Estimation of Dermal Exposure to Volatile Organic Compounds (VOCs) from Feminine Hygiene Products: Integrating Measurement Data and Physiologically Based Toxicokinetic (PBTK) Model.

This study included the five most prevalent categories of FHPs in the United States, i.e., pads, tampons, feminine washes, wipes and moisturizers, as identified in our previous studies. 2 , 13 The measurements of VOCs in these FHPs were also obtained from our previous study, which investigated 74 FHPs (22 pads, 22 tampons, 13 feminine washes, 12 wipes, and 5 moisturizers) covering a variety of types and brands across above categories. 2 We measured VOCs concentrations in urine samples from 25 nonsmoking female participants 22–25 y of age [interquartile range (IQR)] in our previous study. 13 The participants were recruited from the University of Michigan School of Public Health in September 2018 after responding to recruitment emails and completing a brief eligibility survey. Eligibility criteria of participants included ages being between 20 and 49 y, having had a regular menstrual cycle in the past 3 months (menstrual period within 7 d, and an average cycle length of 21–40 d), and self-identifying as White, Black, or Asian. We excluded women who had a vaginal infection, uterine fibroids, polycystic ovarian syndrome, endometriosis or who were pregnant or breastfeeding. The final study samples included 11 White, 6 Black, and 8 Asian women, all of whom provided written informed consent to participate in the study, which was approved by the institutional review board at the University of Michigan. 13 We collected first-morning-void urine samples from participants at four time points: 7 d before menstruation and 3, 7, and 14 d after the first day of menstruation. Finally, from October 2018 to February 2019, we collected 99 urine samples, with one participant failing to collect the first urine sample. 13 VOCs in urine samples were collected using a purge and trap method. 13 In brief, a 10 -mL aliquot of each urine sample was heated at 40 ° C and purged with 1,000 mL pure nitrogen gas. The VOCs purged from the urine were collected in a thermal sorbent tube containing 250 mg of anhydrous sodium sulfate (to remove water vapor) and 160 mg of Tenax-GR (to trap the VOCs), followed by immediate instrument analysis. The tube preparation and instrument analysis are described in the section below titled “Field Study of Inhalation Exposure,” and in our previous studies. 13 , 26 – 29 To ensure methodological robustness, we targeted eight VOCs in this study based on their frequency of detection in urine samples ( > 40 % ), availability of necessary parameters for PBTK modeling, and exclusion of substances with endogenous generation in human bodies. These target VOCs [methylene chloride, methyl ethyl ketone, benzene, toluene, styrene, p -cymene, hexane, and n -nonane (Table S1) 30 ] were grouped by their chemical structures as halohydrocarbon, ketone, aromatic, and alkane. All eight target VOCs possess the potential to cause skin and mucosal corrosion/irritation, with some being recognized or suspected carcinogens, such as methylene chloride, benzene, toluene, and styrene. 31 – 34 To calculate the external dose of dermal exposure, we conducted a survey of participants to collect parameters regarding the use pattern of FHPs in our previous study, including the number and frequency of use. 13 Based on the survey, some females reported using all categories of FHPs. Therefore, to conduct a conservative risk assessment, we assumed that all FHP categories were used except tampons within a 28-d menstrual cycle, because pads and tampons serve similar purposes, with pads generally containing higher VOC levels. 2 The specific use pattern is assumed as follows (Figure S1): pads or tampons were used six times daily for 7 d during a 7-d menstrual period, followed by a postmenstrual duration of 21 d, constituting a complete 28-d menstrual cycle. In addition, feminine washes and moisturizers were assumed to be applied once daily, whereas wipes were used six times daily. During the first 0.1 h (6 min) of the day, when the wash was applied, no other products were used. The maximum VOC concentrations in FHPs from our previous study 2 (Table S1) were applied to simulate a best-guess exposure scenario, representing the worst case for conservative assessment. The specific parameter values relevant to exposure estimation are shown in Table S2, and the initial exposure amounts input into the model are detailed in Table S3. In addition, we simulated urinary VOC concentration dynamics from the use of a single product category, including tampons, to assess the burden attributed to a single product type. All products were used on vulvar tissues and nearby skin areas, except tampons, which were in contact with vaginal tissues. Farage et al. reviewed previous studies to quantify uncertainty factors for mucosal exposures. 35 Two in vivo studies demonstrated that the vulva (labium majus) exhibited a 7-fold increase in permeability to hydrocortisone, a lipophilic substance, in comparison with normal skin. 36 , 37 Given the lipophilicity of the target VOCs ( Table 1 ) and the potential for even greater permeability in vaginal tissue, 35 a 10-fold permeability coefficient was reasonably applied to simulate the enhanced absorption of VOCs by mucosa tissue in comparison with normal skin when using tampons. No participants reported discomfort related to these products, 3 so we did not consider the potential effects of irritation on product use behavior or any changes in skin physiology. Values of toxicokinetic parameters of the target VOCs. Note: K m , concentration of compound in liver tissue, at which the biotransformation rate is half maximum, also called the Michaelis constant; K p , dermal penetration rate of compound into the skin, log   K p = 0.71 × log   K ow − 0.0061 × M W − 2.72 41 , 42 ; K renal , renal excretion coefficient; log   K ow , logarithm of the octanol-water partition coefficient; P tissure  /  blood , partition coefficient of compound between tissue and blood; v max , the maximum biotransformation rate in liver; VOC, volatile organic compound. The dermal absorption through using FHPs was calculated as follows ( Equations 1 − 3 ) 38 – 42 : ( 1 ) M skin = K p × M product × R T l , ( 2 ) M product = M product ,   t = 0 × e − K p × R T l × t , and ( 3 ) log   K P = 0.71 × log   K ow − 0.0061 × M W − 2.72 where M skin is the amount of compounds absorbed via skin per hour (ng/h); K p is the permeability coefficient (cm/h); M product is the amount (ng) of compounds remaining in products at any time during the exposure; M product ,   t = 0 is the initial VOC mass in FHPs (ng), which is calculated by multiplying the initial chemical concentration in products ( C , ng/g) by the mass of products ( M , g) ( Equation 2 ); R T is the transition rate of chemical from the product to the skin ( R T , dimensionless), 40 which is assumed to be 1 in this study due to the lack of available data on the target VOCs, simulating the maximum potential dermal exposure; l is the skin thickness, taking a value of 0.2 cm 43 ; K ow is the octanol-water partition coefficient; and MW is the molecular weight (g/mol). Most previous PBTK studies focused on the inhalation exposure of the target VOCs, 44 – 46 resulting in a lack of permeability coefficient data for these VOCs in the literature. Therefore, the permeability coefficient was calculated using Equation 3 , as suggested by the US Environmental Protection Agency (US EPA) 41 and adapted from Guy and Potts. 