Occurrence and Health Risk Assessment of Organophosphate Esters in Seafood from Southeastern China: Impacts of Cooking Methods and Dietary Exposure

Abstract

Food ingestion is recognized as a major pathway for human exposure to organophosphate esters (OPEs). However, studies investigating dietary external exposure and human internal exposure within the same region remain limited. In this study, 17 OPEs were measured in seafood collected from China’s coastal waters and 16 OPEs were measured in serum samples from local residents. ΣOPE concentrations ranged from 1.88 to 9.23 ng/mL (mean: 4.96 ng/mL) in serum and 15.63–39.61 ng/g dw (mean: 24.54 ng/g dw) in untreated seafood. The emerging OPE 3IPPDPP was detected in untreated seafood at a concentration of 0.07 ng/g dw. Among the processing methods, cooking exhibited the most evident reduction in OPE concentrations, with a mean level of 14.57 ng/g dw. TCIPP and TDCIPP were identified as the dominant compounds in the seafood and serum, respectively. Under the current available toxicological data for OPEs and the assumptions of the exposure model, noncarcinogenic risks from dietary exposure to OPEs in processed seafood were within acceptable limits. These findings provide new insights into dietary exposure to OPEs and their associated health risks.

1. Introduction

Organophosphate esters (OPEs) are characterized by a central phosphate group bonded to alkyl or aromatic side chains. , In recent years, OPEs have been increasingly used as alternatives to polybrominated diphenyl ethers (PBDEs) in a wide range of consumer products, including furniture, plastics, building materials, textiles, electronics, and automotive parts. − However, growing evidence indicates that OPEs exhibit persistence, ecological risks, and toxicity, suggesting that they are not entirely safe or suitable replacements for PBDEs. − Moreover, these compounds have been widely detected in various environmental matrices such as air, water, sediment, soil, dust, and even human blood. Marine contamination by OPEs is mainly derived from terrestrial sources, entering oceanic environments through river discharge and atmospheric deposition. As a result, OPEs enter and accumulate in aquatic organisms, and are subsequently transferred through food webs. ,,

Food ingestion is considered the primary pathway for human exposure to OPEs. However, most existing studies have focused either on OPE pollution levels in environmental media or on assessing exposure levels in humans. − Few studies have simultaneously quantified OPE concentrations in both seafood and human biological samples from the same geographic region. Such parallel measurements are important for improving understanding of the contribution of dietary intake to internal exposure. This gap introduces significant uncertainty into the assessment of human dietary exposure risks associated with OPEs.

Most seafood is processed before consumption, such as by cooking or salting. Studies have demonstrated that processing methods can alter the concentrations of organic contaminants in seafood. , For example, OPE levels frequently change during processing, partly due to volatilization or thermal degradation. Highly volatile OPEs (e.g., triethyl phosphate, TEP) may evaporate readily upon heating, while thermal treatment can also promote their transformation into corresponding diesters, with these behaviors likely governed by their physicochemical properties such as volatility and hydrophobicity. , In salting processes, media such as salt and alcohol modify the solubility and adsorption properties of OPEs, thereby influencing their distribution within seafood. Given the dietary exposure and potential toxicity of OPEs to humans, understanding their migration and transformation during food processing is essential for accurately assessing associated health risks.

In this study, we investigated the occurrence of OPEs in both untreated and processed seafood, as well as local human serum samples, and characterized the dominant OPEs in these matrices. To estimate dietary exposure and potential health risks associated with seafood consumption, including processed seafood, we applied concentration data from seafood combined with assumed consumption patterns. This research enhances the understanding of potential health risks associated with OPE exposure through seafood consumption

2. Materials and Methods

2.1. Chemicals and Reagents

Seventeen OPEs were analyzed in seafood samples, including 11 traditional OPEs TEP, tri-n-butyl phosphate (TNBP), tripropyl phosphate (TPrP), trimethylphenyl phosphate (TMPP), triphenyl phosphate (TPHP), 2-Ethylhexyl diphenyl phosphate (EHDPP), 2,2-bis­(chloromethyl)-propane-1,3-diyltetrakis­(2 chloroethyl) bisphosphate (V6), tri­(2-butoxyethyl) phosphate (TBOEP), Tri­(1,3-dichloro-2-propyl) phosphate (TDCIPP), tri­(2-chloroethyl) phosphate (TCEP), Tri­(1-chloro-2-propyl) phosphate (TCIPP), and 6 emerging OPEs (3-isopropylphenyl diphenyl phosphate (3IPPDPP), 3-tert-Butylphenyl diphenyl phosphate (3tBPDPP), cresyl diphenyl phosphate (CDPP), bis­(3-tert-butylphenyl) phenyl phosphate (B3tBPPP), bis­(3-isopropylphenyl) phenyl phosphate (B3IPPPP), bisphenol A bis (diphenyl phosphate) (BDP)). In serum samples, Tris­(2-ethylhexyl) phosphate (TEHP) and Resorcinol bis (diphenyl phosphate) (RDP) were also quantified in the analysis. In contrast, TEP and V6 were not detected, and TNBP was omitted from further analysis due to background interference identified in the procedural blanks. In total, 16 OPEs were analyzed. Seven (M₆TBEP, TEP-d ₁₅, TDCIPP-d ₁₅, TCEP-d ₁₂, TPRP-d ₂₁, TNBP-d ₂₇ and TPHP-d ₁₅) isotope-labeled standards were obtained from Wellington Laboratories (Guelph, ON, Canada). Organic reagents, including acetone (pesticide grade), HPLC grade acetonitrile (ACN) and methanol (pesticide grade) were purchased from Fisher Scientific (Waltham, USA). Dichloromethane (DCM) and n-hexane (pesticide grade) were obtained from J. T. Baker (Phillipsburg, NJ, USA). Detailed information on chemicals and materials is listed in Tables S1–S2 and Text S1 of the Supporting Information (SI).

