Emerging Contaminants in Raw and Cooked Marine Mussels: The QuEChERS Approach Combined With High‐Performance Liquid Chromatography Coupled With Tandem Mass Spectrometry

TL;DR

This study uses a new method to detect emerging contaminants in raw and cooked mussels, finding higher contaminant levels after boiling.

Contribution

This is the first study to investigate how cooking affects concentrations of emerging contaminants in mussels.

Findings

Caffeine was the most frequently detected contaminant in mussel samples.

Boiled mussel samples showed consistently higher contaminant concentrations than raw samples.

UV filters were commonly found, likely due to increased sunscreen use during the sampling period.

Abstract

Mussel aquaculture has experienced substantial growth in recent decades, with global production exceeding 2.17 megatons (live weight), more than doubling since the early 21st century. Representing nearly 94% of the total mussel production, aquaculture plays a crucial economic and ecological role. Mussels accumulate xenobiotics through their filter‐feeding behaviour, providing valuable insights into potential human exposure to the contaminants. However, the high lipid and protein content in their tissue can introduce analytical challenges, requiring rigorous clean‐up procedures to mitigate matrix effects. Herein, we applied a QuEChERS‐based extraction method coupled with high‐performance liquid chromatography‐tandem mass spectrometry (HPLC–MS/MS) to investigate the occurrence of emerging contaminants (ECs) in raw and boiled
Mytilus galloprovincialis
samples. Samples were collected from three aquaculture farms supplying mussels to fish markets in Liguria (Italy), aiming to provide a representative overview of contamination across different geographical sources. A total of 36 samples were analysed, detecting ECs in 26 samples. Caffeine was the most frequently detected contaminant, consistent with its widespread consumption. Additionally, UV filters were also commonly found in the samples, likely due to the sampling period at the end of summer, when sunscreen use is highest. This is the first study to investigate the impact of cooking on the concentrations of different classes of ECs in mussels, reflecting real consumption conditions. Box and whisker plots revealed consistently higher contaminant concentrations in boiled samples, suggesting that thermal processing may influence contaminant release. This study aims to offer insights into contaminants distribution and preliminary information for human exposure assessment of potential risks to human health.

Figures

Graphical overview of the analysed Mytilus galloprovincialis samples: boiled (B) and raw (R), from the following three different regions: Spain (E), Sardinia (S) and Liguria (L). Icons created by Ylivdesign, sourced from Flaticon (https://www.flaticon.com/free‐icons).

Concentration of ECs profiles expressed as μg per g of dry weigh t of each sample of Mytilus galloprovincialis , both boiled (B) and raw (R), from the following three different regions: Spain (E), Sardinia (S) and Liguria (L).

Box and whiskers plot A represents contamination from all ECs (graph on the left) and UV filters (graph on the right) by region (Spain, Sardinia and Liguria), while box and whiskers plot B is a graphical representation of contamination from all ECs (graph on the left) and UV filters (graph on the right) by cooking method (boiled and raw samples).

Principal component analysis results: score plot and biplot considering PC1 versus PC2.

Full text

Introduction

1

The global aquaculture industry has expanded significantly in recent decades, with the production of shelled molluscs, such as mussels, clams, scallops and oysters, playing a crucial role. In 2018, the worldwide production of shelled molluscs was estimated at 17.7 million tonnes (live weight), valued at USD 34.6 billion, largely from mariculture and coastal systems [1, 2, 3, 4]. Mussels alone accounted for 2.17 megatons (live weight), more than doubling since the beginning of the 21st century, with over 94% originating from aquaculture [1, 2, 3]. Leading producer nations include China and Chile, while Europe, led by Spain, Italy and France, collectively contribute approximately 20% to global output [2, 5].

Mussels are not only an economically valuable resource but also fundamental in regional cuisines, particularly within Europe, where the average annual per capita consumption is estimated at 1.28 kg. However, the potential presence of contaminants residues in mussels and other edible marine species constitutes a significant public health concern that necessitates rigorous monitoring through targeted analytical investigations. Because of their filter‐feeding behaviour, mussels have the ability to accumulate a wide range of xenobiotics from their environment, thereby increasing the risk of human exposure to environmental contamination [3, 6, 7, 8].

