Multiple Organic Contaminants Determination Including Multiclass of Pesticides, Polychlorinated Biphenyls, and Brominated Flame Retardants in Portuguese Kiwano Fruits by Gas Chromatography

Global production of exotic fruits has been growing steadily over the past decade and expanded beyond the originating countries. The consumption of exotic and new fruits, such as kiwano, has increased due to their beneficial properties for human health. However, these fruits are scarcely studied in terms of chemical safety. As there are no studies on the presence of multiple contaminants in kiwano, an optimized analytical method based on the QuEChERS for the evaluation of 30 multiple contaminants (18 pesticides, 5 polychlorinated biphenyls (PCB), 7 brominated flame retardants) was developed and validated. Under the optimal conditions, satisfactory extraction efficiency was obtained with recoveries ranging from 90% to 122%, excellent sensitivity, with a quantification limit in the range of 0.6 to 7.4 µg kg−1, and good linearity ranging from 0.991 to 0.999. The relative standard deviation for precision studies was less than 15%. The assessment of the matrix effects showed enhancement for all the target compounds. The developed method was validated by analyzing samples collected from Douro Region. PCB 101 was found in trace concentration (5.1 µg kg−1). The study highlights the relevance of including other organic contaminants in monitoring studies in food samples in addition to pesticides.


Introduction
The consumers' interest in new and exotic fruits has intensified, mainly due to the growing knowledge regarding their bioactive composition and biological activities with prohealthy effects. Kiwano (Cucumis metuliferus E. Mey), belonging to the Cucurbitaceae family, is a plant naturally occurring in South Africa, Nigeria, Namibia, Botswana, and Southern Sahara, being also sporadically found in Yemen [1]. In the last years, its exportation has grown in countries such as Kenya, New Zealand, France, and Portugal [1,2].
The ripe kiwano fruit is characterized by an orange skin with many blunt thorns on its surface and green, jelly flesh inside [1][2][3][4]. Kiwano fruit has low levels of carbohydrates and calories but high contents of water, minerals including magnesium, calcium, potassium, iron, phosphorus, zinc, copper, and complex B vitamins, vitamin C, and β-carotene [1,2]. Some pharmacological properties of this exotic fruit have been recently revised by Vieira et al. [3], including anticardiovascular, antidiabetic, antiulcer, antioxidant, anti-inflammatory, antimalarial, and antiviral activities.
Due to these beneficial properties, its production, exportation, and consumption have increased, leading to intensive cultivation. As such, these particular fruits contribute directly and importantly to food security and nutrition in most producing zones, however, safety [37]. To the best of our knowledge, there are no analytical methods developed or monitoring studies that report the chemical safety in terms of pesticides and other environmental contaminants, namely plastic-related chemicals and others associated with anthropogenic sources, in kiwano fruit samples. Therefore, the aim of this study was to optimize and validate an extraction methodology for the simultaneous analysis of 30 multiple contaminants (6 OPP, 12 OCP, 5 PCB, and 7 BFR) from kiwano fruit samples using QuEChERS method and d-SPE clean-up to detect trace levels of these contaminants using GC techniques.

Samples
Ten kiwano fruits were supplied by a local farm located at Cinfães, Douro, Portugal. The mature fruits were collected in February 2019 from 10 different plants (random sampling) to obtain a representative set of fruits. The pulp of kiwano was separated from the orange skin, ground in a miller, homogenized, and finally, stored at −18 • C.

