Effect of plastic composition in the combustion material on the Persistent Organic Pollutant content in smoked chicken meat Giang Do Hoang, Uyen Nguyen Thu, Luyen Nguyen Thi, Duong Hoang Thuy, Le Bui Thi Nhat, Nam Vu Duc, Xuyen Nguyen Thi, Thuy Le Minh ABSTRACT Traditional food smoking in Vietnam often uses biomass fuels, but the incidental inclusion of plastics can generate toxic pollutants. This study evaluated the effects of polyethylene (PE), polystyrene (PS), and polyvinyl chloride (PVC) compared with clean wood on persistent organic pollutants (POPs) in smoked chicken, including PCBs, PCDD/Fs, and PAHs. Samples were smoked under controlled conditions, analyzed by GC–MS/MS, and assessed with univariate and multivariate statistics. Clean wood yielded low contamination, while PE introduced lighter PAHs and dioxins, PS was associated with mid-weight PAHs and specific PCBs (66, 195), and PVC caused the most severe contamination with 12 PCDD/Fs, 21 PCBs (e.g., 189, 101, 180, 153), and multiple carcinogenic PAHs (e.g., benzo[a]pyrene). Statistical tests confirmed significant differences among fuel groups (p < 0.01). The contamination gradient (W < PE < PS < PVC) highlights the urgent need to prohibit plastic-contaminated fuels in traditional smoking to safeguard food safety and public health. Introduction The smoking method has been applied for centuries in traditional food processing, serving not only as a means of preservation but also as a way to impart distinctive sensory qualities to food. This process typically involves burning natural biomass such as wood, straw, or agricultural residues, producing smoke rich in antibacterial and antioxidant compounds, as well as flavor-enhancing agents [ 1 ]. In Vietnam, smoking is a long-standing culinary practice used for buffalo, chicken, and pork, particularly in rural kitchens. Traditionally, fuels include paper, wood, or straw; however, in practice, packaging materials or waste wood coated with paint are sometimes used as substitutes due to convenience or cost. Such materials often contain plastics, paints, or adhesives, which release hazardous by-products upon combustion. When fuel sources include plastic components, the combustion process generates a wide spectrum of toxic compounds, notably dioxin-related compounds (DRCs), including polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), and dioxin-like polychlorinated biphenyls (dl-PCBs) [ 2 ]. These compounds are typically formed during incomplete combustion of chlorine-containing substances at elevated temperatures, especially in the presence of synthetic polymers such as polyvinyl chloride (PVC) [ 3 ]. Numerous studies have confirmed that burning plastic-contaminated fuels elevates the levels of PCBs, dioxins, and furans in smoke, which subsequently deposit onto and penetrate food matrices [ 1 , 3 , 4 ]. The presence of these contaminants in smoked foods poses serious risks to food quality and human health. The World Health Organization (WHO) classifies dioxins, furans, and PCBs as Group 1 carcinogens [ 5 ]. They are persistent organic pollutants (POPs), highly stable in the environment and capable of bioaccumulating in fatty tissues [ 6 ]. Chronic dietary exposure has been linked to endocrine disruption, immunotoxicity, reproductive and developmental impairments, as well as elevated risks of cancers, particularly hepatocellular carcinoma and lung cancer [ 7 , 8 ]. Moreover, these compounds adversely affect the nervous system, impairing cognitive development in children and increasing vulnerability during pregnancy [ 9 ]. Long-term exposure has also been associated with metabolic and chronic diseases, including type 2 diabetes, cardiovascular disorders, and autoimmune syndromes [ 9 , 10 ]. Despite these findings, important knowledge gaps remain. Most previous studies have focused on individual pollutant groups or specific fuel types, and systematic comparisons across different plastic materials under controlled conditions are still limited. In particular, there is a lack of studies that simultaneously evaluate multiple classes of persistent organic pollutants (PCBs, PCDD/Fs, and PAHs) and characterize their congener-level fingerprints in relation to fuel composition. Although a related study from our group examined smoked pork under comparable combustion scenarios [ 11 ], the present work is based on an independent smoked chicken experiment with separately generated samples, measurements, and analyses. No data, figures, or tables are duplicated