HRMS analysis of pesticides in vegetables from Shanghai and risk assessment

ABSTRACT A rapid analytical method for the simultaneous determination of 550 pesticide residues in vegetable samples was developed based on ultra-high performance liquid chromatography-tandem Q/Orbitrap high-resolution mass spectrometry (UPLC-Q/Orbitrap-HRMS). To investigate the risk of exposure to pesticide residues through vegetable consumption, 704 leafy vegetable samples from Shanghai were analysed for multiple residues using this method. A total of 54 pesticide residues were identified in these vegetable samples and 302 samples contained one or more pesticide residue. The levels of the detected pesticides did not pose a health risk in the long term and were acceptable according to the results of the chronic dietary risk assessment. Risk rankings displayed that most of the pesticides were low to medium risk. The findings of this study provide a reference for future pesticide monitoring programmes.


Introduction
China's yearly production and consumption of fruits and vegetables is about 700 million tons, accounting for 40% of the world volume (Guancha Syndicate 2019).Vegetables are rich in micronutrients such as dietary fibres, vitamins (vitamins A, B, and C and carotenoids), and minerals (Fe, K, Ca, Mn, Zn, etc.), and their consumption is essential for maintaining human health and physical development (Ermolaeva et al. 2019;Li et al. 2021;Zhang et al. 2021).To improve agricultural productivity and crop production and quality, pesticides and insecticides are often used to control agricultural pests to prevent crop infestation by diseases and insects (Guan et al. 2010;Bakırcı and Hışıl 2012;Bakırcı et al. 2014).Despite the extensive benefits of using pesticides in agriculture, overuse and incorrect use of pesticides can lead to high levels of pesticide residues in food, which can be potentially harmful to the environment in the long run and can pose serious threats to human health, such as acute neurotoxicity, neurodevelopmental disorders, immune, reproductive and endocrine system dysfunction, chronic kidney disease and can even induce cancer (Sinha et al. 2011;Bakırcı and Hışıl 2012;Nagy et al. 2020).Given the potential risks of pesticides to human health, the use of pesticides in agriculture needs to be continuously monitored (Łozowicka et al. 2012).
While monitoring of pesticides provides information on their residue levels in the matrix, additional information is needed in the context of food safety to obtain more realistic levels of pesticide ingestion toxicity (Wei et al. 2019).To evaluate the potential health risks associated with pesticide residues in food, government departments develop dietary pesticide exposure estimates and compare them to, among other things, acute reference doses (ARfD) and acceptable daily intake (ADI) to assess acute and chronic risk (Lehmann et al. 2017;Czaja et al. 2020;Galani et al. 2020;Wang et al. 2022).Dietary exposure to pesticide residues in food is estimated based on average daily food consumption per person, average body weight, and pesticide residue levels in food.If the estimated daily intake (EDI) of a compound exceeds the ARfD or ADI, consumer exposure is of concern.To ensure quality safety and regulatory supervision of agricultural products, regulatory agencies in many countries have established maximum residue limits (MRLs) for various agricultural products to minimise pesticide residue levels.
At present, multi-residue analysis is mostly used to detect pesticide residues in food and agricultural products, main methods including gas chromatographymass spectrometry (GC-MS) (Borrás et al. 2011;Brondi et al. 2011;Alamgir Zaman Chowdhury et al. 2013) and liquid chromatography-mass spectrometry (LC-MS) (McManus et al. 2019;Schwanz et al. 2019;Végh et al. 2022), where tandem mass spectrometry techniques ensure the specificity and accuracy of the detection method.
However, due to the complexity of food matrices, low-resolution mass spectrometry can release false positives in multi-component residue analysis, resulting in misclassification of results.The development of electrostatic field Orbitrap high-resolution mass spectrometry (Q-Orbitrap) has largely remedied this deficiency by screening for unknowns with much greater reliability.It is widely used for non-targeted rapid screening of multi-pesticide residues, resulting in better options for risk assessment of pesticide exposure to consumers from locally produced leafy vegetables and to support effective data for regulation and management by local authorities.The aim of this study was using ultra-high performance liquid chromatography-tandem Q/Orbitrap high-resolution mass spectrometer (UPLC-Q/Orbitrap-HRMS) to investigate the distribution of pesticide residues in 704 batches of vegetables (Green cabbage, Chinese cabbage, and amaranth) produced from local vegetable bases in Shanghai in 2021 and to calculate consumer exposure and potential health risks using hazard quotient (HQ) and hazard index (HI) methods.