Extraction Procedure: Optimization and Validation
The 30 multiple contaminants were extracted from the kiwano samples based on the previously reported QuEChERS method with d-SPE clean-up [22]. The procedure, whose schematic illustration is shown in Figure 1, included five steps: (1) 5 g of kiwano pulp sample was weighed into a 50 mL polypropylene tube, (2) 8 mL of MeCN and 2 mL of UPW were added, and the tube was thoroughly vortexed for 1 min, EN QuEChERS (4 g MgSO 4 , 1 g NaCl, 1 g NaCitrate, 0.5 g disodium citrate sesquihydrate) were added, the tubes were shaken for 1 min with a vortex, and centrifuged for 5 min at 2490 rcf at room temperature, (3) 1 mL of the supernatant was transferred to the 2 mL d-SPE clean-up tube (150 mg of MgSO 4 , 50 mg of PSA, and 25 mg of GCB) and the tubes were vortexed for 1 min and centrifuged for 5 min at 2490 rcf at room temperature, (4) 900 µL of the final extract was transferred to a labelled vial, the extract was dried under nitrogen flow, and it was redissolved in 900 µL of n-hexane, and finally, (5) the sample was vortexed and 150 µL of the extract with the addition of 100 µg L −1 of the IS was added in the vial and was placed in the autosampler for the gas chromatography (GC) analysis. The IS was used to control the analytical quality of the GC analysis. Extractions were performed in triplicate. tubes were shaken for 1 min with a vortex, and centrifuged for 5 min at 2490 rcf at room temperature, (3) 1 mL of the supernatant was transferred to the 2 mL d-SPE clean-up tube (150 mg of MgSO4, 50 mg of PSA, and 25 mg of GCB) and the tubes were vortexed for 1 min and centrifuged for 5 min at 2490 rcf at room temperature, (4) 900 µL of the final extract was transferred to a labelled vial, the extract was dried under nitrogen flow, and it was redissolved in 900 µL of n-hexane, and finally, (5) the sample was vortexed and 150 µL of the extract with the addition of 100 µg L −1 of the IS was added in the vial and was placed in the autosampler for the gas chromatography (GC) analysis. The IS was used to control the analytical quality of the GC analysis. Extractions were performed in triplicate. For the optimization of the methodology, pre-spiking and post-spiking experiments were carried out to evaluate the extraction efficiency. The procedure for pre-spiking was the same as described above (Figure 1), with the difference that the sample in step 1 was contaminated with 7.5 µg kg −1 from the mixture of 30 multiple contaminants. The following steps remained the same, as shown in Figure 1. The procedure for the postspiking had a change in step 4. Before injection in the GC, 7.5 µg kg −1 of the 30 multiple contaminants was added to the vial and redissolved in the kiwano fruit extract. The extraction efficiency was studied in terms of recoveries percentages comparing the results obtained between the pre-spiking and post-spiking studies.
The validation of the method developed was performed following the Eurachem guidelines and SANTE/11312/2021 document by studying several analytical parameters, such as the linearity, recovery at three spiking levels (7.5, 11.2, 14.9 µg kg −1 ) and 5 replicates matrix effects, and intra-day and inter-day precision (experiments with the 7.5 µg kg −1 spiking level by five repeated measurements in the same and intercalary days). Quantification was performed using matrix-matched calibration (linearity between 1.5-18.7 µg kg −1 ) and solvent calibration (linearity between 10-125 µg L −1 ). The analytical validation was performed in the GC coupled to an electron capture detector (GC-ECD) and GC coupled to a flame photometric detector (GC-FPD), and with the regression analysis, the linearity was evaluated, and the limits of detection and quantification (LOD and LOQ) were determined.

Equipment
The GC analysis was performed according to Dorosh et al. [22]. Briefly, the halogenated organic compounds (5 PCB, 7 BFR, and 12 OCP) were analysed using GC-ECD (GC-2010, Shimadzu, Quioto, Japan) and OPP using a GC -FPD (GC-2010, Shimadzu, Quioto, Japan). The presence of contaminants was confirmed by GC/MS. Confirmation was based on a comparison of sample GC retention time and product ion abundance ratios (mass to charge ratio, m/z) against those obtained for a reference standard. The system control and the data acquisition were performed in Shimadzu's GC Solution software in GC-ECD and GC-FPD and Xcalibur software in GC/MS. The GC analysis was performed in triplicate. For the optimization of the methodology, pre-spiking and post-spiking experiments were carried out to evaluate the extraction efficiency. The procedure for pre-spiking was the same as described above (Figure 1), with the difference that the sample in step 1 was contaminated with 7.5 µg kg −1 from the mixture of 30 multiple contaminants. The following steps remained the same, as shown in Figure 1. The procedure for the post-spiking had a change in step 4. Before injection in the GC, 7.5 µg kg −1 of the 30 multiple contaminants was added to the vial and redissolved in the kiwano fruit extract. The extraction efficiency was studied in terms of recoveries percentages comparing the results obtained between the pre-spiking and post-spiking studies.

