Sociodemographic and Lifestyle Factors Associated with Historical Exposure to Persistent Flame Retardant Concentrations in a Spanish Cohort

1. Introduction

Persistent organic pollutants (POPs) are a myriad of chemicals that share a high resistance to degradation. As a result, they have a high potential for bioaccumulation in living organisms and biomagnification over the food chain [1,2]. Ingestion, dermal contact and inhalation are considered the main exposure routes in the general population [2,3,4]. Some POPs, such as Polybrominated diphenyl ethers (PBDEs) and Dechlorane plus (DP) are frequently used as flame retardants in a variety of products such as textiles, plastics, construction materials or electronics [5,6,7].

PBDEs are a group of 209 congeners (designated BDE-1 to BDE-209) that have been widely used since 1970s. PBDEs have been produced in different commercial mixtures, including Penta-BDE, Octa-BDE and Deca-BDE [8,9]. Penta-BDE is predominantly composed of lower-brominated BDE congeners, with BDE-47 and -99 accounting for 38% and 49%, respectively; Octa-BDE refers to a mixture of PBDEs with 6-10 bromine substitutions (e.g., BDE-183); and Deca-BDE is composed primarily of BDE-209 (90%) [10]. DP was introduced in the market in the 1960s and it was used as an alternative to PBDEs and organochlorine pesticide Mirex [6,11]. DP was commercialized as a mixture of their two stereoisomers, being predominant anti-over syn-DP [12,13].

Currently, most of the abovementioned chemicals have been listed for global elimination on different amendments of the annex A of the Stockholm Convention in 2009, 2017, and 2023, and regulated by the UE [14,15]. However, these pollutants are still frequently detected in the general populations [4,16,17], with some of them considered as endocrine disrupting chemicals (EDC), i.e., xenobiotics that can interfere with the endocrine system [18]. Frequent human exposure to low doses of EDCs has been linked to an elevated risk of developing several chronic non-communicable diseases, such as cancer, thyroid disease, as well as reproductive, neurobehavioral, and developmental disorders [19,20,21,22,23,24,25].

Considering the ubiquity of PBDEs/DP, human biomonitoring has emerged as a crucial tool to accurately evaluate individual chemical exposures, accounting for most exposure sources [26]. In addition, human biomonitoring allows the identification of trends in the exposure levels of the population [27]. In this regard, serum has been primarily used as the exposure matrix in most human biomonitoring studies [28,29,30,31,32,33]. Despite the challenges in the sample collection, adipose tissue has been demonstrated to provide relevant information related to long-term exposure to moderately-to-high lipophilic chemicals [3,34,35,36]. Thus, adipose tissue may help identify population groups with elevated accumulated exposure, thereby contributing to the adequate design of effective public health campaigns.

Based on the abovementioned, the aim of this study was to estimate historical exposure to a selection of PBDEs and DP in a subsample of a Spanish adult cohort by analyzing their residues in adipose tissue samples, as well as to identify the potential sociodemographic and lifestyle factors associated.

2. Materials and Methods

2.1. Study Design and Population

This cross-sectional study was conducted on a subcohort of the GraMo cohort (Granada province, Southern Spain), which aims to identifying environmental factors affecting the development of several chronic diseases. A thorough description of GraMo cohort has been performed elsewhere [4,5,6,7,8]. Briefly, participants were recruited between July 2003 and June 2004 from two public hospitals: the Clínico San Cecilio University Hospital in the city of Granada and the Santa Ana Hospital in the coastal town of Motril. Granada province includes rural, semirural and urban areas. The area of Granada city (238,292 inhabitants in 2004) and its suburbs are economically based in the service sector and light industry; while Motril (55,078 inhabitants in 2004) and its semirural area in the Mediterranean coast are surrounded by a great presence of greenhouses and other agricultural activities [37,38]. These two areas are separated by approximately 70 km. Participants were selected from patients who underwent non-cancer-related surgeries. The inclusion criteria were as follows: age older than 16 years, no cancer diagnosis, no hormonal disease related to the hypothalamic axis, no hormone therapy, and at least 10 years living in the referral areas of the recruiting hospitals. 387 of the 409 people contacted (94.6%) agreed to participate and provided biological samples. 179 participants provided sufficient biological sample for PBDEs/DP analysis, among them 134 (74.96%) had complete information about sociodemographic and lifestyle factors and constituted the study population in the present work (Figure 1).

