Permeability of dimethyl phthalate through human skin models – Health Risk Assessment

Dimethyl phthalate (DMP) is used widely in cosmetics and personal care products as a solvent, skin penetrant, moisturizer and softener as well as an anti-cracking agent. Dermal absorption is one of the major exposure routes for lower molecular weight phthalates such as DMP and assessing their dermal permeability is important for evaluating the impact and toxicity of such compounds in humans. The aim of this study was to evaluate DMP permeability through different human skin models (Strat-M ® synthetic membrane, in vitro reconstructed human epidermis and human dermatomed skin) using solutions containing various types of surfactants, as a simulation of the types of mixtures occurring in cosmetics and personal care products, in order to ascertain the impact that surfactants can have on skin permeability. The results have shown that human skin is the least permeable of the used skin models, and that surfactants (in particular cationic and non-ionic) lead to a signi�cant increase of DMP permeability through all skin models. The performed risk assessment however shows that, for all tested models, the margin of safety was not exceeded.


Introduction
Dimethyl phthalate (DMP) belongs to a group of compounds named phthalate esters, which are produced by esteri cation of phthalic acid with methanol.It is classi ed as a low molecular-weight (LMW) phthalate, containing a single carbon ester side-chain (HSDB 2009; Otero et al. 2015).DMP is primarily used in the production of cosmetics and personal care products (PCP), mainly as a solvent and perfume xative or as a sealant in hair spray (FDA 2010; Li et al. 2021;Nicolopoulou-Stamati et al. 2015; Wang et al. 2019).According to available literature, the reported DMP concentration in cosmetics (deodorants, perfumes, hair sprays and conditioners, lotions, face powders and foundations) ranges from 0.00004% up to 34% (in combination with its analogue DEP) (SCCP 2007;NICNAS 2008;CPSC 2010;NICNAS 2014).Giovanoulis and co-authors (2020) have concluded that use of cosmetics and PCPs may have a signi cant impact on human transdermal exposure to phthalates, especially for LMW phthalates such as DMP.Due to the toxicological properties of DMP (low acute toxicity, low skin and eye irritation, low skin sensitizing potential), the occurrence, among consumers, of adverse acute effects resulting from use of cosmetics containing this phthalate is relatively low.However, long-term exposure through leave-on cosmetics containing DMP is related to potential health risks (NICNAS 2014).
To perform health risk assessment (HRA) of phthalates exposure via the dermal route, researchers use data on the percutaneous penetration of these compounds (Olkowska and Gržinić 2022).Data on the dermal permeation behavior of DMP is limited.Elsisi et al. (1989), during their investigation of DMP absorption via male rat skin (5-8 mg/cm 2 ), observed that the cumulative percentage dose excreted in urine and faeces over one week was 20-40% (excretion rate 6-7.5%).Compared with diethyl phthalate (DEP), DMP is considered to be slowly absorbed via the skin membrane.Dermal absorption of DMP through rat skin is about 10-20 times higher than through human skin (410 vs 39.5 µg/cm 2 /hr) (NICNAS 2014).Data on DMP absorption through two skin models was reviewed by Koniecki et al. (2011): in vitro rat skin model (25.57± 8.5%) and human skin (4%).According to the Australian Department of Health HRA protocol, based on difference in absorption through rat skin and the higher permeability rate of DMP vs. DEP through human skin, the dermal availability of DMP was suggested to be 10% (NICNAS 2014).
The investigation of dermal absorption of chemicals present in cosmetics under realistic use conditions is vital for reducing uncertainties in risk estimates.Skin penetration may depend on the physicochemical properties of the chemical, its dose, surface of exposed skin, time of exposure, skin condition and presence of speci c chemical compounds (Jiang at al. 2020).Surface active agents or alcohols, which are ingredients of cosmetics and PCPs, can disturb the natural protective skin barrier and result in increased permeability to other compounds.
The aim of this study was to examine DMP permeability through different human skin models exposed to DMP solutions, with and without surfactants, as a simulation of the types of mixtures occurring in cosmetics.The study also compares DMP permeation through three skin models (Strat-M® membrane (SMM), in vitro epidermal model (Reconstructed Human Epidermis, RHE) and human ex vivo human skin (XenoSkin H, HS)).The DMP permeability results obtained with the different skin membranes were used to evaluate HRAs exposure via the dermal route for cosmetics containing DMP.
Experimental work has been conducted under conditions which limit contamination of glassware and other laboratory equipment with phthalates (treatment with ethanol, acetone and/or high temperature).
Plastic containers were eliminated from protocols unless they had phthalate-free certi cates.
In transdermal diffusion testing, three types of human skin models were used: Strat-M® membrane (SMM), in vitro epidermal model (RHE) and ex vivo human skin (HS).The synthetic Strat-M® membrane (25 mm, non-animal based model) was obtained from Merck KGaA, (Darmstadt, Germany).This membrane has already been used as a permeation barrier for predicting phthalate penetration through skin (e.g.diethyl phthalate, dibutyl phthalate, diisononyl phthalate) (Pan et al., 2014).RHE membranes (Fraunhofer ISC-TLC in vitro epidermal 25 mm models in 6-well format) with the supporting cell culture medium were ordered from the Translational Center Regenerative Therapies TLC-RT, Fraunhofer Institute for Silicate Research ISC (Würzburg, Germany).Frozen abdominal dermatomed human skin XenoSkin H (art. no.H-D20D-24, 24 mm) was obtained from Xenometrix AG (Allschwil, Switzerland) under strict ethical restrictions and with informed consent.No sensitive personal information regarding the patients was retained.

