Probabilistic health risk assessment of zinc oxide nanoparticles from consumer products in adult populations

Abstract

Zinc oxide nanoparticles (n-ZnO) are one of the most ever-increasing utilized nanomaterials in consumer products. Due to their antibacterial properties and superior efficiency in absorbing ultraviolet radiation, they are widely used as additives in food packaging and sunscreens. There is thus a need for scientific understanding of risks to the health of adult populations associated with n-ZnO. However, due to inadequate data in relation to characterizing hazards and exposure, there is a substantial uncertainty in risk assessment. In the present study, probabilistic approaches, including Monte Carlo and bootstrap methods, were integrated to assess the relative uncertainties and risks of n-ZnO to the health of males and females. Two major exposure pathways, oral from food packaging and percutaneous from sunscreen-based comestics, were evaluated by considering the uncertainty and variability involved in the exposure assessment. Given the cumulative uncertainties of all the extrapolation factors, the results showed that the individual margin of exposure (IMoE) of n-ZnO exhibited a minimal risk through oral exposure, with a minimum value of 786 for males and 96.2 for females (5th centile). However, within the entire range of IMoE values by Monte Carlo simulation through dermal exposure, the IMoE values in 11.45% of exposure scenarios for males and 18.87% for females were lower than the upper limit of the acceptable risk (IMoE ≤ 1). Intra-species, inter-species, and subacute-to-chronic extrapolation factors in the hazard assessment process contributed up to 97% of the uncertainty. These findings provided a scientific basis for understanding risks to the health of adult populations that could help allow regulatory acceptance of consumer products containing n-ZnO and highlighted the need for additional studies on hazard and exposure assessments of nanotechnologies.

---

Environmental significance

Nanomaterials are being used to drive the development and production of new technologies as a result of their infinite sizes, shapes, types and compositions. Meanwhile, the release of these materials into the environment poses a potential hazard to human health. In addition, their innovative and varied properties pose challenges to existing tools. Therefore, it is useful to assess the potential risks of n-ZnO to human health to guarantee that nanotechnology-based consumer products are used in a secure and sustainable way. This paper puts forward an approach to introduce probability into health risk assessment and bridge the gap resulting from inadequate data of human health on nanomaterials.

1. Introduction

According to a statistical report by the Nanotechnology Consumer Products Inventory in 2015, due to the rapid development of nanotechnology, more than 1814 nanomaterial-based consumer products have been transferred from the workbench of the laboratory to the shelves of stores, a 30-fold increase from the 54 products listed in 2005.^1,2 There are currently 5224 nano-related products in The Nanodatabase developed by the Technical University of Denmark (DTU) Environment, the Danish Ecological Council and the Danish Consumer Council (https://www.nanodb.dk/).³ Among the commonly utilized nanomaterials, zinc oxide nanoparticles (n-ZnO), which exhibit superior anti-bacterial and anti-ultraviolet properties, are used in a broad range of applications, including personal care products, food additives, drug delivery vehicles, packaging and biosensors.^4,5 There are continuous concerns about the environmental release of n-ZnO and their potential hazards to the health of humans. For instance, several in vivo and in vitro studies indicated that n-ZnO caused adverse effects when migrating through cells, blood and tissues (e.g. liver, lungs, and brain).⁶ The mucosal cells of the human gastrointestinal tract and excrement containing n-ZnO influenced the biological diversity of human intestinal flora and indirectly affected human health.⁷ However, existing health risk assessments of nanomaterials in humans (HHRA) mainly focused on nano-TiO₂ in sunscreen products,⁸ nano-Ag in sterilization sprays,⁹ flame retardant material-coated multi-walled carbon nanotubes,¹⁰ nano-scale copper oxide (CuO) and basic copper carbonate (Cu₂(OH)₂CO₃) in both ionic and micronized wood.^11–13 The risk assessment associated with n-ZnO is still inadequate, thus hampering further regulatory acceptance of these nanomaterials.¹⁴ It is thus necessary to assess the risks of n-ZnO based on the recent development in research on characterizing exposures and potential effects of nanomaterials.

Hazards of n-ZnO to human health have been extensively studied in recent years.^15–17 Several studies suggested that n-ZnO can cause cytotoxicity, neurotoxicity, mutagenicity and oxidative stress damage to the liver and pancreas.^18–20 N-ZnO had the potential to generate reactive oxygen species (ROS) in rats after 14 days of exposure, which resulted in mitochondrial-mediated apoptosis. Moreover, accumulation of n-ZnO in the liver at a concentration of 300 mg kg⁻¹ wet mass could damage cells, with serum alanine aminotransferase (ALT) of 98 U L⁻¹ and alkaline phosphatase (ALP) of 624 U L⁻¹.²¹ Feeding male and female rats 536.8 mg kg⁻¹ n-ZnO for 90 days induced moderate or mild pancreatitis.²² The effects of n-ZnO on human health were also manifested in damage to the microbiome of the gut, which not only led to dose-dependent changes in the composition and diversity of the intestinal flora but also changed their key functional pathways.²³ The proposed toxic mechanism of n-ZnO was by free Zn²⁺ produced by dissolution, but some influencing factors, such as particle size, surface charge, and environmental conditions (pH and dissolved organic matter), should be taken into account.^24–28 Therefore, the risk assessment of n-ZnO is still technically feasible if the uncertainties in the hazard assessment can be completely considered.

Humans can be exposed to n-ZnO through oral ingestion, inhalation, and percutaneous contact during occupational exposures or daily consumption through non-occupational exposures. Aerosol particles with n-ZnO are produced during the spraying and polishing process, and can reach the alveoli through breathing and cause adverse lung reactions or aggravate existing lung diseases in workers.²⁹ They are also widely used in consumer products, such as sunscreens and food packaging owing to their superior absorption of ultraviolet rays (UVs) and antibacterial properties.³⁰ Other characteristics of n-ZnO particles include good aggregate performance, which suggests that they cannot penetrate into deep skin and present a well-tolerated performance for the skin.^31,32 Collectively, existing studies showed that adding n-ZnO to sunscreens was more beneficial than harmful to human health with respect to protecting the skin from aging and carcinomatous changes caused by UV radiation.³³ However, there are still concerns related to the toxic effects of n-ZnO particles. Kim et al. compared the toxic effects of four metal oxide particles (Al₂O₃, CeO₂, TiO₂, and ZnO) on epithelial cells in lungs of humans, A549 cancer, and L-132 cells. N-ZnO were found to produce a strong cytotoxic effect in terms of cell proliferation, cell viability, membrane integrity, and colony formation.³⁴ Food packaging is also an important vector for oral intake of n-ZnO. Some applications in food packaging feature a coating of n-ZnO, which releases Zn²⁺ to inhibit growth of bacteria. This prolongs the quality guarantee period of the food, but it can also result in exposure of consumers.³⁵ The media of exposure, including release of Zn²⁺ from n-ZnO, physiological differences (e.g. age and sex) of those exposed, and intake capacity or frequency make the HHRA of n-ZnO more complicated and challenging.

Based on an existing deterministic framework for HHRA,³⁶ the aim of this study was to comprehensively apply a probabilistic HHRA of n-ZnO in adult populations exposed to primary nanomaterial-based consumer products. Uncertainties were explicitly considered to allow for robust risk management decisions. Considering the particularity of percutaneous exposure, independent assessment processes were made for males and females. Due to insufficient data for the HHRA, multiple extrapolation factors and estimated values were inevitably introduced, which produced great uncertainties for the results of the risk assessment. Herein, the present study applied probability approaches (e.g. Monte Carlo and bootstrap) to quantify the associated uncertainties and to visualize the distributions of risks. This approach can be applied to promote understanding for regulatory bodies to govern the uses of nanomaterials in consumer products.