The need for reliable analytical methods to detect and quantify contaminants in bivalves is essential. In this context, high performance liquid chromatography coupled to electrospray ionisation (ESI) and tandem mass spectrometry (HPLC‐ESI‐MS/MS) plays a crucial role as a highly selective and sensitive technique, capable of determining multiple contaminants [9]. However, the complex biochemical composition of bivalves, characterised by high concentrations of lipids and proteins, presents unique challenges for contaminant analysis. Operating in multiple reaction monitoring (MRM) mode is a valid option to provide the necessary specificity for the target analytes, while reducing the detection of matrix‐related interferents. Still, if sample preparation and purification are not carefully evaluated, matrix effects related to ESI processes can lead to signal suppression or enhancement, potentially compromising analytical accuracy [3, 10, 11]. Traditional methods, such as solid–liquid extraction, require extensive sample purification to mitigate these effects, highlighting the need for more efficient and reliable extraction techniques.

The QuEChERS (Quick, Easy, Cheap, Effective, Rugged and Safe) method has gained importance during the years as a versatile approach for pesticide residue analysis. Initially developed for plant‐based matrices, its spread owes to the applicability to a wide range of analytes, minimal solvent usage, versatility and compatibility with high‐throughput workflows [3, 12, 13, 14, 15, 16, 17]. The method offers several advantages over traditional approaches, including reduced sample preparation time and minimal solvent use, with acetonitrile serving as the primary extraction solvent due to its ability to solubilise both nonpolar and medium‐polar analytes [12, 14, 16]. Despite its many advantages, the QuEChERS method has its own limitations when applied to complex matrices. The procedure steps need to be tuned to obtain high recovery of analytes with different physico‐chemical characteristics, while avoiding interferents. Regarding bivalve tissues, the presence of high lipid (up to 17%) and protein (up to 27%) concentrations can interfere with analytical accuracy, necessitating robust clean‐up procedures to mitigate matrix effects [3, 14]. The QuEChERS method, when coupled with HPLC–MS/MS, offers a powerful solution by combining efficient matrix purification with the high sensitivity and specificity of HPLC–MS/MS. This allows reliable detection and quantification of trace levels of contaminants in complex biological matrices [18].

Herein, we applied an optimised QuEChERS extraction followed by HPLC‐MS/MS analysis to investigate the presence of emerging contaminants (ECs) in raw and boiled samples of M. galloprovincialis . Samples were collected from various aquaculture regions commonly supplying mussels to fish markets in Liguria (Italy), with the aim of providing a representative overview of contamination across different geographical sources. Besides, this work pursues the objective of potential human exposure to ECs through mussel consumption evaluating the possible ECs human intake by mussels' consumption, using an efficient, selective and sensitive methodology. This is one of the few studies to investigate the impact of cooking on the concentrations of ECs in mussels, reflecting the conditions under which these products are usually consumed. In particular, this research is the first report in the scientific literature investigating ECs such as the pesticide omethoate, the lifestyle compound sucralose and the pharmaceutical drugs metoprolol and clenbuterol in mussel matrices. By addressing these novel aspects, the present study provides valuable insights into the contamination profile of aquaculture mussels and expands the current understanding of the occurrence and behaviour of ECs in edible bivalves.

Materials and Methods

2

Standards and Reagents

2.1

The analytical standards were purchased from Sigma‐Aldrich, Fluka Analytical and Alfa Aesar. Ethyl hexyl methoxy cinnamate (EHMC), carbamazepine (CBZ), hydrochlorothiazide (HCTZ), perfluorooctanoic acid (PFOA), perfluoro octane sulfonate (PFOS), benzophenone‐3 (BP‐3), octyl dimethyl p‐aminobenzoate (OD‐PABA), octocrylene (OC), sucralose (SCL), metoprolol (MTPL), clenbuterol (CLBT), ethylhexyl salicylate (EHS), chloramphenicol (CMPH), omethoate (OMT) and triclosan (TCS) were acquired from Sigma‐Aldrich (St. Louis, MO, USA). Caffeine (CAFF), theobromine (TBR) and diclofenac (DCF) were obtained from Fluka Analytical (Saint Gallen, Switzerland). All standards were high purity grade (≥ 97%). Stock standard solutions were prepared in methanol or methanol–water (1:1) mixture.

