely to originate from indirect sources, such as the environment or from transfer/carry over from animal feed to milk. Therefore, regular milk products were not included in our intake estimation of added TiO₂.

The average Ti concentration of the processed dairy products (including vanilla custard, pudding, ice cream and yoghurt drinks) was lower: 0.12 mg Ti/kg product, as it also contained several products with a Ti level below the LOD (see Table 1). Also, the Ti levels measured in many of the other typical Dutch food product was low: the average total-Ti concentration in the non-dairy food products ranged from 0.06 (soft drinks) to 0.43 mg Ti/kg product (salad dressing) (Table 1). They do not indicate E 171 has been added to these products as this usually results in much higher values. The highest concentrations were indeed detected in powdered coffee creamer (1640 mg Ti/kg product) and one of the multivitamin food supplements (1458 mg Ti/kg product), which both had TiO₂ listed on their ingredient list (See Supplementary Material Table SM-2).

Long-term oral intake estimation of TiO₂ via foods, food supplements and toothpaste

Table 2 shows the estimated long-term intake of TiO₂ for the different age groups. The intake is lowest in elderly (70+) and highest in young children (2–6 years old): the mean long-term intake to TiO₂ ranges from 0.06 in elderly to 0.67 mg/kg bw per day in young children (2–6 years old). Corresponding numbers for the P50 were 0.03 and 0.59 μg/kg bw per day, and for the P95 0.23 and 1.29 mg/kg bw per day, respectively (Table 2, left columns).

Table 2. Estimated long-term intake, and lifelong daily intake, of TiO₂ and TiO₂ nanoparticles from food products, food supplements and toothpaste (only children 2–6 years old) ingested by the Dutch population, assuming that the population would vary in the products consumed and thus would be exposed to an average TiO₂ concentration of products.

	TiO2			TiO2 nanoparticles (TiO2 NPs)
	Percentiles of intake^a (mg/kg bw per day)		Mean^a	Percentiles of intake^a (μg/kg bw per day)		Mean^a
Age groups	P50	P95	(mg/kg bw per day)	P50	P95	(μg/kg bw per day)
2–6 years old	0.59 (0.57–0.60)	1.29 (1.19–1.40)	0.67 (0.66–0.70)	1.90 (1.80–1.94)	4.16 (3.84–4.52)	2.16 (2.13–2.26)
7–69 years old	0.08 (0.07–0.08)	0.50 (0.47–0.54)	0.17 (0.16–0.18)	0.26 (0.23–0.26)	1.61 (1.52–1.74)	0.55 (0.52–0.58)
70+	0.03 (0.02–0.03)	0.23 (0.20–0.28)	0.06 (0.05–0.07)	0.10 (0.06–0.10)	0.74 (0.65–0.90)	0.19 (0.16–0.23)
Lifelong daily intake^b	0.11	0.52	0.19	0.36	1.67	0.62

^aNote: 2.5% lower – 97.5% upper confidence limits of intake are reported in parentheses.

^b The lifelong daily intake consists of the weighted average of the estimated long-term intake values of the different age groups, assuming no TiO₂ intake during the first year of life, the same intake in the second year of life as in the 2–6 years old age group, and an average life expectancy of 80 years, as reported for the Dutch population by the Dutch Central Bureau of Statistics (CBS) for 2012 (CBS, 2014).