Chemicals and reagents
All standard solutions of pesticides were purchased from Alta Technology Co., Ltd.(Tianjin, China) and stored at −20°C in darkness.The standard stock solution was prepared as a set of 1 µg/mL mixed standard solutions containing all pesticides using acetone as solvent, stored at 4°C in darkness and reconstituted every 2 months.

Sample collection
As shown in Figure 1, from July to September 2021 a total of 704 samples of Green cabbage (n = 330), Chinese cabbage (n = 250) and amaranth (n = 124) were collected from 10 agriculture-related districts in Shanghai (Baoshan District, Jiading District, Jinshan District, Qingpu District, Songjiang District, Fengxian District, Pudong District, Chongming District, Minhang District, and Jingan District).Each sample was collected from approximately 1 acre and was taken at a sample weight of not less than 3 kg, whereas larger species like cabbage and cauliflower should satisfy at least four plants and a minimum of 3 kg simultaneously.The collected samples were homogeneously mixed and reduced by the method of coning and quartering, before treatment with a homogeniser to take 250-500 g, according to the Agricultural Industry Standard of the People's Republic of China (2004), which were stored, sealed, and uniquely numbered.

Sample preparation
The homogenised sample (10 ± 0.02 g) was placed in a 50 mL polypropylene centrifuge tube, and 10 mL acetonitrile was added and vortexed for 20 min.After that, 5 g NaCl was added to the centrifuge tube and shaken vigorously for 1 min and centrifuged at 5000 rpm for 5 min.After centrifugation, 1 mL of supernatant was mixed with 1 mL ultrapure water and filtered through a 0.22 µm filter membrane to be ready for injection in the UPLC-Q/Orbitrap-HRMS system.

UPLC conditions
The chromatographic column was a Thermo Accucore TM aQ column (150 × 2.1 mm, 2.6 µm) and an Accucore aQ guard column (10 × 2.1 mm, 2.6 µm), obtained from Thermo Fisher Scientific (Waltham, MA, USA).Mobile phase A was 0.1% formic acid in water (containing 5 mmol/L ammonium formate and 2% methanol) and mobile phase B was 0.1% formic acid in methanol (containing 5 mmol/L ammonium formate and 2% water).Formic acid could improve ionisation efficiency and further increase the sensitivity of the analytes.The addition of ammonium formate to the mobile phase was motivated by the fact that the buffer facilitated the retention and separation of most pesticides (Wang et al. 2019).The flow rate was 0.4 mL/min, the injection volume for each sample was 5 µL, and the column temperature was 25°C.Mobile phase B ramped up linearly from 0% to 20% in 4 min and rose to 40% at 5.5 min, after which mobile phase B reached 100% at 10.5 min and was maintained for 2.4 min and then mobile phase B was restored to the initial ratio and reequilibrated for 5 min at 2.1 min.

Q/Orbitrap-HRMS conditions
A heated electrospray ion (H-ESI) source was applied in positive and negative ion acquisition mode at +3.2 kV and −2.8 kV, with sheath gas of 40 arb, auxiliary gas of 10 arb, purge gas of 1 arb, auxiliary gas heating temperature of 350°C, and ion transport tube temperature of 325°C.Full scan/data dependent-MS 2 (dd-MS 2 ) was applied, with a mass range of m/z 100-1000 and a resolution of 70,000 FWHM for the primary full scan and 17, 500 FWHM for the secondary scan (dd-MS 2 ).The top three strongest ions were collected in a single pass, and the collision energy in the C-trap was normalised using a gradient of 20, 40, and 60 eV.

Quality control and quality assurance
Reagent blank and blank control samples were analysed to control the sample pre-treatment procedure.When testing samples, the 20 µg/kg standard solution was injected once every 15 samples.The response values and retention times of this solution were compared with each other and if the deviation between the intensity responses and retention times were ±5% and ±0.25 min, respectively, the instrument quality control was regarded to be compliant with the guideline.Data were collected by selecting precursor ions with aspecific m/z or that meet defined criteria (full MS intensity, dynamic rejection 3-8 s and only single charge ions) and separating them for MS/MS scanning, also known as data-dependent acquisition (DDA).Precursor ions for over 800 compounds were written into the DDA inclusion list to accelerate non-targeted screening.Complete non-target screening included comparison of fragment, RT, and ion ratios with known databases and comparison of predicted suspects with online high-resolution MS/MS spectra, known as tentative confirmation.Confirmation and quantification of suspects were re-verified with purchased standards, followed by refinement of chromatographic and fragmentation information in the database.The database information and the principles of screening and quantification are referred to in previous studies (Si et al. 2021) and 158 new pesticides were added to the previously established database of 392 pesticides, creating a total database information of 550 pesticides.

