Introduction

Polychlorinated biphenyls (PCBs) are a group of 209 congeners that contain two coupled biphenyl rings but differ in the number and position of the chlorine atoms. These compounds were mainly used as dielectrics, cooling fluids, lubricants, flame retardants, oil additives to paints (Carpenter 2006; Lehmler & Robertson 2001), or they can be formed during waste incineration process (Esposito et al. 2022). They do not occur naturally and were commercially manufactured from approximately 1930 until 1979. Unfortunately, PCBs have been shown to have many adverse health effects in animals (McConnell 1985) and humans, including carcinogenicity (Lauby-Secretan et al. 2013), neurological effects (Nakajima et al. 2006; Schantz et al. 2003), and respiratory effects (Emmett et al. 2001). Therefore, PCB production has been banned in most countries. However, owing to their low water solubility and their high lipid solubility, PCBs continue to persist and bioaccumulate in the fatty tissues of living organisms (Esposito et al. 2022), characteristics which classify them as persistent organic pollutants (POPs) according to the Stockholm Convention (Magulova & Priceputu 2016). According to EU Regulation 2023/915 (Commission Regulation (EU) 2023/915), it is recommended to assess the contamination level of 6-Indicator non-dioxin-like (NDL) PCBs (ICES-6). Other studies also include dioxin-like (DL) PCB 118 as the 7th Indicator PCB (Renieri et al. 2019).

Because of urbanization and industrialization, coastal regions are recipients of many pollutants, including PCBs (Lao et al. 2021; Sobek et al. 2014). Because of their hydrophobic nature, PCBs bind with suspended organic particles in water and settle on seabed. One of the best bio-indicators to assessing the PCB contamination of coastal areas are filter feeding mollusks like mussels (Olenycz et al. 2015) and by extension fish (Ranjbar Jafarabadi et al. 2019; Visha et al. 2018). The main source of human exposure to PCBs is dietary intake, and more specifically the main contributor to this exposure is seafood because of their high contamination level (Ravenscroft & Schell 2018; Sirot et al. 2012). Especially in some countries, where seafood consumption is a main dietary habit for coastal population, recurrent seafood consumption may cause health risk (Judd et al. 2004). Therefore, simple and quick analytical methods for the analysis of PCBs in seafood must be developed, in order to ensure food safety especially near areas with extended pollutant load.

Fish and mussels are considered complex environmental samples, so they require suitable sample preparation to avoid matrix effect and other interferences (Lopes et al. 2012). In recent years, many analytical procedures have been used for the analysis of PCBs in fish and shellfish, including Soxhlet extraction (Wang et al. 2010), pressurized liquid extraction (Suchan et al. 2004), ultrasound-assisted extraction (Zhou et al. 2010), ultrasound-assisted dispersive liquid–liquid microextraction (Asati et al. 2018), accelerated solvent extraction (Ottonello et al. 2014), and micro-matrix solid-phase dispersion (Carro et al. 2017). The main disadvantages of some of these methods are that they are time consuming and require large amount of sample and solvents (Baduel et al. 2015). Recently, the QuEChERS (quick, easy, cheap, effective, rugged, and safe) extraction technique has been introduced (Anastassiades et al. 2003) and successfully applied for the extraction of a wide range of analytes, including PCBs, from complex matrices such as fish (Norli et al. 2011) and marine mammal blubber (Pedersen et al. 2023). The QuEChERS procedure is inexpensive, fast, and environmentally friendly using small amount of salts, solvents, and samples (González-Curbelo et al. 2015).

Thus, the purpose of the present study was the development, optimization, and validation of a QuEChERS procedure followed by gas chromatography coupled with mass spectrometry (GC–MS) for the determination of 14 PCB congeners (IUAPC number PCBs 18, 20, 28, 31, 44, 52, 101, 105, 118, 138, 149, 153, 170, and 180) in mussel (Mytilus galloprovincialis) and muscle tissue of gilt-head bream (Sparus aurata) samples. Those included 12 NDL-PCBs (6 Indicator PCBs) and 2 DL-PCBs. Different QuEChERS procedures, including “original” QuEChERS, “AOAC 2007.01” QuEChERS, and “Buffered CEN 15662” QuEChERS, where evaluated. Additional factor for evaluation was the extraction solvent, sample weight, and the clean-up salts. The method was validated and the expanded measurement uncertainty was estimated. Finally, the proposed method was applied for the determination of the selected PCBs in mussel and fish samples from aquacultures located in NW Greece.