Sodium chloride was purchased from Sigma Aldrich (St. Louis, MO, USA) and magnesium sulphate from Carlo Erba Reagenti (Rodano, MI, Italy). Primary secondary amine (PSA) loose sorbent was sourced from Phenomenex (Torrance, CA, USA), whereas end‐capped C18 bonded silica loose sorbent was acquired from Supelco (Bellefonte, PA, USA). Methanol (MeOH) and acetonitrile (ACN), both HPLC–MS grade, were bought from Merck (Darmstadt, Germany). Ultrapure water was procured from a Millipore Q‐Gard system equipped with a Millipak 0.22‐μm filter (Millipore, Watford, UK).

LC–MS/MS Analysis

2.2

The analyses were performed using a 1200 SL Liquid Chromatograph coupled to an Agilent 6430 Triple Quadrupole mass spectrometer (MS) equipped with an electrospray ionisation (ESI) ion source (Agilent Technologies, Santa Clara, CA, USA). MS operating conditions were drying gas N_2_ (purity > 98%), temperature 300°C, 11 L/min flow; nebuliser gas pressure 15 psi; capillary was set at voltage 4000 V.

Chromatographic separation was obtained using a Kinetex F5 analytical reversed‐phase column (100 mm × 2.1 mm i.d.; 2.6‐μm particle size, Phenomenex, Torrance, CA, USA). Separation and analysis of target compounds were performed using two distinct chromatographic methods, for analytes ionizing in positive and negative polarity, respectively; details are present in Supporting Information (Section S1) (Table S1 and Table S2).

Dynamic‐multiple reaction monitoring (d‐MRM) mode was used to enhance the selectivity and sensitivity of the MS detection method. The quantifier ion was set to be the most abundant MRM transition, whereas additional transitions were used as qualifiers. Detailed MS conditions and MRM transitions are provided in Supporting Information (Table S3). The peak area ratio between the qualifier (q) and the quantifier (Q) was used to confirm compound identities. Data acquisition, qualitative, and quantitative analyses were performed using MassHunter 10.0 software from Agilent.

Sample Preparation

2.3

The sample preparation method was adapted from a previous study in which we optimised the extraction of ECs from the Antarctic scallop (Adamussium colbecki) using a QuEChERS‐based approach [19]. The QuEChERS procedure consisted of two main steps: a salting‐out assisted extraction of the analytes followed by a clean‐up using PSA and C18 sorbents.

For the extraction, 100 mg of M. galloprovincialis tissue were placed in a 50‐mL centrifuge tube and 12 mL of ACN:H_2_O (50:50, v/v) were added. The tube was vortexed for 1 min before the addition of 3.4 g of MgSO_4_ and 0.9 g of NaCl to facilitate the salting‐out process. The mixture was then vortexed again and centrifuged (3500 rpm, 5 min) to promote the phase separation between the organic and aqueous phases.

Finally, 2 mL of supernatant was subjected to the d‐SPE with 300 mg of MgSO4, 85 mg of PSA and 100 mg of C18. The sample was vortexed for 1 min and centrifuged under the same conditions (3500 rpm, 5 min). Subsequently, 1 mL of the purified supernatant was evaporated under nitrogen stream and reconstituted in 1 mL of MeOH:H_2_O (50:50, v/v). The final extract was filtered through a 0.2 μm polytetrafluoroethylene (PTFE) filter, diluted 1:5 with MeOH:H_2_O (50:50, v/v) and analysed via HPLC–MS/MS.

Analytical Performance

2.4

The analytical performance of the method was assessed through the evaluation of key parameters, including linearity, matrix effect (ME%), limit of detection (LOD) and quantification (LOQ).