2. Materials and methods

2.1 Problem formulation

Nanomaterial, as defined by the European Commission, is a term used to describe a natural, incidental or manufactured material consisting of solid particles that are present either on their own or as identifiable constituent particles in aggregates or agglomerates, where 50% or more of these particles in the number-based size distribution fulfil at least one of the following conditions: (a) one or more external dimensions of the particle are in the size range 1 nm to 100 nm; (b) the particle has an elongated shape, such as a rod, fiber or tube, where two external dimensions are smaller than 1 nm and the other dimension is larger than 100 nm; (c) the particle has a plate-like shape, where one external dimension is smaller than 1 nm and the other dimensions are larger than 100 nm.^37,38 N-ZnO are usually synthesized and manufactured as nanoparticles.³⁹ The diameter of n-ZnO particles is inversely proportional to the surface area, with smaller particles providing superior performance for absorbing UVs.⁴⁰ During interactions with bacterial surfaces, n-ZnO can release Zn²⁺ and thus have effective antibacterial properties, which allows for extensive applications in food packaging and delivery of biopharmaceuticals.^41–43 To provide decision makers with valuable information and improve the regulatory acceptance of these products, in the present study, a probabilistic approach was used to perform an HHRA of n-ZnO from major consumer products, including sunscreen-based cream, lipstick and packaging for solid and liquid food.

2.2 Hazard assessment

2.2.1 Hazard identification. The toxicities of n-ZnO were derived from databases of the Web of Science (https://www.webofscience.com/) and the National Knowledge Infrastructure of China (https://www.cnki.net/). The terms ‘zinc oxide nanoparticles’ or ‘n-ZnO’ and ‘toxic*’ were used to search the Web of Science Core Collection with the publication years from January 1975 to May 2021. 31 related studies were obtained and 18 papers were further excluded after metrological analysis (ESI,† Table S1). There were eleven papers on ecotoxicity, three papers on toxicity to bacteria, three papers on chemical reaction kinetics and one paper on toxicity to tumor cells. A Klimisch experimental evaluation was conducted to score all the studies using an in vivo experimental system (ESI,† Table S2). Then, an expert panel, based on the following criteria: adequacy, reliability, relevance, quantity and significance of toxicity endpoints, scored each of the identified studies on a scale from 1 to 5 and screened the toxicity testing of enough quality (ESI,† Table S3).^44,45 Studies with an average score of more than 3 were used to derive CED_animal, within which seven toxicity studies were used to calculate the BMD, AIC, test 4 P-value, and D.O.F. (ESI,† Table S4). The detailed physicochemical properties of n-ZnO, selected for calculating the critical CED, are summarized in the ESI† (Table S5).

2.2.2 Dose–response assessment based on the benchmark dose.
2.2.2.1 Oral toxicity test. Estimation of the benchmark dose (BMD) was performed by applying the dose–response models available in the Benchmark Dose Software (BMDS) developed by the United States Environmental Protection Agency (U.S. EPA) and the National Institute for Occupational Safety and Health (NIOSH) of the United States. The BMD was calculated with the standard deviation (SD) of BMR type I as the benchmark response factor (BMRF), a 95% confidence interval, a normal distribution type and a non-constant type of variance which is decided by the different variances of each dose group. Dose–response curves were fitted by several functions, including exponential, Hill, polynormal, power and linear. Goodness of fit was evaluated by calculating the Akaike information criterion (AIC). More than 100 000 iterations were calculated and the resulting CED distributions were divided by inter-species (AF_inter), intra-species (AF_intra), and sub-acute-chronic assessment factors (AF_{subacute-chronic}) to derive individual human critical effect dose (ICED) distributions (eqn (1)). The mean and SD of CED_animal were predicted by means of the bootstrapping technique with the R project for statistical computing (Version 4.1.2).

ICED = CED_animal/(AF_inter × AF_intra × AF_{subacute-chronic}) (1)

AF_inter = AF_inter1 × AF_inter2 (2)

AF_inter1 = [human bw/animal bw]^{1−scaling power} (3)

where ICED represents the individual human critical effect dose, in mg kg⁻¹ d⁻¹, CED_animal represents the critical effect dose for toxicity testing, in mg kg⁻¹ d⁻¹, and AF_inter represents the product of two sub-factors AF_inter1 and AF_inter2 (eqn (2)). AF_inter1 corrects for body mass between test animals and humans and is calculated using eqn (3). AF_inter2 estimates the substance-specific physiologically based toxico-kinetic and toxico-dynamic (PBTK/TD) differences between test animals and humans.

Variability in sensitivities among individual humans was represented by AF_intra, which was defined as a lognormal distribution with a specific geometric mean (GM) and a geometric standard deviation (GSD). Due to the lack of estimates of n-ZnO, default AF_intra values were used in this study.⁴⁶ AF_{subacute-chronic} was estimated by a lognormal distribution, which was empirically derived from subacute and chronic data.⁴⁷

Another recommended approach to compute the RfD(o) by the U.S. EPA was used as a reference approach, which is shown in eqn (4):⁴⁸

RfD = NOAEL (or LOAEL or BMDL or BMD)/UFs (4)

where RfD is defined as an estimate of daily oral/dermal exposure of the human population, in mg kg⁻¹ d⁻¹, no observed adverse effect level (NOAEL) is the highest exposure level at which no statistically or biologically significant increases are seen in the frequency or severity of adverse effects between the exposed population and its appropriate control population, lowest-observed-adverse-effect level (LOAEL) is the lowest exposure level, and BMDL represents the lower confidence limit of BMD (ESI,† Fig. S1). To develop the BMDL, a predetermined change in the response rate of an adverse effect (called the benchmark response or BMR; generally in the range of 1 to 10% depending on the power of a toxicity study) is selected, and the BMDL is a statistical lower confidence limit on the dose that produces the selected response. UFs are uncertainty factors that concern the variability and uncertainty that are reflected in possible differences between test animals and humans and variability within the human population.

2.2.2.2 Dermal toxicity test. The main evidence for transdermal penetration in relation to ZnO nanoparticles is systemic availability as shown by detection of the stable isotope Zn in blood/urine.⁴⁹ In most cases, the toxicological database does not include detailed testing on all possible routes of administration, with possibly significant differences in factors such as mode of action and bioavailability. Consideration is given to potential differences in absorption or metabolism resulting from different routes of exposure, and whenever appropriate data are available, the quantitative impacts of these differences on the risk assessment are delineated.⁵⁰ The U.S. EPA assumes that oral toxicity data can be transformed into dermal risk assessment when toxicity data on dermal absorption are not available. The dermal dose was calculated as follows (eqn (5)):^51,52

RfD(d) = RfD(o) × Absorption factor (5)

RfD(d): skin contact reference dose, mg kg⁻¹ d⁻¹; RfD(o): oral intake reference dose, mg kg⁻¹ d⁻¹; absorption factor: absorption efficiency factor, non-dimensional.

This can be calculated by eqn (6):

BMD(d) = BMD(o) × Absorption factor (6)

2.2.3 Exposure assessment. Exposures of the general population to n-ZnO, but not in an occupational situation, is through nasal and transdermal intake from sunscreen products and oral intake from food packaging.

Estimation through dermal exposure to n-ZnO from sunscreen products was calculated using eqn (7):⁵³

V_sunscreen = (DA × EV × ED × EF × SA)/(BW × AT) (7)

where V_sunscreen represents the dermally absorbed dose, expressed in mg kg⁻¹ d⁻¹, DA is the absorbed dose per event, in mg cm⁻² per event, SA represents the surface area available for contact, in cm², EV represents the event frequency, in events per d, ED represents the exposure duration, in y, F represents the frequency of exposure, in d y⁻¹, and AT represents the average exposure duration, in d.