GC-ECD
The validation of the method developed was performed following the Eurachem guidelines and SANTE/11312/2021 document by studying several analytical parameters, such as the linearity, recovery at three spiking levels (7.5, 11.2, 14.9 µg kg −1 ) and 5 replicates matrix effects, and intra-day and inter-day precision (experiments with the 7.5 µg kg −1 spiking level by five repeated measurements in the same and intercalary days). Quantification was performed using matrix-matched calibration (linearity between 1.5-18.7 µg kg −1 ) and solvent calibration (linearity between 10-125 µg L −1 ). The analytical validation was performed in the GC coupled to an electron capture detector (GC-ECD) and GC coupled to a flame photometric detector (GC-FPD), and with the regression analysis, the linearity was evaluated, and the limits of detection and quantification (LOD and LOQ) were determined.

Equipment
The GC analysis was performed according to Dorosh et al. [22]. Briefly, the halogenated organic compounds (5 PCB, 7 BFR, and 12 OCP) were analysed using GC-ECD (GC-2010, Shimadzu, Quioto, Japan) and OPP using a GC -FPD (GC-2010, Shimadzu, Quioto, Japan). The presence of contaminants was confirmed by GC/MS. Confirmation was based on a comparison of sample GC retention time and product ion abundance ratios (mass to charge ratio, m/z) against those obtained for a reference standard. The system control and the data acquisition were performed in Shimadzu's GC Solution software in GC-ECD and GC-FPD and Xcalibur software in GC/MS. The GC analysis was performed in triplicate.

GC-ECD
The analysis was performed using a capillary GC column Zebron-5MS (30 m × 0.25 mm × 0.25 µm) (Phenomenex, Madrid, Spain). The oven temperature was programmed at 40 • C for 1 min, increased to 120 • C at a rate of 15 • C/min where it was kept for 1 min. Then, the temperature was increased once more at a rate of 10 • C/min to 200 • C, where it was kept for 1 min, and lastly, the temperature was increased from 7 • C/ min to 290 • C and held for 10 min. The injection was performed in splitless mode. The temperatures of the injector and ECD were 250 • C and 300 • C, respectively. Helium was used as a carrier gas (1.3 mL/min), and nitrogen as a makeup gas (30 mL/min).

GC-FPD
The GC-FPD column was the same as the one described in Section 2.4.1. The carrier gas was helium at 1 mL/min with a linear velocity of 25.4 cm s −1 . The detector was at 250 • C in injection was performed in splitless mode, and the analytes were detected at 290 • C. The column was programmed at 100 • C, which was kept for 1 min before increasing it to 150 • C at a rate of 20 • C/min, where it was held for 1 min. Following, the temperature was increased to 180 • C at 2 • C/min and kept for 2 min, and finally, increased at 20 • C/min to 270 • C, where it was kept for 1 min.

GC/MS Analysis
According to SANTE guidelines, confirmation of samples should be performed by MS detector. GC/MS analysis was performed with similar conditions of GC-ECD only in the positive samples observed in GC-ECD in order to have confirmation. GC/MS instrument, TRACE GC Ultra (Thermo Fisher Scientific, Austin, TX, USA) gas chromatograph coupled with a Polaris Q ion trap mass spectrometer was used. The transfer line and the ion source temperature were 260 and 270 • C, respectively. Data acquisition was performed first in full scanning mode from 50 to 500 m/z to confirm the retention times of the analytes. All standards and sample extracts were analyzed in selective ion monitoring (SIM) mode. PCB101 confirmation was performed with the identification of three m/z ions 326 > 324 > 286.