Although the data collection for this cross-sectional study was conducted in 2003-04, the analysis of PBDEs and DP was carried out in 2024 following the subsequent funding received from the Instituto de Salud Carlos III, Spain (PI20/01568). The original study protocol and ethical aspects were approved by the Ethics Committee of each participating hospital in 2002, and the current study received approval from the Provincial Ethics Committee of Biomedical Research of Granada (Comité de Ética de Investigación Clínica de Granada, 8/2016). All participants included in this study were duly informed and signed the corresponding informed consent.

2.2. Laboratory Analyses

The concentrations of 12 brominated flame retardants [bromo diphenyl ether (BDE) -28, -47, -99, -100, -153, -154, -183, -196, -197, -203, -209 and 1,3,5-tribromo-2-[2-(2,4,6-tribromophenoxy) ethoxy] benzene (BTBPE)], and the 2 DP isomers (syn-DP and anti-DP) were analyzed in adipose tissue samples. 5-10g adipose tissue samples were collected during each participant´s surgery. The samples were immediately coded and frozen at -80°C until chemical analyses.

The sample preparation procedure was adapted from others previously described [39,40]. Briefly, Adipose tissue samples (in general, between 150-250 mg) were spiked internal standard (BDE 79 and BDE138) and after 30 min equilibration time, transferred to a 10 ml glass tube for extraction. In order to ensure complete disintegration of the sample and facilitate deeper penetration of the solvents [41], adipose tissue was in the first stage, crushed and homogenized with an Ultra-Turrax® (T25, IKA) using 4 mL of n-hexane:Acetone (1:1) as solvent extraction. Then ultrasound-assisted extraction was carried out in an ultrasonic bath for 15 min at 40 ºC. The extract obtained after centrifugation (4000 rpm, 4 °C, 10 min) was evaporated at 40 ºC with a gentle flow of nitrogen up to 0.5 mL. The extract was cleaned in a multilayer silica gel column, packed in a 6 mL empty glass column and formed by 0.8 g of activated silica gel, 3 g of silica modified with sulfuric acid (44%, w/w), and 0.3 g of anhydrous sodium sulphate. Analytes were eluted with 8 mL of n-hexane and 6 ml of dichloromethane (DCM) (1:1, v/v). The extract was evaporated to dryness and finally reconstituted with of 100 µL of 13C-labeled BDE-209 solution in isooctane.

Based on previous study [42], instrumental analysis was performance by gas chromatography (GC) (HP 7890A Series, Hewlett-Packard, Palo Alto, CA) equipped with a multimode inlet (MMI) and coupled to mass spectrometer operating in negative chemical ionization (NCI, 5975C Agilent). Thus, 5 µL were injected in large volume injection (LVI) and chromatography separation was carried out by DB-5MS column (15 m x 0.25 mm x 0.10 µm, Agilent). The GC oven temperature was programmed from 80 ºC (held for 2.6 min) to 200 ºC at 30 ºC/min, and then up to 275 ºC at 5 ºC/min, and finally to 315 ºC (6min) at 50ºC/min. Helium was used as the carrier gas, ramping the flow from 1.5 mL/min (19.8 min) up to 4.5 mL/min (9 min) to avoid BDE 209 column degradation. Mass spectrometer used methane as reagent gas and 70eV of electron energy. Transfer line, quadrupole, and ion source temperatures were set at 300, 150, and 150 ºC, respectively. Quantification was performed in selected ion monitoring (SIM).

Quantification based on nine-point calibration curves from 0.05 to 20.00 ng/mL, except to PBDE 209, i.e. 0.5–200 ng/mL, with satisfactory linearity (R² > 0.996) was obtained. For internal quality control purposes, procedure blanks and quality controls at 0.1 ng/mL and 1 ng/mL (1 and 10 ng/mL for PBDE 209) processed in the same way as the samples were included in every analytical series i.e. twelve samples. As confirmation criteria, retention times and the maintenance of the two selected ions ratio within 15% of the standard value were used.

The lipid content in adipose tissue samples was quantified by gravimetry, as reported Rivas et al. [43], which was used for normalizing PBDE/DP concentrations, as reported by Phillips et al. [44]. The final concentrations were then expressed as nanograms per gram of lipid (ng/g lipid).