The products that contributed most to the TiO₂ intake differed between the age groups (see Table 3; note the food products listed are separate items within the DNFCS, but in reality, these items may be similar, e.g. cappuccino with caffeine and cappuccino instant ready to drink). In young children, toothpaste contributed most (57%), followed by a food item consisting of a specific type of hard candy with sugar (4%). In persons aged 7–69, the food item chewing gum contributed most to the TiO₂ intake (14%), followed by the item coffee creamer (11%). In persons aged 70+, the item coffee creamer was the major contributor (13%). Overall, based on the top 10 list of items in Table 3, the products contributing most to the TiO₂ intake in young children (2–6 years old) are toothpaste, confectionary (sweets, chocolate products, chewing gum) and fine bakery wares (biscuits). These products differ from those in the other age groups (Table 3). In 7–69-year-olds and elderly (70+), the food items in the top 10 are comparable, although the ranking is different: confectionary (chewing gum), coffee creamer-related milk products (creamer and milk powder, cappuccino), sauces (dressings and savoury sauces) and fine bakery wares (cake for age group 7–69 year old, pastries for age group 70+). The top 10 of food items, food supplements and toothpaste covers about 78%, 55% and 42% of the added TiO₂ intake in young children, 7–69-year-olds and elderly, respectively. Most food products only contribute <5% to the TiO₂ intake, indicating the intake is spread over many products. A clear exception is the toothpaste for young children, contributing to over 57%, and the number of items in the other age groups related to coffee creamer (>20%). The intake calculations for added TiO₂ are therefore sensitive for these contributions, respectively.

Table 3. Top 10 of food items reported in the Dutch National Food Consumption Survey (DNFCS), food supplements and toothpaste (only children 2–6 years old) contributing most to the TiO₂ intake using average TiO₂ concentrations. The percentage of the contribution per item is indicated in parentheses.

Top 10	DNFCS-children (2–6 years old)	DNFCS-2007–2010 (7–69 years old)	DNFCS-older adults (70+)
1	Toothpaste (57%)	Chewing gum (14%)	Coffee creamer (13%)
2	Hard candy with sugar (4%)	Coffee creamer (11%)	Thickened milk for coffee, powdered (8%)
3	Sugar-coated chocolate confectionery (3%)	Mayonnaise normal (7%)	Chewing gum (4%)
4	Chewing gum with sugar (2%)	Sauce, garlic, mayonnaise-based (5%)	Cappuccino with caffein (3%)
5	Sugar-coated chocolates (2%)	Cappuccino (4%)	Cappuccino instant ready to drink (3%)
6	Biscuit with flavoured milk-filling (2%)	Thickened milk for coffee, powdered (3%)	Cappuccino (3%)
7	Biscuit with milk-filling (2%)	Iced cake “glacekoek” (3%)	Sauce, garlic, mayonnaise-based (2%)
8	Marsh mellow big (2%)	Cappuccino instant ready to drink (3%)	Pastries with cream (2%)
9	Biscuits with coffee glaze (2%)	Cappuccino with caffein (3%)	Sauce based on mayonnaise (2%)
10	Chewing gum sugar-free (2%)	Coffee liquid with sugar and milk (machine) (2%)	Mayonnaise (2%)

Long-term oral intake estimation of TiO₂ NPs from foods, food supplements and toothpaste

As NPs are especially of interest, the intake estimates of Table 2 (left columns) were converted to intake estimates of TiO₂ NPs, by multiplying with the factor 0.0031 (see Methods). Table 2 (right columns) shows the estimated long-term oral intake of TiO₂ NPs for the different age groups, assuming that the Dutch population varies in the products consumed and thus be exposed to the average TiO₂ concentration of products. The mean intake ranges from 0.19 μg/kg bw per day in persons aged 70+ to 2.16 μg/kg bw per day in young children (2–6 years old). Corresponding numbers for the P50 were 0.10 and 1.90 μg/kg bw per day, and for the P95 0.74 and 4.16 μg/kg bw per day, respectively (Table 2, right columns).

Discussion

The dietary intake of TiO₂ has been calculated in previous years by Penttilä et al. (1988), by the consortium of the Expochi project for EFSA (Huybrechts et al., 2010), and by Weir et al. (2012). The dietary intake varied 1300-fold between 0.01 and 13.0 mg TiO₂/kg bw/d. Upon evaluation, we decided not to use these assessments as they were either too old (>20 years), unclear in the precise steps and underlying assumptions or not sufficiently relevant for the Dutch or European population. Instead, we performed a new estimation of TiO₂ and TiO₂ NPs intake, based on the DNFCS (Ocké et al., 2008,2013; van Rossum et al., 2011).

