Human exposure assessment of organophosphate esters (OPEs) through dust ingestion and dermal absorption in Colombian cities

Concentrations of 19 organophosphate esters (OPEs) were determined in dust samples collected from house and car indoor microenvironments in three Colombian cities. ∑OPE concentrations ranged from 1.31 to 599 μg/g. Mean concentrations of dust homes were 82.6, 48.3, and 46.7 μg/g for Cartagena, Bogotá, and Medellín, respectively. The pollution inside cars was somewhat higher than in houses, with a mean value of 231 μg/g. Sixteen compounds were detected, being TPHP, DCP, TEP, and TCEP the most frequently detected. As for OPEs with higher levels in houses, we found (mean ± SD) 35.2 ± 37.1 μg/g for TDCIPP in Cartagena, 35.6 ± 80.2 μg/g for TPHP in Cartagena, 15.9 ± 31.4 μg/g for DCP in Cartagena, 35.7 ± 19.1 μg/g for TBOEP in Bogotá, 15.7 ± 14.8 μg/g for 4IPPDPP in Medellín, and 17.5 ± 22.9 μg/g for TCEP in Cartagena, while the highest OPE value found in cars was 176 ± 144 μg/g for TDCIPP. The estimated daily intake (EDI) of OPEs through dust ingestion ranged from 0.001 ng/kg bw/day for adults to 110 ng/kg bw/day for toddlers, while dermal absorption ranged from Page 2 of Olivero-Verbel et al. J Environ Expo Assess 2022;1:8 https://dx.doi.org/10.20517/jeea.2021.08 17 0.02 ng/kg bw/day for adults to 42.7 ng/kg bw/day for infants. Overall, the EDIs of dust ingestion were three times greater than those of dust dermal absorption. The estimated EDIs were several orders of magnitude below the corresponding reference doses. However, the incremental lifetime cancer risk (ILCR) for TCEP ranged from 1.1 × 10 for infants in Bogotá to 4.3 × 10 for adults in Cartagena, while ILCR for TEHP ranged from 8.8 × 10 for infants in Bogotá to 1.1 × 10 for adults in Bogotá. These estimated ILCRs were higher than the safe limit value of 1 × 10 and showed that these populations are exposed to moderate cancer risk.


INTRODUCTION
Flame retardants (FRs) are chemical additives that are incorporated into different materials such as plastics, textiles, foams, furniture, and electronic materials [1,2] . Their function is to prevent combustion or reduce the spread of fire after ignition [2] . Organophosphate esters (OPEs) have come to the fore in recent years due to their increasing use as an alternative to legacy brominated FRs [3][4][5] . Global demand for OPEs increased from 500,000 tons in 2011 to 680,000 tons in 2015 [6,7] . OPEs are also widely used as plasticizers in furniture, textile coatings, upholstery, electronics, paints, polyurethane foams, lubricants, and hydraulic fluids [8] .
OPEs have been found in different environmental matrices such as wastewater [9] , surface water [10] , drinking water [10,11] , air [3] , and indoor dust [12] , as well as in biotic samples such as river fish [13] and marine mammals [14] . In addition, toxicology and epidemiology studies indicate that these compounds are associated with immunotoxicity [15] , cardiotoxicity [16] , neurotoxicity [17] , adverse reproductive effects, and endocrine disruption [18] , and they are carcinogenic [19] . Tri-n-butyl phosphate (TNBP) causes sick building syndrome, which is related to the effects produced by chemical, physical, biological, and ergonomic agents, often related to the structure, distribution, facilities, and equipment of the building, as described by the World Health Organization [20] . OPEs are also considered to have neurodevelopmental problems after long-term exposure [21] . There is evidence that tris(1,3-dichloro-2-propyl) phosphate (TDCIPP) causes adverse effects on thyroid hormone (TH) imbalance in aquatic and avian organisms [22] and thyroid function and hormone homeostasis in mammals [23] . Tris-2-chloroethyl phosphate (TCEP) exhibits carcinogenicity and was eliminated by the EU in 1980 [24,25] . Furthermore, TDClPP may be carcinogenic [19] .
Humans spend their lives in a variety of indoor microenvironments, such as homes, offices, and cars. Research reports that OPEs are widely used in various industrial products such as plastics, building materials, textiles, furniture, electronic parts, and vehicle parts [26] . OPEs can be released from a multitude of products, which inevitably lead to them being found in these indoor microenvironments. Indoor environments are considered a hotspot for human exposure to OPEs [27] , which can occur from several routes including air inhalation, dust ingestion, dust dermal absorption, and dietary intake [5,28] . However, the pathways related with dust and air have been identified as the most important matrices for estimation of non-dietary exposure in the indoor microenvironment [26,28] , probably because of the large surface area in dust, becoming important deposits of OPEs [5,29] . In general, OPE levels in indoor dust are significantly higher than those in outdoor dust [30,31] . Total OPE levels have been reported in indoor dust where concentrations were generally at the nanogram per gram level, but in some cases they reach values at the microgram per gram level [12,32] .
This study aimed to evaluate the presence of different OPEs in dust samples from microenvironments of houses and cars, collected from the three most populated, industrialized, and polluted cities (Bogotá, Medell ín, and Cartagena) in Colombia. In addition, we assessed the dust pathway as non-dietary human exposure to these compounds through ingestion and dermal routes for several age groups (infants, toddlers, children, teenagers, and adults) as recommended by US-EPA [33][34][35] . To the best of our knowledge, this is the first study to show the occurrence of these FRs and plasticizers in several cities in Colombia.