2.2. Sample Collection

All seafood samples were purchased from markets in 2024, which are located along the southeastern coast of Zhejiang Province in areas predominantly characterized by autoparts enterprises. Food categories were defined according to the Chinese standard. Based on this guideline, all seafood samples were classified into four groups: untreated food, cooked fishery food, dried fishery products, and salted fishery food. To ensure representativeness of local dietary habits, a field survey was conducted to identify frequently consumed seafood through consultation with local residents and vendors. Untreated seafood, dried fishery products, and salted fishery food were purchased from local markets. Samples within each category were randomly collected and processed using consistent laboratory procedures. It should be noted that comparisons among dried fishery products and salted fishery food are observational in nature, as these samples were not specimen-matched with untreated seafood. The number and composition of samples for each processing method were as follows: untreated food (n = 22: fish (n = 5), crustacean (n = 3), mollusk (n = 11), algae (n = 3)), cooked fishery food (n = 22: fish (n = 5), crustacean (n = 3), mollusk (n = 11), algae (n = 3)), dried fishery products (n = 10: fish (n = 4), crustacean (n = 1), mollusk (n = 3), algae (n = 2)), salted fishery food (n = 5: fish (n = 1), crustacean (n = 1), mollusk (n = 3)). After being rinsed with distilled water, the samples were wrapped in aluminum foil, sealed in clean poly­(tetrafluoroethylene) bags, and immediately transported frozen to the laboratory.

The edible portions were separated, and the 22 cooked fishery food were obtained by steaming the corresponding 22 untreated seafood on a one-to-one matched basis. The samples were then freeze-dried, ground to a fine powder, wrapped in aluminum foil, and stored at −20 °C until analysis. Detailed information can be found in SI (Table S3).

Venous blood samples (n = 206) were collected from hospitals in Yuhuan City, Zhejiang, China, during March 2022-July 2023. The study was approved by the Ethics Committees of the People’s Hospital of Yuhuan (Approval No: Yuyilunshen-2022–055), and all of the participants provided written informed consent before enrollment. To explore age-related differences in OPE exposure profiles, participants were categorized into three age groups: young adults (18–44 years), midlife adults (45–59 years), and older people (≥60 years), according to their chronological age. , This classification was applied for subsequent multivariate analyses, including principal component analysis (PCA). Following standardized biospecimen collection procedures, venous blood was drawn into sterile, single-use vacuum tubes. Samples were promptly subjected to refrigerated centrifugation at 3000 g for 10 min at 4 °C to separate the serum fraction. After isolation, serum samples were transported to the analytical laboratory for further processing and stored at −80 °C until pretreatment. Due to the design of clinical sampling, individual-level dietary surveys were not conducted for specific participants. All participants were long-term residents of the coastal area. Given the temporal stability of OPE concentrations in marine environments, , seafood samples collected within a two-year window of serum sampling were used as a regional proxy for comparison of congener profiles. This study compared seafood and serum at the compound level rather than through paired individual analysis, aiming to contrast overall distribution patterns and congener profiles between matrices to explore potential exposure characteristics in the study region.

2.3. Analytical Procedures

100 mg of the freeze-dried seafood samples were weighed in a 15 mL polypropylene (PP) tube with 400 mg of NaCl and 400 mg of NaSO₄ and spiked with 10 ng internal standard (IS) mixture (TEP-d ₁₅, TPHP-d ₁₅, TDCIPP-d ₁₅, TCEP-d ₁₂, TNBP-d ₂₇, TPRP-d ₂₁, and M₆TBEP). Subsequently, samples were added with 10 mL of ACN, vortexed 1 min, ultrasonicated in an ultrasonic-cleaner for 1 h, shaked on the oscillator at 300 rpm for 15 min and centrifuged at 5000 rpm for 10 min. The extraction was repeated 3 times, concentrated to approximately 1 mL via rotary evaporation and purified through a glass column containing 5 g of anhydrous granulated sodium sulfate, 1 g PSA, 5 g of neutral silica gel, 5 g of florisil, 2 g of silica and 0.2 g of anhydrous granulated sodium sulfate. The column was activated sequentially with 40 mL acetone/DCM mixture (1:1, v/v), 40 mL n-hexane and eluted by using 80 mL of an acetone/DCM mixture (1:1, v/v). The eluate was concentrated by rotary evaporation, evaporated to near dryness under a gentle nitrogen stream, and reconstituted in 200 μL of methanol prior to instrumental analysis.