In the present study, we were interested in an “average” scenario, assuming that the Dutch population would vary in the brands consumed and thus be exposed to the average TiO₂ concentration of the brands available for each food product, or at least not consume the worst case brand for each type of food products. For food concentration levels, we used average measurements of Ti and TiO₂ particle concentrations in various food products reported in the literature (Lomer et al. (2000); Weir et al. (2012); Peters et al. (2014)). Furthermore, additional analyses were performed in specific food products in order to obtain more representative data for the Dutch situation, or to correct for naturally occurring Ti and TiO₂ levels in milk. Food supplements and toothpaste were included in the intake estimation as they are known to contain TiO₂ and may contribute to the oral intake to TiO₂. The estimated TiO₂ intake from food products, food supplements and toothpaste (only young children) ingested were multiplied by 0.0031 to estimate the long-term TiO₂ NP intake (by mass).

General uncertainties and assumptions

In the present study, efforts are made to derive an accurate estimate of the intake of TiO₂ and its nanofraction from food products, food supplements and toothpaste. Obviously, some uncertainties in the dataset exist, and several assumptions had to be made:

• Dataset: Although an impressive dataset of Ti or TiO₂ levels in foods, supplements and toothpaste could be gathered with measured data for many products in the most contributing product groups, this dataset was not complete. However, with the many food products in which E 171 is authorised, gathering a complete dataset would be virtually impossible. Ideally, the mean concentration should be based on the average concentration of all food products on the Dutch market weighed to market shares of products and brands. Analysis of each specific food product would be very time and resource consuming, and information on market shares are not publicly available. Not all food products were analysed (e.g. not all types of cookies) and neither all brands of the same food product. Data on representative other products were used as a substitute. This may have resulted in an over- or underestimation of intake, depending on the actual market shares of E 171 containing products and brands.

• Inclusion criteria used: We included products of which the determined Ti or TiO₂ levels are likely to originate from added TiO₂ (based on Weir et al. (2012), Lomer et al. (2000) and Peters et al. (2014) and on the search in the Innova database). For product groups with limited information on the Ti concentration and that potentially contributed considerably to the intake of TiO₂, (milk products, fine bakery wares, confectionary, sauces and beverages) in a pre-study, more (representative) Ti concentration levels specific for the Dutch market were desired. The selection of these specific products was based on representativeness for the Dutch market. They were generally available products, of common brands and bought at large Dutch supermarket chains. The selected products turned out to be mainly products without TiO₂/E 171 on the label and the TiO₂ concentration of these products was generally lower than the values for comparable products from literature (see Supplementary Material Table SM-1). Probably, the low TiO₂ levels originate from background levels instead of added TiO₂ levels (see below). The decision to use these low values may have caused an underestimation of the added TiO₂ intake and may explain the lower intake of TiO₂ calculated as compared to the other studies (see below), as application of TiO₂ is authorised in certain product groups (fine bakery wares, confectionary, sauces and beverages), and other products within the same product group may contain added TiO₂.