Statistical Analysis
Two-way ANOVA statistical analysis was applied to estimate significant differences among different analytical procedures using GraphPad software. Multiple comparisons were performed where each mean value was compared to each group of contaminants.

Results and Discussion
The extraction and clean-up steps for kiwano' matrices were a challenging part of the method development due to its rich composition in carotenoids, steroids, alkaloids, saponins, glycosides, flavonoids, tannins, and phenolic compounds [1,3]. The optimization of analytical methods for the determination of 30 contaminants in kiwano samples included the two crucial steps of the QuEChERS procedure: (1) Sample extraction and (2)  The study of the evaluation of the method's efficiency was carried out according to the guidelines of the SANTE document [38], being the range of recovery established 70 to 120%. In Figure 3, poor extraction recoveries were observed for some of the chemical families using QuEChERS AOAC. The OCP, PCB, and BFR compounds presented recoveries of less than 70% using the QuEChERS AOAC and CL1, while for QuEChERS AOAC and CL2 only the PCB compounds. Since recovery percentages after the clean-up CL1 (150 mg of MgSO 4 , 50 mg of PSA, and 50 mg of GCB) for QuEChERS AOAC evaluation were not satisfactory, the approach testing test other QuEChERS contents (EN) and another d-SPE clean-up (CL2) was followed. After reducing GCB in the CL2 clean-up and using QuEChERS EN, an improvement in extraction recoveries for all targeted multiple compounds was stated. The most evident result on extraction efficiency is the negative influence of the amount of GCB used in the second step of the extraction. As previously reported, GCB adsorbs compounds such as pigments, anthocyanins, and carotenoids, as well as planar compounds [23,33]. Therefore, reducing its quantity in the cleaning step is one of the optimizations of this process. Although the lower amount of GCB did not absorb all the coloring compounds like the previous CL1 clean-up, the samples were still suitable for GC analysis. ANOVA statistical analysis was used to compare the mean recoveries of each cleaning test (CL1, CL2) between the target chemical groups (OCP, OPP, PCB, BFR). The two-way ANOVA statistical study showed that the recoveries are significantly different comparing the two different clean-up sets (CL1 and CL2) for OCP and BFR using QuEChERS AOAC while for QuEChERS EN all chemical groups were statistically different. Overall, the results showed that most of the compounds are in the 70-120% range when QuEChERS EN and CL2 are used. Figure 4 shows a summary of the results of the recovery studies. It was observed that in the satisfactory range 70-120%, the highest number of contaminants was achieved with QuEChERS EN and CL2. The study of the evaluation of the method's efficiency was carried out according to the guidelines of the SANTE document [38], being the range of recovery established 70 to 120%. In Figure 3, poor extraction recoveries were observed for some of the chemical families using QuEChERS AOAC. The OCP, PCB, and BFR compounds presented recoveries of less than 70% using the QuEChERS AOAC and CL1, while for QuEChERS AOAC and CL2 only the PCB compounds. Since recovery percentages after the clean-up CL1 (150 mg of MgSO4, 50 mg of PSA, and 50 mg of GCB) for QuEChERS AOAC evaluation were not satisfactory, the approach testing test other QuEChERS contents (EN) and another d-SPE clean-up (CL2) was followed. After reducing GCB in the CL2 clean-up and using QuEChERS EN, an improvement in extraction recoveries for all targeted multiple compounds was stated. The most evident result on extraction efficiency is the negative influence of the amount of GCB used in the second step of the extraction. As As previously reported, a detailed optimization is an extremely important step as it reveals which compounds show the best results. As reported by Fernandes et al. [22][23][24]35], this extraction method is suitable but needs to be optimized and studied for each group of compounds and matrices.
The results, displayed in Figures 3 and 4, allowed us to assess that the best extraction and cleaning procedures for kiwano were QuEChERS EN with a clean-up CL2 (150 mg of MgSO 4 , 50 mg of PSA, and 25 mg of GCB), and this was selected for all further investigations.