Linearity was determined by constructing calibration curves, in which the peak areas of the target analytes were plotted against their corresponding concentrations. Calibration standards were prepared over the concentration range spanning from 0.1 to 50 ng/mL, specifically at 0.1, 0.5, 1, 5, 10, 20 and 50 ng/mL. The linearity of the response was evaluated by calculating the coefficient of determination (R ^2^), ensuring a robust quantification over the specified range.

ME% was obtained by analysing the extracts spiked with a known quantity of analytes (PA A) and comparing the signal with the pure standard in MeOH:H_2_O. The following formula was used to calculate ME%:

$$[eqn]$$

where PA_NS_ corresponds to the peak area of the nonspiked sample and PA_P_ is the peak area of the pure standard solution.

The LOD and LOQ were experimentally derived from the procedural blanks. The LOD was defined as the lowest analyte concentration distinguishable from the matrix background signal, corresponding to a signal‐to‐noise (S/N) ratio of 3. Similarly, the LOQ was determined as the lowest concentration at which the analyte could be quantified with acceptable precision and accuracy, corresponding to a S/N ratio of 10. All LOD and LOQ values are reported as mean ± standard deviation, calculated from replicate analyses of procedural blanks.

Sample Collection, Pooling and Storage

2.5

M. galloprovincialis specimens were purchased in September 2024 from local fish markets, sourced from aquaculture farms in the following three distinct regions: Liguria (Italy), Sardinia (Italy) and Spain. These territories are the most common supply areas for mussels in Italy. The seafood was handled under consumer conditions, including manual handling by workers and transportation in plastic bags to the laboratory.

Upon arrival, the mussels were thoroughly rinsed to remove any potential surface contaminants and only the edible tissue was retained for analysis. Bivalves from each region were then divided into two groups: one designated for raw analysis and the other for boiling. The boiled samples were cooked under typical household conditions, using a casserole dish and moderate flame before further processing.

Following this step, a total of 36 pooled samples, each consisting of tissues from five individuals, were obtained from the purchased mussels of the three geographical origins. In particular, six pooled raw samples and six pooled boiled samples for each region were considered, as shown in the scheme in Figure 1. The pooling process was performed to ensure representative sampling and minimise biological variability.

Graphical overview of the analysed Mytilus galloprovincialis samples: boiled (B) and raw (R), from the following three different regions: Spain (E), Sardinia (S) and Liguria (L). Icons created by Ylivdesign, sourced from Flaticon (https://www.flaticon.com/free‐icons).

Finally, all the specimens were frozen at −20°C for 24 h and subsequently freeze‐dried to preserve sample integrity until analysis and facilitate further processing.

Data Analysis

2.6

Box and whisker plots provide a comprehensive summary of a dataset by visualising key statistical parameters, including the smallest and largest nonoutlier observation, the lower quartile, median and the upper quartile. The whiskers, extending from the hinges of the box, are defined as 1.5 times the interquartile range (IQR), unless minimum or maximum values are reached when outliers are excluded. Any data points outside this range are represented as individual markers, indicating potential outliers [20]. All box and whiskers plots presented in this paper were generated using Microsoft Excel 2024. Data analysis was performed using descriptive and exploratory statistical approaches only. No inferential statistical tests were applied due to the high proportion of nondetected values. Therefore, differences discussed refer to the observed trends.

To explore the multidimensional relationship between contaminants, geographical origins and cooking treatments, a principal component analysis (PCA) was carried out. This multivariate technique was applied to discern patterns, highlight potential clustering and determine the analytes most responsible for potential group differentiation. The PCA was performed using the open‐source software Chemometric Agil Tool (CAT), which facilitated the visualisation of sample distribution and variance within the dataset [21].

Results and Discussion

3

This study was conducted on mussel samples collected from aquaculture farms in three different geographical regions: two Italian sites (Liguria and Sardinia) and one Spanish site. The primary aim was to compare contamination levels in these regions. Given the significant role of seafood in human nutrition, particularly bivalves, a further objective was to evaluate the impact of cooking on contaminants concentrations, as thermal processing may alter both the chemical composition of the food and the levels of environmental pollutants [22]. Consequently, the study took into account raw and boiled mussels to quantify ECs in both sample types.

