Investigating a human pesticide intoxication incident: The importance of robust analytical approaches

A human intoxication incident attributed to pesticide abuse was investigated using cutting-edge analytical methodologies. An LC-ESI-MS/MS method, based on a hybrid solid-phase extraction protocol (hybrid-SPE), was applied for the detection and quantification of several pesticides and metabolites in human biological fluids. Concomitantly, an UHPLC-HRMS method was applied to investigate potential metabolites, assisted by a complementary GC-MS method to elucidate the presence of plausible pesticides co-formulants. The LC-ESIMS/MS method exhibited acceptable mean recoveries at the lower limit of quantification (LLOQ) and three additional levels, varying from 85 to 106% for all analytes and matrices. In serum, urine, and gastric fluid samples, the suspect compounds, namely chlorpyrifos and myclobutanil, predominated. Gastric fluid samples contained the highest concentrations of chlorpyrifos (39,800 ng/mL) and myclobutanil (18,800 ng/mL), while the neonicotinoid imidacloprid was also quantified, below 30 ng/mL. Notwithstanding, the UHPLC-HRMS analysis unveiled several metabolites of chlorpyrifos and myclobutanil. In parallel, GC-MS analysis, corroborated the presence of several co-formulants in gastric fluid samples, exemplified by mand o-xylene, and cyclohexanone. Overall, three analytical methods were implemented to elucidate the chemical causality of a human intoxication incident. The presence of suspected active substances, one additional, and several metabolites and co-formulants were documented.


Introduction
The pesticide intoxication incidents are reported worldwide and hold a significant share among human poisoning cases, especially in developing countries [1]. It is estimated that over 150,000 people decease each year due to pesticide intoxication, despite the significant efforts toward pesticide restriction policies related to poisoning [2,3]. Suicide attempts usually by ingestion of pesticide formulations are observed, in particular, in rural communities. The main route of accidental or intentional acute pesticide poisoning is through ingestion and secondarily by inhalation of vapors. Poisoning incidents through dermal exposure to pesticides, including those with a fatal outcome, have been reported [4], but their contribution to the overall percentage is by far lower and usually occur when personal protective equipment is not at all or properly used by those involved in agricultural activities [5]. In addition, intoxication via contaminated food consumption should not be neglected [6].
Currently, exploitation of diagnostic biomarkers such as chemical markers in biofluids is sporadic leading to a knowledge gap on the record of human pesticides intoxication cases, thereby diminishing their importance and projection. If this knowledge gap was to be covered it could be considered as the scientific basis to minimize and prevent poisoning incidents through raising awareness activities, especially in rural communities and triggering the need for development of appropriate legislation aimed at restricting access of the general public to highly hazardous pesticides [7].
The European Commission has published a notice on the guidance for monitoring and surveying the impact of pesticide use on human health and the environment, which designates the necessity of robust biomonitoring data [8] covering aspects from human cohort studies to isolated intoxication incidents. Therefore, scientific data derived from incident reports, which not only include the description of the case, but also chemical analytical data, are of utmost importance. Among the pesticide superfamily, organophosphorus (OP) compounds are the predominant culprits behind most of these intoxications, even though the last decade measures have been taken to substitute highly toxic OPs with safer ones. Despite the efforts, banned pesticides (including some OPs) are still reported in European poison control centres [9,10].
Considering the extensive and sometimes rapid in vivo metabolism that several active substances undergo, and the toxic profile of metabolic products, it is imperative to include both the parent compounds and the potential metabolites in the analytical methods. Nevertheless, and since it is practically impossible to acquire all analytical standards even if they are available, the non-targeted analytical methods and metabolomics parallel elucidation of xenobiotics and metabolites have emerged as a pivotal tool in the investigation of such incidents, and the overall metabolic profiling of biological fluids [11,12]. Last but not least, the contingent presence of non-active ingredients of pesticide formulations, namely co-formulants, such as solvents, stabilizers, etc., in biological fluids also needs to be investigated. The latter stems from reports that documented an irritative potential or potentiation of toxicity of pesticides, such as OPs, or the known herbicide glyphosate in the presence of such chemicals [13][14][15].
In this work, we present the investigation of a pesticide intoxication incident, in which an 86-year-old female was hospitalized after suspected use of pesticides' formulations. A modern and universal sample preparation coupled with targeted LC-ESI-MS/MS and nontargeted UHPLC-HRMS analyses was implemented, managing to detect and quantify the alleged active substances and identify their metabolites in biological fluids. Finally, GC-MS analysis disclosed the presence of several coformulants.