Limits of quantitation (LOQ) for the PBDE congeners, BTBPE and DPs ranged from 0.0029 to 0.0785 ng/g lipid (Supplementary Material, Table S1). The chromatography laboratory at the Centro Nacional de Sanidad Ambiental has participated in the HBM4EU QA/QC program, resulting in its qualification as EU laboratory for the analysis of PBDE and DPs in human serum.

2.3. Sociodemographic, Lifestyle and Clinical Information

Sociodemographic, dietary habits, occupation, perceived health status and chemical exposure information were gathered from face-to-face interviews by trained staff during the participants’ hospital stay at recruitment. Dietary habits were assessed using a semi-quantitative questionnaire of food consumption, including the following food groups: meat, cold meats, fats, fish, eggs, dairy products (excluding milk and cheese), milk, cheese, vegetables, legumes, fruits, bread, and pasta. A participant was considered as smoker at any level of daily tobacco consumption (≥ 1 cigarette/week), and alcohol consumer at any level of weekly alcohol consumption (≥ 1 glass/week). Body mass index (BMI) was expressed as weight/height squared (kg/m²). Participants’ area of residence was classified as urban (> 100,000 inhabitants), semi-rural (10,000-100,000 inhabitants), or rural (< 10,000 inhabitants). Following the Goldthorpe social class classification [45], participants were assigned to six occupational categories. Because of sample size limitations, social classes I–III were grouped as manual workers, and social classes IV–V as non-manual workers.

2.4. Statistical Analyses

Descriptive analyses included absolute values and percentages for categorical variables; mean, standard deviation (SD) and median and interquartile range (IQR, 25th and 75th percentiles) for continuous variables. Chemical concentrations were reported in ng/Kg lipid for clarity in the descriptive analysis, whereas these variables were entered as ng/g lipid in the regression models to improve the interpretability of the model´s coefficients.

For PBDEs and DP with quantification rates above 70%, values below the LOD were imputed by assigning a random number between 0 and the LOQ. BTBPE, BDE-196, and BDE-203 were excluded from analyses because their quantification rates were below 70%. Concentrations were natural log-transformed to relax the linearity assumption. Consequently, the beta coefficients (β) reflect the average change in the natural log-transformed pollutant concentration (ng/g lipid) associated to a 1-unit increase in continuous independent variables. For categorical variables, β indicates the average difference in the log-transformed concentrations between that category and the reference category. Associations of sociodemographic, lifestyle, occupation and perceived chemical exposure (independent variables) with PBDEs and both DP isomers (dependent variables) were identified using multivariable linear regression with manual step forward-backward model variable selection. First, variables were selected based on the available literature and biological plausibility as potential factors of PBDEs/DP exposure [46,47,48,49,50]. Then, in the forward stage, all new independent variables with a p-value ≤ 0.20 were included in the same multivariable model. Finally, in the backward stage, all independent variables with p-value ≥ 0.10 and/or reduction or no-significant increase of Bayesian information criterion (BIC) were excluded.

The significance level was stablished at p ≤ 0.05 and, given the exploratory nature of the present study, we also consider a p-value of ≤ 0.10 as borderline-significant [51,52].

Statistical analyses were performed using R Statistical Software v4.5.0 (R Core Team 2025) [53]. Linear regressions were performed by using base R and descriptive tables by “gtsummary” package (v2.2.0) [54].

3. Results and Discussion

3.1. Description of the Study Population and Adipose Tissue Pollutant Concentrations

The main characteristics of the study population are summarized in Table 1. This subcohort included slightly fewer females than males. The mean BMI in our cohort was 27.9 kg/m²; with no relevant difference among sexes. This was higher than the average BMI reported for the Andalusian population in 2003 [55], likely because our hospital-based cohort included participants with obesity-related conditions. The main source of adipose tissue samples was hernia surgery, followed by other surgeries and gallbladder surgeries. In our population, approximately one third of the participants were smokers at recruitment, and 69.9% of males and 29.5% of females reported being alcohol consumers. These figures are similar to those reported for the Andalusian population [55]. In our study, manual workers were predominant, while a higher proportion of females (but not males) resided in semirural areas vs urban and rural areas.