Analysis Results of the Environmental Samples

3.1

The selected sample treatment was a QuEChERS procedure optimised on bivalve mollusc tissue. An exhaustive multivariate optimisation had been carried out to maximise recovery and minimize ion suppression by using spiked Adamussium colbecki (also called Antarctic scallop) material [19, 23]. The method had been originally tested on a wider range of ECs but was validated for the 18 ECs listed in paragraph 2.1. The sample composition of mussels was expected to be rather similar to that of A. colbecki, given their belonging to the same animal class and the comparable main components of mussels and scallops (approximately 20% proteins, 1%–4% lipids and 6% carbohydrates). To ensure the accuracy of the method, matrix effects were carefully evaluated in the mussel extracts obtained by the QuEChERS technique and analysed by HPLC–MS/MS. For 17 out of the 18 analytes, matrix effects were observed within the optimal range of 79%–109%. DCF exhibited a slightly lower ME%, of 74%, though associated with a higher standard deviation (detailed values are reported in Table S4). Given these results, LC–MS signal normalisation was unnecessary, and quantification was performed using external calibration, to avoid the time‐consuming standard addition method. Notably, DCF was not detected in any of the samples. Besides the accuracy evaluation, LODs and LOQs specifically related to the mussel samples were assessed and are reported in Supplementary Information (Table S4).

In the considered monitoring campaign, 10 out of the 18 analytes were not detected in any of the samples, including THB, OMT, MTPL, CLBT, HCTZ, SCL, CMPH, PFOS, DCF and TCS. The absence of these compounds in mussel tissues could be attributed either to their absence in the waters or to the mussels' ability to metabolise them efficiently.

Contrariwise, several contaminants were found ubiquitously across almost all samples, with UV filters being among the most prevalent pollutants. The complete quantification results for the environmental samples analysis are presented in Table S54. All concentration values are expressed as mean ± standard deviation, calculated from replicate pooled samples. The primary UV filters detected were OD‐PABA, EHS, OC, EHMC and BP‐3, as reported in Figure 2. Among them, BP‐3 exhibited the lowest concentration (71 ng/g), appearing in only one sample, whereas OD‐PABA was the most widespread, with concentrations ranging from 7 ng/g to a maximum of 2044 ng/g, varying based on the geographical origin of the mussels. EHS was quantifiable only in certain samples, reaching a peak concentration of 546 ng/g, while in 10 other samples, it remained below the LOQ. Finally, the other UV filters detected were OC and EHMC, both at higher concentrations than EHS (respectively 626 and 592 ng/g) but less frequently detected.

Concentration of ECs profiles expressed as μg per g of dry weigh t of each sample of Mytilus galloprovincialis , both boiled (B) and raw (R), from the following three different regions: Spain (E), Sardinia (S) and Liguria (L).

These findings align with previous research indicating that UV filters persist in aquatic environments for extended periods [24, 25]. Their variable distribution in mussel samples may depend on factors such as the composition and consumption of personal care products (PPCPs) in the investigated regions, proximity to recreational bathing sites and intrinsic different persistence of these contaminants.

Caffeine was another widely detected contaminant, exhibiting the highest concentrations among all quantified analytes, with values ranging from 163 to 2672 ng/g. Because caffeine is extensively consumed worldwide through beverages such as coffee, tea and soft drinks, as well as various food products, its ubiquity in the aquatic environment is not surprising [26]. Estimates suggest that approximately 90% of the global population ingests at least one caffeine‐containing product daily, with an average daily intake of 80–400 mg per person [26, 27]. Consequently, caffeine serves as a reliable tracer for anthropogenic contamination in aquatic systems.