• Background intake to TiO₂ not taken into account: Not all food product groups have been included in the intake calculation (such as meat and vegetables) because E 171 is not expected to be added to these products and there are no reliable data to prove this assumption wrong. Because of the Ti present in milk, which suggests that TiO₂ particles may originate from indirect sources or natural background, it is possible that, although TiO₂ has not been added, Ti (and TiO₂ particles) may be present in certain food products. For instance, in other milk-derived products, but possibly also in meat, or in plant-derived products. To obtain an indication of the contribution of background levels of TiO₂ by milk products, we performed a second intake estimation in which measured levels of TiO₂ in regular milk products were added to the intake estimation of TiO₂ from food products, food supplements and toothpaste (data not shown). The outcome of this estimation was almost identical (differing maximally 1.6%) to the results presented in Table 2 (i.e. the intake estimation not taking background levels of TiO₂ in regular milk products into account). This indicates that the contribution of background levels of TiO₂ via milk products, although consumed extensively in the Netherlands, was minimal. Yet, as no reliable data are available on possible background levels in other foods like meat, vegetables and cheese and the number of NPs in the background levels are unknown, no definite conclusions on this contribution can be drawn. Since the measured Ti levels in processed dairy products are in the same range as raw milk (see Table 1), of which in some samples the Ti levels were even below the LOD, these levels may also be background levels. However, since the use of E 171 is authorised for these products, we decided to assume these Ti levels to originate from added TiO₂ and included these levels in our intake estimation. It should be noted, though, that the contribution to the total intake is minimal.

• Assuming all measured elemental Ti originates from added TiO₂: the levels in food, food supplements and toothpaste were measured as the element Ti, and for some products, reported in the literature, as TiO₂ particles. For the products of which only Ti was measured and in which the presence of E 171 seemed plausible, it has been assumed all Ti comes from added TiO₂ (E 171). Based on the data of about 20 products in which both Ti and TiO₂ particle content were measured (Peters et al., 2014), this assumption seems to be justified. However, as other speciations containing Ti are possible, the presented TiO₂ intake results may be overestimated.

• Differences in TiO₂ concentration levels between regions: In the present study, we included literature data of food products in which Ti levels have been measured. However, as the authorisation or level of application of E 171 can be, or could have been different between regions, the food industry may apply different TiO₂ concentration levels per country or continent for a specific product type. It is thus not known to which extent the TiO₂ concentration levels (from Weir et al. (2012) and Lomer et al. (2000)) are representative for the Netherlands. The levels in the Netherlands for a specific application might be higher, lower or even be zero when no TiO₂ has been added. This may cause an over- or underestimation of the TiO₂ intake presented in this study.

• Ti-concentration of food supplements: In the present study, the Ti concentration of two food supplements has been determined. The two measurements are very different, resulting in an average Ti concentration (± SD) of 744 (±1009) mg/kg products. Measurements in more different food supplements are needed to obtain a more reliable average value for food supplements. The presented TiO₂ intake based on the Ti concentration of two food supplements may be over- or underestimated.

• The broad range in the estimation of nanofraction of E 171: the nanofraction of E 171 was estimated at 0.012–0.31% (by mass) by Peters et al. (2014). Although the maximum value of 0.31 weight% of Peters et al. was chosen to calculate the nanofraction (<100 nm) of TiO₂ (TiO₂ NPs) we acknowledge that there is uncertainty around the true exposure to TiO₂ NPs as the other reported values were higher (Warheit et al., 2015; Weir et al., 2012; Yang et al., 2015), the percentage of 0.31% is possibly an underestimation and may have led to an underestimation of the intake of TiO₂ NPs in the present study.

• Intake of toothpaste: We assumed no intake from toothpaste by children >6-year-old, adults and elderly, which might result in an underestimation.

• Intake of other TiO₂ containing products. We have only considered the oral intake from food, supplements and toothpaste (in young children only). However, there may be additional intake from other sources (e.g. lip balm, drugs), thus the total oral intake of TiO₂ may have been underestimated. Oral intake through drinking water is expected to be negligible. In addition to oral exposure, dermal exposure from e.g. sunscreens does not appear to lead to an uptake in the body (SCCS, 2013), and inhalatory exposure, excluding potential occupational settings, seems to be very limited compared to oral exposure (Peters et al., in preparation).

Overall, the intake of added TiO₂ in the Dutch population as estimated in the present study aims at a realistic intake, but may have been underestimated.