Prior to serum sample preparation, 0.8 mL of serum was spiked into a mixed internal standard solution containing TNBP-d ₂₇, TPHP-d ₁₅, TEP-d ₁₅, TPRP-d ₂₁, TCEP-d ₁₂, TDCIPP-d ₁₅, and M₆TBEP at a concentration of 20 ng/mL. The spiked serum was then subjected to clean up using a glass Oasis HLB cartridge coupled with a Florisil-silica composite cartridge, following procedures described in detail in our previous work.

Instrumental analysis of OPEs was performed with a high-performance liquid chromatography triple quadrupole mass spectrometer (HPLC-MS/MS, LCMS-8050; SHIMADZU, Japan). Additional information on analytical conditions is provided in SI (Table S4).

2.4. Quality Control

All glassware and laboratory tools were sequentially rinsed with acetone, DCM, and n-hexane prior to use and between sample preparations to minimize cross-contamination. To assess potential contamination during sample preparation and analysis, one procedural blank was included for every batch of 12 samples. Analyte concentrations in the method blanks were subtracted from those of the corresponding samples. The method detection limits (MDLs) were determined as the concentrations corresponding to a signal-to-noise ratio of 3. When analytes were not detected in procedural blanks, MDLs were estimated as three times the standard deviation (3SD) of the response from the lowest calibration level. The concentrations below the MDL were substituted with the MDL/2 for statistical analysis in PCA and correlation analyses. For concentration calculation and risk assessment, values > MQL were measured values, values < MQL were replaced by MDL/2, and not detected were treated as zero in the descriptive concentrations and EDI/HQ/CR calculations. Spike recovery experiments showed absolute recoveries of the target OPEs ranging from 65% to 115% in seafood samples and from 54% to 113% in serum samples. The quantification was based on internal standard method using eight concentration levels (0.1, 0.5, 2, 5, 10, 20, 50, and 100 ppb), with all calibration curves exhibiting coefficients of determination (R ²) above 0.999. Detailed data on method recoveries and MDL are provided in Tables S5,S6.

2.5. Dietary Exposure Estimation

Based on the concentrations of OPEs detected in marine products, the estimated daily intake (EDI) values and hazard quotient (HQ) for each compound were derived for the Zhejiang adult population under the conservative assumption of complete gastrointestinal absorption (100%), which may lead to a potential overestimation of exposure (Table S7).

$$\mathrm{EDI}=\frac{\mathrm{PR}\times \mathrm{C}}{\mathrm{BW}}$$ 2-1

$$\mathrm{HQ}=\frac{\mathrm{EDI}}{\mathrm{RfD}}$$ 2-2

where EDI is the estimated daily intake of OPE (ng/kg bw/day), PR is the per capita annual consumption of aquatic products in Zhejiang Province (81.10 g/person/day for residents in Zhejiang) (National Burean of Statistics of China, 2024), C is the 95th percentile concentration of each OPEs (ng/g ww) detected in seafood samples (Table S7), BW is the average body weight of residents (kg), 60 kg was used for adults. The 95th percentile concentration used for EDI estimation was calculated separately for untreated food, cooked fishery food, dried fishery products and salted fishery food to represent method specific high-exposure scenarios. In addition, the total EDI was calculated using the 95th percentile concentration derived from all seafood samples pooled together. RfD is the reference dose for OPEs (ng/kg bw/day, Table S7). A HQ value larger than 1 indicates a significant probability of noncarcinogenic risk (NCR) upon OPE exposure.

The carcinogenic risk (CR) of OPEs is calculated as follows

$$\mathrm{CR}=\mathrm{EDI}\times \mathrm{SFO}$$ 2-3

where SFO is the oral cancer slope factor (ng/kg bw/day) ^–1, (Table S7). A CR value higher than 1 × 10^–6 indicates a high probability of a person developing cancer from the lifetime exposure.