42 The calculated values are shown in Table 1 . Based on previous studies, VOCs are rapidly absorbed through the skin. 47 , 48 In addition, materials such as the waterproof backing of pads can prevent VOCs from evaporating. 40 Therefore, we assume no evaporation of VOCs during dermal exposure in this study, which also simulates the worst case for conservative assessment. Although inhalation exposure is identified as the main exposure route of VOCs, 2 it was not the target exposure route aimed in this study. Therefore, it was only considered in the model validation. VOCs concentrations in outdoor air were much lower than those indoors, 26 and given that individuals spend over 80% of their time indoors, 49 measured indoor air VOCs concentrations were used to simulate the inhalation exposure. The inhalation absorption dose was calculated as below ( Equation 4 ): ( 4 ) M inhalation = Q P × C air , where M inhalation is the amount of compound absorbed via inhalation per hour (ng/h); Q P (mL/h) represents the alveolar ventilation volume per hour, set at 460 ,   588 mL / h in this study for a 74.9 -kg female; C air is the concentration (ng/mL) of the target VOCs in indoor air. Q P was calculated based on 390,000 mL / h for a 60 -kg , seated, awake female 50 and adjusted according to the reported correlation between Q P and body weight 0.75 ( BW 0.75 ). 51 Because body weight (BW) was not measured in the studied population, 3 we used 74.9 kg , the average BW of US females 20–29 y of age during the period 2015–2018, 52 corresponding to the age range of the studied population (IQR of the age of the female participants in this study being 22–25). 13 To assess the concurrent inhalation exposure, we conducted a field study involving VOC measurements in indoor air samples ( n = 20 ) from Ann Arbor and Detroit, Michigan, the same cities where the 25 female participants lived, though not from the participants’ residences. We collected these air samples during the same period when we collected urine samples from participants. Sixteen samples were obtained between January and June 2019 from a hotel located in Ann Arbor, and four samples were collected in September 2018 from a house located in Detroit. The concentrations of VOCs detected in these samples are shown in Table S1. We collected the air samples using passive sampling with thermal desorption tubes, following establishing protocols. 26 , 27 These 10 -cm -long stainless sorbent tubes with a 0.4 cm inner diameter, were packed with 160 mg Tenax-GR (60/80 mesh), with a diffusion gap of 0.5 cm . Glass wool plugs were placed at both ends to secure the sorbent Tenax-GR (Figure S2). Prior to sampling, the tubes were conditioned at 325°C for 6 h with purging ultrapure nitrogen gas at 30 mL / min . Once conditioned, the tubes were sealed with stainless-steel caps containing prebaked Teflon seals, wrapped in baked aluminum foil, stored in a clean glass jar with activated carbon pack, and placed in a refrigerator kept at 4 º C . During sampling, the cap was removed and the tube was placed in a stand at breathing height ( ∼ 1.5 m ; Figure S2) to allow air to passively diffuse. The sampling uptake rate was calculated thorough a diffusion model. 26 , 27 After sampling, the tubes were resealed and stored under the same conditions as they were under before sampling, followed by instrument analysis. Instrument analysis has been well described in our previous studies. 28 , 29 In brief, each tube was spiked with internal standards (fluorobenzene, p -bromofluorobenzene, and 1,2 -dichlorobenzene-d 4 ) before being analyzed using an automated thermal desorption system (TD; Scientific Instrument Services) coupled to a gas chromatography–mass spectrometer (GC-MS; Agilent Model 6890/5973), which was equipped with a cryogenic trap. The instrument setting and detailed conditions are presented in Table S4. The PBTK model used in this study ( Figure 2 ) featured five compartments representing various tissues, including blood, liver (the metabolic tissue), adipose tissue (the storage tissue), kidney (the excretory tissue), and other tissues (rest of the body). Absorption via dermal exposure was input into the PBTK model to investigate VOCs dynamics in human tissues from the use of FHPs. Indoor inhalation exposure was included only during the model validation, regarding its significant contribution to internal exposure of the target VOCs, as stated above. Because oral exposure was not the focus of this study, it was not considered in the model, and thus gut tissue was merged into the rest of the body. The major kinetic equations ( Equations 5 – 11 ) are provided below to predict the VOCs dynamics in different tissues. 38 , 53 – 56 ( 5 ) d C B d t = 1 V B Q L C L P liver / blood + Q AT C AT P adipose / blood + Q K C K P kidney / blood + Q RB C RB P rest of body / blood − Q C C B + M skin + M inhalation − M exhalation , Schematic diagram of PBTK model implemented in this study. Concentrations in skin and lung (in dashed boxes) were not predicted, and they are shown in the schematic diagram to indicate the exposure routes. Note: C B , chemical concentration in blood; K m , Michaelis constant; K p , permeability coefficient; K renal , the renal excretion coefficient; PBTK, physiologically based toxicokinetic; Q AT , Q C , Q K , Q L , and Q RB , blood flow to adipose tissue, cardiac output, kidney, liver, lung and rest of body, respectively; Q P , alveolar ventilation volume per hour; v max , maximum biotransformation rate in liver. and ( 11 ) C urine = Q K C B K renal V urine , where C AT , C B , C K , C L , C RB , and C urine are the concentrations (ng/mL) of compounds in the adipose tissue, blood, kidney, liver, rest of body, and urine, respectively; Q C is the total cardiac output per hour (mL/h); Q AT , Q K , Q L , and Q RB are the blood flow (mL/h) to adipose tissue, kidney, liver, and rest of body, respectively; V AT , V B , V K , V L , and V RB are the volumes (mL) of adipose tissue, blood, kidney, liver, and rest of body, respectively; V urine is the volume of urine produced per hour (mL/h); P adipose / blood , P kidney / blood , P liver / blood , P rest of body / blood , and P blood / air are the partition coefficients (dimensionless) of corresponding tissues to blood and blood to air ( Table 1 ) 38 , 57 , 58 ; v max is the maximum biotransformation rate (ng/mL tissue/h) in liver, 54 , 59 – 62 shown in Table 1 ; K m is the Michaelis constant, i.e., the concentration (ng/mL) of a compound in liver tissue when the biotransformation rate of liver reaches half of its maximum ( Table 1 ) 54 , 59 – 62 ; M exhalation is the amount of compound exhaled per hour (ng/h); and K renal is the renal excretion coefficient (dimensionless, Table 1 ). 38 Physiological values such as blood flow to various tissues and tissue volumes are related to BW, set at 74.9 kg in this study, as explained above in the section titled “External Exposure Scenario and Dermal Absorption.” According to Brown et al., 63 cardiac output ( Q C ) correlates with BW 0.75 . In addition, the International Commission on Radiological Protection (ICRP) reports that a female weighing 60 kg has a cardiac output of 5.9 L / min . 50 Based on this, we applied the relationship: cardiac output L / h = 16.4 L / h / kg 0.75 × BW 0.75 kg , as demonstrated by Smith et al. 51 According to the ICRP, 50 percentages of blood flow to adipose tissue ( Q AT ), kidney ( Q K ), and liver ( Q L ) are 8.5%, 17%, and 27% of cardiac output ( Q C ), respectively. The masses of adipose tissue, blood, kidney, and liver are 30%, 6.8%, 0.458%, and 2.33% of body weight, respectively, 50 which were used to determine the tissue volumes ( V AT , V B , V K , and V L ) in this study (Table S2). Partition coefficients between various tissues and blood were primarily derived from the tissue/air partition coefficients reported by Meulenberg and Vijverberg. 