Estimation of oral exposure to n-ZnO from food packaging was performed using eqn (8).⁵³

V_food = (C × IR × EF × ED)/(BW × AT) (8)

where V_food represents the orally absorbed dose, expressed in mg kg⁻¹ d⁻¹, C represents the concentration of n-ZnO in the diet, in mg kg⁻¹, IR represents the daily intake, in kg d⁻¹, F represents the frequency of exposure, in d y⁻¹, ED represents the exposure duration, in y, BW represents the body weight, in kg, and AT represents the average exposure duration, d. Estimates of exposure were reported as point values that were then converted to distributions of individual exposure levels (IEXP). The IEXP involving cumulative exposure were calculated by summing up the daily intakes of n-ZnO released from various products (eqn (9)):⁵⁴

IEXP = (U₁V₁ + U₂V₂ + U₃V₃…… + U_nV_n) (9)

where V_1,2,…n is the variability of the assessed n-ZnO linked to product 1, 2, …, n, while U_1,2,…n is the uncertainty of n-ZnO intake in relation to product 1, 2, …, n.

2.2.4 Risk characterization. Risk is expressed as the individual margin of exposure (IMoE) derived from a person's ICED and IEXP (eqn (10)). Risks are not acceptable when IMoE ≤ 1. The associated probability of IMoE is discussed with respect to the probable uncertainty of estimates of risks to the health of humans.

IMoE = ICED/IEXP (10)

To validate the results from the probabilistic risk assessment (PRA), another deterministic approach was also introduced by the U.S. EPA. The margin of exposure (MoE) is used to determine if exposure to a chemical can be expected to cause an adverse effect. The MoE is calculated by dividing the toxicological point of departure (POD) by the estimated daily dose to which humans will be exposed, as shown in eqn (11).⁵⁵

MoE = POD/Daily Dose (11)

The POD is the lower confidence bound on the lowest experimental dose that showed an effect in a dose–response study. The dose is determined from dose–response data and marks the beginning of extrapolation to determine the risk associated with environmentally relevant human exposures. Commonly, it is the NOAEL from an animal toxicity study in the laboratory, which represents the dose at which no adverse effects were observed in laboratory animals.

After calculating the MoE from the POD and daily dose, the U.S. EPA evaluates the risk from exposure to a chemical by comparing the calculated MoE to a target MoE.⁵⁵ To explain the incomplete characterization in the n-ZnO toxicity studies, an additional uncertainty factor (UF_D), which accounts for the incomplete characterization of nanotoxicities, together with AF_inter and AF_intra, should be introduced. In order to extrapolate chronic toxicity (90 days) from subacute toxicity (14 days), a three-fold assessment factor was used to account for greater toxic potency to the liver and bioaccumulation. Based on these considerations, a target MoE was estimated using eqn (12).

target MoE = AF_inter × AF_intra × UF_D × T_chronic/T_subacute (12)

If calculated MoE > target MoE: risk is not of concern and mitigation is not required; if calculated MoE ≤ target MoE: risk is of concern and mitigation is required. In this case, mitigation measures such as engineering controls and/or personal protective equipment are employed until the calculated MoE exceeds the target MoE.

2.2.5 Uncertainty analysis. A probabilistic approach based on Monte Carlo sampling was used to estimate the contribution of uncertainties in the HHRA using Crystal Ball v11.1 (Oracle Corporation, Redwood Shores). The hazard and exposure data were imported into Crystal Ball and distributions of ICED and IEXP were simulated. There was a framework of HHRA of n-ZnO with uncertainty analysis by sensitivity forecasting (Fig. 1).

Fig. 1 A framework describing the process of human health risk assessment of n-ZnO and the uncertainty analysis.

3. Results and discussion

3.1 Hazard identification

Searches of multiple bibliographic databases resulted in 13 in vivo studies (ESI,† Table S1). According to an expert scoring panel⁴⁴ and a weight-of-evidence approach,⁴⁵ the quality of data was evaluated and a toxicity study with the greatest quality scores was selected for hazard assessment of n-ZnO (ESI,† Tables S2–S4).⁵⁶ Healthy Sprague Dawley rats of both sexes aged between 8 and 9 weeks with body masses of 180–220 g were randomly divided into six groups, with five females and five males in each group, and then exposed to 20 nm n-ZnO with distilled water. The range of actual particle sizes was 15–120 nm, with a mean size of particles of 63 nm. Rats were not fed overnight before the test and then fed 5, 50, 300, 1000 or 2000 mg kg⁻¹ bw. The toxicity in vivo was determined based on Guideline 423 of the Organization for Economic Cooperation and Development. Two weeks later, rats of both sexes were euthanized to measure the biochemical indicators of blood serum.⁵⁶ All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of the “Organization for Economic Co-operation and Development (OECD) Guideline 423 (OECD, 2001)” and approved by the Animal Ethics Committee of the “International Institute of Biotechnology and Toxicology (IIBAT)”. Due to the particle size of n-ZnO in daily consumer products being generally less than 100 nm,⁵⁷ toxicity endpoints with an average particle size of 63 nm were used for further analysis (ESI,† Table S5).

3.2 Dose response assessment

Changes in the total number of inflammatory cells in the liver can be considered as a critical endpoint for assessment,¹² in which alanine aminotransferase (ALT) activity is the most sensitive indicator of liver inflammatory changes.⁵⁸ A linear model was used to fit the logarithm of toxicity data with the activity of ALT, which was estimated to be associated with CED_animal values of 43.5 mg kg⁻¹ d⁻¹ for females and 202.0 mg kg⁻¹ d⁻¹ for males (Fig. 2A). Bootstrap sampling generated a normal CED_animal probability distribution defined by SD values of 2 and 19. The 5th to 95th percentiles of CED_animal span from 39.14 to 47.54 mg kg⁻¹ d⁻¹ for females (Fig. 2B) and from 162.1 to 238.2 mg kg⁻¹ d⁻¹ for males (Fig. 2C).

Fig. 2 (A). Dose–response relationship between the concentration of n-ZnO and alanine aminotransferase (ALT) activity; (B). Probability distribution of CED_animal for males (CED: 43.50; CEDL: critical effect dose lower-39.14; CEDU: critical effect dose upper-47.54); (C). Probability distribution of CED_animal for females (CED: 202; CEDL: critical effect dose lower-162.1; CEDU: critical effect dose upper-238.2).

Based on CED_animal, extrapolation factors were applied to estimate the distribution of ICED (ESI,† Table S6). Specifically, the AF_inter1 accounted for the difference between the mean body mass of a healthy Sprague Dawley rat (200 g) and that of an adult Chinese person (the average weight is 60 kg).⁵³ Considering the scaling power of the AF_inter1 was defined within the range of 0.65–0.75, 0.70 was determined with a SD of 0.033. The AF_inter1 was defined as a log-normal distribution with a geometric mean (GM) of 5.54 and a geometric standard deviation (GSD) of 1.2. The AF_inter2 was estimated as a log-normal distribution with a GM of 1.00 and a GSD of 2.00 and AF_intra was a log-normal distribution with a GM of 0.60 and a GSD of 1.60. The AF_{subacute-chronic} was defined as a log-normal distribution with a GM of 4.10 and a GSD of 4.40. These values were empirically derived from subacute and chronic data.⁴⁷

3.2.1 Oral study. Probabilistic distributions of the ICED for males had a range with a mean of 286.9 mg kg⁻¹ d⁻¹, a 5th centile of 0.896 mg kg⁻¹ d⁻¹ and a 95th centile of 969.1 mg kg⁻¹ d⁻¹. By comparison, the ICED for females had a similar probability distribution with a higher mean of 1366 mg kg⁻¹ d⁻¹, a 5th centile of 4.437 mg kg⁻¹ d⁻¹ and a 95th centile of 4788 mg kg⁻¹ d⁻¹ (ESI,† Fig. S2).