Method validation
The method was validated according to European Commission guideline SANTE/11312/2021 (SANTE 2021).Method validation of the developed method was carried out for specific validation parameters including linearity, limit of detection (LOD), limit of quantification (LOQ), accuracy, and intraday precision.The instrument was calibrated daily prior to linearity analysis to ensure correct quantification.The linearity of all target pesticides was verified by adding pesticide standards at different concentrations (0.001 mg/kg-0.100mg/kg) to blank matrix.LOD and LOQ were calculated as the signal-tonoise ratio (S/N) for 3× and 10× quantitative ion pairs, respectively.The accuracy and precision of the method were assessed by spiked recovery experiments, and six repetitions were carried out at spiked levels of 0.010 mg/ kg and 0.100 mg/kg, respectively.Recoveries and relative standard deviations (RSDs) of the spiked recoveries were obtained by applying the previous method.Chromatograms of reagent blank, blank control samples, and samples at LOQ levels were analysed to assess the specificity of the method.Measurement uncertainty assessment is described in detail in Section 1 of the Supporting Materials.The measurement uncertainties for the 54 pesticides ranged from 0.11 to 0.19 and the results are presented in Table S1.

Dietary risk assessments
Dietary exposures and risk quotient values for pesticide residues in food for humans are used primarily to assess possible exposure pathways and measurement levels and to specify actual versus expected exposure dose levels and sensitive populations that may be at risk.Health risk assessment is carried out through the Hazard Quotient (HQ) and Hazard Index (HI) (Wu et al. 2017;Park et al. 2022).The chronic dietary risk entropy number (HQc) is calculated from the estimated daily intake (EDI, mg/kg bw) and the acceptable daily intake (ADI, mg/kg bw).The relevant equations are EDI = C i × F/bw and HQc = EDI/ ADI, where C i (mg/kg) is the average residue concentration of the pesticide detected, F (kg/d) is the average consumption of vegetables and bw (kg) is the average weight for different age groups (Liu et al. 2020).ADI values were taken from the Joint Meeting of the Food and Agriculture Organization (FAO) and the World Health Organization (WHO) on Pesticide Residues (JMPR) document (WHO 2020).When HQ is below 1 the risk is considered acceptable and does not pose a health risk on the long term.The higher the HQ value, the greater the exposure risk (Kumari and John 2019;Wu et al. 2021).The cumulative effect of pesticide residues is assessed by calculating the hazard index (HI), using the formula HI=∑HQ.The risk is considered acceptable when HI < 1 and does not pose a health risk in the long term and an unacceptable risk when HI > 1.

Risk ranking system
The veterinary drug residue risk ranking matrix was employed to rank residual risks (Veterinary Drug Residue Commission 2019).The values for each index are listed in Table 1.According to the LD 50 , pesticides were classified into four classes, defined as extremely high toxicity, high toxicity, mild, and low toxicity.The LD 50 values of the pesticides were sourced from a JMPR report (FAO 2016).The scores are a combination of toxicity scores and exposure scores.Toxicity scores include separate scores for A and B, while exposure scores include four separate scores for C, D, E, and F. The definitions and scores for categories A-F are shown in Table 1.
The frequency of dose of pesticides (FOD) was calculated according to FOD = T P �100, where T is the frequency of insecticide use during planting and P is the plant growth cycle (days).The residual risk score (S) of each pesticide in the sample is calculated by S = (A+B) × (C+D+E+F), where S is the pesticide residue risk score, A is the toxicity score, B is the pesticide efficacy score, C is the proportion of vegetables in the diet score, D is the frequency of pesticide use during cultivation, E is the number of highly exposed people score, and F is the residue level score.The higher the mean of the pesticide residue risk score, the higher the risk.