57 This report consolidated data from multiple previous studies, providing comprehensive data including P kidney / blood . Partition coefficients for methyl ethyl ketone was sourced from a previous PBTK study. 58 Because there are no relevant human data for p -cymene, its partition coefficients were calculated using a generic PBTK model developed by Jongeneelen and Ten Berge. 38 The partition coefficient between muscle and blood was used as the coefficient for the rest of the body/blood ( P rest of body / blood ), which was also sourced from these studies, together with P blood / air . 38 , 57 , 58 The maximum biotransformation rates in the liver ( v max ) were derived from previous PBTK modeling studies and are as below: 2,575; 122; 504; and 874 μ mol / kg tissue/h for methylene chloride, benzene, toluene, and styrene, respectively 60 ; 30 μ mol / min based on 1.5 kg of liver for methyl ethyl ketone 59 ; and 1.35 mg / h / kg BW for hexane. 54 No in vivo v max data for p -cymene and n -nonane in humans are available. An in vitro study by Meesters et al. identified the v max for p -cymene metabolism to three metabolites via four CYP enzymes. 61 We averaged the reported v max values, resulting in 4.372  mmol / min / mmol CYP enzymes, 61 and then multiplied this by the averaged CYP concentration in microsomes ( 0.345  nmol / mg ) and the microsomal content in the liver ( 32 mg / g ), 64 , 65 yielding a calculated v max for p -cymene of 3.89 × 10 5   ng / mL tissue/h. A similar v max deviation for n -nonane was derived from an in vitro v max of 429  nmol / min / mg human liver microsomes, 62 resulting in a calculated v max of 1.06 × 10 8   ng / mL tissue/h for n -nonane. Correspondingly, Michaelis constants ( K m ) were also obtained from these studies, with values of 6.8, 2.0, 40, 5.84, 3.46, and 62.5 μ mol / L for methylene chloride, 60 methyl ethyl ketone, 59 benzene, 60 toluene, 60 styrene, 60 and n -nonane, 62 respectively, and 0.40 mg / L for hexane. 54 The averaged K m for p -cymene was 399   μ mol / L by different CYP enzymes. 61 Because renal excretion coefficients ( K renal ) were generally not included in previous PBTK modeling studies, we used a generic method ( Equation 12 ) for their calculation, as developed by Jongeneelen and Ten Berge. 38 ( 12 ) K renal = G l o m F i l t r × F r W sol × ( 1 − F r tubular ) and ( 13 ) F r W sol = 0.993 0.993 + 0 . 007 × 10   log   K ow , where, GlomFiltr is the fraction of glomerulus filtrate (dimensionless), set as 0.2 based on Carlson, 66 which is close to the average value of 0.18 derived from the three values (0.08, 0.16, and 0.3) provided by Jongeneelen and Ten Berge 38 , 67 ; F r W sol is the fraction of the compound in blood that is dissolved in water (dimensionless), calculated using Equation 13 38 ; F r tubular is the fraction of tubular resorption (dimensionless), set as 0.94 for methyl ethyl ketone and 0.99 for other target VOCs, as dependent from their log   K ow . 67 We validated the model performance by comparing the estimated concentrations with the measurements obtained from urine samples. The urinary VOC concentrations in first-morning-void urine samples, collected from our previous study involving 25 female participants, 13 were used as the measured values. Urine samples were collected at four time points: 7 d before menstruation and 3, 7, and 14 d after the first day of menstruation. 13 We assumed that the first-morning-void urine represented the previous 8 h [from 2400 hours to 0800 hours (midnight to 8 A.M.)]. Therefore, these four time points corresponded to the 505th to 512th hour, the 73rd to 80th hour, the 169th to 176th hour, and the 337th to 344th hour of the 28-d menstrual cycle in the model, accounting for both dermal exposure to FHPs and inhalation exposure. Consequently, we compared the mean urinary concentrations of the 25 females at each time point with the average estimated concentrations during the corresponding 8-h intervals from the model. The target VOCs may also be contained in other non-FHP personal care products, 68 , 69 which were not included in the current model verification due to the lack of relevant exposure data among the female participants, 2 , 13 potentially leading to underestimation by the PBTK model. Among the target VOCs, benzene, toluene, styrene, and p -cymene had the most readily available data in the literature on additional exposure sources. 68 , 69 To conduct a further model verification by including a more comprehensive range of exposure sources, we incorporated additional exposure data from the literature for these four substances: 48.7, 94.6, and 171  ng / kg / d for dermal exposure to benzene, toluene, and styrene from sunscreen 69 and 42  ng / kg / d for inhalation and 508  ng / kg / d for dermal exposure to p -cymene from non-FHP fragrance products. 68 For our uncertainty analysis, we adopted probability bounds analysis, which investigated two extreme exposure scenarios beyond the best-guess scenario. 70 This method could efficiently capture the full range of potential predictions in a resource-efficient manner. The fifth and 95th percentile values of parameters in the PBTK model were used (Table S5). For the lowest exposure scenario, the 95th percentile BW among females 20–29 y of age in the United States was used. 52 In addition, associated physiological data (e.g., blood flows in tissue and tissue volumes) were considered. To simulate this scenario, the VOC amount in FHPs was reduced to 25% of the original value, reflecting that the mass of a panty liner was only 25% that of a regular pad, as measured in our previous study. 2 This adjustment was also applied to other FHPs in the absence of specific data, relying on empirical estimates. Conversely, for the highest exposure scenario, the fifth percentile BW among females 20–29 y of age in the United States was used, 52 along with associated physiological data. The amounts of VOCs in FHPs were doubled (because the mass of the super-size pad was twice that of the regular size) to simulate this scenario. 2 , 52 We conducted further assessments of noncancer risks [i.e., hazard ratio (HR) for adverse nonneoplastic health effects] and cancer risks (CR) associated with dermal exposure to VOCs from the use of FHPs using Equations 14 – 16 . These estimations were performed under three exposure scenarios (best-guess, lowest, and highest), as previously mentioned. A noncancer risk is considered notable when the HR exceeds 1 (or 0.1 in a more conservative scenario), and a cancer risk is deemed notable when the CR surpasses 10 − 6 (or 10 − 7 in a more conservative scenario). ( 14 ) D a i l y   d o s e = T i m e - c u m u l a t i v e   a b s o r p t i o n   × 10 − 6 × 13 × D E B o d y   W e i g h t   ×   28 × 13 × D E , ( 15 ) H a z a r d   r a t i o   H R   o f   d e r m a l   e x p o s u r e = D a i l y   d o s e R f D , and ( 16 ) C a n c e r   r i s k   C R   o f   d e r m a l   e x p o s u r e = D a i l y   d o s e × C S F where time-cumulative absorption is the cumulative mass (ng) absorbed via skin by using the four categories of FHPs; Body Weight is the BW (kg) of the US females 20–29 y of age 52 10 − 6 is to convert ng to mg; 13 is the number of menstrual cycles in 1 y, if a menstrual cycle is 28 d; D E is the exposure duration (year), herein referring to the number of years women experience menstrual cycles; 28 is the days of a menstrual cycle; RfD is the reference dose (mg/kg/d), an estimate of a daily exposure, under which there is unlikely an appreciable risk of adverse noncancer health effects 71 ; and CSF is the cancer slope factor (d/kg/mg), which is the upper bound (approximately a 95% confidence limit) for the increased cancer risk from a lifetime exposure expressed as dose (mg/kg/d). 