3.2.2 Dermal study. According to a report of the Scientific Committee on Consumer Safety (SCCS),⁵⁹ analysis with electron microscopy revealed that penetration of ZnO nanoparticles was limited to the outer surface of the stratum corneum (SC), desquamating cells of the upper SC only. There was no evidence of penetration of nanoparticles in the lower SC layers or viable epidermis. The SCCS considers that 0.03% of the applied dose may be absorbed after topical application of ZnO nanoparticles on the skin. Therefore, the dermal absorption factor is set to 0.03% in a triangular distribution (ESI,† Table S6). Probabilistic distributions of the ICED for males had a range with a mean of 0.084 mg kg⁻¹ d⁻¹, a 5th centile of 0.0002 mg kg⁻¹ d⁻¹ and 95% centile of 0.292 mg kg⁻¹ d⁻¹. By comparison, the ICED for females had a similar probability distribution with a mean of 0.439 mg kg⁻¹ d⁻¹, a 5th centile of 0.0012 mg kg⁻¹ d⁻¹ and a 95th centile of 1.422 mg kg⁻¹ d⁻¹ (ESI,† Fig. S3).

3.3 Exposure assessment

In 2020, global production of metal oxide nanoparticles, specifically those incorporated into skincare products, was estimated to be 1000 tons per year.⁶⁰ The general population is also exposed to n-ZnO-based food packaging with antibacterial properties.^61,62 Therefore, the assessment of incorporated included four sources of exposure, namely, sunblocking cream (P1), sunscreen-based lipstick (P2), solid food packaging (P3) and liquid food packaging (P4) (Table 1). The total released n-ZnO per day from each product was 7.62 ± 1.71 μg cm⁻² (sunblocking cream),⁶² 0.9 mg kg⁻¹ (sunscreen-based lipstick),⁵⁹ 15.51 ± 0.160 mg kg⁻¹ (solid food packaging),⁴¹ and 0.16 ± 0.007 μg L⁻¹ (liquid food packaging).⁶³

Table 1 Intake of zinc oxide nanoparticles (n-ZnO) from selected consumer products by both sexes

	P1	P2	P3	P4
a Based on expert judgement. b 25%: the maximum percentage of nano materials allowed to be added to consumer goods.
Release setup	Sunblocking cream	Sunscreen-based lipstick	Solid food (frozen fish packaging)	Liquid food (orange juice packaging)
Exposure route	Dermal	Oral/dermal	Oral	Oral
Daily release	7.62 ± 1.71 μg cm^−2	≥0.9 mg kg^−1	15.51 ± 0.160 mg kg^−1	0.16 ± 0.007 μg l^−1
Test duration (d)	1	—	14	28
Daily intake in females kg^−1 body mass (mg kg^−1 d^−1)	0.9030 ± 0.2027	0.90*25%^b	0.00262 ± 0.000027	5.4 × 10^−8 ± 2.36 × 10^−9
Probability of intake percentage in females^a	5–6–7%	Oral: 10–20–30% dermal: 4.5–5–5.5%	40–50–60%	40–50–60%
Daily intake in males per kg body mass (mg kg^−1 d^−1)	0.1272 ± 0.0287	—	0.00229 ± 0.0000236	4.72 × 10^−8 ± 2.07 × 10^−10
Probability of intake percentage in males^a	4–5–6%	—	40–50–60%	40–50–60%

---

According to the SCCS Notes of Guidance for the Testing of Cosmetic Ingredients and Their Safety Evaluation (SCCS/1416/11), the daily exposure to sunscreen-based lipstick is 0.9 mg kg⁻¹ d⁻¹. It is expected that exposure to sunscreen-based lipstick that contains high SPF factors (e.g. ZnO nanomaterials) is less than exposure to ‘regular’ sunscreen-based lipstick. Typically, these products are used only in specific time frames, e.g. during outdoor holidays. The value of 0.9 mg kg⁻¹ d⁻¹ is considered a conservative one.⁵⁹ In terms of n-ZnO intake, the probability of intake from sunblocking cosmetics was less than indirect oral intake through food packaging. This result is consistent with the fact that there is little evidence of systemic exposure from penetration of skin by the metal oxide nanoparticles used in sunblocking cream. However, a question has been raised for injured skin. In addition, with respect to differences between sexes in utilizing sunscreens, the frequency of utilization was set to two times per day for males and three times per day for females, which was recommended by the DHI Water and Environment, Denmark.⁶⁴ With reference to the basic exposure parameters provided in a Chinese document named ‘Exposure Factors Handbook of Chinese Population (Adults)’, the standard body mass of adult males is recommended to be 65 kg, with the exposed area with sunblocking cream being 0.44 m², which consists of 0.12 m² on the face, 0.24 m² on arms and 0.08 m² on hands. By comparison, the body mass of adult females is 56.8 kg, with the exposed area with sunblocking cream being 0.91 m², which consists of 0.12 m² on the face, 0.24 m² on arms, 0.08 m² on hands and 0.47 m² on legs.⁶⁵ Daily intake of solid and liquid food was recommended to be 0.25 kg and 0.25 L, respectively. The average daily intake of n-ZnO was estimated to be 0.1272 ± 0.0287 mg kg⁻¹ d⁻¹ for males and 0.9056 ± 0.2027 mg kg⁻¹ d⁻¹ for females (Table 1). This study formulated a worst case involving a person exposed to all these products simultaneously and exposed to the maximum amount of released n-ZnO.

The probability of intake from consuming different products was defined as a triangular distribution (ESI,† Table S6). For oral exposure, the predicted mean of IEXP for females was 0.046 mg kg⁻¹ d⁻¹, ranging from a 5th centile of 0.030 mg kg⁻¹ d⁻¹ to a 95th centile of 0.064 mg kg⁻¹ d⁻¹. By comparison, the predicted mean of IEXP for males (0.0011 mg kg⁻¹ d⁻¹) was lower, ranging from a 5th centile of 0.0010 mg kg⁻¹ d⁻¹ to a 95th centile of 0.0013 mg kg⁻¹ d⁻¹ (ESI,† Fig. S5). Through the dermal exposure pathway, the predicted mean of IEXP for females was 0.0065 mg kg⁻¹ d⁻¹, ranging from a 5th centile of 0.0045 mg kg⁻¹ d⁻¹ to a 95th centile of 0.0087 mg kg⁻¹ d⁻¹. By comparison, the predicted mean of IEXP for males was 0.0006 mg kg⁻¹ d⁻¹, ranging from a 5th centile of 0.0004 mg kg⁻¹ d⁻¹ to a 95th centile of 0.0009 mg kg⁻¹ d⁻¹ (ESI,† Fig. S7).

3.4 Characterization of risk

The forecast of ICED, IEXP, and IMoE in different sexes and exposure pathways was performed by the Crystal Ball software (ESI,† Tables S7–S10). The probability distribution functions of IMoE were fitted by a sigmoidal-logistic model, with acceptable coefficients of correlation (R² ≥ 0.999) (Fig. 3 and Table S11†).

Fig. 3 The cumulative probabilities of individual margins of exposure (IMoE). The curves were fitted by a sigmoidal-logistic model (ESI,† Table S10). The red area shows unacceptable risk where IMoE is lower than 1. The minimum IMoE through oral exposure within the 5–95% interval was greater than that through dermal exposure, for both males and females. Females exhibit a greater probability (18.87%) than males (11.45%) to have risks to human health through dermal exposure.

The higher values of IMoE indicated a lower level of risk. Herein, the IMoE through oral exposure was always higher than 1 (minimum = 786.2 for males and minimum = 96.23 for females within the 5–95% interval), indicating no likely risk to humans associated with lip products and food packaging containing n-ZnO (ESI,† Fig. S6). By comparison, the IMoE through dermal exposure was always partially lower than 1 (minimum = 0.3941 for males and minimum = 0.1800 for females within the 5–95% interval), triggering a non-acceptable dermal risk of 18.87% for females and 11.45% for males (ESI,† Fig. S7). Therefore, it is still necessary to reduce the frequency of n-ZnO sunscreen-based product use and to find safe alternatives to reduce intake of n-ZnO.