In addition to UV filters and caffeine, other contaminants found in mussel samples included CBZ, TCS and PFOA. CBZ is an anticonvulsant and analgesic drug used for the treatment of epilepsy, bipolar disorder and psychosis [28]. A concentration of 28 ng/g was revealed in one sample, while in eight others, it was detected below quantification levels. TCS, an antimicrobial agent banned by the European Union in soap products since 2016 but still permitted within specific limits in certain PPCPs, as reported by Sinicropi et al., was found in a boiled Spanish mussel sample at a concentration of 64 ng/g [29]. Finally, PFOA, a perfluorinated compound used in a variety of applications including the production of water and oil repellent surfaces, surfactants and firefighting foams, as described by Teaf et al., was detected at 39 ng/g in one boiled Spanish mussel sample and present below LOQ in other five [30].

Overall, this study demonstrates that UV filters represent the most abundant class of contaminants in farmed mussels, surpassing the other pollutants. This result is consistent with the timing of the sampling campaign, which was conducted at the end of the summer season, a period characterised by peak tourist activity. Furthermore, the high detection rates of these compounds align with the known bioaccumulation patterns of these compounds in marine organisms [31, 32].

Univariate and Multivariate Data Analysis

3.2

In order to compare results obtained for boiled and raw samples, the box and whiskers plot was chosen for data visualisation. This visualisation method is suitable for nonnormally distributed datasets, as they do not require data transformation and can effectively handle a substantial proportion of nondetected values, allowing a direct comparison of different groups [33]. It was used to represent (i) the sum of the concentrations of the detected ECs and (ii) the sum of the concentrations for UV filters class only, to assess differences between raw and boiled mussels and to evaluate variations in the contamination levels across the three geographical areas under investigation.

The graphical representations in Figure 3 indicate no substantial evident differences in mean contamination levels (denoted by the “X” in Figure 3A) for UV filters and the total set of ECs across the three regions. However, Sardinia exhibited the narrowest range of contaminants' concentrations, whereas Liguria displayed the widest variability (Figure 3A).

Box and whiskers plot A represents contamination from all ECs (graph on the left) and UV filters (graph on the right) by region (Spain, Sardinia and Liguria), while box and whiskers plot B is a graphical representation of contamination from all ECs (graph on the left) and UV filters (graph on the right) by cooking method (boiled and raw samples).

When comparing raw and boiled mussel samples, the mean total contaminants' concentration (denoted by the “X” in Figure 3B) was consistently higher in boiled samples. Moreover, for all analytes, the third quartile in raw samples was significantly larger than the second quartile (which was zero), indicating a positively skewed distribution. Furthermore, boiled samples exhibited a higher frequency of outliers compared to raw samples, suggesting a more heterogeneous distribution of contaminants after cooking. In both cases, UV filters and total ECs, the overall distribution of data in boiled samples was broader than in raw samples.

A plausible explanation for the phenomena may be linked to the metabolic cycle of mussels, particularly the process of glucuronidation. This metabolic reaction serves as a key detoxification mechanism by transforming lipophilic compounds through conjugation with glucuronic acid. The resulting glucuronides are typically stable; however, exposure to elevated temperatures and variation in pH during cooking may lead to their cleavage, converting the conjugated compounds into their active forms and consequently increasing their measurable concentrations. These findings validate the hypothesis that the observed increase in contaminant concentrations post‐cooking may, at least partially, be induced by the thermal breakdown of glucuronide conjugates, which lead to the release of their parent compounds within the mussel tissue postcooking [34, 35].