71 Herein, we did not average the daily dose into a lifetime to assess the cancer risk as a scenario of an early onset cancer. The evaluation of health risks was conducted using RfDs and CSFs sourced from CalTOX ( Table 2 ) 72 or alternatively using data from oral exposure combined with gastrointestinal absorption factors (GIABS). 2 , 20 Health risks of estimated skin absorption of VOCs by using feminine hygiene products during a 28-d menstrual cycle in three scenarios. Note: RfD and CSF were sourced from CalTOX, 72 or alternatively using data from oral exposure and GIABS. 2 , 20 The best-guess scenario is the general scenario using the mean value of body weight of females 20–29 y of age in the United States, along with the corresponding kinetic and physiological data based on body weight. The 95th and fifth percentile body weights of females 20–29 y of age in the United States were applied to simulate the lowest and highest exposure scenarios, respectively, along with the corresponding kinetic and physiological data based on body weight. One-fourth and double the dermal exposure amounts from the best-guess scenario were used to simulate the lowest and highest exposure scenarios, respectively. —, no data; CFS, cancer slope factor; GIABS, gastrointestinal absorption factors; RfD, reference dose; VOC, volatile organic compound. Derived from oral exposure and GIABS. The modeling was performed using R (version 4.4.0; R Foundation for Statistical Computing; https://www.R-project.org/ ), and the source code is provided in the Supplemental Materials, “Model Code in R Software.” Internal VOC levels stabilized quickly with regular FHP use, typically reaching equilibrium by the sixth menstrual cycle (Figure S3). Therefore, most results below, except for the distribution in the “Results” section titled, “Distribution of VOCs in Different Tissues Based on PBTK Model,” are based on the modeling outcomes of the sixth menstrual cycle. The distribution results in section titled, “Distribution of VOCs in Different Tissues Based on PBTK Model,” in the “Results” section of this article focus on the first menstrual cycle to simulate the accumulation process. Measurements (VOCs in air, FHPs, and urine) below the MDL (listed in Table S1) were recorded as 0 and labeled as ND (not detected). A single missing urine sample was excluded from the analysis. The descriptive statistics provided in the main text and tables included the median, mean, standard deviation (SD), and range. Group differences were assessed using the Mann-Whitney U -test. Spearman correlation analysis was conducted to examine the relationship between log   K ow and the partition proportion in adipose tissue. All statistical tests were two-sided, with a type-I error rate set at 0.05. Analyses were performed using SPSS (SPSS, Inc.), and the results were visualized using Origin 2024b (OriginLab Corporation) and R (version 4.4.0).

The skin absorption amounts of VOCs from four FHP categories over a 28-d menstrual cycle were estimated using Equations 1 – 3 , based on measured VOC concentrations in FHPs. The estimated skin absorption of most VOCs, except for methylene chloride, exhibited a consistent increase throughout the duration of one menstrual cycle ( Figure 3A ). Methylene chloride, mainly existing in pads rather than other target FHPs, 2 was estimated to reach its peak absorption on the seventh day of the menstrual period. Overall, the cumulative estimated skin absorption of all target VOCs amounted to 2.08 × 10 5  ng during one menstrual cycle of 28 d, with individual VOCs ranging from 54.2  ng (methylene chloride) to 1.60 × 10 5   ng ( p -cymene) (Table S6). (A) The estimated skin absorption amount dynamics of eight target VOCs from four categories of feminine hygiene products during a 28-d menstrual cycle. (B) Contribution dynamics of four categories of feminine hygiene products to the estimated skin absorption amounts of eight target VOCs during a 28-d menstrual cycle. All the data are shown in Supplemental Excel Tables S2–S3. Note: VOC, volatile organic compound. In a 28-d menstrual cycle, all methylene chloride exposure was estimated to be attributed to pad usage. Pads also contributed to more than half of the estimated exposure to toluene (50.0%) and hexane (51.4%) ( Figure 3B ). Feminine wash products were estimated to contribute predominantly to the absorption of the total target VOCs (84.1%, 1.75 × 10 5   ng ; Figure 4 ), especially styrene (73.4%, 2,858  ng ), p -cymene (95.1%, 1.52 × 10 5   ng ), and n -nonane (82.8%, 1.23 × 10 4   ng ) (Table S6). Feminine wipes were estimated to account for 87.0% of the total absorption of methyl ethyl ketone ( 1,732   ng ), whereas moisturizers contributed 82.5% of benzene ( 2,370  ng ). (A) Sankey diagram illustrating the relative contributions of VOCs absorbed from four categories of feminine hygiene products during the first 28-d menstrual cycle and their distribution in human body at the end of a menstrual cycle based on PBTK model. (B) Scatter diagram between log   K ow and proportion in adipose after skin absorption of VOCs from four categories of feminine hygiene products during the first 28-d menstrual cycle and the trend line based on PBTK model. The correlation was tested by Spearman analysis. All the data are shown in Supplemental Excel Tables S4–S5. Note: log   K ow , logarithm of the octanol-water partition coefficient; PBTK, physiologically based toxicokinetic; VOC, volatile organic compound. During a 28-d menstrual cycle, the highest estimated concentration was 2.23  ng / mL of p -cymene in the adipose tissue, estimated 395 min after the end of menstrual period. Across different tissues, adipose tissue generally exhibited higher VOC levels than other tissues, whereas the urine showed the lowest estimated concentrations for most target VOCs ( Figure 5A ). The exception to this trend was methyl ethyl ketone, which was estimated to have the highest levels in urine and the lowest in liver. (A) VOC concentration dynamics in different tissues using four categories of feminine hygiene products in the first 8 h of a menstrual cycle based on PBTK model. (B) VOC concentration dynamics in different tissues using four categories of feminine hygiene products during 28 d of a menstrual cycle based on PBTK model. All the data are shown in Supplemental Excel Table S6. Note: PBTK, physiologically based toxicokinetic; VOC, volatile organic compound. Given that pads are almost exclusively used during menstruation, 73 the model estimated that VOCs such as hexane, which exhibited significant proportions of dermal absorption through pads ( Figure 3 ), showed increased concentrations in tissues (excluding adipose) and urine during menstruation in comparison with the postmenstrual period ( Figure 5B ). Methylene chloride experienced a dramatic increase at the beginning of the menstrual period and then maintained stable but slow increases and decreases in tissues along with the pad use during the menstrual period, followed by a large decrease during the postmenstrual period. For the VOCs present in the wash, the estimated concentrations in the blood, liver, kidney, and urine exhibited a sharp rise after wash use, usually peaking at 6 min before rapidly declining as the wash application stopped ( Figure 5A ). Afterward, the estimated concentrations of all target VOCs increased again due to the use of other FHPs, but the subsequent rises were less pronounced. For