Nevertheless, the minimum IMoE values in females through oral exposure were 0.04 within the 0–5% interval (<1) (ESI,† Fig. S8), indicating that n-ZnO posed a potential risk of liver damage to a very small number of the general population.

3.5 Uncertainty analysis

The contributions of the sources of uncertainty are mainly from AF_intra, AF_inter2 and AF_{subacute-to-chronic}, which account for 96% of the total uncertainty (Fig. 4). The dermal absorption factor contributes more than 1% of the overall uncertainty (1.9% for females and 1.7% for males), while the uncertainty due to the variability of intake behaviour (V_n and U_n) was minor (0 to 0.5%).

Fig. 4 Relative contributions of various sources of uncertainty in assessing the risk of n-ZnO.

3.6 Contribution of uncertainty in the hazard assessment

The determination of the toxicity endpoint was a foundation for the HHRA of n-ZnO. Among the reported chronic toxic effects of n-ZnO, injury to the liver was the most concerning toxicity endpoint.⁵⁶ Activities of ALT in liver cells were 1000- to 3000-fold greater than that in serum, and a 1% death of liver cells can double the activity of ALT in blood serum.^66,67 Therefore, ALT was selected as a sensitive biomarker of damage to liver cells. N-ZnO exhibited toxicity to mice at lesser doses, indicating an inverse dose-dependent increase observed for activities of ALT in blood serum in individuals exposed to n-ZnO (Fig. 2A). Compared to micro-sized zinc oxide, incidences of microscopic lesions in liver were greater at lesser doses of n-ZnO (≤2000 mg kg⁻¹ bw), which was consistent with the finding of Seok et al.²² The three extrapolation factors, AF_intra, AF_inter, and AF_{subacute-chronic}, contributed more to the uncertainty of the IMoE, and uncertainties in estimations of exposure were comparatively less (<4%).^12,46 To reduce the overall uncertainty in the step of hazard assessment, toxicity tests on more organisms are needed to provide solid information for exploring the dose–response relationships. Deriving empirical parameters from other substances reduces the accuracy of risk assessment. Importantly, several transdermal experiments are also needed to further explore the dermal mechanism of nanoparticles.

3.7 Contribution of uncertainty in the exposure assessment

Some uncertainties mainly came from the variability of exposure pathways and the estimated probability for the intake of n-ZnO. The predicted distribution of IEXP for females within the 5–95% interval ranged from 0.0300 to 0.0640 through oral exposure and from 0.0045 to 0.0087 through dermal exposure, while it ranged from 0.0010 to 0.0013 through oral exposure and from 0.0004 to 0.0009 through dermal exposure for males (ESI,† Fig. S3 and S7). However, there was no significant difference between the sexes through food packaging.

The exposure assessment was conducted for worst case scenarios, when many measures (e.g. rubbish/washing off the sunscreen by accident, wiping off lipstick when they eat) that may reduce their intake concentrations were not considered. Therefore, one source of uncertainty in the exposure assessment is the proportion of absorbed sunscreen. Particle size is another potential impact factor to determine the dissolution kinetics. Other sources of uncertainty in the exposure assessment may result from the fact that only external doses were considered in the present study and the uptake and translocation in the organism were not included.

The exposure from drinking water was not taken into consideration. After flocculation/sedimentation, microfiltration and ultrafiltration treatment, the concentration of n-ZnO detected in ultimate drinking water samples was lower than 55 μg L⁻¹.^68,69 Therefore, compared to other exposure pathways, the intake of n-ZnO from drinking water can be neglected.

3.8 Contribution of uncertainty in the risk characterization

A contribution to the uncertainty in the IMoE estimates is expressed in Fig. 4, which explains possible determinants for all of the exposure scenarios. 96% of the variation in the IMoE distributions was influenced by uncertainties in AF_inter, AF_intra, and AF_{subacute-chronic}, while the remaining uncertainty was mainly from the dermal exposure. In addition to the incompleteness of the nanotoxicity database, the effects of n-ZnO due to their physicochemical properties (e.g. particle size, surface area, charge) were ignored after consuming nano products.^44,70

The minimum IMoE values for females and males through oral exposure were 0.04 and 3.098 within the 0–5% interval (<1) (ESI,† Tables S8 and S10), indicating that n-ZnO posed a potential risk of liver damage to a very small number of females through consuming lip products.

3.9 Comparison between the IMoE, deterministic margin of exposure (MoE), margin of safety (MoS) and low tier PRA

By comparing the BMDL with exposure data (ESI,† Fig. S4 and S7), the values of MoE for both sexes were calculated using eqn (13)–(16).⁷¹

Through oral exposure:

Through dermal exposure:

The target MoE was calculated to be 6429 in eqn (12). The values of MoE through both oral and dermal exposures were less than the target MoE, which indicates that the risk of n-ZnO from consumer products is unacceptable. Compared to the probabilistic IMoE assessment, the deterministic MOE approach overestimated the risk, which indicates that MoE is apparently a more conservative approach. The different conclusions are mainly from great uncertainties of these assessment factors. For example, the factor UF_D that accounts for the incomplete characterization of nanotoxicities contributed a great uncertainty to obtain the target MoE. Therefore, to reduce the uncertainty of UF_D, more toxicity tests of n-ZnO need to be performed.

The risk of n-ZnO in sunscreen and lipstick is further assessed by an MoS method from the SCCS, where the MoS value of n-ZnO was 7.4 in sunscreen and 3.7 in lipstick.⁵⁹ If the calculated MoS value is more than 1, the risk is identified to be acceptable. The finding is consistent with the result of the probabilistic IMoE assessment for most scenarios.

A low tier probabilistic analysis was performed using the APROBA-Plus tool released by the WHO International Programme on Chemical Safety.⁷² It applies lognormal uncertainty distributions to different aspects of the hazard assessment (e.g. point of departure, assessment factors), resulting in a confidence interval for the human dose associated with specified protection goals, as well as a probabilistic health-based guidance value. The uncertainty in exposure, by including the option to insert a quick and approximate estimate of the exposure uncertainty, is then graphically compared with the probabilistic outcome from the hazard characterization.⁷³ The approximate approach based on lognormal distributions is much quicker and easier to apply, and simulations indicate that the resulting confidence bounds are a reasonable approximation to those obtained in the fully probabilistic approach.

The exposure assessment was divided into four groups, oral exposure for males, dermal exposure for males, oral exposure for females and dermal exposure for females. The plot in Fig. S10† shows the exposure uncertainty on the x-axis and the HD_M^I uncertainty on the y-axis. HD_M^I is depicted as a solid (vertical) line representing the confidence interval. For individual exposure scenarios, the results showed that when the whole ellipse is in the green area of the plot, then the exposure most likely is below HD_M^I, and the protection goals are most likely to be met. This is consistent with our independent assessment results. But the APROBA-Plus approach could not take uncertainties of cumulative exposure into account.

Therefore, among the three approaches, the IMoE approach presents incomparable advantages that we can obtain critical information on risks from a probabilistic perspective.

3.10 Advantages and limitations

Emerging nanotechnologies with high uncertainty challenge existing governance processes to identify, assess, and manage risks. The conventional deterministic risk-based paradigm is not essential for decision makers to evaluate the impact on human health since quantitative data for risk assessment remain incomplete or limited. A PRA approach is introduced to avoid overestimating risks due to the multiplication of several uncertainty factors in deriving point estimates of toxicity values. It is critical to use a higher tier PRA when a deterministic assessment cannot provide sufficient information for decision-makers. In the present case study on n-ZnO, the overall viability and uncertainty came from certain extrapolation factors when performing hazard and exposure assessment. Therefore, obtaining more intra-species and inter-species toxicity data is critical to improving the accuracy of the risk assessment. It is also worth noting that information about consumer exposure characteristics, e.g. gender, age range, weight, and exposure route of concern, needs to be collected in a scientifically rigorous manner.