Comparison of measured Ti levels to Ti levels reported in literature

The Ti-concentrations measured in Dutch, regular dairy products (average 0.47 mg Ti/kg product), including several milk and yoghurt samples, were in line with values reported by others including the one low-fat milk sample analysed by Weir et al. (0.26 mg Ti/kg product), or market milk analysed by Lavi and Alfassi (0.25 mg Ti/kg product) (Lavi & Alfassi, 1990; Weir et al., 2012). Also, the raw milk samples analysed in the present study contained similar levels of Ti (average 0.31 mg Ti/kg product) and therefore the Ti levels in the regular dairy products can be explained by the background level of Ti originating from cow milk. Ti has been measured in raw milk of cows, in levels similar to those found in the present study, e.g. 0.11 mg Ti/kg product (Anderson, 1992) or 0.40 (Grebennikov et al., 1964) mg Ti/kg product.

According to the EU legislation, TiO₂ (E 171) may not be added to unflavoured milk and unflavoured fermented milk products, only to “flavoured dehydrated milk as defined by Directive 2001/114/EC”, “other creams, only flavoured creams”, “flavoured, fermented milk products including heat-treated products”, “flavoured unripened cheeses”, “edible cheese rind”, “whey cheeses”, “flavoured processed cheeses”, “flavoured unripened cheese products” and “dairy analogues, including whitening beverages”.

All 11 “regular” dairy products (i.e. milk, semi-skimmed milk, skimmed milk, long-life milk and yoghurt), to which addition of TiO₂ as E 171 is not allowed, demonstrated an average Ti concentration of 0.47 mg Ti/kg product, while the 10 processed dairy products, to which addition of E 171 in principle cannot be excluded, demonstrated an average Ti concentration of 0.12 mg Ti/kg product (Table 1). As also all the raw milk samples contained Ti, with an average concentration of 0.31 mg Ti/kg product, and the results of de TiO₂ particles analysis demonstrated that TiO₂ particles (size 70-500 nm) were present both in samples of raw milk and in regular milk products, we concluded that the Ti levels in these regular milk products are likely to originate from other sources, such as the environment or animal feed, rather than from direct addition of E 171. Likely, also the Ti concentrations detected in processed dairy products and cheese may be (partly) explained by this background level of Ti present in cow milk.

Concerning products other than dairy, Ti levels presently analysed in energy, soft and sports drinks were lower than the Ti concentrations Weir et al. (2012) detected in two of such beverages (0.18 and 0.37 mg Ti/kg product). High Ti concentrations were detected in a specific food supplement with TiO₂ listed on its ingredient list, as well as in powdered coffee creamer. These values are within the range reported before by Peters et al. (2014).

Taken together, the Ti concentrations presently measured in products are in good agreement with literature data, if available. Low levels of TiO₂ particles are present in milk products that can be attributed to the background. Also in some other products, low levels have been observed that may not originate from added TiO₂.

Contribution of food products, food supplements and toothpaste to intake of TiO₂

The products contributing most to the TiO₂ intake in young children (2–6 years old) are toothpaste, confectionary (sweets, chocolate products, chewing gum) and fine bakery wares (biscuits). The products differ from those in the other age groups (see Table 3). In 7–69-year-olds and elderly (70+), the products in the top 10 are comparable although the ranking is different: confectionary (chewing gum), milk products (creamer and milk powder, cappuccino), sauces (dressings and savoury sauces) and fine bakery wares (cake for age group 7–69 year old and pasties for age group 70+). The top 10 food products, food supplements and toothpaste covers 78%, 55% and 42% of the TiO₂ intake in young children, 7–69-year-olds and elderly, respectively. Most food products only contribute <5% to the TiO₂ intake, indicating the intake is spread over many products. A clear exception is the toothpaste for young children, contributing to over 57%. The intake calculation for young children is, therefore, sensitive for this contribution. We applied a worst-case scenario there, with a full bar of toothpaste that is completely ingested. When brushing teeth twice a day, this results in a daily intake of 2.3 g of toothpaste per day. If a smaller portion (the advised pea size portion of approximately 0.6 g used twice daily) is taken and it is assumed 30% of it is ingested (SCCP, 2005), this would result in a daily ingestion of 0.36 g of toothpaste, a factor 6 lower. When considering lifelong exposure, the contribution of toothpaste is limited (assuming toothpaste is not ingested by persons >6 year old), and the discussion on a pea or bar sized portion of toothpaste less relevant.