example, after the use of other FHPs, hexane and n -nonane peaked within 2 min, whereas benzene and p -cymene peaked within 8–14 min (Table S7). Other VOCs showed a more gradual increase, with their peak times ranging from 37 to 50 min. In the rest of the body, concentrations rose more slowly, peaking 9–49 min later than those in the blood. After peaking, concentrations of styrene, p -cymene, hexane, and n -nonane were estimated to decrease to ∼ 10 % or even lower until the next FHP use, whereas the concentrations of other target VOCs declined at a slower rate ( Figure 5A ). In contrast to the significant fluctuations estimated in other tissues and urine, the estimated levels of VOCs in adipose tissue displayed a very small undulation, with some VOCs, including methylene chloride, benzene, toluene, and hexane, maintaining a gradual upward trend through the end of the menstrual period. Even after 7 d of the menstrual period, the levels of VOCs in adipose tissue were estimated to experience only a slight decrease, with methylene chloride being the exception ( Figure 5B ). Throughout the subsequent 21-d postmenstrual period, most VOC concentrations in adipose tissue exhibited a more pronounced, yet still minor, pattern of increase and decrease. After exposure from using FHPs during the first 28-d menstrual cycle, the PBTK model predicted that 53.1% of the total target VOCs ( 1.10 × 10 5   ng ) were metabolized via liver. In addition, 25.8% ( 5.35 × 10 4   ng ) were estimated to be exhaled, and 0.02% ( 31.7  ng ) were excreted via urine, as illustrated in Figure 4A . The remaining VOCs were primarily distributed in adipose tissue, although constituting only 20.1% of the total target VOCs ( 4.18 × 10 4   ng ). The estimated adipose distribution primarily comprised p -cymene ( 4.11 × 10 4   ng ), n -nonane ( 538  ng ), and hexane ( 53.7  ng ). These distribution disparities may explain the higher estimated concentrations of most target VOCs in adipose tissue. In addition, the model predicted that the accumulation typically peaked within hours after the end of the menstrual period, echoing the peak concentration of p -cymene occurring about 395 min post menstruation, as discussed above. Regarding individual VOCs, their estimated proportions in adipose tissue varied widely, ranging from 0% ( 0  ng out of 54.2  ng for methylene chloride) to 25.7% ( 4.11 × 10 4   ng out of 1.60 × 10 5   ng for p -cymene). We further examined the logarithm of the octanol-water partition coefficient ( log   K ow ) of the individual VOCs and their estimated relative mass proportions in adipose tissue in relation to total skin absorption ( Figure 4B ), revealing a significantly positive correlation ( p = 0.037 , Spearman analysis). Throughout the 28-d menstrual cycle involving four categories of FHPs, the PBTK model estimated that VOC concentrations in urine were relatively low. p -Cymene presented the highest estimated level among all targeted VOCs ( Figure 5 ), which was merely up to 0.005  ng / mL and primarily resulted from using a wash product ( 3,053  ng / g of p -cymene in wash). The estimated urinary dynamics of all VOCs mirrored those in most tissues, exhibiting similar peak times and percentage changes (Table S7). For example, the urinary concentration of p -cymene was estimated to rapidly increase to the peak of 0.005  ng / mL after using FHPs, then decline to about 2.73 × 10 − 4   ng / mL over the next 24 h before the subsequent use cycle of wash products, with minor increases from using other FHPs like pads and wipes in between ( Figure 5 ). As for the difference between VOCs, the estimated urinary concentrations of hexane and n -nonane were consistently the lowest across all FHP categories ( Figure 6 ; Figure S4), despite their significant skin absorption amounts ( 1.98 × 10 4 and 1.49 × 10 4 ng, respectively). Although the estimated dynamics of these urinary VOCs aligned with those in most tissues, showing increases and decreases ( Figure 5 ), the changes in methyl ethyl ketone were less pronounced, attributed to its lower dermal penetration rate of 0.0011 cm / h . This difference was particularly evident when using wipes and moisturizers ( Figure 6 ), because the low dermal absorption allowed the residue to remain on the skin and then to continue the skin absorption until next use. A similar situation applied to methyl ethyl ketone in pads, because we assumed that the pads were always in contact with the skin, inducing the absorption to continue to occur. A 10 times of absorption coefficient was employed for using a tampon, to simulate the easy penetration of mucosa, so there was a small increase in methyl ethyl ketone when using tampons. VOC concentration dynamics in urine using individual feminine hygiene products during first 8 h of menstrual period based on PBTK model. All the data are shown in Supplemental Excel Table S7. Note: PBTK, physiologically based toxicokinetic; VOC, volatile organic compound. As indicated above, inhalation exposure was additionally considered when validating the modeling. The detection frequency (DF) of the target VOCs was measured to be low in indoor air samples (Table S1). Only benzene and toluene had DFs above 50%, with the highest measured concentrations at 1.38 and 14.6   μ g / m 3 , respectively. The measured indoor VOC levels in the Detroit samples were not significantly different from those sampled from a hotel in Ann Arbor, except for p -cymene, which was significantly higher in Detroit samples than that in Ann Arbor samples ( 0.23 ± 0.09   μ g / m 3 > 0.009 ± 0.037   μ g / m 3 , p = 0.001 ), as shown in Figure S5. Over a 28-d menstrual cycle, inhalation was estimated to account for over 96% to the total intake for most target VOCs (Table S6), except for p -cymene, where inhalation contributed only 22.4%. These estimates assume complete absorption of inhaled VOCs. A comparison between the estimated and measured urinary concentrations is shown in Figure 7A . The predictions of the model tended to slightly underestimate concentrations, typically within 1 order of magnitude. For example, estimates for methylene chloride, benzene, toluene, and styrene ranged from 22.1% to 33.7% of the measured values, with some cases reaching up to 48.3%. Methyl ethyl ketone estimates were lower at only 7.70% of the measured concentrations. This finding suggests that inhalation exposure to indoor air and dermal absorption from FHPs together contributed only about 8%–34% (up to 49% in some cases) of the overall internal exposure levels of VOCs. The underestimation of p -cymene, hexane, and n -nonane by the model was more significant than that of the other VOCs. Even though p -cymene estimates were only 1.82% of measured concentrations, hexane and n -nonane estimates were approximately 3 orders of magnitude lower than the measurements for other VOCs. These discrepancies indicate the presence of other, potentially larger, sources, especially for p -cymene, hexane, and n -nonane, such as the potential use of other wide ranges of personal care products and gasoline exposure, etc., for which relevant data were not available among the studied female participants. 2 , 13 Benzene, toluene, styrene, and p -cymene had available additional external exposure estimates in the literature as impurities or fragrance ingredients in non-FHP personal care products. 68 , 69 As stated above, these exposure source estimates were additionally added into the PBTK modeling for further model validation on these substances. It was found that the updated urinary predictions aligned better with the measurements in comparison with when these non-FHP personal care product exposure sources were not considered (Figure S6). The slight overestimation of p -cymene is due to the combined effects of the conservative exposure assumptions in this study and the broader exposure assessment reported in the literature. 