Herein, although the probabilistic HHRA presented the uncertainty that was needed in supporting a conservative assessment of risks, it should be further developed in three areas, including the following: 1) the n-ZnO products chosen to be included in this HHRA exclude some additional products, such as cosmetics, reusable face masks, food supplements, and liquid food supplements. More work is needed to fill incomplete data for exclusive consumer products with nano-sized ZnO. 2) Risk assessment of n-ZnO in children, who might have greater consumption to body weight ratios, is not a focus of the present study. 3) Relationships between the physico-chemical properties of n-ZnO (size, geometric configuration, and aggregation) and their toxic effects are highly desirable to balance the development of dose–response assessment with the intended use.

4. Conclusions

With the ever-increasing number of nano-enabled consumer products, a risk governance approach is required to quantitatively estimate risks to human health and associated uncertainties. Traditionally, risk assessments have been deterministic and “conservative,” relying on multiple safety factors and often overestimating risks. The framework of the high tier PRA developed here provides a foundation for evaluating potential risks associated with product use based on hazards and exposure, and allows for dealing with input information in a scientifically rigorous manner and supporting democratic decision making in governing emerging technologies. Future work could include multi-criteria decision analysis to characterize and balance the risks, benefits, costs, and societal implications, and investigation of probabilistic exposure models capable of linking to physiologically-based pharmacokinetic (PBPK) models to provide better estimates of the exposure ranges, dose, and risk in individuals in the population. Eventually, a robust decision-support tool based on such work should be developed for a broader suite of nanomaterials. These approaches will facilitate better and more cost-effective risk governance of emerging technologies.

Author contributions

Y. S. M. and X. L. contributed equally to this work. Y. S. M. and F. C. W. conceived the project and designed the model. P. H. C. and C. F. P. collected and analyzed the data. X. L., P. H. C., C. F. P., and H. Z. C. performed statistical analysis and validated the QSAR-SSD model. Y. S. M. and F. G. Z. wrote the paper. J. P. G. revised the paper and gave constructive suggestions.

Conflicts of interest

The authors have no conflict of interest to declare.

Acknowledgements

This research was supported by the Fundamental Research Funds for the Central Universities and the Research Funds of the Renmin University of China (21XNLG25 and 2020030262). The authors would like to thank Dr. Kenneth M. Y. Leung (City University of Hong Kong) and three anonymous reviewers for providing comments on prior drafts. Language editing service was provided by Editage, Cactus Communications Inc. Prof. Giesy was supported by the Canada Research Chairs Program, the 2012 “High Level Foreign Experts” (#GDT20143200016) program provided by the State Administration of Foreign Experts Affairs, P.R. China to Nanjing University, the Einstein Professorship Program of the Chinese Academy of Sciences, and a Distinguished Visiting Professorship in the Department of Environmental Sciences, Baylor University in Waco, TX, USA.