Sampling collection
Dust samples were collected in three Colombian cities from house and car interiors in the cities of Bogotá (n = 20 houses), Medellín (n = 15 houses and n = 5 cars), and Cartagena (n = 20 houses) in 2018.
These cities have different climatic characteristics. Bogotá is the capital and the largest city in Colombia, located in the mountains of the Andes at 2586 m above sea level (asl), with moderately cold temperatures (an average of 14 °C) that fall below 5 °C in the rainy season. The relative humidity varies between 77% and 83%. Medellín is the second-largest city in Colombia, located in the Aburra Valley in the Andes at 1495 m asl, with humid weather (subtropical relative humidity between 63% and 73%) and moderate temperatures of 21.6 °C on average. Cartagena is a coastal city at sea level on the shores of the Caribbean Sea, humid (on average 78%-82%) with temperatures varying from 24 to 31 °C but generally higher than those cities located in the Andes.
Dust samples were collected from single or multiple bedrooms and living rooms of homes and apartments, selected for sampling. Floor dust samples were obtained from vacuum cleaner bags in each of the sampling sites. Samples were taken at the room normal conditions to observe the current human exposure. Prior to the analysis, samples were manually cleaned, removing particulate matter from stones, plastic, iron, grains, wood, and shavings. Samples were packaged in glass bottles, previously sterilized, labeled. and free of any organic or inorganic contamination, and immediately stored at -18 °C.

Sample preparation
Throughout all analysis processes, plastic material was avoided due to potential contamination, as our analytes are used as plasticizers. We tried to minimize as much as possible blank signals, i.e., heating all the non-volumetric material at 340 °C and rinsing with ethanol and hexane:acetone (1:1) just before use. For each batch of samples, a blank was included. Blank levels were subtracted from corresponding samples.
The analytical method applied was previously developed [13] . The dust sample (between 0.5 and 1 g) was spiked with internal standards (100 ng of d 12 -TCEP, d 15 -TDClPP, d 27 -TNBP, d 15 -TPHP, and 13 C 2 -TBOEP) and loaded into a 22 mL extraction cell previously filled with 0.5 g of powdered copper (particle size < 63 μm) and 8 g of diatomaceous earth sorbents (Hydromatrix). Extraction was done by pressurized liquid extraction (PLE) using an ASE350 system (Dionex, Sunnywale, CA, USA), with hexane:acetone (1:1) as solvent extraction and working at 1600 psi and 50 °C. After PLE, extracts were concentrated to incipient dryness and re-dissolved with methanol for a final volume of 500 μL.