72 This finding not only reinforces the likelihood of other significant exposure sources but also supports the reliability of the PBTK model in estimating the VOC toxicokinetics. The results of the uncertainty analysis are shown in Figure 7B . The difference in urinary concentrations between the best-guess exposure scenario and the lowest or highest exposure scenarios ranged from 11.7% (methylene chloride and hexane) to 19.5% ( n -nonane), with the exception of p -cymene. For p -cymene, the disparities in estimates were 65.9% for the best-guess exposure scenario in comparison with the lowest exposure scenario and 170% in comparison with the highest exposure scenario. Comparable concentration differences were observed in other tissues as well (Figure S7). (A) Comparison of the estimated and measured VOC concentrations in urine. The estimates were derived from using best-guess exposure scenario with both dermal exposure from using feminine hygiene products and inhalation exposure from indoor air, with the error bars showing the highest and lowest exposure scenarios. These estimates correspond to the mean levels in 505th to 512th h, the 73rd to 80th h, the 169th to 176th h, and the 337th to 344th h of the 28-d menstrual cycle in the model. The measurements ( n = 25 for each mean value, except n = 24 for the first time point) were from the urine of 25 females collected at four time points: 7 d before menstruation and 3, 7, and 14 d after the first day of menstruation. Each dot (mean with SD) in the comparison represents the value at each time point. (B) Uncertainty analysis results of urinary concentrations using probability bounds analysis based on PBTK model. The column is the estimated concentration using best-guess exposure scenario with both dermal exposure from using feminine hygiene products and inhalation exposure from indoor air. The minus bar indicates the lowest exposure scenario when using 95th body weight value among the US females 20–29 y of age, the associated kinetic and physiological data, and only one-fourth exposure dose. The plus bar indicates the highest exposure scenario when using fifth body weight value among the US females 20–29 y of age, the associated kinetic and physiological data, and double exposure dose. All the data are shown in Supplemental Excel Tables S9–S10. Note: PBTK, physiologically based toxicokinetic; SD, standard deviation; VOC, volatile organic compounds. We used the time-averaged daily absorption amount as the dermal exposure dose to assess the health risks ( Table 2 ). Under the best-guess exposure scenario, both HRs (noncancer risk) and CRs (cancer risks) were much lower than 1 and 10 − 6 (the recommended levels of noncancer and cancer risks), respectively, even lower than 0.1 and 10 − 7 (more conservative recommended levels of noncancer and cancer risks). Therefore, normal use of the FHPs was determined as unlikely to pose significant noncancer risks, such as skin and mucosal irritation, or cancer risks attributed to exposure to the target VOCs, whereas, under the highest exposure scenario, the HRs of n -nonane were 0.07, approaching 0.1, and the cancer risk of benzene exceeded 10 − 7 , reaching 1.82 × 10 − 7 . n -Nonane and benzene originated mainly from washes and moisturizers, respectively ( Figure 4 ; Table S6).

In this study, we integrated measurement data and PBTK modeling to reveal VOC toxicokinetics in FHP users. The modeling results suggest that FHPs can contribute to increasing levels of VOCs in all body tissues via dermal exposure, but mostly at low levels and decreasing within hours. However, it is worth noting that certain VOCs, especially p -cymene and n -nonane, were estimated to present a cumulative trend in adipose, which led to several orders of magnitude higher concentrations predicted in adipose than those in other tissues. Regarding the health risks, n -nonane occasionally contributed a potential noteworthy noncancer risk when using wash products, and benzene showed a significant cancer risk when using moisturizers. The estimated dynamics of VOCs in most body tissues including blood were characterized by an increase within minutes to one hour, followed by a significant decline of over 90% within hours (for most target VOCs). This pattern was consistent with findings from previous studies. Animal studies have shown that blood concentrations of benzene, toluene, styrene, and hexane rise quickly, often peaking within the first hour or even within 15 min after exposure. 74 – 76 These levels drop rapidly in the absence of continued exposure 71 ; for example, toluene concentrations can decrease to 15% of their peak within 2 h. 77 Georgopoulos et al. used a PBTK model to simulate human exposure to benzene in a household setting and found a sharp increase in skin absorption during a shower, followed by a rapid decrease. 78 However, a benzene exhalation study on a female rhesus monkey exposed to benzene through whole-hand immersion in a solution showed peak exhalation occurring approximately 2 h post exposure. 79 Other VOCs exhibited similar dynamics. Studies on chloroform absorption during showers showed that exhaled concentrations peaked within 20–30 min, as confirmed by both human monitoring and PBTK models. 80 , 81 Kim et al. combined in vivo studies with PBTK modeling to investigate dermal exposure to naphthalene among US Air Force personnel, showing an increase in blood concentrations during the first hour and peaking at around 62 min. 24 , 82 The estimated dynamics of rapid fluctuation are primarily attributed to the physicochemical properties and kinetic values of target VOCs. 83 Most target VOCs possess a skin absorption rate K p exceeding 0.01 cm / h ( Table 1 ), facilitating rapid absorption through the skin upon FHP use, particularly for p -cymene, hexane, and n -nonane. The absorption rates, combined with high partition coefficients between tissues and blood, resulted in sharp increases across various tissues ( Figure 5 ; Table S7). We found it notable that both skin absorption rates and partition coefficients of substances correlated with their log   K ow (Figure S8). 38 , 41 The subsequent rapid decline, especially for certain VOCs, was attributed to highly efficient metabolization by the liver, as seen with p -cymene and n -nonane ( Table 1 ) and partitioning into exhaled air, as in the case of hexane. The maximum reaction velocities ( v max ) of target VOCs are notably high, ranging from 9.53 × 10 3   ng / mL tissue/h for benzene to 1.06 × 10 8   ng / mL tissue/h for n -nonane ( Table 1 ). 54 , 59 – 62 In addition, the Michaelis constants ( K m ) even range from 144  ng / mL (methyl ethyl ketone) to 53,554  ng / mL ( n -nonane). 54 , 59 – 62 In comparison with other target VOCs, methyl ethyl ketone has a lower skin permeability coefficient ( 0.0011 cm / h ) 41 , 42 ; lower tissue partition coefficients, particularly in adipose tissue ( P adipose / blood = 0.88 ) 58 ; and higher renal excretion coefficient of 1.18 × 10 − 2 . 57 These properties led to its relatively slower skin absorption and greater urine excretion rather than accumulation in body tissues, as reflected by a later peak time (Table S7) and consistently higher levels in urine ( Figures 5 and 6 ). In adipose tissue, the estimated dynamics of VOCs showed much smaller fluctuations and even a gradual increase during the whole menstrual period. The estimated levels in adipose were 1–4 orders of magnitude higher than those estimated in other tissues ( Figure 5 ). Furthermore, several VOCs such as p -cymene and n -nonane were estimated to have significantly larger distribution in adipose ( Figure 4 ), suggesting a potential accumulation. A similar accumulation phenomenon was observed in rats exposed to n -nonane in a closed chamber. 