In the intake calculation for 7–69-year-olds and elderly (70+) we did not take toothpaste into account because of the assumption that children >6 years old, adults and elderly do usually not swallow toothpaste and, data on toothpaste have not been recorded in the DNFCS and DNFCS-older adults. The SCCS, however, advises a retention factor of 5% for toothpaste exposure calculation for adults (SCCS, 2012). According to probabilistic modelling of European consumer exposure, people (aged 17–74 years) use on average 2.09 g toothpaste per day (P50 = 2.10; P95 = 2.96 g per day), or 29.85 mg/kg bw per day (P50 = 28.67, P95 = 48.61 g/kg bw per day) (Hall et al., 2007; McNamara et al., 2007). Following these values, the SCCS guidance, and the Ti concentration of toothpaste as determined by Peters et al. (2014), this would result in an additional mean oral intake of 0.009 mg TiO₂/kg bw per day (P50 = 0.009, P95 = 0.015 mg TiO₂/kg bw per day) in people >6 years old. For 7–69-year-olds this would result in an additional 5%, 11% and 5% TiO₂ intake to the mean, P50 and P95 values in Table 2, respectively; for elderly (70+) this would result, respectively, in an additional 15%, 29% and 6% TiO₂ intake.

Results of the present study compared to other intake estimations

In Table 4, the calculated P50 intake levels of TiO₂ as calculated by this study and other studies are presented. There are two other intake estimates in the Dutch population: an intake estimate for Dutch children (Huybrechts et al., 2010), giving a median intake of 12–13.0 mg/kg bw per day, and a very recent intake estimation of E 171 with use levels provided by the industry, giving a median long-term intake of 1.4, 0.7 and 0.5 mg/kg bw per day for the age groups 2–6 years, 7–69 years and 70+, respectively (Sprong et al., 2015). The intake levels in the present study were substantially lower than those reported by Huybrechts et al. (2010) and Sprong et al. (2015) and also lower than the TiO₂ intake estimations for the German, UK and US population (Bachler et al., 2015; Weir et al., 2012) (Table 4). It should be noted that the results of the different studies cannot be compared directly to each other, due to the specific assumptions and choices that have been made in each study. Differences in TiO₂ estimations can be caused by differences in product groups and products included, in used Ti and TiO₂ levels of products, the difference in the assignment of those levels to products consumed in the food consumption surveys, and differences in the intake scenarios and the calculation methods used.

Table 4. Intake estimations of TiO₂ per age group as calculated by the present study, and by other comparable studies.

Intake to TiO2 (mg/kg bw per day)	Details	Source
Young children
0.59	2–6 years old, the Netherlands, P50 (point value), average TiO2 concentrations	Present study
1.4	2–6 years old, the Netherlands, P50, tier 4	Sprong et al., 2015
12.0–13.0	<10 years old, the Netherlands, P50, average measured TiO2 levels when assuming 5–20% of product consists of coating	Huybrechts et al., 2010
9.31 (3.89–13.8)	<10 years old, average EU, P50, average measured TiO2 levels when assuming 5–20% of product consists of coating (range found over nine countries)	Huybrechts et al., 2010
1–2	<11 years old, USA	Weir et al., 2012
2–3	<11 years old, UK	Weir et al., 2012
Older children, adults
0.08	7–69 years old, the Netherlands, P50 (point value), average TiO2 concentrations	Present study
0.7	7–69 years old, the Netherlands, P50, tier 4	Sprong et al., 2015
0.2–0.7	≥11 years old, USA	Weir et al., 2012
1	≥11 years old, UK	Weir et al., 2012
0.5–2.0 (Ti intake)  0.8–3.3 (TiO2 intake)^a	Range over six age classes, Germany, P50	Bachler et al., 2015
Older adults
0.03	70+, the Netherlands, P50 (point value), average TiO2 concentrations	Present study
0.5	70+ years old, the Netherlands, P50, tier 4	Sprong et al., 2015