Instrumental analysis
An online sample purification and analysis method based on turbulent flow chromatography (TFC) (Thermo Scientific TurboFlow™ system) in combination with a triple quadrupole (QqQ) tandem mass spectrometry (MS-MS) and a heated-electrospray ionization source (H-ESI) was used. Two liquid chromatography (LC) quaternary pumps and three LC columns were employed: Cyclone™-P (0.5 mm × 50 mm) and C18-XL (0.5 mm × 50 mm) for purification and Purosphere Star RP-18 (125 mm × 0.2 mm) with a particle size of 5 μm for separation [13] . Detailed conditions used for purification and chromatographic separation are included in Supplementary Table 1. Selective reaction monitoring mode was used for all compounds with two transitions monitored for each analyte. The most intense transition was used for quantification, while the second provided confirmation. Instrumental working parameters such as retention times, transitions, declustering potential, and collision energies are summarized in Supplementary Table 2.
Confirmation criteria for the detection and quantification of OPEs should include the following: (1) retention time for all transition monitored for a given analyte should maximize simultaneously ± 1 s, with signal-to-noise ratio ≥ 3 for each; and (2) the ratio between the two monitored transitions should be within 15% of the theoretical. Quantification was carried out by internal standard method based on the use of labeled OPE standards with d 15 -TDClPP, d 27 -TNBP, d 12 -TCEP, 13 C 12 -TBOEP, and d 15 -TPHP as internal standards.
Analytical parameters such as recoveries, limits of detection (LODs), and limits of quantification (LOQs) are summarized in Supplementary Table 3. These quality parameters were calculated by analyzing four replicates of spiked samples. Samples were spiked with 10 ng of OPE analytes and extraction was performed following the same procedure as for regular dust samples. Previously to final reconstitution, 10 ng of labeled internal standards of OPEs were spiked. Our analytical methodology provided recoveries ranging between 59% and 118%, LODs between 0.001 and 0.042 μg/g, and LOQs between 0.003 and 0.141 μg/g.

Statistical analysis
For the statistical analysis, a logarithmic transformation was performed on the data, which facilitated their use as some values obtained exceed the mean obtained by several orders of magnitude. Statistical analyses were carried out with R statistical software environment and Statgraphics Centurion XVIII, together with Excel. Significant statistical differences were considered when the P-value was ≤ 0.05.

Human risk assessment
Previous studies reported that humans can be exposed to OPEs in dust mainly via dust ingestion and dermal absorption. To evaluate human exposure, estimated daily intakes (EDIs) were calculated, expressed in μg/kg body weight (bw)/day, using Equations (1)-(3): where C (μg/g) represents the mean OPE concentration in dust samples, IR (g/day) is the daily dust ingestion rate, FT (%) is the fraction of time spent at the microenvironment (home or car), BW (kg) is the body weight, BSA (cm 2 /day) is the body surface area, DA (mg/cm 2 ) is the dust adhered to the skin, and AF is the fraction of OPEs absorbed by the skin [36][37][38][39] .
where ED (year) is the exposure duration, EF (day/year) is the exposure frequency, AT (day) is the average time, and SF (mg/kg/day) -1 is the slope factor for cancer risk. SF values were only available for the oral route for TCEP and TEHP [41][42][43] . SF values were not available for dermal route, and therefore oral SF values were used for all routes. Exposure to OPEs (TCEP and TEHP) through dust ingestion and dermal absorption was considered as cancer risk when the estimated ILCR was ≥ 1 × 10 -6 [41][42][43][44] .
Non-cancer and cancer endpoints, RfD, and SF of target OPE compounds are collected in Supplementary  Table 4.

Probabilistic model
The estimated EDI, HQ, LADD, and ILCR values on the health risk exposure of OPEs are represented as single values. The input values for calculation may contain some degree of uncertainty arising from multiple and different sources. Thus, simple point estimations are inaccurate. Therefore, an uncertainty analysis was performed associated relative errors of the factors through Monte Carlo simulation (MCS). The probabilistic analysis of MCS allows estimating the uncertainties of all possible outcomes and assessing the impact of risk, and therefore better decisions can be made under uncertainty conditions. In this study, MCS was performed using The Crystal Ball software, version 11.1.2.4 (Oracle, Inc. Redwood. US), and the number of simulations is defined as 10,000. A probability distribution of the exposure factors was assumed, and the detailed description of abbreviations, sources, probability distribution, and units for C, IR, FT, BW, BSA, DA, AF, FT, ED, EF, AT, RfD, and SF are available at Supplementary Tables 4 and 5. The uncertainty analysis for EDI, HQ, LADD, and ILCR was performed at 10,000 random repetitions for all calculations.