2.6. Statistical Analyses

SPSS 27.0 software (IBM, Armonk, NY), Python (version 3.6.6), R (version 4.5.1) and origin 2025 software (OriginLab Corporation, MA) were used for statistical analysis and graphics production in our study. Descriptive statistical analysis revealed that the data set in this study deviated from a normal distribution. Kruskal–Wallis test was applied to test the difference between concentrations in four groups (untreated food,cooked fishery food, dried fishery products and salted fishery food) of different processing method seafood samples. The results were considered to be statistically significant at p < 0.05. Additionally, Spearman rank correlation coefficients were calculated to assess the within-matrix correlations among OPE concentrations in seafood products, aiming to explore potential co-occurrence patterns. Partial correlation analysis was performed in R to examine the within-matrix relationships among OPEs in human serum. While the field samples in this study were specifically sourced from markets along the southeastern coast of Zhejiang, a broader assessment of potential risks across China’s coastal regions was achieved by integrating multicity literature data through Monte Carlo simulation (MCS). The simulation was implemented using Oracle Crystal Ball software (version 11.1.30), with 10,000 times performed per run to characterize the health risks associated with seafood consumption across six distinct geographical regions in China.

Statistical analyses were performed using Python (version 3.6.6), with the Scikit-learn package (version 0.24.2) employed for the Random Forest (RF) component. This method aimed to investigate the within-matrix feature-importance of each OPE monomer in seafood and serum samples to their own total OPE content (ΣOPEs). Considering the multidimensional nature of OPEs and their potential nonlinearity relationship, we employed RF model to evaluate the within-matrix feature-importance of each feature. Detailed information is provided in Text S2 of the SI.

3. Results and Discussion

3.1. Occurrence of OPEs in Seafood and Serum

The concentrations of OPEs in seafood are presented in Table . Among the 17 target OPEs, 13 were detected in the untreated seafood samples, excluding TCEP, TPRP, B3IPPPP, and V6. The detection frequencies (DF) ranged from 4.5% (B3tBPPP, CDPP) to 100% (TNBP, TCIPP). The total OPE (ΣOPEs) concentrations in the untreated seafood samples ranged from 15.63 to 39.61 ng/g dry weight (dw), with a mean value of 24.54 ng/g dw. Across the four seafood categories, ΣOPE levels followed the order: fish (range: 23.35–31.41 ng/g dw, mean: 27.20 ng/g dw) > mollusks (range: 15.63–39.61 ng/g dw, mean: 24.30 ng/g dw) > crustaceans (range: 23.00–23.67 ng/g dw, mean: 23.35 ng/g dw) > algae (range: 17.49–27.83 ng/g dw, mean: 22.14 ng/g dw). The predominant compounds, each with a DF > 60%, were TCIPP (mean: 9.85 ng/g dw), TNBP (7.83 ng/g dw), TPHP (1.07 ng/g dw), and TBOEP (0.79 ng/g dw), contributing 40.6%, 32.3%, 4.4%, and 3.3% to the total OPEs, respectively. Overall, the ΣOPE levels found in this study were lower than those reported in fish samples from Laizhou Bay, Shandong Province, China (23.3–107.1 ng/g dw), and from Weizhou Island in the South China Sea (58.2 ng/g dw).

1. Detection Frequencies (DF) and the Mean Concentrations (ng/g dw) of OPEs in Untreated Food (n = 22).

	mean concentrations (ng/g dw)
compounds	fish (n = 5)	crustacean (n = 3)	mollusk (n = 11)	algae (n = 3)	total DF (%)
TEP	2.79	1.88	3.34	1.32	40.9
TNBP	7.37	9.41	7.79	6.77	100
TBOEP	1.14	0.58	0.68	0.76	59.1
TPrP	ND	ND	ND	ND	0.0
TPHP	1.50	0.47	1.38	0.93	68.2
3tBPDPP	ND	ND	0.08	0.11	18.2
CDPP	ND	ND	0.06	ND	4.5
TMPP	0.07	ND	0.12	0.06	27.3
B3tBPPP	ND	ND	0.01	ND	4.5
B3IPPPP	ND	ND	ND	ND	0.0
BDP	0.39	ND	0.57	0.28	13.6
3IPPDPP	0.06	0.04	0.09	0.06	54.5
EHDPP	0.31	ND	0.33	0.45	36.4
TCIPP	10.77	10.98	8.64	9.02	100
TCEP	ND	ND	ND	ND	0.0
TDCIPP	2.81	ND	1.22	2.37	40.9
V6	ND	ND	ND	ND	0.0

^a: not detected.

Of other 16 targeted OPEs analyzed, 12 were detected in the serum samples (Table ). The mean serum concentrations of ΣOPEs were 5.44 ng/mL in older adults, 5.10 ng/mL in midlife adults, and 4.72 ng/mL in young adults, respectively. The ΣOPEs ranged from 1.88 to 9.23 ng/mL, with a mean value of 4.96 ng/mL. The predominant compounds, each with a DF greater than 60%, were TDCIPP (mean: 1.79 ng/mL), TCIPP (1.19 ng/mL), TBOEP (0.78 ng/mL), TEHP (0.45 ng/mL), EHDPP (0.29 ng/mL), TPHP (0.29 ng/mL), and TMPP (0.03 ng/mL). These compounds accounted for 36.0%, 24.0%, 15.7%, 9.0%, 5.9%, 5.9%, and 0.6% of the ΣOPEs, respectively. The concentration levels measured in this study were higher than those reported for the general population in Hengshui, Hebei, China (0.94–4.37 ng/mL), but lower than those found in pregnant women from Kunshan, Jiangsu, China (1.71–90.1 ng/mL)