Materials and Methods

Standards and Materials

A mixture of the 14 PCB congeners’ standard solution at a concentration of 10 μg mL−1 was obtained from CPAchem Ltd (Bulgaria). All solvents used were of high purity, including acetonitrile and methanol (Fisher Scientific, Leicester, UK) and n-hexane (Labscan, Dublin, Ireland). Dilutions of the standard mixture were prepared in n-hexane to appropriate concentration levels. All working solutions were stored in amber glassware at 4 °C. To perform QuEChERS extraction, polypropylene centrifuge tubes of 50 mL and 15 mL were purchased from Sarsdedt (Germany) while the salt-sorbent materials included as follows: anhydrous magnesium sulfate (MgSO4), LiChroprep RP-18 (40–64 μm), and tri-sodium citrate dehydrate purchased from Merck KGaA (Darmstadt, Germany); sodium acetate and sodium chloride (NaCl) of analytical reagent grade from Riedel de Haën (Hannover, Germany); sodium citrate dibasic sesquihydrate from Sigma–Aldrich (Bellefonte, PA, USA); and Primary Secondary Amine (Bondesil-PSA; 40 μm) from Agilent Technologies (Waldbronn, Germany). Finally, Millex-HV (PTFE, 0.22 μm) syringe filters were purchased from Millipore (Cork, Ireland).

Sampling and Sample Preparation

Fish (Sparus aurata) and mussel (Mytilus galloprovincialis) samples were obtained from aquacultures in the Ionian Sea (see section “Application to Real Samples Impacted by Aquacultures,” Fig. S1). After collection, samples were transferred to the laboratory in portable freezers. Regarding fish, the samples were filleted, and the muscle tissue was homogenized and stored at − 20 °C until analysis. Concerning mussel samples, they were cleaned, the shell was removed, and the mussel meat and the adductor muscle were homogenized and then stored at − 20 °C until analysis. All samples used for the optimization and validation were obtained from an inactive mussel farm from the same region and were previously analyzed and found to be free of the target PCBs.

The homogenized samples were thawed at room temperature, then 1 g was weighed into a 50-mL polypropylene tube and an appropriate volume of the standard solution was added, and left dry for 15 min. Afterwards, 5 mL of ACN was added and the tube was shaken by a vortex mixer for 1 min, then, the extraction salts were added (2 g MgSO4/1 g NaCl) and the tube was immediately shaken manually and followed by vortex for 1 min. Next, the sample was centrifuged at 4000 rpm for 5 min and 1 mL of the supernatant organic layer was collected, followed by dSPE clean up.

The aliquot of the organic layer (1 mL) was then transferred to a 15-mL polypropylene centrifuge tube containing 150 mg MgSO4, 25 mg PSA, and 25 mg C18. The tube was vigorously mixed and centrifuged at 4000 rpm for 5 min. The final aliquot was collected, evaporated to dryness, and then reconstituted in 200 μL of n-hexane. The final step prior to GC–MS analysis was the filtration of the extract with a polytetrafluoroethyene filter (PTFE, 0.22 μm) and its transfer to an autosampler vial.

GC–MS Analysis

GC–MS analysis was performed using a SHIMADZU QP-2020 GC–MS system equipped with an AOC-20i autosampler. The column was a MEGA – 5MS column (30 m × 0.25 mm, 0.25 μm). The ion source temperature was 220 °C while the injector temperature was 250 °C. Spitless mode was used with a sampling time of 2 min. The carrier gas was helium with a constant flow of 1.5 mL min−1. The column oven temperature program starts at 50 °C for 2 min and then it is heated up to 260 °C with a rate of 10 °C min−1, where it is held for 7 min. The total program time is 30 min. The injection volume was 1 μl injected by the autosampler. GC–MS operated at SIM (selected ion monitoring) mode. Quantification was performed by matrix matched multi-level calibration..

The retention times, target ions, and qualifier ions are shown in Table 1. A SIM chromatogram of a standard solution of the selected PCBs at a concentration of 100 ng mL−1 is shown in Fig. S2.