Matrix Effects
In the present work, the matrix effect was evaluated by comparing the slope obtained with the calibration curves of each compound in the matrix phase and n-hexane. This evaluation was complemented by comparing the retention times of the chromatograms with the same concentration in the matrix phase and n-hexane, and no significant differences were observed. It is well described in the literature that some analytes in fruit extracts exhibit a matrix signal enhancement/suppression effect when analyzed by GC [23,39]. This effect occurs when interferences from fruit matrices (such as pigments, lipids, acids, etc.) compete with the target analytes in the GC injector [40]. Figure 5 shows that the different chemical families (OCP, OPP, PCB, and BFR) analyzed in kiwano fruits presented different matrix effects behaviors. The signal enhancement was observed with the use of both QuEChERS AOAC and EN with the CL2 cleaning step. Additionally, with QuEChERS AOAC and CL2 clean-up, the mean matrix factor value was higher than 1.2 in all the chemical families. The BFR are those with the highest signal increase. The QuEChERS EN showed a satisfactory matrix factor with CL1 clean-up. However, as shown in Section 3, the extraction efficiency was not acceptable with this extraction procedure. In any case, this study confirmed that the matrix effect was more evident when the lowest amount of GCB sorbent was used.  Figure 4 shows a summary of the results of the recovery studies. It was observed that in the satisfactory range 70-120%, the highest number of contaminants was achieved with QuEChERS EN and CL2.    As previously reported, a detailed optimization is an extremely important step as it reveals which compounds show the best results. As reported by Fernandes et al. [22,23,24,35], this extraction method is suitable but needs to be optimized and studied for each group of compounds and matrices.
The results, displayed in Figures 3 and 4, allowed us to assess that the best extraction and cleaning procedures for kiwano were QuEChERS EN with a clean-up CL2 (150 mg of MgSO4, 50 mg of PSA, and 25 mg of GCB), and this was selected for all further investigations.

Method Validation
Method validation is an important requirement in the practice of an analytical method process. The reliability and robustness of the method to be used for real sample analysis should be studied considering several analytical parameters. Linearity, extraction recovery at three spiking levels (7.5, 11.2, 14.9 µg kg −1 ), precision, LODs and LOQs obtained by the regression analysis (based on the standard deviation of the response of the curve and the slope of the calibration curve), as well as matrix effects, were the parameters studied for the validation of analysis of multiple contaminants in kiwano samples. Table 1 summarizes the analytical parameters in order of retention time obtained by GC-ECD and GC-FPD.
Considering the matrix effects described in the previous section, the analytical validation process was carried out in kiwano extract. Matrix-matched calibration curves were obtained in kiwano extracts of the 30 target analytes with a coefficient of determinations greater than 0.991. LODs and LOQs ranged from 0.2 to 2.2 and 0.6 to 7.4 µg kg −1 , respec-tively ( Table 1). The mean recoveries at the three spiking levels of 7.5, 11.2, and 14.9 µg kg −1 ranged from 90% and 122% (99% on average) with relative standard deviation (RSD) values between 8% and 15%. The method precision was determined through intra-day and inter-day repeatability experiments by five repeated measurements, and the results were less than 15% of RSD, which is suggested as the acceptable precision (Table 1). When compared to other studies on exotic fruits [30], we can say that for organochlorine pesticides, for example, the analytical parameters, namely the LOD and LOQ, are much better in the present work. As for the BFR, a study in capsicum cultivars [23] already reported presents higher LOD and LOQ values than those obtained for Kiwano.
with the calibration curves of each compound in the matrix phase and n-hexane. This evaluation was complemented by comparing the retention times of the chromatograms with the same concentration in the matrix phase and n-hexane, and no significant differences were observed. It is well described in the literature that some analytes in fruit extracts exhibit a matrix signal enhancement/suppression effect when analyzed by GC [23,39]. This effect occurs when interferences from fruit matrices (such as pigments, lipids, acids, etc.) compete with the target analytes in the GC injector [40]. Figure 5 shows that the different chemical families (OCP, OPP, PCB, and BFR) analyzed in kiwano fruits presented different matrix effects behaviors. The signal enhancement was observed with the use of both QuEChERS AOAC and EN with the CL2 cleaning step. Additionally, with QuEChERS AOAC and CL2 clean-up, the mean matrix factor value was higher than 1.2 in all the chemical families. The BFR are those with the highest signal increase. The QuEChERS EN showed a satisfactory matrix factor with CL1 clean-up. However, as shown in Section 3, the extraction efficiency was not acceptable with this extraction procedure. In any case, this study confirmed that the matrix effect was more evident when the lowest amount of GCB sorbent was used.