Notes and references

1. M. E. Vance, T. Kuiken, E. P. Vejerano, S. P. McGinnis, M. F. Hochella, Jr., D. Rejeski and M. S. Hull, Nanotechnology in the real world: Redeveloping the nanomaterial consumer products inventory, Beilstein J. Nanotechnol., 2015, 6, 1769–1780.
2. J. Park, S. Ham, M. Jang, J. Lee, S. Kim, S. Kim, K. Lee, D. Park, J. Kwon, H. Kim, P. Kim, K. Choi and C. Yoon, Spatial-temporal dispersion of aerosolized nanoparticles during the use of consumer spray products and estimates of inhalation exposure, Environ. Sci. Technol., 2017, 51, 7624–7638.
3. S. F. Hansen, O. F. H. Hansen and M. B. Nielsen, Advances and challenges towards consumerization of nanomaterials, Nat. Nanotechnol., 2020, 15, 964–965.
4. L. Voss, P. E. J. Saloga, V. Stock, L. Bohmert, A. Braeuning, A. F. Thunemann, A. Lampen and H. Sieg, Environmental impact of ZnO nanoparticles evaluated by in vitro simulated digestion, ACS Appl. Nano Mater., 2020, 3, 724–733.
5. S. M. Youn and S. J. Choi, Food additive zinc oxide nanoparticles: dissolution, interaction, fate, cytotoxicity, and qral toxicity, Int. J. Mol. Sci., 2022, 23, 6074–6092.
6. B. Wang, W. Y. Feng, M. Wang, T. C. Wang, Y. Q. Gu, M. T. Zhu, H. Ouyang, J. W. Shi, F. Zhang, Y. L. Zhao, Z. F. Chai, H. F. Wang and J. Wang, Acute toxicological impact of nano- and submicro-scaled zinc oxide powder on healthy adult mice, J. Nanopart. Res., 2007, 10, 263–276.
7. H. Bouwmeester, M. V. D. Zande and M. A. Jepson, Effects of food-borne nanomaterials on gastrointestinal tissues and microbiota, Wiley Interdiscip. Rev.: Nanomed. Nanobiotechnol., 2017, 1–12.
8. U.S. EPA, Nanomaterial case studies: nanoscale titanium dioxide in water treatment and in topical sunscreen (final), Report EPA/600/R-09/057F, Research Triangle Park, NC, 2010.
9. U.S. EPA, Nanomaterial case study: nanoscale silver in disinfectant spray (external review draft), Washington, DC, 2010.
10. U.S. EPA, Nanomaterial case study: a comparison of multiwalled carbon nanotube and decabromodiphenyl ether flame-retardant coatings applied to upholstery textiles (draft), Research Triangle Park, NC, 2012.
11. V. Cazzagon, E. Giubilato, L. Pizzol, C. Ravagli, S. Doumett, G. Baldi, M. Blosi, A. Brunelli, C. Fito, F. Huertas, A. Marcomini, E. Semenzin, A. Zabeo, I. Zanoni and D. Hristozov, Occupational risk of nano-biomaterials: Assessment of nano-enabled magnetite contrast agent using the BIORIMA Decision Support System, NanoImpact, 2022, 25, 100373.
12. D. Hristozov, L. Pizzol, G. Basei, A. Zabeo, A. Mackevica, S. F. Hansen, I. Gosens, F. R. Cassee, W. de Jong, A. J. Koivisto, N. Neubauer, A. Sanchez Jimenez, E. Semenzin, V. Subramanian, W. Fransman, K. A. Jensen, W. Wohlleben, V. Stone and A. Marcomini, Quantitative human health risk assessment along the lifecycle of nano-scale copper-based wood preservatives, Nanotoxicology, 2018, 12, 747–765.
13. L. Pizzol, D. Hristozov, A. Zabeo, G. Basei, W. Wohlleben, A. J. Koivisto, K. A. Jensen, W. Fransman, V. Stone and A. Marcomini, SUNDS probabilistic human health risk assessment methodology and its application to organic pigment used in the automotive industry, NanoImpact, 2019, 13, 26–36.
14. K. Schilling, B. Bradford, D. Castelli, E. Dufour, J. F. Nash, W. Pape, S. Schulte, I. Tooley, J. van den Bosch and F. Schellauf, Human safety review of "nano" titanium dioxide and zinc oxide, Photochem. Photobiol. Sci., 2010, 9, 495–509.
15. H. C. Chuang, H. T. Juan, C. N. Chang, Y. H. Yan, T. H. Yuan, J. S. Wang, H. C. Chen, Y. H. Hwang, C. H. Lee and T. J. Cheng, Cardiopulmonary toxicity of pulmonary exposure to occupationally relevant zinc oxide nanoparticles, Nanotoxicology, 2014, 8, 593–604.
16. B. M. Johnson, J. A. Fraietta, D. T. Gracias, J. L. Hope, C. J. Stairiker, P. R. Patel, Y. M. Mueller, M. D. McHugh, L. J. Jablonowski, M. A. Wheatley and P. D. Katsikis, Acute exposure to ZnO nanoparticles induces autophagic immune cell death, Nanotoxicology, 2015, 9, 737–748.
17. S. Keerthana and A. Kumar, Potential risks and benefits of zinc oxide nanoparticles: a systematic review, Crit. Rev. Toxicol., 2020, 50, 47–71.
18. X. Hu, S. Cook, P. Wang and H. M. Hwang, In vitro evaluation of cytotoxicity of engineered metal oxide nanoparticles, Sci. Total Environ., 2009, 407, 3070–3072.
19. G. Yan, Y. Huang, Q. Bu, L. Lv, P. Deng, J. Zhou, Y. Wang, Y. Yang, Q. Liu, X. Cen and Y. Zhao, Zinc oxide nanoparticles cause nephrotoxicity and kidney metabolism alterations in rats, J. Environ. Sci. Health, Part A: Toxic/Hazard. Subst. Environ. Eng., 2012, 47, 577–588.
20. T. Kong, S. H. Zhang, C. Zhang, J. L. Zhang, F. Yang, G. Y. Wang, Z. J. Yang, D. Y. Bai, M. Y. Zhang, J. Wang and B. H. Zhang, Long-term effects of unmodified 50 nm ZnO in mice, Biol. Trace Elem. Res., 2019, 189, 478–489.
21. V. Sharma, P. Singh, A. K. Pandey and A. Dhawan, Induction of oxidative stress, DNA damage and apoptosis in mouse liver after sub-acute oral exposure to zinc oxide nanoparticles, Mutat. Res., 2012, 745, 84–91.
22. S. H. Seok, W. S. Cho, J. S. Park, Y. Na, A. Jang, H. Kim, Y. Cho, T. Kim, J. R. You, S. Ko, B. C. Kang, J. K. Lee, J. Jeong and J. H. Che, Rat pancreatitis produced by 13-week administration of zinc oxide nanoparticles: biopersistence of nanoparticles and possible solutions, J. Appl. Toxicol., 2013, 33, 1089–1096.
23. T. Zhang, X. Zhu, J. Guo, A. Z. Gu, D. Li and J. Chen, Toxicity assessment of nano-ZnO exposure on the human intestinal microbiome, metabolic functions, and resistome using an in vitro colon simulator, Environ. Sci. Technol., 2021, 55, 6884–6896.
24. A. Nicolosi, L. Cardoit, P. Pasquereau, C. Jaillet, M. Thoby-Brisson, L. Juvin and D. Morin, Acute exposure to zinc oxide nanoparticles critically disrupts operation of the respiratory neural network in neonatal rat, Neurotoxicology, 2018, 67, 150–160.
25. N. Singh, M. K. Das, R. Gautam, A. Ramteke and P. Rajamani, Assessment of intermittent exposure of zinc oxide nanoparticle (ZNP)-mediated toxicity and biochemical alterations in the splenocytes of male Wistar rat, Environ. Sci. Pollut. Res., 2019, 26, 33642–33653.
26. H. Rokbani, F. Daigle and A. Ajji, Long- and short-term antibacterial properties of low-density polyethylene-based films coated with zinc oxide nanoparticles for potential use in food packaging, J. Plast. Film Sheeting, 2019, 35, 117–134.
27. J. Du, J. Tang, S. Xu, J. Ge, Y. Dong, H. Li and M. Jin, ZnO nanoparticles: recent advances in ecotoxicity and risk assessment, Drug Chem. Toxicol., 2020, 43, 322–333.
28. J. Yu and S. J. Choi, Particle size and biological fate of ZnO do not cause acute toxicity, but affect toxicokinetics and gene expression profiles in the rat livers after oral administration, Int. J. Mol. Sci., 2021, 22, 1698–1718.
29. M. R. Cooper, G. H. West, L. G. Burrelli, D. Dresser, K. N. Griffin, A. M. Segrave, J. Perrenoud and B. E. Lippy, Inhalation exposure during spray application and subsequent sanding of a wood sealant containing zinc oxide nanoparticles, J. Occup. Environ. Hyg., 2017, 14, 510–522.
30. K. B. Kim, Y. W. Kim, S. K. Lim, T. H. Roh, D. Y. Bang, S. M. Choi, D. S. Lim, Y. J. Kim, S. H. Baek, M. K. Kim, H. S. Seo, M. H. Kim, H. S. Kim, J. Y. Lee, S. Kacew and B. M. Lee, Risk assessment of zinc oxide, a cosmetic ingredient used as a UV filter of sunscreens, J. Toxicol. Environ. Health, Part B, 2017, 20, 155–182.
31. E. Spisni, S. Seo, S. H. Joo and C. Su, Release and toxicity comparison between industrial- and sunscreen-derived nano-ZnO particles, Int. J. Environ. Sci. Technol., 2016, 13, 2485–2494.
32. R. Ridwan, T. Rihayat, S. Suryani, A. S. Ismi, N. Nurhanifa and S. Riskina, Combination of poly lactid acid zinc oxide nanocomposite for antimicrobial packaging application, International Conference on Innovation in Engineering and Vocational Education 2019 (Icieve 2019), Pts 1–4, 2020, vol. 830.
33. G. J. Nohynek and E. K. Dufour, Nano-sized cosmetic formulations or solid nanoparticles in sunscreens: A risk to human health?, Arch. Toxicol., 2012, 86, 1063–1075.