Materials and methods
An 86-year-old female was admitted to the hospital on day 1 (16 Oct 2019) after reportedly receiving organophosphates (chlorpyrifos) and an additional formulation with potential active ingredient myclobutanil (triazole fungicide). She had an altered level of consciousness and was gasping for air. She was intubated at the emergency department, and mechanical ventilation was initiated. Soon after the intubation, the patient went into cardiac arrest and was resuscitated immediately. After receiving vasopressors and pralidoxime, she was then transferred to the intensive care unit (ICU). On day 2, she received charcoal. The patient demonstrated a gradual improvement in her level of consciousness and her muscular strength. She did not demonstrate any evidence of miosis, bradycardia, salivation, nor did she need any atropine. Based on the above observation and after contacting poison control center, the infusion of pralidoxime was discontinued on day 3. In the evening of day 4, the patient demonstrated agitation and the next morning she had a sudden change in her level of consciousness along with bradycardia and evidence of miosis on physical examination. She gradually received 6 amps of intravenous atropine without a response, and the pralidoxime drip was restarted. This resulted in an improvement in her clinical condition later at the same night. Finally, at day 15, she was discharged upon further improvement.
Standard solutions were prepared in acetonitrile and methanol in volumetric flasks and kept at −18°C away from the light. Acetonitrile, methanol, ammonium formate, and formic acid of LC-MS grade were obtained from Fisher Scientific (Hampton, NH, USA). Dichloromethane (DCM) was of pesticide residues grade and was obtained from Sigma Aldrich (Seelze, Germany). Sodium sulfate anhydrous (Na 2 SO 4 ), and sulfuric acid (purity >95%) were obtained from Fisher Scientific. Hybrid SPE-phospholipid cartridges (30 mg, 1 mL) were purchased from Supelco (Bellefonte, PA, USA). All reagents and chemicals were of analytical grade.

Samples
Serum and urine used as control samples were obtained from the stock reserved at Benaki Phytopathological Institute for biomonitoring studies performed in-house. With regard to real samples analyzed in this study, urine samples collected at day 1 (no 1) and day 7 (no 2), blood and gastric fluid samples (before (no 1) and after (no 2) gastric lavage) that had been collected during day 1 (all samples were kept in freezer) were sent to the analytical laboratory at day 7.

Liquid chromatography 2.3.1 Sample preparation
Sample preparation was envisaged by the work of Honda et al. [16]. Briefly, serum, urine, and gastric fluid (after being stored at −20°C) were thawed at room temperature. Then, 0.25 mL of the fluid was transferred to an Eppendorf tube, diluted with 0.75 mL of MeOH containing 1% ammonium formate (w/v), the mixture was hand-shaken for 10 s, and centrifuged at 4,500 rpm for 5 min. Consequently, the supernatant was loaded on a pre-washed (with 1.5 mL of MeOH containing 1% ammonium formate (w/v)) Hybrid-SPE cartridge. The eluate was collected, filtered, and directly injected into the LC-ESI-MS/MS and UHPLC-HRMS systems.

Gas chromatography 2.4.1 Sample preparation
Sample preparation was based on a liquid-liquid extraction (LLE) bibliographic protocol [17] with minor amendments, using DCM as extraction solvent. Briefly, 1 mL of the biological fluid was mixed with 1 mL sulfuric acid (1 M) and 0.1 g of Na 2 SO 4 . The mixture was vortexed for 1.5 min (complete salt dissolution was observed). Then, 4 mL of DCM was added, and the mixture was agitated for 2 min (with a magnetic stirrer) and vortexed for an additional 1 min. Consequently, the sample was centrifuged for 4 min at 4,000 rpm, the DCM layer was collected, dried over Na 2 SO 4 , and a portion of it (3 mL) was evaporated to dryness under a gentle stream of nitrogen. The dry residue was reconstituted in 30 μL of DCM, filtered, and then 1 μL injected to the GC-MS system.

Chromatographic conditions 2.5.1 LC-ESI-MS/MS system and operating conditions
Same chromatographic and operating conditions (see Supplementary material) with previous work of our group were encompassed [18]. The resolution for full-scan analysis was set on 70,000, whereas for the data-dependent acquisition mode, the resolution was 35,000, allowing for MS/MS fragmentation of the three most intense ions. Stepped normalized collision energy was set at 35, 60, and 100. The column temperature was kept at 40°C, while the sample tray temperature was set at 4°C. Data analysis was performed using Compound Discoverer 2.1 software (Thermo Fisher Scientific™ -San Jose, CA, USA). The detection of the metabolites was in accordance with the data from literature applying m/z tolerance of 5 ppm and included the composition rules of H/C, NOPS/ C, RDBE, and the isotopic pattern.