Method validation
Calibration curves showed satisfactory linearity with coefficients of determination (R 2 ) between 0.9877 and 0.9997.LODs and LOQs for the method ranged from 1 to 5 µg/kg and 2 to 15 µg/kg, respectively.The accuracy of the analytical method was assessed by adding standards of the pesticides to blank matrix for recovery experiments.The recoveries shown in Table S1 were calculated by solvent standard curves.Most pesticides have recoveries between 70% and 120% and intraday RSDs of less than 20%, which is acceptable.In addition to the recovery of 37.8% for propamocarb and 1.4% for RSD at the 0.010 mg/kg level, 123% for bifenthrin and 14.2% for RSD and 61.8% for nitenpyram and 2.0% for RSD and the recovery of Propamocarb at the 0.100 mg/kg level was 47.6% and RSD was 3.6%.For exceptions where the required recoveries (70-120%) were not achieved, additional matrix-matched calibration with a concentration close to the quantification result of the solvent standard curve was performed to correct the results.The reported contents of propamocarb, bifenthrin, and nitenpyram have been corrected by matrix matching standards.The recoveries of propamocarb, bifenthrin, and nitenpyram were 72.6%, 77.0%, 71.2% at 0.010 mg/kg level, respectively, and the recovery of propamocarb at 0.100 mg/kg level was 78.4%.All together the method provided the required performance in terms of linearity, limit of detection (LOD), limit of quantification (LOQ), accuracy, precision, and specificity, as in accordance with the guideline (SANTE 2021).HRMS produced accurate masses of MS and MS2 with both high selectivity and specificity.Even if precursor ions of close mass are separated at the same time, different MS2 can accurately exclude false positives.In this study, the responses around the corresponding retention times of pesticides in both blank control samples and reagent blanks were below 2e 5 and most of the full MS channels showed no response, which meets the SANTE criteria (2021).
Of the 54 pesticide residues detected, for 38 pesticides, MRLs were established in Chinese regulations.However, still for 15 pesticides (Rotenone, Dichlorobenzamide, Spinetoram L, Oxadixyl, Lufenuron, Triadimefon, Atrazine, Fenobucarb, Fluroxypyr, Haloxyfop-methyl, Paclobutrazol, Prometryn, Fenbuconazole, Nitenpyram, Buprofezin) no MRLs in leafy vegetables have been published (Table 2).Among the 54 pesticides found, acetamiprid and cypermethrin were detected at values higher than the MRLs of the Chinese national standards.Four samples contained acetamiprid and one sample contained cypermethrin above the respective MRLs, while residues in the remaining samples were all below the MRLs.
The distribution of multi-pesticide residues is shown in Figure 3.In the 704 samples, 402 (57%) contained no residues, 167 (24%) contained 1 pesticide, 50 (7%) contained 2 pesticides, 28 (4%) contained 3 pesticides, 19 (3%) contained 4 pesticides and 38 samples (5%) contained more than 5 pesticides.Multiple pesticides were detected in one single sample, which could be due to use of multiple pesticides in the production process or inclusion of multiple pesticide ingredients in one recipe (Wang et al. 2022).

Dietary chronic risk assessment
Chronic risk was assessed by estimating dietary exposure to pesticides in food and comparing the exposure estimates to ADI values to estimate HQ (Park et al. 2016).
The results for long-term intake and chronic exposure risk are presented in Table 3.For chronic risk assessment, all HI values were less than 1 in different age and gender groups, indicating that the concentration of pesticide residues in leafy vegetables is acceptable in Shanghai.It is notable that the highest HI values were found in the group of children aged 2-7 years.In the group of children aged 2-7 years, the risk values for cypermethrin, indoxacarb, and bifenthrin were high in the green vegetables with HQ values of 0.211, 0.106, and 0.102, respectively, all greater than 0.10.The risk value for indoxacarb in the Hanging cabbage sample was higher with an HQ of 0.117.In the amaranth samples, the risk values for cypermethrin and carbendazim were higher with HQs of 0.111 and 0.109, respectively.These high HQ pairs constituted a major influence on the high HI values of the group.
For each vegetable, the long-term dietary intake of HI ranged from 0.511-0.924for green vegetables, 0.467-0.845for Chinese cabbage, and 0.466-0.843for amaranth, respectively.Among the different populations in China, the risk of long-term dietary exposure to pesticides is in the order of children aged 2-7 years > adolescents aged 8-12 years > females aged 13-19 years > females aged 20-50 years > females aged 65+ years > females aged 51-65 years > males aged 20-50 years > males aged 51-65 years > males aged 13-19 years > males aged 65+ years.For adults, women have higher HI values than men, demonstrating that women are at greater risk of exposure to pesticides relative to men.