To further explore potential correlations among the considered analytes and samples, a PCA was conducted. The analyte concentrations were used as initial variables, while the M. galloprovincialis samples were considered as objects. Prior to analysis, data were scaled and centred. The first three principal components (PCs) explained 59% of the total variance, demonstrating a reasonably decent correlation among the original variables. By focusing only on the first two PCs, 43.9% of the variance was explained; no significant difference was observed between the score plots obtained in two or three dimensions; thus, only PC1 and PC2 will be considered for an easier graphical interpretation. The score plot in the PC1 and PC2 dimensions (Figure 4) revealed a certain clustering of the samples, suggesting the presence of two groups and one isolated sample. As highlighted by the biplot in Figure 4, the first group, containing most samples, was characterized by higher and rather homogeneous concentrations of TCS and OD‐PABA. The second, less defined, group included four samples (2 from Liguria, 1 from Spain and 1 from Sardinia) distinguished by elevated levels of CAFF and EHS. These samples also exhibited similarities in their content of CLNT and CBZ. Lastly, one boiled sample from Sardinia was detached from the others, because it exhibited the highest concentrations of OC and BP‐3, as illustrated in the biplot shown in Figure 4. All samples from the second group (with the exception of RL1) and BS3, the isolated sample, were identified as outliers in the Box and Whiskers Plots that represented the total contamination levels of ECs. Despite these groupings, PCA did not reveal any significant clear correlation between contaminant concentrations and either the cooking method or the geographical origin of the samples.

Principal component analysis results: score plot and biplot considering PC1 versus PC2.

Comparison With Previous Studies

3.3

This study aimed to assess contamination present in mussels' tissue to evaluate potential health risks associated with their consumption. A similar investigation was conducted by Álvarez‐Ruiz et al. (2021), who tested three different QuEChERS extraction methods on raw commercially purchased mussels. Although their study included several of the same analytes investigated in this work (CAFF, DCF, PFOA, PFOS and TCS), none of the target compounds were detected in the analysed samples. The authors attributed these results to the commercial purification process applied to mussels in certain regions, such as Valencia, where the samples were bought. Notably, similarly to the present study, no compounds exceeded the LOQs, but trace signals for PFOA were observed in some chromatograms, suggesting the possible presence of residual contaminants at concentrations below quantification limits. The LOQ for PFOA reported by Álvarez‐Ruiz et al. was 16 ng/g, whereas the analytical method employed in the present work achieved a lower LOQ of 12 ng/g [36].

McEneff et al. investigated both raw and cooked wild mussels collected in Ireland, allowing a direct comparison between preparation methods. Their results showed significantly higher concentrations of CBZ than those observed in the present study, with 12.1 μg/g in raw and 13.6 μg/g in cooked mussels, while, in our samples, CBZ was quantified at 28 ng/g in raw mussels and remained below the LOQ in cooked specimens [35]. To the best of the authors' knowledge, no further studies have specifically focused on aquaculture mussels and the influence of cooking on contaminant concentrations, while previous research has primarily utilised mussels as biomonitoring organisms, focusing exclusively on raw specimens. Accordingly, comparison with literature data was conducted solely on the raw mussel samples analysed in this study, as no comparable cooked sample datasets are currently available.

Several studies, including those conducted by Picot‐Groz et al. (2014) and Rodil et al. (2019), have documented the presence of several UV filters in mussels using QuEChERS‐based extractions, though with variations in reported concentrations [37, 38]. Notably, in the present study, a comparison limited to raw mussel samples revealed differing maximum concentrations for certain contaminants that were detected compared to those reported by Rodil et al. (2019), who analysed wild mussels collected during low tides in the Galician coast (Spain). For instance, OC and EHMC were found below the LOQ in our samples, whereas Rodil et al. reported higher concentrations (141 ng/g for OC and 94 ng/g for EHMC). Conversely, OD‐PABA was detected at 110 ng/g, exceeding the highest concentration recorded by Rodil et al. of 12 ng/g. Finally, BP‐3 concentrations were comparable between the studies, with our findings (71 ng/g) closely aligning with those of Rodil et al. (63 ng/g) [38].

A comparison with the findings of Picot‐Groz et al. (2014) indicates lower contamination levels for OC and EHMC in the present study. In their analysis of mussel samples collected from four beaches in southern Portugal, maximum concentrations reached 3992 and 1765 ng/g, respectively, which were markedly higher than those observed in our analysis. Similarly, for OD‐PABA, we detected a maximum concentration of 110 ng/g in raw mussel samples, which is lower than the 833 ng/g reported by Picot‐Groz et al. [37].

Rodil et al. documented the presence of PFOA at a maximum concentration of 17 ng/g, whereas a slightly higher concentration of 39 ng/g of this persistent pollutant was detected in the present study, even though it was quantified in a single sample [38].