OPE levels in indoor dust
The mean concentrations of OPEs in dust samples collected from different homes and cars in Colombian cities are summarized in Table 1 (for individual sample results, see Supplementary Table 6). OPEs were detected in all analyzed samples, indicating widespread contamination by these emerging pollutants, with levels ranging between 1.31 and 599 μg/g. ∑OPE concentrations at homes varied between cities, being the highest values those obtained from Cartagena (mean ± SD), 82.6 ± 113 μg/g, followed by Bogotá and Medellín with similar values (48.3 ± 34.3 and 46.7 ± 48.1 μg/g, respectively) [Supplementary Figure 1]. Concentration levels inside cars were higher, with a mean ± SD of 231 ± 247 μg/g [ Table 1]. The higher concentrations found in Cartagena may be related to temperature differences between cities, being 24-31 °C in Cartagena, 16-26 °C in Medellín, and 3-19 °C in Bogotá. As OPEs are semi-volatile organic compounds, temperature changes may cause different partitions between air and dust, being OPEs with higher vapor pressure more sensitive to temperature changes. Therefore, the higher OPE concentrations found in Cartagena may be due to easier release of OPEs from products to the indoor environment at higher temperatures.
Different studies have been conducted on OPE occurrence in dust samples [Supplementary Table 7]. Comparison with published data must be carried out with caution because analytical methodologies included different OPEs (3-20 compounds) as well as differences in the sampling methodology. A study on indoor dust from homes in 12 different countries (China, Colombia, USA, India, Japan, and Vietnam, among others) reported values of 0.05-249 μg/g [12] . A similar range of concentrations has been observed in dust from homes in USA [45] , ranging from 16 to 224 μg/g. The most recent research [46] studied OPE levels in dust from homes, offices, and schools, obtaining levels between 5.82 and 148 μg/g. The values reported in these studies are very close to those found in our research, where we reported values between 1.31 and 446 μg/g. Regarding studies in cars, Tokumura et al. [47] found OPE concentrations up to 640 μg/g, values very close to those found in our study with values up to 599 μg/g.

OPE patterns
Sixteen of nineteen analyzed compounds were detected in dust samples and three were not detected in any samples (TNBP, IDPP, and THP). TPHP was the compound with the highest frequency detection in home dust samples (87%), followed by DCP (81%), TEP (72%), and TCEP (60%). In the case of car dust, TEP was detected in all analyzed samples (100%), followed by DCP (80%), and TPHP and TCEP were detected in the 60% of analyzed samples.
The persistence of TPHP in dust samples could be related to their widespread use in polyvinyl chloride (PVC)-based products, plastics, flooring, electronic products, hydraulic fluids, and flexible polyurethane foam for car upholstery [48] . Zhou and Püttmann [49] also reported a high TPHP occurrence (93%) in household samples and building materials markets. A study also related high TPHP values in dust to its presence in mattresses [50] . The high frequency detection of DCP could be related to their common use as plasticizer for PVC [20] . TEP follows as the most frequently detected compound, and this OPE is used in resins, plastics, rubbers, defoamers, and as flame retardant in rigid urethane foam [51] . Regarding TCEP, this compound is being replaced by TCIPP due to its carcinogenicity [52] , and, consequently, we also observed a high contribution of TCIPP in the total concentration of dust samples.
Zhou et al. [32] analyzed indoor dust samples from different indoor microenvironments in Germany, and they found that TBOEP and TCIPP were the most abundant OPEs. Another study of indoor dust showed that the most abundant OPEs were TCIPP, TCEP, and TPHP, with mean values of 0.52, 0.29, and 0.18 μg/g, respectively [53] . Wang et al. [54] analyzed dust samples collected from a college library, and they found the highest concentrations for TCEP (0.03-68.4 μg/g), TCIPP (0.11-64.6 μg/g), and TBOEP (0.10-58.2 μg/g). Thus, the different published studies, as well as the results of our study, showed the prevalence of the same OPE compounds.