Franz Diffusion Cell Experiments
DMP permeation experiments, with and without the addition of surface active agents, were performed according to OECD guidelines (OECD 2004).A 6-cell manual diffusion cell system with 2mag-Magnetic-Drive (2mag-AG, München, Germany) and circulating waterbath HE4 (JULABO GmbH, Seelbach, Germany) was obtained from Hanson Research (Chatsworth, CA, USA).All studies were performed using 7 mL vertical diffusion cells (donor medium) with open cell top (1.8 cm 2 diffusion area, 1 mL of donor medium) and cap.The temperature and stirring parameters were set to 32ºC and 350 rpm, respectively.Except for the Strat-M® membrane, skin models were hydrated before diffusion cell experiments (Sugino et al. 2017).PBS solution with addition of 10% of ethanol was used as donor and acceptor media for all tested mixtures, which had a positive effect on DMP solubility and the solvent is compatible with aqueous buffer (Katakam and Katari 2021).The applied DMP dose per cm 2 of skin model was 1078 µg/cm 2 .The amounts of surface active agents added were based on average values used in cosmetics formulations.The composition of the tested mixtures can be seen in Table 1.Fresh mixtures were prepared for every round of diffusion experiments and measurements were carried out in three replicates with each mixture, for every type of membrane.After preparation, the diffusion cells ( lled with acceptor medium, with mounted skin model, covered with the cell top) were checked to ensure that there were no air bubbles between the skin model and the receptor medium.Aliquots (500 µL) were collected from the receptor sections at speci ed time intervals (0 h, 1 h, 2 h, 3 h, 4 h, 6 h, 8 h) by injecting warmed PBS solution using a 1.0 mL chromatographic syringe.Next, samples were extracted using liquid-liquid extraction (LLE) and analyzed by gas chromatography coupled with mass spectrometry (GC-MS).

Chromatographic Analysis
The determination of DMP migration through the skin models was performed based on analytical methodologies proposed by Guo et al. (2010).Stock standard solutions of DMP and d 4 -DMP (internal standard) at 1000 µg/mL were prepared separately.A calibration curve for both analytes, ranging from 0.025 µg/mL to 50 µg/mL, was prepared by dilution of the mentioned stock solution.Validation samples at concentrations of 1, 10 and 25 µg/mL were also prepared as above.Each sample was analyzed in triplicate with methanol injections between every three samples.
The amount of DMP in acceptor and donor samples, after LLE (double extraction with 1 mL of dichloromethane after addition of 10 µL d 4 -DMP at 5 µg/mL), was investigated with a GC-2010 PLUS gas chromatograph coupled with an AOC-20ia auto injector and MS-TQ8040 mass spectrometer from Shimadzu Corp. (Kyoto, Japan).Before analysis, the extracts were evaporated and the residue was dissolved in 500 µL of methanol.The injection temperature was set to 280°C in splitless mode.The GC oven temperature was programmed as follows: at 50°C initially and held for 1 min, then ramped to 310°C and held for 2.5 min.The separation was carried out on a fused silica capillary column GC Zebron ZB-5MS (30 m, 0.25 mm, 0.25 mm) with helium as carrier gas (99.999995% pure, ow rate of 1.0 mL/min).The ion source temperature was maintained at 220°C and the transfer line was heated to 310°C.The MS was operated in electron impact mode with electron energy of 70 eV.The target compounds were determined in full scan (SCAN) and selected ion monitoring (SIM) mode.LabSolution Analysis software (Shimadzu Corp., Kyoto, Japan) was used for GC-MS control and data acquisition.The identi cation was performed by using similarity search in the National Institute of Standards and Technology MS database (NIST 11).

Quality Assurance And Data Analysis
The analytical method has been evaluated using the following validation parameters: detection limit (DL), quanti cation limit (QL), linearity, recovery, and precision (ICH 2021).DL and QL were calculated based on the standard deviation of the linear response and the slope of the calibration curve as well as on a signal to-noise ratio of 3:1 and 10:1, respectively.The linearity was evaluated based on the coe cient of determination (R 2 ) of a 6-point calibration curve.Recovery of the target compound was performed at three different levels to evaluate the accuracy of the proposed protocol.Precision (as percent relative standard deviation, %RSD) was investigated by carrying out six independent sample analyses for three consecutive days.
The skin permeability of DMP was calculated from the quantity of target analyte, which permeated through the skin membrane, divided by the membrane surface and the time duration.The permeability coe cient (k p ) was determined from the steady-state ux (J ss ) and DMP levels in the donor phase.
Phthalate ux was calculated from the slope of the penetration amount vs. time pro le.Additionally, the ratio of total amount of DMP (Eq. 1) in the receptor uid was compared to the amount of DMP in the donor phase to determine the total absorption rate (Hopf et  The results of the non-cancer HRA were taken as the margin of safety (MoS) (Eq.3), with values above 100 indicating a safe value (MFDS 2017; SCCS 2016; Kim et al. 2020).If MoS is < 100 for an investigated compound, the ingredient is considered to be a potential cause of adverse health effects and there are safety concerns in terms of its use.
All calculations were done using Microsoft® Excel® 2016 MSO.