Adipose tissue PBDE/DP concentrations are summarized in Table 1. Males showed consistently higher levels of PBDEs than females. PBDE/DP detection frequencies ranged from 82.8% (BDE-209) to 100.0% (BDE-99, -153, -154, -183 and anti-DP) (Supplementary material, Table S2). In accordance with previous studies, BDE-153 was the predominant congener [9,31,33,56,57,58,59,60,61,62]. This finding may be explained by the following: (i) increased relative exposure to BDE-153 compared to other PBDEs, and (ii) slower metabolism in humans [9,31]. In fact, lower brominated BDEs, such as BDE-153, have been found to have longer half-lives and are therefore more likely to accumulate at higher concentrations in human tissues. Congruently, in our study, DBE-209, with a relatively high degree of bromination, was found at the lowest detection rates [63].

No differences in bioaccumulation were observed between the two DP isomers. In our study, anti-DP was the predominant form with an average anti-DP fraction (anti-DP/total DP) of 0.75 ± 0.11 (data not shown), consistent with the reported proportions in commercial DP products ranging between (0.60-0.80) and in Brasseur et al observations [12].

3.2. Factors Associated with Adipose Tissue PBDEs and DP Concentrations

As described in the materials and methods section, we fitted a model for each individual pollutant by entering all the variables found relevant during its individual backward selection. For clarity purposes, each model`s data are split into 3 tables, corresponding to sociodemographic (Table 2), perceived exposures (Table 3) and dietary (Table 4) variables.

Regarding sociodemographic and anthropometric characteristics (Table 2), males showed higher BDE-28, -47, -153 and -209 adipose tissue concentrations than females. Our results are in agreement with those from previous studies [33,64,65,66,67,68,69], since the observed lower levels of PBDEs in females may be the result of a clearance process during pregnancy and breastfeeding, as Bramwell et al. hypothesized [33]. Concretely, internal flame retardants’ levels may be reduced during breastfeeding due to the high affinity of flame retardants for lipid-rich structures. In addition, PBDEs cross the placenta and can selectively accumulate in the fetus and the umbilical cord [70,71]. These processes are consistent with previous studies that have detected flame retardants in breast milk [9,72,73], and umbilical cord blood [71,74,75,76].

The higher PBDE concentrations in males vs. females in our study population may be related to lifestyle differences, as previously observed for other POPs in GraMo cohort [37,77,78]. Indeed, in our study, males reported an increased consumption of fatty food from animal origin, such as pork (23.3% vs. 6.6%), while chicken consumption was predominant in females (41.0% vs. 26.0%). Additionally, 20.55% of males, and 1.64% of females acknowledge involvement in industrial activities such as operators of installations, machinery and assemblers, craftsmen and, workers in industry construction and mining.

Interestingly, age was inversely linked to BDE-47 and -99. These findings are consistent with prior studies reporting inverse associations of age with BDE-47 and -99, with lower levels for middle-aged adults compared to younger or older adults [79,80]. The explanation may be the relatively recent manufacture and environmental release of PBDEs, and consequently, younger individuals have therefore spent most of their lives exposed to higher environmental concentrations, whereas older adults were largely unexposed for much of their lives [79,80].

BMI was negatively associated with BDE-153, -183, -197 and syn- and anti-DP (Table 2). A sensitivity analysis entering BMI categorized in three categories as < 25 kg/m², 25-30 kg/m² and ≥ 30 kg/m², corroborated this negative trend (data not shown). Similar inverse associations for BDE-153 were described in previous studies using serum as the matrix for exposure assessment [30,33,67,81]. This negative trend may be attributable to a dilution effect of PBDEs in obese individuals as initially proposed by Wolff et al. [30,82]. This might be particularly relevant in our populations, with 39,6% of overweight and 27,6% of obese individuals.

Manual workers showed lower BDE-28 and higher BDE-154 and anti-DP and syn-DP concentrations compared to non-manual workers (Table 2). These findings are consistent with previous studies identifying specific manual activities such as industrial manufacturing industry worker, construction or e-waste recycling sites have been identified as sources of PBDEs [83,84].

Regarding perceived exposures (Table 3), higher BDE-28, -47, -99, -100 and -153 levels were associated with perceived exposure to paints. Furthermore, elevated BDE-154 and -183 concentrations were associated with perceived exposure to solvents. In addition, some paints, which often contain PBDEs in their composition [7,85,86], have shown positive correlations with these pollutants as reported in a Korean study that used umbilical cord blood as biological matrix [87].