Method Validation
Method validation is an important requirement in the practice of an analytical method process. The reliability and robustness of the method to be used for real sample analysis should be studied considering several analytical parameters. Linearity, extraction Although the European Union legislation for pesticides [41] does not include the kiwano fruit, the analytical parameters obtained for this method meet the requirements. As for the other studied compounds, most of them are not included in the food legislation, despite being frequently detected in food products. As an example, EFSA recommends BFR monitoring studies in food samples [42].

Kiwano Sample Analysis
After the method validation, the optimized method was applied to evaluate possible contamination in kiwano samples. Since the study was carried out on the kiwano pulp, as it is the edible part, the results are presented by pulp mass. The screening of the 30 multiple contaminants in a total of 10 kiwano samples led to the identification and quantification of PCB 101 (5.1 µg kg −1 in the kiwano pulp) in a single sample. GC/MS analysis confirmed the presence of PCB 101 ( Figure 6). It was also confirmed that, except for one sample, the kiwano fruit samples are safe in terms of 12 OCP, 6 OPP, 7 BFR, and 5 PCB studied. The presence of pesticides is well reported in the literature on fruits [28,[43][44][45], concerning other contaminants, the works are less represented. However, PCBs, mostly associated with anthropogenic sources, have been reported in grapes, and other several fruits [46,47] and BFR in red fruits [24], capsicum cultivars [23], among others [48]. This work was performed in a small number of samples, and Portugal is still in the beginning regarding this crop. However, it shows the great importance of including these fruits in monitoring studies and that it should be extended to a larger number of samples from different production sites. Furthermore, the results suggest the importance of including other organic contaminants in monitoring studies on food samples in addition to pesticides. Table 1. Data including correlation to the matrix-matched calibration curve, the limit of detection (LOD) and limit of quantification (LOQ), mean recoveries (from three spiking levels), and precision obtained for the 30 target contaminants.

Conclusions
An analytical methodology based on an optimized QuEChERS technique was effectively applied for the simultaneous analysis of 30 multiple contaminants (12 OCP, 7 OPP, 5 PCB, and 7 BFR) in kiwano samples. The optimized QuEChERS procedure encompassed the study of two QuEChERS compositions (QuEChERS AOAC and EN) in addition to two d-SPE clean-up compositions (CL1 and CL2). Although matrix effects were observed, it was found that QuEChERS EN, in combination with CL2 clean-up, offered an improvement in overall extraction recovery of the multiple target contaminants. Based on these results, it can be concluded that analytical method optimization studies are crucial for the analysis of multiple compounds in complex matrices. The methodology meets the analytical requirements in terms of accuracy, sensitivity, and precision. The novelty of this study allows the evaluation of multiple contaminants in kiwano samples, ensuring their safe commercialization in terms of the presence of pesticides and other organic contaminants. The presence of PCB 101 in one kiwano fruit reinforces the need for monitoring studies of organic contaminants, such as PCBs and BFRs.