Daily intake estimations
EDI values were calculated based on exposure factors and OPE concentrations in dust from homes. Humans can be exposed to OPEs by dust through non-dietary routes such as ingestion, air inhalation, and dermal absorption. However, ingestion and dermal absorption have been reported as predominant pathways [12] . The EDIs of OPEs through dust ingestion (EDI Ingestion ) and dermal absorption (EDI Dermal ) were estimated. The results are presented in Supplementary Tables 8 and 9. Because body weight, ingestion rate, and body surface area vary with the age, we estimated the EDIs for five age groups: infants (< 1 year), toddlers (1-3 years), children (4-10 years), teenagers (11-18 years), and adults (≥ 19 years).
The obtained mean EDI Ingestion values ranged from 0.03 ng TPPO/kg-bw/day for teenagers and adults from Bogotá to 110 ng TDClPP/kg-bw/day for toddlers from Cartagena. Regarding EDI Dermal , mean values were between 0.49 ng TPPO/kg-bw/day for adults from Bogotá and 42.7 ng TCEP/kg-bw/day for infants from Cartagena. EDIs varied depending on the pathway, the age group, and the city. Among the cities, EDIs decreased for the most of OPEs in the following order: Cartagena > Medellín > Bogotá. As regards the age group, in general, EDIs decreased with the age. Toddlers were the most exposed group for dust ingestion, being their EDI Ingestion values on average 50 and 70 times higher than those for teenagers and adults, respectively. Dust ingestion is an important pathway of pollutant exposure for toddlers because they spend a significant amount of time on the floor, where they touch different objects with their fingers and transfer them to the mouth, as a result of the hand-to-mouth behavior. On the other hand, infants presented the highest EDI Dermal values, being 50 times greater than those for teenagers and adults. Moreover, dust ingestion was the dominant pathway for infants and toddlers, whereas dermal absorption was the dominant pathway for teenagers and adults [ Figure 2].  [55] .