Risk score for the detected pesticides
According to the national standards for the rational use of pesticides, each pesticide should be used a maximum of 3 times on these vegetables.The growth cycle of leafy vegetables is 25-45 days and the frequency of dosing is calculated according to Equation (4), which shows that the frequency of dosing of each pesticide is between 2.5% and 20%, therefore D scores of 1.Although there are differences in the frequency of vegetable consumption among different populations in China, there are no data on the density of highly exposed populations.According to Table 1, the score for the highly exposed population (E) is equal to 3 (Wang et al. 2022).The risk scores for pesticide residues in vegetables are listed in Table S2.According to the risk score of pesticide residues in vegetables, pesticides are classified into three categories: high risk pesticides (S ≥ 20), medium risk pesticides (15 ≤ S < 20), and low risk pesticides (S < 15).Among the pesticides detected, there were 9 high risk pesticides, 2 medium risk pesticides, and 43 low risk pesticides, respectively.Of the pesticides with high detection rates, rotenone is a high-risk pesticide (S = 36) and acetamiprid, dimethomorph, indoxacarb, and propamocarb are low risk pesticides.For high risk pesticides, the use of these pesticides in vegetable production should be minimised.
Among high risk pesticides, rotenone, dichlorobenzamide, and spinetoram L have high risk values, but the lack of corresponding ADIs when conducting chronic risk assessments had made it impossible to calculate HQs.Tolfenpyrad, cypermethrin, imidacloprid, and cyhalothrin had the next highest risk values, they were not detected in vegetables at high rates, and the corresponding HQ values were low in the chronic risk assessment except for cypermethrin; however, the chronic risk of these pesticides is acceptable.The results show that the overall risk score does not always coincide with the detection rate and chronic risk, which means that the overall risk of a pesticide is a combination of all factors.
From the point of view of human health issues and pesticide control measures, monitoring of fruits and vegetables is carried out in many countries (Jardim and Caldas 2012;Bakırcı et al. 2014;Li et al. 2015;Ong-Artborirak et al. 2017;Khoshnam et al. 2022) and monitoring of pesticides in other foods has also been conducted, such as in rice in southern and southwestern China by Chen et al. (2023).In our study, the more frequently found pesticide residues were rotenone, acetamiprid, dimethomorph, indoxacarb, and propamocarb, whereas Algharibeh and AlFararjeh (2019) and Luo et al. (2023) in their risk assessment of pesticide residues in vegetables and fruits concluded that acetamiprid was the most detected pesticide and assessed whether this pesticide residue posed a risk to human health.In general, the risk indices for leafy vegetables produced in Shanghai were low and the health risks to children and adults from the detected pesticides were within acceptable limits.However, this does not mean that the overall health risk is acceptable.For example, Lv et al. (2022) found that fruits and vegetables containing carbofuran did not pose chronic health hazards to children and adults, but some fruits and vegetables containing carbofuran had unacceptable acute risks.Moreover, in comparison, the hazard index for children aged 2-7 years was higher, especially for cypermethrin, bifenthrin, and indoxacarb, which requires more attention, as also concluded by Omwenga et al. (2021) for organophosphate and carbamate pesticide residues in commonly consumed vegetables in Kenya.Li et al. (2015) found bifenthrin to be a potential acute health hazard for children in the risk assessment of fruits.In  contrast, Szpyrka et al. (2015) showed that pesticide residues in vegetables and fruit did not pose a health risk to adults and children.The results of their study also pointed to a greater risk of pesticide exposure in women than in men, which is consistent with a previous study by Zhang and Si et al. (2021).
were moderate and low risk pesticides.The pesticide residues detected in Green vegetables, Chinese cabbage, and amaranth were assessed to pose no serious health risk to consumers.The results of our study provide a reference for future regulation and risk assessment of pesticides in vegetables to better protect consumer health.

Figure 1 .
Figure 1.Regional distribution of sampling in Shanghai, China.

Figure 2 .
Figure 2. Detection rates of pesticide residues in vegetables (a) and the number of pesticides detected in different vegetables (b).

Figure 3 .
Figure 3.The distribution of multi pesticide residues.

Table 1 .
Definition and score of A-F indices for risk scoring.
Nd: No evidence of detectable residues.

Table 2 .
Detection of pesticide residues in vegetable samples.

Table 3 .
Results of the long-term intake and chronic exposure risk.