Table 1 GC–MS parameters of the selected PCBs
Full size table

Validation Characteristics

Method validation was performed in order to evaluate sensitivity, linearity, recovery (as a measure of trueness), precision (repeatability and within-laboratory reproducibility), limit of quantification (LOQ), matrix effects (ME), and measurement uncertainty (MU) according to the Commission Regulation (EU) 2017/644 (Commission Regulation (EU) 2017). The linearity of the optimized method was investigated over proper spiking ranges (from LOQ to approximately 100 LOQ), depending on the analyte, using weighted least-squares regression and expressed as the determination coefficient (r2). The method recoveries were evaluated at three concentration levels of 10 ng g−1, 50 ng g−1, and 100 ng g−1 (n = 5), by fortifying the sample at the beginning of the extraction. Precision of the method was expressed as follows: five replicates (n = 5) of fortified mussel samples with the PCB mixture at the three concentration levels were analyzed in the same day (intra-day) to calculate the repeatability (RSDr %) and over five consecutive days (inter-day) to obtain the reproducibility (RSDR %). Commission regulation guidance documents establish that in a quantitative method mean recoveries should be within the range 60–120%, with an associated RSD ≤ 20%, for all PCBs. The limits of quantification (LOQs) were determined as the minimum concentration of analytes in sample extracts (after QuEChERS), which provides a signal to noise ratio (S/N) ≥ 10.

To estimate matrix effects (ME) for the analytes, a calibration curve prepared in sample extracts was compared to a calibration curve prepared in solvent, at the same concentration range. Calculations were based on the following formula (Nannou et al. 2019) (Eq. 1):

$$\mathrm{\%ME}=[\left(\frac{\mathrm{slope\;of\;calibration\;curve\;in\;matrix}}{\mathrm{slope\;of\;calibration\;curve\;in solvent}}\right)-1] \times 100$$
(1)

In the present study, an expanded MU was estimated based on the five replicates’ analysis in different days, at 10, 50, and 100 ng g−1. Calculations were based on guidelines provided by SANTE 11312/2021 (SANTE 11312/2021), the Working Group for Measurement Uncertainty in PCDD/F and PCB Analysis (Group for Measurement Uncertainty in PCDD and Analysis 2017), so as the EURACHEM/CITAC Guide (CITAC Guide number 4) and the NORDTEST Handbook (Magnusson et al. 2007). The expanded uncertainty, expressed by U, is calculated from the relative standard uncertainty U′ in Eq. 2 and Eq. 3 (Kosma et al. 2021):

$$U=k\times {U}^{\mathrm{^{\prime}}}$$
(2)
$${U}^{\mathrm{^{\prime}}}=\sqrt{ {{{U}^{\mathrm{^{\prime}}}({\text{bias}})}^{2}{+U}^{\mathrm{^{\prime}}}({\text{precision}})}^{2}}$$
(3)

Where U is the expanded uncertainty, U′ is the combined measurement uncertainty, U′(bias) is the uncertainty component arising from method bias, U′(precision) is the uncertainty component for the precision, and k is the coverage factor of 2 (at 95% confidence level).

More specifically, U′(bias) is the uncertainty resulting from the bias of the fortified samples (SANTE 11312/2021) and it is given by Eq. 4:

$${U}{\prime}\left({\text{bias}}\right)=\sqrt{{{\text{mean}}}_{{\text{bias}}}^{2}+{SD.P}_{{\text{bias}}}^{2}}$$
(4)

where mean bias is the mean of the bias and SD. P bias is the population standard deviation of the bias.

Concerning U′(precision), the precision component from within-laboratory reproducibility is used, as shown in Eq. 5:

$${U}{\prime}\left({\text{precision}}\right)={{\text{RSD}}}_{R}$$
(5)

Furthermore, concerning residue analysis, another usual practice for estimating acceptable measurement precision includes the Horwitz equation or fit-for-purpose relative standard deviation (FFP-RSD) as well as the calculation of the Horwitz ratio (HorRat) as a simple performance parameter of measurement precision. When the ratio equals 1, there is an exact correspondence, and when the ratio is < 1, the precision is better than expected and poorer if it is greater than one. The empirical acceptable range is 0.5–2.0 (Horwitz and Albert 2006).