2. DF and the Mean Concentrations (ng/mL) of OPEs in Serum Samples.

	mean concentrations (ng/mL)
compounds	young adults (n = 96)	midlife adults (n = 87)	older people (n = 23)	total DF (%)
TPrP	ND	ND	ND	0.0
TEHP	0.44	0.46	0.42	98.1
TBOEP	0.82	0.68	0.95	95.6
TCIPP	1.14	1.20	1.36	100.0
TCEP	0.04	0.06	0.03	25.2
TDCIPP	1.67	1.84	2.06	94.7
TPHP	0.23	0.36	0.32	85.4
TMPP	0.03	0.03	0.03	68.0
CDPP	0.005	ND	ND	0.5
EHDPP	0.28	0.31	0.29	98.5
RDP	ND	ND	ND	0.0
BDP	0.02	0.003	ND	1.5
3IPPDPP	0.01	0.01	0.002	18.4
B3IPPPP	ND	ND	ND	0.0
3tBPDPP	ND	ND	ND	0.0
B3tBPPP	0.05	0.16	ND	9.7

^a: not detected.

3.2. Relationship Between OPEs in Seafood and Serum

The composition of OPEs in seafood and serum was shown in Figure . Spearman correlation analysis was used to evaluate the relationship between OPE concentrations in seafood. Partial correlation was conducted to assess the relationships among OPE concentrations in human serum after adjusting for smoking, alcohol consumption, sex, BMI, and age. OPEs were selected primarily based on a detection frequency >60% in both seafood and serum matrices; additionally, 3IPPDPP was included as it was identified for the first time in seafood from Zhejiang Province. TDCIPP was retained in the seafood analysis due to its comparatively high concentrations in human serum and its detectable occurrence in seafood samples. Consequently, TCIPP, TPHP, TNBP, TDCIPP, TBOEP and 3IPPDPP were selected for untreated seafood (n = 22), while TCIPP, TPHP, TDCIPP, TBOEP, EHDPP, TMPP, and TEHP were analyzed for serum (n = 206). The data were aligned at the congener level.

1.

Spearman rank correlation, partial correlation, and principal component analysis (PCA) of OPEs in seafood and human serum. (a) Spearman rank correlation heatmap for OPEs in seafood. (b) Partial correlation analysis for serum OPEs adjusted for smoking, alcohol consumption, sex, BMI, and age. (c, d) PCA of OPEs in seafood and serum.

In seafood, chlorinated OPEs (Cl-OPEs, e.g., TCIPP, TDCIPP) showed significant positive correlations with alkyl-OPEs (e.g., TNBP, TBOEP) and aryl-OPEs (e.g., 3IPPDPP) (r = 0.51–0.74) (Figure a), reflecting a consistent co-occurrence pattern of these congeners within marine organisms. In contrast, the correlation of OPEs in human serum was generally low (r = −0.25–0.51) (Figure b). These distinct internal correlation structures within each matrix suggest different accumulation profiles; however, it should be noted that similar or differing internal correlations do not establish a direct dietary contribution from seafood to human serum. The lower correlations in serum likely reflect the complexity of multipathway exposure (e.g., dietary intake, dermal contact, and inhalation) and individual metabolic differences. , The lower correlation in human serum may be also due to metabolic differences of different types of OPEs in the human body. Cl-OPEs with longer half-lives and wider availability (such as TCIPP and TDCIPP) may accumulate in the body over a long period, while shorter-half-lives OPEs (such as TBOEP and TNBP) are more affected by recent intake and metabolic rate. TNBP was reported to be the fastest metabolized among common OPE congeners. ,

To further analyze the distribution of OPEs in seafood and human serum, PCA was performed (Figure c, d). In seafood, two dimensions (PC1 and PC2) represented the two main factors that explained the most variance, with cumulative contributions of 65.9% (45.9% and 20.0%) to the distribution of OPEs. PC1 was characterized by high loadings of TDCIPP and TBOEP, which showed a significant positive correlation (p < 0.05) with each other, while PC2 was primarily driven by TCIPP and TPHP. In human serum, subjects were grouped by age (young adults, midlife adults and older people) to explore age-related differences in OPE profiles. PC1, dominated by TPHP and EHDPP, accounted for 30.6% of the variance, and these two OPEs showed a significant positive correlation (p < 0.05). PC2 was mainly driven by TCIPP and TDCIPP and explained 24.4% of the total variance. The observed age-related differences in serum OPE profiles suggest potential variability in internal exposure across life stages. Such variability may reflect differences in dietary habits, metabolic capacity, and cumulative exposure history. Although TDCIPP and TCIPP exhibited relatively high loadings in the PCA results for both seafood and serum, this observation is strictly descriptive of the variance structure within each independent data set. It does not imply a direct exposure linkage or source-to-target contribution between the two matrices.