84 The concentration in fat decreased only slightly during nonexposure periods and increased to higher levels with subsequent exposures, indicating significant accumulation. 84 These results are attributed to the log   K ow values of substances. Most of the target VOCs possess log   K ow values higher than 1 ( Table 1 ), indicating a preference for partitioning in lipid-rich tissues such as adipose tissue. Consequently, the partition coefficients between adipose and blood for target VOCs were correlated with log   K ow ( p < 0.001 , Spearman analysis; Figure S8) and much higher than those between other tissues and blood, such as liver and kidney ( Table 1 ). 38 , 58 The phenomena of higher levels and larger distribution estimated in adipose tissues were particularly significant for VOCs with log   K ow values exceeding 3.5, such as p -cymene, hexane, and n -nonane. Hamelin et al. also observed hexane accumulation in adipose tissue among occupational populations using physiologically based pharmacokinetic (PBPK) modeling and attributed this to its high lipophilicity. 85 The phenomena, including differences in distribution between tissues within the human body, as discussed above, extend beyond VOCs. Many chemicals have lipophilic properties, such as organochlorine pesticides (OCPs) and polycyclic aromatic hydrocarbons (PAHs). As a result, these chemicals tend to distribute more in body tissues such as adipose tissue rather than in blood or urine. 86 , 87 However, we usually collect urine or blood for quantifying the internal levels of chemicals in human body, given the ethical challenges and operational difficulties in obtaining samples from other body tissues. 88 , 89 Moreover, such quantification is typically based on a single type of biological sample. All these limitations may result in a bias of the quantification of internal chemical levels in the human body, especially when considering individual variations in body composition. Our study proposes a promising solution by integrating measurements with a PBTK modeling, offering a more comprehensive understanding of chemical distribution and accumulation within the body. This approach could yield a more precise evaluation of exposure levels and facilitate more thorough investigations into potential health effects. We have identified n -nonane and benzene to induce potential notable noncancer ( ∼ 0.07 ) and cancer risks ( ∼ 1.82 × 10 − 7 ), respectively, in certain cases by the use of FHPs ( Table 2 ). n -Nonane is an alkane usually included in products as solvents and lubricants. 26 An extremely high level of n -nonane, 1.44 × 10 4   ng / g , was detected in personal care products. 2 Prolonged exposure to n -nonane has the potential to cause irritation to the skin and mucous membranes, 90 whereas short-term exposure can result in neurotoxic effects. 33 Benzene, which usually undergoes rapid metabolism in the liver after absorption, resulting in low levels in all tissues and in urine, is a well-known carcinogen in humans that is associated with genetic defects. 91 A range of reactive species such as benzene oxide, benzoquinones, and benzene diolepoxide are produced after metabolism 92 and tend to distribute to lipid-rich tissues like bone marrow, based on their fat content and perfusion rate by blood, thereby causing multiple adverse health effects. 93 In previous studies, the external exposure doses of both n -nonane and benzene from the use of various personal care products have been estimated to yield significant noncancer and cancer risks. 2 , 94 In addition, both of these substances have shown significant positive associations between the measured levels in menstrual products and urinary concentrations in women in our previous study. 13 The results from the current study further reenforce the importance of reducing the existence and controlling the exposure to n -nonane and benzene via using FHPs. The underestimation of urinary concentrations could have occurred for several reasons. First, there might be other important sources of exposure. 68 , 69 Epidemiological studies have shown that females typically encounter higher levels of chemicals than males, primarily because they use a greater variety of personal care products, 95 – 97 which were not quantified or considered in this study due to absence of data. 2 , 13 p -Cymene is a widely used fragrance additive in personal and household care products and is even ubiquitous in food. 2 , 26 , 68 , 98 , 99 Hexane and n -nonane are both alkanes and usually included in many products as solvents and lubricants, including detergent and floor cleaners, 26 and have also been detected in food. 100 As indicated above, when additional exposure to non-FHP personal care products was considered in the modeling, the model performance was significantly improved (Figure S6). Second, the underestimation particularly existed among the chemicals with higher partition coefficients between adipose and blood, i.e., p -cymene, hexane, and n -nonane. This factor suggests that the disparities in distribution may result in the bias of estimation. Furthermore, we cannot obviate the possible limitations of PBTK model. A refined model compartment and structure that better simulates the actual physiological environment of human bodies would ensure a more accurate model prediction. However, the increasing number of parameters, stemming from the rise in the quantity of tissues and complexity of the model, would also introduce higher uncertainties. Due to the limitation of data, it is often difficult to carry out calibration and verification. The current study had several limitations. We were unable to fully evaluate each participant’s complete exposure profile when conducting model validation, including personal inhalation exposure due to sampling condition limitations and potential exposure from other sources due to lack of available data. For example, using other personal care products may increase the exposure; however, the lack of relevant data makes it difficult to consider this in the current study. Future research could validate the model by conducting a comprehensive assessment of aggregate exposure across more complete exposure sources and pathways and by employing longitudinal data derived from population-based controlled exposure studies. However, the current scenario has considered the most important routes with available data for model evaluation. The assignment of exposure parameters introduces uncertainty, e.g., variations in BW and the frequency of product usage, which were modeled using extreme exposure scenarios rather than actual measurements from the study population. Assumptions were made to simulate the worst-case scenario for a conservative assessment, including the assumed 10-fold increase in permeability to the target VOCs for vaginal tissue in comparison with normal skin and the complete transition of chemicals from products to the skin. Interactions between chemicals in mixture exposure scenarios and their potential impact on toxicokinetics were not explored due to insufficient data for parameterization. The metabolites of VOCs were not included in the simulation. Nonetheless, to the best of our knowledge, this study represents the most comprehensive research conducted under these specific conditions and is the first investigation to unveil a comprehensive assessment of VOC exposure in FHPs, spanning from product composition to urinary excretion levels in population. Although ADME processes in humans of these VOCs have been assessed in previous studies, a majority of them considered only inhalation exposure. 