Results and Discussion

Optimization of QuEChERS Method

The matrix used for optimization purposes was mussel meat. For the first step of the extraction optimization procedure, three primary extraction protocols were tested. These were protocol 1 (“original” QuEChERS), protocol 2 (“AOAC 2007.01” QuEChERS), and protocol 3 (“Buffered CEN 15662” QuEChERS). The fortification level for optimization purposes was 100 ng g−1 wet weight. In the first protocol, MgSO4 and NaCl are used as sorbents, and ACN as the elution solvent. Protocol 2 combines MgSO4 and NaOAc as sorbents and ACN with 1% acetic acid as elution solvent. Finally, the buffered protocol includes a combination of MgSO4, NaCl, and sodium citrate for sorbents and ACN as the elution solvent. Protocol 1 produced better recoveries than the other two protocols, ranging from 71 to 81% as illustrated in Fig. 1.

Fig. 1
figure 1

Recoveries for all selected PCBs using the three protocols at the spiking level of 100 ng g−1 (ACN as extraction solvent, dSPE. cleanup with MgSO4 and PSA)

Full size image

The next step in the optimization procedure was the evaluation of the extraction solvents. The most common solvent used in the QuEChERS method is acetonitrile (ACN), which covers a wide range of analytes with different polarities. The other two solvent systems that were evaluated were a mixture of ACN and MeOH at 1:1 ratio, so as ACN with the addition of 0.1% acetic acid that usually promotes the salting-out effect (Rizzetti et al. 2016). The results are shown in Fig. 2, showing that ACN without any other addition produced the best recoveries for all the analytes whereas the mixture of ACN and MeOH exhibited the lowest recoveries for all analytes.

Fig. 2
figure 2

Recoveries for all selected PCBs after extraction with different solvents at the spiking level of 100 ng g.−1 (protocol 1 for extraction, dSPE cleanup with MgSO4 and PSA)

Full size image

Regarding sample weight, three values were tested: 1 g, 2.5 g, and 5 g. The recoveries among the three weights were similar (Fig. S3). However, 1 g produced a cleaner chromatogram, since it contained a lower amount of lipid content and because recoveries were slightly enhanced, the lower sample amount was chosen as the optimum.

Equally important is the optimization of the clean-up salt mixture. The first combination included 150 mg MgSO4 with 25 mg PSA, and the second included 150 mg MgSO4, 25 mg PSA, and 25 mg C18. The combination containing C18 shown to have better recoveries, due to the fact that C18 is used in QuEChERS clean-up to remove fats and other non-polar interferences, so in combination with the PSA, which is used for polar matrix components and free fatty acids, the results were significantly better, as shown in Fig. 3, with recoveries ranging from 77 to 91% in all cases.

Fig. 3
figure 3

Recoveries for all selected PCBs after the two clean up procedures at the fortification level of 100 ng g.−1 (protocol 1 for extraction, ACN as solvent, 1-g sample)

Full size image

To complete the optimization procedure, the volume of the solvent and quantity of the extraction salts had to be evaluated. To make the procedure environmental friendly and low cost, and because only 1 g of the wet weight sample was used, the volume of the solvent was reduced by half, as well as the quantity of the extraction salts, keeping the ratio at 4:1 by adding 2 g of MgSO4 and 0.5 g of NaCl. Thus, the recoveries of the proposed method for all the target PCBs were retained in the same range.

The final optimum conditions for the proposed QuEChERS-GC/MS method are described in detail in section “Sampling and Sample Preparation.”

Validation Characteristics of QuEChERS Method

After the selection of the best procedure, the proposed method was validated at three spiking levels (10, 50, and 100 ng g−1) for mussel samples. It was also evaluated without any changes in the procedure and validated for fish tissue samples since recoveries obtained in this case were also in the same range. Validation performance characteristics for mussels and fish are shown in Tables 2 and 3, respectively. In all cases, the method exhibited excellent linearity for all compounds (r2 > 0.9962), obtained from the calibration curves for each matrix constructed following least-squares linear regression analysis (LOQ-500 ng g−1). Quantification of the selected PCBs was conducted by matrix matched standard calibration curves, prepared by extracting blank mussel and fish samples, using the proposed method. The extracts were spiked with the selected analytes at appropriate concentration. To ensure the absence of the targeted PCBs in the blank samples, they were previously analyzed with the proposed method.