3.3. Effects of Processing Techniques on OPEs in Seafood

Previous research has primarily focused on OPE contamination in unprocessed seafood, often overlooking the complete dietary pathway through which OPEs may enter the human body via consumption of processed seafood products. In this study, we further investigated the influence of common seafood processing methods on OPE levels to better understand how processing may alter dietary exposure (Table S8). As shown in Figure a, lower ΣOPE concentrations were observed in three processed seafood products. Kruskal–Wallis test found no statistically significant differences in OPE concentrations between untreated seafood and the three processed groups. The ΣOPE levels in cooked fishery food, dried fishery products and salted fishery food were 8.86 to 22.99 ng/g dw (mean: 14.57 ng/g dw), 12.86 to 22.66 ng/g dw (mean: 17.08 ng/g dw) and 8.93 to 27.31 ng/g dw (mean: 17.34 ng/g dw), respectively. These values were considerably lower than those in untreated seafood (range: 15.63 to 39.61 ng/g dw; mean: 24.54 ng/g dw). The effect of processing on ΣOPE levels varied across different seafood categories. As illustrated in Figure b, cooking consistently reduced ΣOPE concentrations in all types of seafood, which can be attributed to the leaching of OPEs caused by washing during the cooking process and the degradation of OPEs induced by high temperatures. , Notably, crustaceans exhibited a significant decrease in ΣOPE levels under each processing method. In contrast, salting had a relatively minor impact on ΣOPE concentrations in mollusks. The effects of processing on individual OPEs were also evaluated. Compared with untreated seafood, the detection rate of TNBP showed no significant changes after processing (Tables , S8). In contrast, 3IPPDPP, which had a high detection rate of 54.5% in untreated seafood, decreased to only 4.5% after cooking and was not detected at all in dried or salted products, demonstrating excellent removal efficiency.

2.

Occurrence and processing effects of OPEs in fishery products. (a) Boxplots of OPE concentrations (ng/g, dw) in untreated food (UF), cooked fishery food (CFF), dried fishery products (DFP), and salted fishery food (SFP). (b) Distribution of OPEs among different categories of seafood (fish, crustaceans, mollusks, and algae). (c) F _P /F _U of OPEs in different processing methods. a, b, c: The ratio of individual compound concentrations (wet weight) in cooked fishery food, dried fishery products, salted fishery food to that in untreated food, respectively.

In this study, F _P /F _U ratio was defined as the concentration of a compound in processed seafood (F _P ) divided by its concentration in the untreated counterpart (F _U ). Higher values indicate greater concentration effect during processing. All concentrations are converted to wet weight concentrations for F _P /F _U calculation. Cooking generally reduced OPE concentrations, with most F _P /F _U ratios falling below 1. This reduction may be attributed to the leaching of OPEs into cooking water and their volatilization during heating. In addition, changes in lipid structure and partial lipid loss may influence the retention of lipophilic OPEs. , In contrast, F _P /F _U > 1 in dried fishery products does not necessarily imply an increase in the absolute mass of OPEs. Instead, it is primarily driven by the substantial loss of moisture (dehydration) during prolonged processing, which effectively concentrates the remaining stable OPEs within a smaller sample mass. For instance, the emerging 3tBPDPP exhibited substantially higher F _P /F _U values (8.31 and 9.46) compared to TCIPP (1.44 and 3.76) during salting and drying, indicating stronger concentration effect under these conditions. CDPP showed an exceptionally high F _P /F _U value during drying (29.13), indicating a strong concentration effect that merits further investigation. TCIPP and TNBP, both dominant compounds in seafood, also had F _P /F _U values above 1 during drying and salting, underscoring their relative persistence. Furthermore, across all three processing methods, EHDPP, TMPP, and TPHP displayed F _P /F _U values close to or exceeding those of TBOEP, suggesting that more hydrophobic compounds tend to accumulate more readily, a finding consistent with prior observations by Wang et al. The variability in lipid content among seafood species and processing categories may influence the distribution of hydrophobic OPEs, although this mechanism remains speculative. In this study, the concentration effect (F _P /F _U > 1) was most pronounced during drying, followed by salting and cooking (Figure c). Notably, F _P /F _U > 1 does not necessarily imply an increase in the absolute mass of OPEs, rather can be attributed to a greater relative loss of moisture compared to the mass loss of OPE congeners via leaching, volatilization, or thermal degradation. Drying involves extensive dehydration and prolonged processing times, leading to stronger concentration effects. This phenomenon is likely driven by substantial water removal and the subsequent structural reorganization of proteins and lipids, which potentially creates a more hydrophobic environment that potentially traps and concentrates stable OPEs within the remaining lipid-rich matrix, and this hypothesis requires confirmation in future studies. ,