Gas chromatography-mass spectrometry analysis and operating conditions
The GC-MS analysis was performed on a Chromtech Evolution MS/MS triple quadrupole mass spectrometer built on an Agilent 5975 B inert XL EI/CI MSD system, which was operated in full scan data acquisition mode. Samples were injected (injection volume was 1 μL) with a Gerstel MPS-2 autosampler using a 10-μL syringe. Separations were performed on an HP-5 ms UI, length 30 m, ID 0.25 mm, film thickness 0.25 μm (J&W Folsom, USA). Helium (99.9999% purity) was used as the carrier gas at a flow rate of 1.2 mL min −1 . The column oven temperature program initiated from 45°C, staying for 1 min, increased to 250°C at a rate of 5°C min −1 where it stayed for 5 min. The transfer line, manifold, and source of ionization temperatures were 300, 40, and 230°C, respectively. The electron multiplier voltage was set at 2,000 V. The total GC analysis was 47 min. Identified peaks in GC-MS (triplicate analysis) were verified by matching the acquired mass spectra with those in the commercial library of NIST 08.

Analytical method validation
Validation of the analytical method was based on the bioanalytical and forensic toxicology validation guideline documents [19,20]. SANTE guideline was also taken into account [21]. Validation parameters considered were linearity, sensitivity (lower limit of quantitation, LLOQ), selectivity, accuracy (recovery evaluation), precision, and matrix effect (ME). Linear range varied from 5 to 1,000 ng/mL containing at least six calibration levels, using internal standard calibration (preferably matrix matched calibrators). Internal standards use also corrected recovery and assisted the accurate quantification of some of the analytes. LLOQ was determined by injecting three replicates (three times each) of the lowest (non-zero) calibrator (5 ng/mL), evaluating the concordance with the detection/identification criteria. Selectivity was assessed by analyzing consecutive blank samples of biofluids, evaluating respective responses in the retention times of the analytes. LLOQ coincides in this study with the decision point concentration (for reporting positive analytes) of 5 ng/mL, verified in this study. Recovery was assessed at four concentration levels (1, 5, 50, and 200 LLOQ), using five independent replicates per level. Up to three MRM transitions were monitored for the active substances. Since blank (human) gastric fluid sample was not available (it required consent for gastroendoscopic examination from volunteers or patients, not pursued at the time of the incident), results were extrapolated using urine or serum matrix-matched standard solutions. In order to confirm suitability of this approach, an artificial mixture of hydrochloric acid (aqueous solution), potassium chloride, and sodium chloride (the main components of gastric fluid) was prepared to mimic this fluid. Spiking this solution, subsequent extraction and analysis, did not exhibit substantial differences (in terms of analytical figures of merit) in comparison to the spiked urine or serum samples. Lastly, considering that the objective of this work is to present the findings of the specific incident, no data are presented for the rest of the active substances validated with this methodology (more than 70 substances and metabolites) of the analytical methods (will be presented in a forthcoming work of our group).
Ethical approval: The presented investigations were carried out following the ethics rules of the Declaration of Helsinki of 1975, and later amendments, as revised in 2013. More specifically, the ethical committee of Benaki Phytopathological Institute approved the sample preparation and subsequent chemical analysis of the mentioned biological fluids.