^aBachler et al. calculated the Ti intake (0.5–2.0 mg/kg bw per day), not the intake of TiO₂. For comparison, we multiplied the Ti levels by 1.67 to obtain TiO₂ levels (and assuming that Ti levels originate only from added TiO₂).

The intake of certain food additives can differ markedly between countries, as is the case for example for caramel colours (E 150 a,b,c,d) (EFSA, 2004b). The estimated intake values of TiO₂ in the present study are derived specifically for the Netherlands, using Dutch consumption data and additional analyses of products specific for the Dutch market, and therefore one could argue about the applicability of these values to the intake of TiO₂ in other countries, other than being indicative values in addition to other intake estimates as reported in Table 4. Many of the products contributing may contain similar concentrations of TiO₂, and are used in other countries as well. Depending on habits, for example, the finding that toothpaste is the main contributor to the oral intake by young children could easily be applicable to certain other countries as well. On the contrary, for older people, one could imagine a more limited intake via coffee creamer in certain other countries compared to the Netherlands, however, in other countries other particular products could be main contributors. The variability in E 171 intake does not have to be that great, as, for example, the estimated intake of E 171 by children determined in different European countries according to the Expochi study was demonstrated to only vary about three-fold (Huybrechts et al., 2010), although the intake estimation in that study was conducted less sophisticatedly as has been performed in the present study. Still, it would be very interesting to determine TiO₂ intake levels for other populations as well, especially with regard to the limited margin of exposure as calculated for TiO₂ NPs (Heringa et al., submitted).

Although direct comparison is difficult, some reflections on for the lower intake estimations in the present study can be given. In comparison to the present study, Bachler et al. (2015) included drugs as product group and they assumed long-life ingestion of toothpaste in the German population, which may (partly) explain the higher intake levels of Ti in the German population. The difference in intake between the present study and the intake estimations of Weir et al. (2012) may be explained by the difference in consumed products between the US, UK and Dutch population, but also by the TiO₂ levels assigned to the different food products.

In the Expochi study (Huybrechts et al., 2010), TiO₂ intake has been estimated for Dutch children, giving a median exposure of 12.0–13.0 mg/kg bw/d in a conservative scenario. They used maximum reported use levels and assumed 100% use in all foods in which E 171 is authorised, which explains the higher intakes in the Expochi study, in comparison to the intake values in the present study.

Recently, in parallel to the intake estimation presented in this manuscript, the long-term intake of TiO₂ in the Dutch population was also estimated based on authorised use in the EU and use levels provided by the industry (Sprong et al., 2015). Sprong et al. calculated the dietary intake of E 171 in four tiers. In Tier 4, disaggregated food categories and mean values of reported use levels were applied. Using the most sophisticated tier 4, the estimated median (P50) values for long-term exposure to TiO₂ (E 171) are 1.4 mg/kg bw/d for young children (2–6 years old), 0.7 mg/kg bw/d for 7–69-year-old people and 0.5 mg/kg bw/d for elderly (70+) (Sprong et al., 2015). The respective P95 values are 4.9, 2.8 and 1.8 mg/kg bw/d (Sprong et al., 2015). It was concluded by Sprong et al. that many food groups contribute to the intake to E 171; the most important ones are fine bakery wares, dessert and sauces. Fine bakery wares contributed most to the intake of E 171 and ranges from 43% in young children (2–6 years old) to 53% in elderly. The differences in results between Sprong et al. (2015) and the present study are most likely caused by the differences in scenario (respectively, a more conservative scenario versus an average, realistic scenario) and in the TiO₂ concentration levels used. Sprong et al. used the mean of positive use levels thus excluding zero concentrations reported by the industry whereas we use average TiO₂ levels based on measured Ti levels, including non-detects. It is noted that for most product groups, the reported use levels are (much) higher than the TiO₂ concentration levels based on Ti-measurements in the present study. For instance, Sprong et al. used a TiO₂ level of 1138 mg/kg product for fine bakery wares (based on three positive use levels ranging from 40 to 2338 mg/kg) while our average TiO₂ values for fine bakery wares (cake, pies, pastries, cookies and biscuits) were (much) lower and ranged from 0.4 to 443 mg/kg product (see Supplementary Material SM-1). Why these values differ so much is not clear. Most probably the use levels and measured Ti levels are based on different products. Certainly, the inclusion of some products without TiO₂ on the label in the present study, contribute to these differences as levels in these products might be substantially lower than in labelled products. Another explanation might be that reported use levels represent a maximum that can be added, resulting in lower TiO₂ levels when measured.