Determination of dimethyl phthalate -quality
Table 2 summarizes the parameters of the analytical protocol used to estimate DMP concentrations in the acceptor and donor medium, including validation parameters.The calibration curves obtained for DMP showed a good linear range of calibration (R 2 = 0.996).
Table 2 Parameters of dimethyl phthalate determination using LLE-GC-MS protocol The veri cation of the extraction method was performed by estimating the recoveries obtained from the blank samples and samples containing DMP standards.The analytes showed good recovery (more than 90%) and the %RSDs were lower than 10%, con rming good accuracy.With regards to precision, the results were within acceptable ranges according to guidelines (%RSD less than 15%).The skin permeation pro les of DMP through synthetic membrane, in vitro epidermal model and ex vivo human skin, are shown in Fig. 1a-c, respectively.Based on data presented in Fig. 1, the highest skin permeability was observed for the in vitro epidermal model (solution with cationic surfactant).The lowest DMP concentration (presented as µg/cm 2 ) in the receptor phase was observed for human skin.Differences in permeability between the synthetic membrane and human skin are similar to previously reported values (Scott et al. 1987, Scott et al. 1989, Olkowska and Gržinić 2022).The lowest impact on DMP permeation was observed for solutions containing the anionic surfactant.Human skin and RHE permeability were most impacted when a cationic surfactant was used.The non-ionic surfactant increased permeability of DMP through all types of membranes.This surfactant has been reported as causing serious eye damage (ECHA 2022).
The permeation parameters are shown in Table 3, along with the percentage absorption used in HRA.

Conclusion
In this study, we have examined the permeability of DMP with different human skin models.While human skin remains the golden standard, reconstructed human epidermis and synthetic membranes provide results which can, with appropriate corrections, be useful in cosmetics health risk assessment.We have likewise tested different types of surfactants, which have been shown (in particular cationic and non-ionic ones) to increase DMP permeability trough all skin models by a factor of at least two.Since such compounds are often used in cosmetics and personal care products, the assessment of their in uence on skin absorption is crucial in cosmetic risk assessment.Nevertheless, the margin of safety, with all skin models, was not exceeded.Overall, risk assessment points to a relatively low risk related to surfactant use in combination with DMP, however additional research is needed for the determination of the impact of mixtures with other phthalates and of long term risk.Moreover, cumulative HRE assessments should also consider additional exposure routes to DMP such as transdermal absorption from indoor air, which can account for up to 35% of total exposure in indoor environments (Bu et al., 2016).Therefore, from a HRA perspective, efforts should focus on estimating overall DMP exposure via the dermal route resulting from different sources (cosmetics, detergents, clothes, indoor air, households items etc.).

Statements And Declarations
al. 2014; Neri et al. 2022): total absorption (%) = amount of DMP acceptor/total amount DMP in donor × 100 (1) Total absorption of the test compounds can be used as one of the exposure parameters in human HRA.The obtained results were used in non-cancer risk assessment of DMP connected with its potential occurrence in cosmetics or personal care products (MFDS 2017; SCCS 2016; Kim et al. 2020).The systemic exposure dose (SED) was calculated using Eq.(2).SED (mg/kg/day) = B × 1000 mg/g × C/100 × A/100/BW (2) where: SED -systematic exposure dosage for cosmetic ingredients (estimated amount of exposure, per body weight, per day) [mg/kg body weight/day]; B -amount of cosmetic products used in one day [g/day]; C -concentration of target ingredient in evaluated cosmetic products [%]; A -skin absorption rate expressed in real use conditions [%]; BW -average body weight (60 kg) [kg].

Table 1
Composition of tested dimethyl phthalate mixtures with surfactants 4NSAA -non-ionic surface active agent

Table 2
Parameters of dimethyl phthalate determination using LLE-GC-MS protocolDimethyl Phthalate Permeation With Different Skin Models The lowest penetration (1.2-3.0 µg/cm 2 /h) was observed for human skin with all mixtures.Similar ux values (2.5-4 µg/cm 2 /h) were described in CPSC (2010).Flux of DMP across the Strat-M and RHE membrane was similar, and comparable to values obtained for rat epidermis (40-50 µg/cm 2 /h) (CPSC 2010).Such skin models can be a good alternative to in-vivo animal testing.Total absorption for human skin was lower that reported in guidelines (5-10%).Differences can be attributed to variables such as vehicle effects and/or study design (NICNAS 2014).

Table 3
Results of skin model permeation experiments with dimethyl phthalate