3.4. Feature Importance Analysis

To identify the key chemical drivers of total OPE burdens and compare the prominence of specific congeners across matrices, RF models were independently applied to the seafood and human serum data sets. During 10-fold cross-validation, the RF model performed well for seafood (R ² = 0.806 ± 0.220, mean absolute error (MAE) = 0.042 ± 0.040, root-mean-square error (RMSE) = 0.056 ± 0.063) and human serum (R ² = 0.809 ± 0.205, MAE = 0.378 ± 0.113, RMSE = 0.572 ± 0.340). When evaluated on the test set, the model achieved high predictive accuracy, with R ² values of 0.910 (MAE = 0.044, RMSE = 0.059) for seafood and 0.882 (MAE = 0.316, RMSE = 0.429) for human serum. Future studies with larger, more diverse seafood cohorts are needed to further validate the robustness and generalizability of these findings.

SHapley Additive exPlanations (SHAP) analysis (Figure a, b) revealed that TDCIPP had the highest mean absolute impact in serum, whereas TCIPP was the most influential in seafood, underscoring their dominant roles in model predictions within each data set. Other OPE monomers, such as TPHP, EHDPP, and 3IPPDPP, exhibited considerably lower average impacts. Individual SHAP value plots (Figure c, d) further illustrate how specific feature values affect model output. For TDCIPP in serum and TCIPP in seafood, both positive and negative effects were observed; generally, higher concentrations corresponded to positive SHAP values, while lower concentrations were associated with negative effects. In contrast, other OPE monomers in both sample types, such as TPHP, EHDPP, and 3IPPDPP, displayed less variability in SHAP values, reflecting a more stable, though relatively limited, impact on model predictions.

3.

SHAP-based feature importance analysis of OPE monomers in seafood and human serum. (a, b) The bar charts show the average size of the SHAP values and ranks features by their overall impact of seafood and serum samples. (c, d) The bee colony plots provide a detailed view of the SHAP value distribution for each feature, indicating the degree of influence of high or low values on the model output.

Overall, the profiling analysis revealed that TCIPP was the most influential compound in seafood, while TDCIPP contributed the most in serum. Among the major contributors, Cl-OPEs accounted for a high proportion in both seafood and human serum, representing a common occurrence pattern consistent with earlier reports. In aquatic environments, Cl-OPEs are frequently the dominant OPE group, accounting for the highest concentrations in 73% of reported cases. These results indicate that Cl-OPEs (particularly TDCIPP and TCIPP) are among the major contributors to the total OPE burden in both serum and seafood in terms of relative abundance. The observed similarity in congener importance across both matrices suggests a comparable compositional pattern of OPE exposure profiles.

3.5. Health Risk Assessment

Under the high exposure scenarios (95th percentile), the total EDI value of OPEs for Zhejiang residents was 22.9 ng/kg bw/day. Dried fishery products (27.9 ng/kg bw/day) exhibited the highest dietary intake levelsapproximately twice that of untreated food (13.4 ng/kg bw/day) and salted fishery food (16.3 ng/kg bw/day) and over three times that of cooked fishery food (8.6 ng/kg bw/day) (Figure a). The total EDI of OPEs through seafood in this study was comparable to that reported for male youth in the Beibu Gulf of China (23.75 ng/kg bw/day) and urban residents along the South China Sea (26.7 ng/kg bw/day). Among all OPEs in this study, TCIPP exhibited the highest exposure dose via seafood consumption, accounting for 38% of the total (Figure b), which is consistent with findings from Zhuhai Bay, where TCIPP contributed 59% of the total EDI.

4.

Risk assessment of OPEs in seafood. (a, c) Corresponding bar charts of EDI (a) and HQ (c) for four seafood product categories (UF, CFF, DFP, SFP). (b, d) Polar plot showing the estimated daily intake (EDI, b) and hazard quotient (HQ, d) for individual OPEs. (e) Cumulative probability distribution of THQs of OPEs among the Taihu Lake (TH), the Pearl River (ZJK), the Bohai Sea (BH), the Beibu Gulf mariculture farms (BBWY), Lianzhou Bay (BBWL) and this study (YH).

The NCR values of untreated food, cooked fishery food, dried fishery products, and salted fishery food were calculated as 1.1 × 10^–3, 8.4 × 10^–4, 2.8 × 10^–3, and 1.3 × 10^–3, respectively. All values were below the threshold of 1, indicating a low level of NCR. These findings are consistent with previous studies conducted in Laizhou Bay, Belgium, and Baiyangdian Lake, which also reported minimal health risks associated with OPEs in seafood. Notably, the NCR value for dried fishery products was approximately twice as high as those for untreated food and salted fishery food, and 3.5 times that of cooked fishery food alone (Figure c), suggesting that dried fishery products poses the highest potential NCR among the product types examined and should be given particular attention. The dehydration process leads to the concentration of OPEs in dried fishery products compared with untreated seafood. This explains why dried fishery products exhibited the highest OPE levels and HQ values in this study. The CR was estimated based on four OPEs for which SFO were available. The partial CR associated with the four seafood categories was below 1 × 10^–6, indicating a low potential CR under the current assessment framework. Due to the absence of SFO values for some OPEs, the total CR might be underestimated. Compounds such as TDCIPP, TCIPP, and TCEP have been confirmed to be carcinogenic, suggesting that ongoing monitoring and assessment are necessary to manage potential exposure risks effectively. Moreover, TCIPP and TNBP have been classified as typical high-risk OPEs due to their neurotoxicity, reproductive toxicity and other adverse effect.