44 – 46 , 101 , 102 This study, novelly focusing on dermal exposure through using FHPs, benefits from the availability of relevant parameters from the previous publications. Our study also boasts several other notable strengths. We integrated actual measurements with model simulations to elucidate the dynamics of internal VOC exposure from the use of FHPs. Leveraging the PBTK model, we simulated the dermal exposure and the distribution of VOCs in the body by incorporating data from population surveys on FHP categories and usage, along with chemical concentration measurements in FHPs. We also included individual exposure data such as measured urinary concentrations to validate the model. Furthermore, uncertainty analyses were introduced to test the robustness of the model. These strengths underscore the significance of further, more comprehensive follow-up studies. In conclusion, our study using the PBTK modeling approach elucidates that, through skin absorption, FHPs can lead to elevated levels of VOCs in all body tissues, primarily at low concentrations and for short durations. In certain cases, benzene and n -nonane in products could pose significant health risks. It was surprising that our findings also reveal distribution disparities characterized by high levels of VOCs estimated in adipose tissue but low urinary levels, particularly for substances with high log   K ow values like n -nonane. Expanding our findings to other substances, it becomes clear that relying solely on one type of biological samples like urine, blood, or hair to characterize internal exposure in population studies may introduce a quantification bias, particularly for lipophilic chemicals. Evaluating tissue distribution in conjunction with PBTK modeling offers a promising avenue. Therefore, we recommend further studies employing comprehensive assessments that integrate exposure measurement with toxicokinetic modeling to yield more accurate results and inform regulatory measures. Future research should also incorporate extensive data to address comprehensive exposure routes and mixture chemical exposure scenarios in the model, which could enhance the model’s functionality to better address extending scientific issues and risk assessment related to VOC exposure.

Feminine hygiene products (FHPs), including tampons, menstrual pads, washes, wipes, sprays, and powders, are widely used across the world. 1 Over the course of the reproductive years, typically spanning from 12–51 y of age, a woman may use more than 10,000 pads or tampons. 2 Up to 40% of women use washes, wipes, and sprays. 3 The production of most FHPs typically involves the intentional addition of volatile organic compounds (VOCs) as fragrances, moisture barriers, and binders, 4 and VOCs may also be unintentionally introduced as by-products or contaminants from raw materials or during the manufacturing process. 5 An increasing number of studies have reported the detection of diverse VOCs in a broad range of FHPs sold in different markets around the world, such as benzene and styrene detected in sanitary napkins and panty liners sold in South Korea 6 and styrene and chloroform measured in a wide range of FHPs in the United States. 2 Exposure to VOCs in FHPs mainly occurs through dermal permeation 7 , 8 and inhalation. 9 – 11 Elevated concentrations of certain VOCs such as hexane have been detected in some FHPs in comparison with air, indicating dermal exposure as a potentially dominant route of intake, especially for products used on highly permeable and sensitive vaginal and vulvar tissues with efficient VOCs uptake rates. 2 , 12 A few human studies have substantiated a positive association between chemical concentrations in FHPs, due to the usage, and levels in the body of consumers. 3 , 13 Long-term exposure, including dermal exposure to high concentrations of certain VOCs, is relevant with various known or suspected health effects such as mucosa and skin irritation, liver and kidney damage, reproductive effects, and carcinogenicity, based on evidence from both animal 14 , 15 and human studies. 16 Benzene, a recognized carcinogen and common contaminant in FHPs due to production residue, has been identified as one of the top 10 chemicals of public health concern by the World Health Organization (WHO). 17 It frequently induces high exposure risk in multiple categories of FHPs, surpassing the benchmark level of cancer risk ( 10 − 6 ). 2 Dermal exposure to n -nonane can cause cumulative irritation with inflammation effect, as demonstrated in a rat study. 18 In addition, the reported reference dose of n -nonane induced by inhalation can be equivalent to 0.002 mg / kg / d , 19 significantly higher than that induced by dermal exposure ( 0.0003 mg / kg / d ), 2 , 20 also supporting the importance of recognizing dermal exposure to VOCs in assessing health risks. 21 An increasing number of studies have explored the exposure and associated health risks posed by VOCs from the use of FHPs. However, the majority have centered on quantifying external exposure levels. 2 , 6 , 22 Only a limited number of studies have explored the connections between external exposure levels and internal exposure (measured in urine and blood), as well as the importance of dermal exposure. 2 , 3 , 13 There is a scarcity of research depicting the absorption, distribution, metabolism and excretion (ADME) of VOCs within the human body following dermal exposure to FHPs. Consequently, the body burden of VOC exposure stemming from the use of FHPs remains largely uncharted territory. The physiologically based toxicokinetic (PBTK) model is an effective and acknowledged tool for quantifying the ADME process of chemicals, overcoming limitations of single biomarker use in exposure-level quantification. 23 It has been used in several similar investigations, e.g., quantifying the exposure of naphthalene via jet propulsion (JP) fuel 8 among members of the US Air Force, 24 and evaluating the dermal and air-to-skin exposure routes of diethyl phthalate via using personal care products. 25 Such type of models have advantages in the wide applicability to various exposure scenarios and physiological conditions and the achievable visualization of chemical ADME dynamics without in vivo experiments. This study integrates measurement data of VOCs in FHPs with a PBTK modeling approach (study design appears in Figure 1 ), aiming to a ) estimate the toxicokinetics of VOCs in body tissues resulting from dermal exposure to FHPs by linking the external VOC exposure from FHPs to the internal dose in specific tissues and b ) identify the specific chemicals and products posing significant health risks. Furthermore, this study validates its estimates by comparing them with measured urinary VOCs levels in a female population. The schematic study design. The image in the figure was taken by the authors, and the graphics were created using Adobe Illustrator 2020 (Adobe Inc.). Note: C AT , C B , C RB and C urine , chemical concentration in the adipose tissue, blood, rest of body and urine, respectively; K renal , the renal excretion coefficient; P adipose / blood and P rest of body / blood , partition coefficients of corresponding tissues to blood; Q AT , Q K and Q RB , blood flow to adipose tissue, kidney and rest of body, respectively; V AT and V RB , volume of adipose tissue and rest of body; V urine , volume of urine produced per unit time.