Additional comparison can be drawn with studies by Álvarez‐Muñoz et al. (2019) and Gadelha et al. (2019), which investigated the presence of TCS and CBZ in aquaculture oysters ( Crassostrea gigas ) and cockle ( Cerastoderma edule ) using QuEChERS‐based methodologies. In the former study, specimens were collected from coastal sites in Catalonia (Spain), while in the latter, samples were directly obtained from aquaculture bags in the Portuguese coast. In both studies, the target contaminants were below detectable levels. In contrast, in the present study, TCS was not detected in any raw mussel samples, whereas CBZ was quantified at a concentration of 28 ng/g [39]. Furthermore, Álvarez‐Muñoz et al. reported an average CAFF concentration of 99.85 ng/g in their study, somewhat lower than the mean concentration of 209 ng/g revealed in raw mussels in this paper [39].

The observed discrepancies in contamination levels across the studies may be attributed to several factors, including differences in sampling locations, temporal variations and anthropogenic influences. In general, ECs concentrations tend to be higher in aquaculture mussels compared to those collected from open‐sea environments. This pattern may be explained by prolonged exposure to anthropogenic activities, combined with the limited water renewal and absence of strong currents in aquaculture settings, which may facilitate the accumulation of the contaminants.

Conclusions

4

In this study, a rapid, cost‐effective and reliable analytical approach for the extraction and quantification of ECs in Mytilus galloprovincialis samples was presented. The QuEChERS extraction method followed by HPLC–MS/MS analysis was successfully employed to determine the presence and distribution of ECs in raw and boiled mussels from different geographical regions.

ECs were detected in 26 samples out of 36, including three different geographical regions and two cooking methods, with quantifiable levels of contamination in 15 samples. The prevalent contaminant was CAFF, an expected outcome given its widespread consumption. Additionally, UV filters (in particular OD‐PABA, EHS, OC, EHMC and BP‐3) were among the most frequently detected contaminants. These findings align with the timing of the sampling campaign, which was conducted at the end of the summer season, a period characterised by extensive use of sunscreen products that can be consequently released into the marine environment.

Boiled samples consistently exhibited a higher contamination, both considering the summed concentrations of only UV filters and of the total set of ECs, along with a wider data distribution compared to raw samples. Possible correlation between geographical origin of the samples and contaminant concentration was explored using box and whiskers plots and PCA, but no significant noticeable relationship was found. As a future perspective, a wider set of samples may be analysed to strengthen the present outcomes, which suggest that differences among individual mussels' batches may be higher than those related to geographical origin. Further studies could also investigate the presence of glucuronide conjugates to confirm the hypothesis that cooking promotes their cleavage, resulting in increased levels of free contaminants.

Although the obtained data may be considered as preliminary, this study represents the first investigation into the impact of cooking on the concentrations of ECs in mussels, reflecting the actual conditions under which these organisms are consumed. Notably, it is also the first report documenting the investigation of specific ECs, such as OMT, SCL, MTPL and CLBT, in mussel matrices.

By addressing these novel aspects, the present work expands the current understanding of the behaviour and occurrence of ECs in edible bivalves, providing a foundation for future studies. Overall, this work underscores the utility of a standardised analytical protocol for environmental monitoring of ECs; the presented approach provides a robust and efficient tool for assessing trace level contamination in marine biota. Future studies will focus on expanding the dataset and incorporating estimated daily intake calculations under different consumption scenarios, in order to evaluate, offering valuable insights into potential risks to human health.

Supporting information

Table S1: Gradient elution conditions used for the analytes studied in positive mode ionisation. Table S2: Gradient elution conditions used for the analytes studied in negative mode ionisation. Table S3: Details on the mass detection of all considered analytes, including the parameters of the MRM detection. Table S4: Determination coefficient, matrix effect, LOD and LOQ. Table S5: Quantitation results, expressed as ng per g of dry weight of each sample of Mytilus galloprovincialis both boiled and raw from the three different regions (Spain, Sardinia and Liguria).