Results and discussion
The analytical method validation criteria fulfilled requirements of the respective guidelines. More specifically, for the tested biological fluids, linearity was acceptable with the lowest calibrator demonstrating concentrations <±15% of the nominal concentrations (and correlation coefficients values, r > 0.9989 for all analytes derived from matrix-matched calibration curves). The LLOQs were established for chlorpyrifos, myclobutanil, and imidacloprid at 5 ng/mL. Trueness was assessed by the recovery study, verifying sufficient extraction of analytes from all types of biological fluids (recoveries were in the range of 85-106%, at all studied concentrations, RSD% 6-9.2%, see Table S1). Precision (inter and intraday, Table S1) was also acceptable displaying RDS% values up to 10%. For chlorpyrifos and myclobutanil, ME varied from −2.7 to 7.9%, respectively, since no significant interfering peaks were obtained, demonstrating the acceptable selectivity of the method. For imidacloprid, ME fluctuated from 8.3 to 14.8%. Carryover was far below the threshold value of 20% with regard to the LLOQ. Consequently, the implementation of this analytical method in the gastric fluid, serum, and urine samples of the patient hospitalized revealed mainly the presence of chlorpyrifos and myclobutanil, yet, in far lower magnitude imidacloprid (see Table 1 for analytical results, and Figure S1a-d for MRM transitions in real samples). The results designate that the female patient was exposed to high amounts of both active substances, also considering their biological half-life (t 1/2 ). The half-life for chlorpyrifos is 8.15 h, and for myclobutanil, it ranges from 0.6 to 0.9 for low dose and to 30.1 h for high dose [22,23].
Marques et al. reported blood chlorpyrifos levels in a poisoned individual at approximately 52 ng/mL [24]. In a 1993 report regarding chlorpyrifos poisoned persons, the range of serum concentrations varied from 52.6 to 2485.7 ng/mL, which designate that the serum finding of the present study falls within this range [25]. Higher concentrations in serum were reported by a Spanish group in 2004 (at 5,400 ng/mL); yet, in the same work, the gastric fluid concentration was determined at 9,400 ng/mL [26], 3-4 times lower than that in the present case report. Myclobutanil was detected in two serum samples of orchard workers (previous work of our group) at much lower levels [27], although the two cases are not comparable. To emphasize on the severity of this incident, the total amount of chlorpyrifos in gastric fluid is estimated taking into account the highest concentration of chlorpyrifos analytically determined in gastric fluid as worst case (i.e. 39,800 ng/mL), the weight of the patient (70 kg), and the human adult production of 1,500 mL gastric fluid on a single day [28]. The total amount of chlorpyrifos in gastric fluid was estimated to be at least 852.9 μg kg −1 bw, which is 170.6 times higher than the acute reference dose (ARfD) of 5 μg kg −1 bw for chlorpyrifos [29]. For myclobutanil, the respective calculations resulted in exceedance of the ARfD of myclobutanil by 1.3 times [30]. It is acknowledged that this is a rough estimation since the amount detected in gastric fluid is expected to be lower than the actual ingested amount considering that both chlorpyrifos [31] and myclobutanil [30] are readily bioavailable due to rapid and complete absorption via the oral route.
The prevalence of chlorpyrifos and myclobutanil metabolites was assessed using UHPLC-HRMS analysis. For chlorpyrifos, 3,5,6-trichloro-2-pyridinol (TCPy), the principal metabolite found in the plasma, and chlorpyrifos oxon (diethyl(3,5,6-trichloropyridin-2-yl) phosphate)), both were annotated in gastric fluid samples (see Figures 1 and 2), and in the serum sample with mass error values (D (ppm)) below 0.6 ppm. Chlorpyrifos oxon detection is another indication of the high concentration of the parent compound, taking into account its fast in vivo conversion to TCPγ. The dichloro metabolite (3,6-dichloro-2-pyridinol) was annotated in one gastric fluid sample (no 2). Chlorpyrifos-related dialkyl phosphates (DAPs) and the hydrolysis products such as diethyl phosphate (DEP) and diethylthiophosphate (DETP) were also annotated in both the samples of gastric fluid (see indicative LC-HRMS chromatogram, Figure S3 exhibiting D (ppm) far below the 5 ppm threshold (−0.65 to 0 ppm)). Both metabolites though are usually targeted in urine samples are reported in other biofluids as well (such as blood serum) [25,32]. On the whole, these outcomes verify the in vivo degradation of chlorpyrifos to several of its metabolites. Among them, the activated desulfurated oxon counterpart produced in the liver via cytochrome P450 is a potent acetylcholinesterase (AChE) inhibitor, since 99% of it binds irreversibly to blood AChE [33,34].
The other DAPs, such as dimethyl phosphate (DMP) and dimethyl thiophosphate (DMTP), were also annotated, but are not specific, and can be related to other OPs as well. Such findings might also be associated with dietary or environmental uptake related active substances.
Diethyl dithiophosphate (DEDTP) was annotated in one urine and one blood serum sample. In the same samples, both DMP and DEP were additionally detected. For myclobutanil, three metabolites were putatively identified, which are hydroxy (or dihydroxy) and keto derivatives of the parent compound (see Figure 3 for one of them).
Interestingly, both targeted and untargeted analyses (LC-HRMS chromatogram, Figure S2 Table 2. Since pesticides' formulations contain co-formulants (i.e., solvents, surfactants, emulsifiers, inert materials) and adjuvants (such as polyethoxylated alkyl amines (POEA)), GC-MS analysis was implemented to possibly unveil such chemicals. Ignorance of them and their potential toxicity [14] can lead to falsification with regard to the safety of the commercial pesticide [35], and subsequently to the partial assessment of intoxication incidents, where pesticides formulations are implicated.
In this direction, such constituents were identified in gastric fluid samples (see representative GC-MS full scan TIC magnified chromatogram in Figure 4) after successful application of an LLE protocol and exemplary depicted in Table 3 (for gastric fluid sample no. 1). The most abundant components were mand o-xylene, and cyclohexanone. Quantification of the co-formulants was out of the scope of the presented study.
Among the GC-MS revealed chemicals, cyclohexanone was reported to play significant role in the potentiation of toxicity of another key OP, dimethoate, in a minipig model, after oral administration of an emulsifiable concentrate (EC) formulation [13]. Hence, the detection of cyclohexanone in the gastric fluid DCM extract (as one of the major components) can possibly lead to a similar hypothesis concerning the potentiation of chlorpyrifos toxicity. Similar conclusion can be drawn for mand o-xylene, since studies have shown xyleneinduced toxicity in several organisms (indicatively see ref. [36]).
Ethylbenzene was also retrieved at lower relative amounts. No hazard is identified for the general population exposed to ethylbenzene via the oral route.
The substance has been assessed in the frames of the REACH (Regulation (EC) 1907/2006 for Registration, Evaluation, Authorisation and Restriction of Chemicals) and it is registered for use up to 1,00,00,000 tonnes per annum [37]. In the REACH dossier, hearing loss has not been considered a specific effect for ethylbenzene but has also been found with other aromatic solvents such as styrene, toluene, or xylenes. According to Gagnaire et al., ethylbenzene led to maximal hearing impairment after about 4 weeks without further deterioration by prolongation of exposure [38].
For other benzene derivatives, such as 1,2,4-trimethylbenzene, there is EU-harmonized classification for serious eye irritation, skin, and respiratory irritation, and is harmful if inhaled [39]. An EU indicative occupational exposure limit value (OELV) of 20 ppm for a reference exposure period of 8-hours' time weighted average (TWA) via the inhalation route has collectively been set for trimethylbenzenes [40], while the need for more specific national values is recognized. Chlorpyrifos is a controversial compound, including its degradation products [41]. In July 2019, the European Food Safety Authority (EFSA) was mandated by the European Commission to provide a statement on the available outcomes of the human health risk assessment in the context of the peer review of chlorpyrifos [42].