Lower concentrations of BDE-47, -153, -154 were associated with perceived exposure to toxic metals (Table 3). Toxic metals can co-occur in the same PBDE sources such as recycling sites or industrial activities [88]. Thus, perceived exposure to toxic metals does not appear to be a useful indicator of highly PBDE/DP exposed individuals. The negative associations observed may be explained by greater implementation or adherence to safety measures among individuals who perceive themselves as being at risk.

Higher BDE-99 levels were observed in former smokers but not in current smokers (Table 3). Lower BDE-209 levels were associated with alcohol consumption. Both tobacco and alcohol can upregulate cytochrome P450 family [89,90,91,92], particularly tobacco consumption induces CYP2B6 [93], which is thought to be the main via of metabolization of BDE-99 [94,95,96]. Thus, smoking would be expected to enhance metabolic clearance of BDE-99. However, we did not observe reduced BDE-99 levels among current smokers. Additionally, BDE-209 metabolic pathways remain incompletely understood [94,97]. We cannot provide a definitive explanation and cannot exclude the possibility that this finding arose by chance.

Regarding dietary variables (Table 4), higher BDE-197 levels were associated with oily fish consumption and elevated BDE-183 and -197 with consumption ≥ 2 portions/week. This may be due to the fact that fish has previously been identified as the main dietary source of PBDEs [50,98,99]. BDE-28 was also linked to meat consumption frequency larger than 2 portions per week. This is consistent with previous epidemiological studies that have associated meat consumption with PBDE levels [50,100,101]. However, in our population, an inverse association was found for BDE-99 and eggs (Table 4). This finding is unexpected, since BDE-99 has been reported as the dominant congener in eggs [102]. We cannot offer a clear explanation for this inverse association, which can even be a chance finding caused by the exploratory design of this study.

Other fatty products from animal-origin, such as dairy products, have been established as source of organic pollutants [103]. In our study, higher BDE-153 concentrations were associated with cheese consumption and daily milk consumption.

In our study, lower BDE-153 levels were associated with vegetable consumption. This is consistent with prior research reporting lower PBDE levels among individuals following vegetarian versus omnivorous diets [100]. In addition, consumption of dietary fiber, characteristic of plant-based diets, may reduce the intestinal absorption of organic pollutants [104,105].

Nonetheless, higher BDE-197 concentrations were positively linked with fruit consumption. Other congeners have been related to fruit through industrial processes such as “hydro-cooling” which involves the use of BDE-209 coated pallets [7,106]. Nevertheless, we cannot exclude the possibility that this finding arose by chance.

This study has potential limitations to take into account in the interpretation of the results. Firstly, the study size, and especially the sex-stratified subsamples, was relatively small, although it was sufficient to detect trends that warrant confirmation in larger cohorts. Secondly, although our hospital-based design may limit generalizability; there is no clear that our participants’ characteristics differ substantially from those of the broader population. Of note, PBDEs and DP levels are unlikely to be influenced by our heterogeneous selection criteria (including conditions such as gallbladder disease, varicose veins, or hernias). The large number of analyses increases the risk of false positives a cause to multiple comparisons. Nevertheless, this study is exploratory and aims to identify population subgroups with higher PBDE/DP exposure; therefore, our findings will require confirmation in future studies.

This study has important strengths. As previously mentioned, the use of adipose tissue for exposure assessment provides a highly accurate measure of long-term PBDE accumulation. This matrix also reduces intra-sample variability compared to other biological specimens such serum, thereby minimizing random error and lowering the sample size needed to detect associations [34]. Furthermore, adipose tissue might be particularly relevant in hospital-based studies conducted in countries with universal public healthcare systems such as Spain, where routine clinical or surgical procedures provide opportunities to collect adipose tissue samples from large and heterogeneous population samples. In addition, this study was performed within the well-characterized GraMo cohort, which provides detailed sociodemographic, lifestyle, metabolic, and exposure information for participants. Notably, our study is among the largest to utilize adipose tissue for PBDE/DP exposure assessment in adults. Importantly, our chemical analyses covered a wide range of PBDE congeners and both stereoisomers of DP, with concentrations quantified using validated, and high-quality analytical methods that reduce exposure misclassification.

4. Conclusions

Our study revealed that exposure to PBDEs and DP in the GraMo cohort, as estimated by analysis of adipose tissue samples, was ubiquitous in the study population. Sex, BMI, occupational social class, self-reported exposure to paints and solvents, and dietary habits (particularly fatty food from animal origin) were found as the main factors of the exposure. These results might help further targeted public-health interventions to reduce exposure.