from any previous publication. Therefore, this study aims to: (i) quantify the concentrations of PCBs, PCDD/Fs, and PAHs in smoked chicken using different fuel types (wood, wood + PE, wood + PS, and wood + PVC); (ii) compare congener profiles and identify characteristic marker compounds associated with each polymer type; and (iii) apply multivariate analyses to discriminate contamination patterns and fingerprints related to fuel composition. The findings highlight the critical importance of monitoring and controlling fuel sources in traditional smoking practices. Materials and methods Samples and smoking procedures Nine whole chickens were purchased from local markets, rinsed thoroughly with tap water, and drained. Each chicken was divided into five equal portions (approximately 300 g each). Four portions were assigned to smoking treatments, while the remaining portion served as an unsmoked control and was oven-dried at 90 °C to constant weight (baseline sample). These control samples were subjected to the same analytical procedures as the smoked samples, and target contaminants were not detected (below the method detection limits). Smoking experiments were conducted using a custom-built box-type smoking chamber (1200 × 600 × 1200 mm, L × W × H) equipped with a separate combustion compartment and an adjustable airflow system. The distance between the combustion source and the sample chamber was maintained at approximately 120 cm to prevent direct flame contact and to ensure that the smoke entering the chamber was sufficiently cooled. The chamber temperature was monitored and maintained within the range of 80–90 °C throughout the smoking process. Relative humidity was not controlled and followed ambient laboratory conditions. The fuel materials consisted of (i) pure wood (Group W), (ii) wood mixed with polyethylene (PE), (iii) wood mixed with polyvinyl chloride (PVC), and (iv) wood mixed with polystyrene (PS), each at a ratio of 99:1 (w/w). The wood used in all treatments was Manglietia conifera , which was washed with water to remove surface impurities and air-dried prior to use. To avoid cross-contamination between treatments, each fuel type was combusted in separate smoking sessions. After each session, the chamber was thoroughly cleaned and ventilated before the next experiment. Smoking was performed for five consecutive days, 8 hours per day, under identical operational conditions for all treatments. Portions originating from the same chicken were assigned to different treatments to minimize biological variability. After treatment, all samples (smoked and control) were freeze-dried, ground into fine powder, and stored at −20 °C until chemical analysis. Experimental The samples were processed and analyzed following the same procedure as described in our previous publication [ 11 , 12 ]. Briefly, samples were freeze-dried, homogenized, and fortified with isotopically labelled internal standards prior to analysis. Total lipid content was determined gravimetrically from a portion of the extract. Persistent organic pollutants (PAHs, PCBs, and PCDD/Fs) were extracted using accelerated solvent extraction, followed by solvent exchange, lipid removal, and multi-step clean-up and enrichment procedures tailored to each compound class. Quantitative determination was carried out by GC–MS/MS for PAHs and PCBs and by high-resolution GC–MS for PCDD/Fs using isotope-dilution calibration. Instrumental data were processed using Xcalibur 4.2 (Thermo Scientific, USA), and statistical analyses were performed in R (R Foundation for Statistical Computing, Vienna, Austria). The analytical procedures applied in this study were validated using standard quality assurance and quality control criteria, as described in the Supplemental Materials. Briefly, method performance was evaluated in terms of linearity, sensitivity, and accuracy using multi-level calibration with native and isotopically labelled standards. Method detection limits (MDLs) and method quantification limits (MQLs) were calculated according to US EPA guidelines and reported on a lipid-weight basis. Extraction efficiency and overall method accuracy were assessed through recovery experiments employing 13 C-labelled internal standards spiked prior to extraction. Acceptance of analytical results was based on established EPA recovery and performance criteria. All analytes met EPA acceptance criteria for linearity, sensitivity, and recovery, confirming the robustness of the analytical approach. Detailed