Human exposure assessment
Monte Carlo analysis has become a powerful tool in environmental risk assessment because it incorporates the quantitative uncertainty analysis into the risk estimate. HQ values were simulated by Monte Carlo technique providing a useful, systematic, and quantitative description of the human health risk of OPE exposure via dust ingestion and dermal absorption. The final result was a probabilistic density function (PDF) of the HQ for OPEs exposure through dust. HQ values were reported for the sum of the ingestion and dermal absorption pathways, which varied depending on the age group, city, and type of OPE [ Table 2, Figure 3]. The obtained mean values ranged from 0 to 0.014, being always below the safe value of 1. Similar results were obtained when exposure to dust car was also included and added to the exposure to dust homes [Supplementary Table 11]. The highest HQ values were found for TCEP: this compound was one of the most frequently detected OPEs (60%), found at high concentration levels in home dust (mean value of 13.7 μg/g), and has the lowest RfD value (7 μg/kg-bw/day). Supplementary Figure 2 shows the PDFs of the risk for TCEP in Bogotá homes, and we observed an approximately lognormal distribution for the different age groups. Expected risks were 5.52 × 10 -3 , 6.13 × 10 -3 , 3.62 × 10 -3 , 2.08 × 10 -3 , and 1.94 × 10 -3 for infants, toddlers, children, teenagers, and adults, respectively. However, there is a roughly 50% probability that the expected risk for the population will be less than these values. In general, TCEP is 100-800 times below the expected risk for the whole population, TCIPP 400-3000 times, TDClPP 200-3000 times, TEHP 5000-1 × 10 6 , TMCP 200-3 × 10 4 times, and TPPO 2000-5 × 10 6 . These results suggest that the contribution of dust ingestion and dermal absorption to OPE exposure do not pose a health risk for infants, toddlers, children, teenagers, and adults in the three Colombian cities.
Some toxicological studies have found that TCEP and TEHP have carcinogenic potential and indoor dusts vary according to the indoor environments studied [56][57][58] . ILCR values were estimated in the three cities and for the different age groups [ Table 3, Figure 4]. The ILCRs estimated by MCS were higher via dust ingestion than those via dust dermal absorption. The guideline recommended by US-EPA as safe limit for ILCR is a value of 1 × 10 -6 [39] . ILCR values estimated for TCEP ranged from 1.14 × 10 -5 for infants in Bogotá to 4.27 × 10 -4 for adults in Cartagena, while ILCRs for TEHP ranged from 8.79 × 10 -7 for infants in Bogotá to 1.03 × 10 -5 for adults in Bogotá. Due to the assumed longer lifetime exposure for adults, ILCR values were 40 and 12 times higher in adults than in infants for TCEP and TEHP, respectively. In a worst-case scenario and assuming the highest dust intake and dermal absorption, ILCRs estimated at 95th percentile were 1.39 × 10 -3 for TCEP in Cartagena and 1.71 × 10 -5 for TEHP in Bogotá. Because mean and 95th percentiles were higher than the safe limit of 1 × 10 -6 , there is a moderate probability of cancer risk through TCEP and TEHP exposure by dust ingestion and dermal absorption, which is a cause of concern for the public health. A recent study in Saudi Arabia [55] estimated a cancer risk from OPE exposure via dust similar to our estimations for the Colombian population. Our ILCR results were three orders of magnitude higher than those reported for Saudi Arabia.
This study only focused on the OPE exposure through contaminated dust. However, it is important to note that OPE exposure also occurs in other indoor environments (workplaces), as well as by other routes, such as air inhalation and food and beverage ingestion. The sum of all these exposures should increase the estimated risk, both for non-carcinogenic (HQ) and carcinogenic (ILCR) values, which would lead to a higher risk of harmful effects on human health. Different research studies have determined that prolonged human exposure to OPEs can cause adverse effects such as neurotoxicity, reproductive toxicity, carcinogenicity, and endocrine disruption.  It is important to make a special mention of the many uncertainties associated with the estimations of OPE exposure because various factors such as personal habits, occupation, dietary preferences, variation of OPE concentrations from room-to-room, continuous use of consumer products containing OPEs, time spent indoors or outdoors, seasonal dust variation, skin exposure, or ventilation can all affect the magnitude of the exposure.

CONCLUSIONS
The occurrence of OPEs in dust from homes and cars in three Colombian cities was reported for the first time. Sixteen OPEs were detected at concentrations up to 599 μg/g, being the levels similar to those reported in other countries around the world. Levels in car dust samples were higher than those of home dust, but the most contributing OPE compounds were the same in both microenvironments: TCIPP, TDClPP, TPHP, TBOEP, and TCEP.
The EDIs through dust ingestion and dermal absorption were estimated for five different age groups. In general, EDI values decreased with age. Moreover, dust ingestion was the dominant pathway for infants and toddlers, whereas dermal absorption was the dominant pathway for teenagers and adults. Non-carcinogenic risk was evaluated by the estimation of HQ values using MCS, which provided quantitative description of the human health risk of OPE exposure via dust ingestion and dermal absorption. Obtained mean values ranged from 0 to 0.014, being always below the safe value of 1. Moreover, the carcinogenic risk was also estimated by the ILCRs obtained through MCS, showing higher values for dust ingestion than those for dermal absorption. ILCRs ranged from 1.14 × 10 -5 to 8.79 × 10 -7 , being for many cases higher than the safe limit value (1 × 10 -6 ). Due to the assumed longer lifetime exposure for adults, ILCR values were forty and twelve times higher in adults than infants for TCEP and TEHP, respectively.
Estimated ILCR values for TCEP and TEHP showed that these populations are exposed to moderate cancer risk, being a cause of concern for the public health. In this context, adopting the precautionary principle, it would appear desirable to reduce overall human exposure to OPEs. Environmental legislation should pay more attention to this health problem and act with measures aimed at reducing the degree of human  Not applicable.