Table 2 Method validation parameters for mussel samples
Full size table
Table 3 Method validation parameters for fish samples
Full size table

Trueness and precision of the method were evaluated by recovery studies with fortified mussel and fish samples at three concentration levels (10, 50, and 100 ng g−1) with five replicates the same day and in five consecutive days (Tables 2 and 3). Mean recoveries for mussel samples ranged in all levels between 72.5 and 109.5%, whereas for fish samples ranged between 72.1 and 88.4%. In Fig. S4, a SIM chromatogram of a spiked mussel sample at a concentration of 50 ng g−1 is shown. Relative standard deviation (RSD) values were in all cases below 7.9% and 6.8% for mussel and fish samples, respectively. Specifically, intra-day precision (RSDr) ranged from 0.3 to 7.9% for mussels, while in fish samples from 0.3 to 3.4%. Inter-day precision (RSDR) ranged from 0.4 to 7.7%, and from 0.7 to 6.8%, for mussel and fish samples, respectively. The LOQs of the method for both matrices are also presented in Tables 2 and 3, and ranged between 1 and 2.5 ng g−1.

Regarding matrix effect (ME), as it is shown in Fig. 4, for mussel samples, %ME values were found to be between − 28.9 and 22.0%, and for fish samples, the values ranged between − 17.8 and 16.0%, demonstrating low ME in all cases.

Fig. 4
figure 4

Matrix effects (%ME) in mussel samples (a) and fish samples (b)

Full size image

The expanded MU was estimated for both matrices at three concentration levels individually for each PCB at twice the value of the uncertainty (confidence level 95%, k = 2) and the results are presented in Table 4. For the low spiking level (10 ng g−1), the MU ranged between 3.2 and 50.8% for the mussel samples, and from 24.1 to 43.0% for fish samples. At concentration of 50 ng g−1, the values scale from 14.7 to 52.2% and from 30.6 to 50.8% for mussel and fish samples respectively. In the high spiking level of 100 ng g−1, the MU ranged up to 51.6% and 52.4% for mussel and fish samples respectively. All the values are within the requirement (< 53%) according to the guidelines. In Table 4, the HorRat value, which was calculated in all spiking levels for each PCB individually, is shown. For both mussels and fish samples, in all cases, the values are < 1, ranging from 0.1 to 0.6 for both matrices.

Table 4 Measurement uncertainty U (%) for mussel and fish samples in three concentrations (k = 2, confidence level 95%) as well as HorRat values
Full size table

Application to Real Samples Impacted by Aquacultures

The optimized and validated method used for the analysis of mussel and fish samples from saltwater aquacultures in the Ionian Sea, NW Greece, to check its applicability. More specifically, twelve in total gilthead sea bream samples were obtained from four cages in three different fish farms located in the Sagiada region in Thesprotia, Epirus, Greece (Figure S1). Moreover, mussels that were grown on the same cages of these three aquaculture facilities were collected. The area of Sagiada accommodates several aquaculture facilities and is regularly submitted to controls concerning several pollutants’ levels as indicative by relevant literature. Therefore, there is evidence of its relatively low pollutant load in the specific area (Aminot et al. 2019; Aznar-Alemany et al. 2018; Boti et al. 2023; Martinaiou et al. 2022). Retention time, target and qualifier ions, and their intensities in real samples were compared to those of the calibration standards to confirm the identity of the target compounds using the criteria of the EU Regulation 2017/644 (EU 2017/644). Method quantification limits achieved were low enough to monitor potential levels of PCBs contamination in such complex matrices and they were in accordance with EU Regulation 2023/915 concerning their concentrations in food matrices. However, application of the proposed method revealed PCB levels below the method detection limits (LODs as shown in Tables 2 and 3). Thus, it was not feasible to express the results according to EU Regulation 2023/915 (EU 2023/915) that suggests a bound-approach.

Conclusions

A simple and fast method based on QuEChERS extraction was established for the determination of fourteen PCB congeners in mussel (Mytilus galloprovincialis) and fish (Sparus aurata) samples. The method was fully optimized and validated in terms of linearity, trueness, precision, and LOQ, with highly satisfactory results. The calculated expanded uncertainty was acceptable according to the guidelines. In addition, the extraction of twenty samples could be performed in less than 2 h, indicating that the proposed method is time-saving, easy-handling, and could be used for routine analysis of PCBs in complex and fatty matrices such as edible aquatic species. The method was applied to the analysis of three mussel samples and twelve fish samples revealing the absence of PCB congeners.