34. I. S. Kim, M. Baek and S. J. Choi, Comparative Cytotoxicity of Al₂O₃, CeO₂, TiO₂ and ZnO Nanoparticles to Human Lung Cells, J. Nanosci. Nanotechnol., 2010, 10, 3453–3458.
35. N. Bumbudsanpharoke and S. Ko, Nano-food packaging: an overview of market, migration research, and safety regulations, J. Food Sci., 2015, 80, R910–R923.
36. D. Romeo, B. Salieri, R. Hischier, B. Nowack and P. Wick, An integrated pathway based on in vitro data for the human hazard assessment of nanomaterials, Environ. Int., 2020, 137, 105505.
37. J. Potočnik, Commission recommendation of 18 October 2011 on the definition of nanomaterial (2011/696/EU), Off. J. Eur. Union, 2011, 275, 38–40.
38. European Commission, Commission recommendation of 10 June 2022 on the definition of nanomaterials (Text with EEA relevance) 2022/C 229/01, 2022, pp. 1–5.
39. W. Moller, K. Felten, K. Sommerer, G. Scheuch, G. Meyer, P. Meyer, K. Haussinger and W. G. Kreyling, Deposition, retention, and translocation of ultrafine particles from the central airways and lung periphery, Am. J. Respir. Crit. Care Med., 2008, 177, 426–432.
40. S. E. Cross, B. Innes, M. S. Roberts, T. Tsuzuki, T. A. Robertson and P. McCormick, Human skin penetration of sunscreen nanoparticles: in-vitro assessment of a novel micronized zinc oxide formulation, Skin Pharmacol. Physiol., 2007, 20, 148–154.
41. M. Heydari-Majd, B. Ghanbarzadeh, M. Shahidi-Noghabi, M. A. Najafi and M. Hosseini, A new active nanocomposite film based on PLA/ZnO nanoparticle/essential oils for the preservation of refrigerated Otolithes ruber fillets, Food Packag. Shelf Life, 2019, 19, 94–103.
42. M. A. Emamhadi, M. Sarafraz, M. Akbari, V. N. Thai, Y. Fakhri, N. T. T. Linh and A. Mousavi Khaneghah, Nanomaterials for food packaging applications: A systematic review, Food Chem. Toxicol., 2020, 146, 111825.
43. X. Li, Y. Xing, Y. Jiang, Y. Ding and W. Li, Antimicrobial activities of ZnO powder-coated PVC film to inactivate food pathogens, Int. J. Food Sci. Technol., 2009, 44, 2161–2168.
44. C. Pang, D. Hristozov, A. Zabeo, L. Pizzol, M. P. Tsang, P. Sayre and A. Marcomini, Probabilistic approach for assessing infants' health risks due to ingestion of nanoscale silver released from consumer products, Environ. Int., 2017, 99, 199–207.
45. H. J. Klimisch, M. Andreae and U. Tillmann, A systematic approach for evaluating the quality of experimental toxicological and ecotoxicological data, Regul. Toxicol. Pharmacol., 1997, 25, 1–5.
46. B. G. H. Bokkers and W. Slob, Deriving a data-based interspecies assessment factor using the NOAEL and the benchmark dose approach, Crit. Rev. Toxicol., 2007, 37, 355–373.
47. H. J. Kramer, W. A. vanden Ham, W. Slob and M. N. Pieters, Conversion factors estimating indicative chronic no-observed-adverse-effect levels from short-term toxicity data, Regul. Toxicol. Pharmacol., 1996, 23, 249–255.
48. U.S. EPA, A review of the reference dose and reference concentration processes, U.S. Environmental Protection Agency, Report EPA/630/P-02/002F, Risk Assessment Forum, Washington, DC, 2002.
49. Danish Ministry of the Environment, Dermal absorption of nanomaterials part of the “Better control of nano” initiative 2012–2015, 2013.
50. U.S. EPA, Reference dose (RfD): Description and use in health risk assessments, Regul. Toxicol. Pharmacol., 1988, 8, 471–486.
51. U.S. EPA, Risk Assessment Guidance for Superfund (RAGS), Volume I: Human health evaluation manual (Part E, Supplemental guidance for dermal risk assessment) interim, Office of Solid Waste Emergency Response, 2009.
52. P. R. Bushel, J. A. L. Bourdon, D. Krewski and S. Auerbach, Technical guide for applications of gene expression profiling in human health risk assessment of environmental chemicals, Regul. Toxicol. Pharmacol., 2015, 72, 292–309.
53. X. L. Duan, Exposure Factors Handbook of Chinese Population, China Environmental Science Press, Beijing, 2013.
54. H. van der Voet, W. Slob and J. van Klaveren, Integration of probabilistic exposure assessment and probabilistic hazard characterization, Toxicol. Lett., 2007, 172, S109–S109.
55. U.S. EPA, Conditional registration of HeiQ AGS-20 as a materials preservative in textiles, 2011.
56. S. Pasupuleti, S. Alapati, S. Ganapathy, G. Anumolu, N. R. Pully and B. M. Prakhya, Toxicity of zinc oxide nanoparticles through oral route, Toxicol. Ind. Health, 2012, 28, 675–686.
57. H. Genc, B. Barutca, A. T. Koparal, U. Ozogut, Y. Sahin and E. Suvaci, Biocompatibility of designed MicNo-ZnO particles: Cytotoxicity, genotoxicity and phototoxicity in human skin keratinocyte cells, Toxicol. In Vitro, 2018, 47, 238–248.
58. Q. Li, L. Chen and Y. Zhou, Diagnostic accuracy of liver stiffness measurement in chronic hepatitis B patients with normal or mildly elevated alanine transaminase levels, Sci. Rep., 2018, 8, 5224.
59. S. C. O. C. Safety (SCCS), Opinion on zinc oxide (nano form) COLIPA S 76, 2012.
60. P. J. Borm, D. Robbins, S. Haubold, T. Kuhlbusch, H. Fissan, K. Donaldson, R. Schins, V. Stone, W. Kreyling, J. Lademann, J. Krutmann, D. Warheit and E. Oberdorster, The potential risks of nanomaterials: a review carried out for ECETOC, Part. Fibre Toxicol., 2006, 3, 11.
61. V. D. Subramaniam, S. V. Prasad, A. Banerjee, M. Gopinath, R. Murugesan, F. Marotta, X. F. Sun and S. Pathak, Health hazards of nanoparticles: understanding the toxicity mechanism of nanosized ZnO in cosmetic products, Drug Chem. Toxicol., 2019, 42, 84–93.
62. A. M. Holmes, I. Kempson, T. Turnbull, D. Paterson and M. S. Roberts, Penetration of zinc into human skin after topical application of nano zinc oxide used in commercial sunscreen formulations, ACS Appl. Bio Mater., 2020, 3, 3640–3647.
63. A. Emamifar, M. Kadivar, M. Shahedi and S. Soleimanian-Zad, Evaluation of nanocomposite packaging containing Ag and ZnO on shelf life of fresh orange juice, Innovative Food Sci. Emerging Technol., 2010, 11, 742–748.
64. F. Stuer-Lauridsen, A. Kamper, P. Borling, G. I. Petersen and A. Baun, Survey of nanotechnological consumer products, 2007.
65. X. Duan, X. Zhao, B. Wang, Y. Chen and S. Cao, Highlights of the Chinese Exposure Factors Handbook, Highlights of the Chinese Exposure Factors Handbook(Adults), 2015.
66. Y. Luo, L. Ma, H. Zheng, L. Chen, R. Li, C. He, S. Yang, X. Ye, Z. Chen, Z. Li, Y. Gao, J. Han, G. He, L. Yang and Y. Wei, Discovery of (Z)-5-(4-methoxybenzylidene)thiazolidine-2,4-dione, a readily available and orally active glitazone for the treatment of concanavalin A-induced acute liver injury of BALB/c mice, J. Med. Chem., 2010, 53, 273–281.
67. V. Forest, J. F. Hochepied and J. Pourchez, Importance of choosing relevant biological end points to predict nanoparticle toxicity with computational approaches for human health risk assessment, Chem. Res. Toxicol., 2019, 32, 1320–1326.
68. T. E. Abbott Chalew, G. S. Ajmani, H. Huang and K. J. Schwab, Evaluating nanoparticle breakthrough during drinking water treatment, Environ. Health Perspect., 2013, 121, 1161–1166.
69. V. S. Sousa and M. Ribau Teixeira, Metal-based engineered nanoparticles in the drinking water treatment systems: A critical review, Sci. Total Environ., 2020, 707, 136077.
70. A. Spinazzè, A. Cattaneo, F. Borghi, L. Del Buono, D. Campagnolo, S. Rovelli and D. M. Cavallo, Probabilistic approach for the risk assessment of nanomaterials: A case study for graphene nanoplatelets, Int. J. Hyg. Environ. Health, 2019, 222, 76–83.
71. U. S. EPA, Conditional Registration of HeiQ AGS-20 as a Materials Preservative in Textiles, December 1, 2011, pp. 1–51.
72. B. G. H. Bokkers, M. J. Mengelers, M. I. Bakker, W. A. Chiu and W. Slob, APROBA-Plus: A probabilistic tool to evaluate and express uncertainty in hazard characterization and exposure assessment of substances, Food Chem. Toxicol., 2017, 110, 408–417.
73. T. P. Pastoor, A. N. Bachman, D. R. Bell, S. M. Cohen, M. Dellarco, I. C. Dewhurst, J. E. Doe, N. G. Doerrer, M. R. Embry, R. N. Hines, A. Moretto, R. D. Phillips, J. C. Rowlands, J. Y. Tanir, D. C. Wolf and A. R. Boobis, A 21st century roadmap for human health risk assessment, Crit. Rev. Toxicol., 2014, 44(Suppl 3), 1–5.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2en00570k

---