validation data, including calibration statistics, MDL/MQL values, and recovery ranges for PAHs, PCBs, and PCDD/Fs, are provided in the Supplemental section. Data processing and statistical methods Instrumental data were processed using Xcalibur version 4.2 (Thermo Scientific, USA), and statistical analyses were performed in R (R Foundation for Statistical Computing, Vienna, Austria). Non-detected (N.D.) values were replaced with one-half of the corresponding method detection limit (MDL/2) prior to calculation of total concentrations and subsequent multivariate/statistical analyses; detailed procedures are provided in the Supporting Information. Because portions from the same chicken were assigned across treatments, differences among fuel groups were evaluated using the Friedman test for repeated measures, followed by paired Wilcoxon signed-rank post-hoc tests with Benjamini–Hochberg correction for multiple comparisons. Statistical significance was defined at p < 0.05. Prior to multivariate analysis, data were standardized, and principal component analysis (PCA) together with hierarchical cluster analysis (HCA; Euclidean distance, Ward’s method) were applied to visualize contamination patterns and group discrimination. Ethics statement Chicken samples were commercially purchased from local markets, and no live animals were involved in the study. Therefore, formal animal ethics approval was not required. Results PCDD/Fs and PCBs concentrations in the samples The analysis of chicken meat smoked with different fuel sources revealed pronounced differences in contamination levels of total PCBs and total PCDDs/Fs ( S1 Table , Fig 1 ). When clean wood (W) was used, the mean total PCB concentration remained low (152 ± 8 ng/kg of lipid), with values ranging from 139 to 165 ng/kg of lipid, and no detectable PCDDs/Fs were observed. In this group, only a limited number of lower-chlorinated PCB congeners were present. 10.1371/journal.pone.0350345.g001 Fig 1 Boxplots of total PCBs (left) and total PCDD/Fs (right) concentrations of the samples. The central line represents the median, the box indicates the interquartile range (IQR), whiskers represent the minimum and maximum values, and “×” denotes the mean. In the PE group, 25 PCB congeners and a single PCDD/F congener were detected. Total PCB levels averaged 465 ± 14 ng/kg of lipid, ranging from 451 to 494 ng/kg of lipid, while total PCDDs/Fs were minimal (1.06 ± 0.11 ng/kg of lipid, range: 0.89–1.23 ng/kg of lipid). This indicates that although PE contains no chlorine, its combustion still promoted PCB formation. The PS group exhibited more pronounced contamination, with nearly all PCB congeners and multiple PCDD/F congeners present. The mean total PCB concentration reached 806 ± 42 ng/kg of lipid (range: 720–856 ng/kg of lipid), and total PCDDs/Fs averaged 5.21 ± 0.44 ng/kg of lipid (range: 4.70–5.88 ng/kg of lipid). Compared to PE, PS combustion yielded both higher levels and a broader spectrum of contaminants, reflecting more complex formation pathways under combustion conditions. The most severe contamination occurred in the PVC group, where all 29 PCB congeners and a wide range of PCDD/F congeners were detected. Total PCBs reached 990 ± 31 ng/kg of lipid (range: 926–1011 ng/kg of lipid), while PCDDs/Fs peaked at 33.90 ± 2.44 ng/kg of lipid (range: 27.9–36.6 ng/kg of lipid). The wider dispersion in PVC samples suggests highly variable but intense formation of toxic compounds, consistent with the chlorine-rich structure of PVC. The hierarchical cluster analysis (HCA) further supported the differentiation of samples according to the type of smoking fuel ( Fig 2 ). The dendrogram clearly separated the PVC group from all others, reflecting its distinct contamination profile with markedly higher levels of both PCBs and PCDDs/Fs. The PS samples formed a separate cluster, positioned between the PVC and the lower-contamination groups, consistent with their intermediate contamination levels. In contrast, the PE and W groups clustered together, although still distinguishable, indicating that PE combustion resulted in elevated PCB concentrations but remained closer in profile to clean wood compared to PS and PVC. 10.1371/journal.pone.0350345.g002 Fig 2 Hierarchical cluster analysis dendrogram of the PCDD/Fs and PCBs in smoked chicken meat, constructed using Euclidean distance and Ward’s method after data standardization. This clustering pattern corroborates the quantitative results: contamination followed the order W <