The present study aimed at estimating a realistic value for the intake of added TiO₂ NPs, and is based on actual measurements of Ti and TiO₂ particle concentrations in products. As a consequence, and based on the comparison with the other studies, we conclude that the intake of added TiO₂ and TiO₂ NPs in the Dutch population calculated in the present study is realistic but might be underestimated. For a realistic to worst-case estimate, the range between the intake values of the current study and the intake values of Sprong et al. should also be taken into account. The separate risk assessment of TiO₂ NPs is conducted using the P95 intake values resulting from the present study (Table 2); the difference compared to risk assessment of the worst case intake based on the values by Sprong et al. will be discussed accordingly (see Heringa et al., submitted).

Conclusion

The dietary intake of the Dutch population to TiO₂ NPs was calculated on the basis of the DNFCS and literature values of the Ti or TiO₂ particle concentrations in food products, food supplements and toothpaste using the OIM method. Additional analysis of the Ti concentration of typical TiO₂ containing Dutch food products, as well as of a range of dairy products, for which also a background level of Ti/TiO₂ has been reported, was performed. The presently estimated mean long-term exposure to TiO₂ from food products, food supplements and toothpaste, used by the Dutch population ranges from 0.06 mg/kg bw per day for elderly (70+), 0.17 mg/kg bw/day for 7–69-year-old people, to 0.67 mg/kg bw per day for young children (2–6 years old). The mean estimated long-term intake of TiO₂ NPs ranged from 0.19 μg/kg bw/day for elderly, 0.55 μg/kg bw/day for 7–69-year-old people, to 2.16 μg/kg bw/day for young children. Ninety-fifth percentile values were about 3.9-, 2.9- and 1.9-fold higher, respectively. Toothpaste dominated the contribution to the dietary intake by young children (57%), followed by sweets and cookies. For 7–69-year-old people and elderly coffee milk and creamer, chewing gum and sauces contributed most to the dietary intake of TiO₂ NPs. Although the investigated dairy products contained Ti and TiO₂ particles, raw cow milk samples demonstrated this to originate from the background or indirect sources. The contribution of this background level in milk and yoghurt to the estimated long-term dietary intake of TiO₂ NPs would be only minimal (<1.6% additional long-term intake). With several assumptions and its limitations as discussed, an up to date realistic estimation for the oral intake of TiO₂ NPs, valid for the situation in the Netherlands, has been performed. The outcome can be used for health risk assessment.

Declaration of interest

This research was commissioned and financed by The Netherlands Food and Consumer Product Safety Authority (NVWA). The authors declare that they have no competing interests.

Supplementary material available online

Acknowledgements

We are grateful to Jacqueline Castenmiller and Dirk van Aken for their contribution to discussions. We thank P. Nobels and W. Schuurmans of Wageningen University for their assistance with the ICP-HRMS analysis.