The point evaluation model assumes that the concentrations of chemicals in food remain constant, whereas residents’ dietary exposure to OPEs may vary from day to day. To enhance the accuracy and reliability of health risk assessments, the MCS was performed to optimize the evaluation process by effectively addressing uncertainties associated with the model. Therefore, we collected data on the concentrations of OPEs in aquatic products from five representative regions in China, including Taihu Lake (TH), , the Pearl River (ZJK), the Bohai Sea (BH), the Beibu Gulf mariculture farms (BBWY), and Lianzhou Bay (BBWL). The NCR associated with OPE exposure through seafood consumption was evaluated using MCS.

Based on the 95th percentile concentrations of OPEs in aquatic products, the THQs for the six regions ranged from 8.37 × 10^–4 to 7.94 × 10^–3, and were all far below 1, suggesting that the NCR of OPE exposure through seafood consumption was low under current assessment assumptions (Figure e). Among them, the Bohai Sea region, including Bohai Bay, Laizhou Bay, and Liaodong Bay, with a THQ of 7.94 × 10^–3, exhibited the highest NCR. The Bohai Sea is a semienclosed Chinese water body, and previous studies revealed that nearly 40 rivers discharge large amounts of domestic and industrial wastewater into it, resulting in elevated OPE concentrations (9.6–1549 ng/L) in the marine environment. Widely used in building materials, electronics, and automobile and motorcycle parts, OPEs are typically additives rather than chemically bound, allowing release via abrasion, volatilization, and leaching. The long-standing dominance of the automobile and motorcycle parts industry in our sampling regions suggests that OPEs enter the marine environment via riverine discharge and atmospheric deposition. , Once in the ocean, these chemicals can disperse, be absorbed by marine organisms, and biomagnify along the food chain, potentially resulting in toxicological effects on human health.

3.6. Implications, Strengths and Limitations

This study provides robust evidence for the widespread occurrence of OPEs in seafood from China’s coastal waters. Among the 13 OPEs detected, TCIPP and TNBP showed the highest detection frequencies and contamination levels, indicating their dominant contribution to the total OPE burden. TCIPP and TDCIPP were identified as the major contributors in seafood and serum, respectively. By analyzing OPEs in seafood and human serum from the same coastal region, this study offers an integrated perspective on environmental occurrence and internal human exposure.

Building upon the occurrence of OPEs in seafood and their potential contribution to dietary exposure, this study further investigated the influence of common seafood processing methods on OPE levels. The impact of processing on total OPE concentrations varied across different seafood categories and cooking consistently reduced ΣOPE levels in all types of seafood. The inclusion of emerging OPEs, such as 3IPPDPP, further expands current knowledge regarding their occurrence in seafood and associated health risks. Based on the estimated dietary intake of OPEs from processed seafood, risk assessment indicated that noncarcinogenic risks to humans remain within acceptable levels. However, several limitations should be acknowledged. This study did not perform a paired dietary and indoor exposure assessment linking individual seafood consumption to serum concentrations, and the sample size was limited to a single coastal region, which may restrict the generalizability of the findings. Although TCEP, TBBPA, and TPP were included in the high-priority substance risk assessment list in 2019, toxicological data and systematic health risk assessments remain largely insufficient for most other OPEs, potentially leading to an underestimation of HQ and CR. Moreover, other exposure pathways were not fully considered, and the role of OPE metabolites formed during processing in human exposure remains unclear. Future studies should incorporate metabolite analysis, toxicokinetic modeling, mixture-effect assessment, and real consumption patterns to better quantify dietary contributions to overall OPE exposure.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/envhealth.6c00114.

• Chemicals and reagents; random forest model description; OPE standards in seafood and serum samples; basic information on food samples; instrumental conditions; method validation for marine samples; method validation for serum samples; risk assessment parameters; and processed seafood results (PDF)

∇.

Y.L. and Q.Z. contributed equally to this work. Y.L.: Conceptualization, investigation, methodology, formal analysis, Writing–original draft. Q.Z.: Methodology, formal analysis. M.L. and M.S.: Conceptualization, methodology, investigation, review and editing. Q.Z.: Resources. M.P.: Review and editing. F.C., X.C., J.L., and G.Z.: Methodology. Y.D. and G.Z.: Investigation.

The authors declare no competing financial interest.