Imidacloprid
Briefly, EFSA concluded that chlorpyrifos is toxic if swallowed and harmful in contact with skin. Inhibition of AChE was the critical effect after repeated oral administration of chlorpyrifos in experimental animals resulting in cholinergic overstimulation. Significant uncertainties were linked to neurodevelopment, since decrease in cerebellum height (corrected by brain weight) was observed in pups from the lowest dose of 0.3 mg/kg b.w./day in a developmental neurotoxicity study in rats. EFSA considered that these concerns were supported by available epidemiological evidence related to the development of neurological outcomes in children [43][44][45][46]. There was no evidence of carcinogenicity, but the genotoxic potential of chlorpyrifos was unclear. Concerns were raised on the possibility of chlorpyrifos to induce chromosomal aberrations and DNA damage through oxidative stress or topoisomerase II inhibition; the latter was considered a molecular initiating event for infant leukemia based on the previous work of the EFSA Panel on Plant Protection Products and their Residues [47]. This mode of action is not supported by more recent research funded by EFSA [48] but more research would be needed before a definite conclusion can be drawn. Concerning myclobutanil, its acute toxicity to mammals is far lower compared to chlorpyrifos, exhibiting an LD50 oral at 1,600 mg kg −1 in rats [49] (respective endpoint value for chlorpyrifos is 66 mg kg −1 [31]).
It is noted, however, that toxic effects after exposure to a mixture of pesticides may be influenced by interactions between individual components of the mixture.  Figure 4. b RI, retention index on HP5-MS UI column (relative to n-alkanes), identification based on mass spectra comparison with the reference databases, and comparison with literature RIs (depicted in parentheses). The mechanism behind these interactions may be linked to the toxicokinetic and/or toxicodynamic properties of each component. At doses exceeding dietary-relevant levels, as it is the case with this study of pesticide abuse, synergism may also occur. Compounds such as OPs and triazoles display synergistic effects in a plethora of organisms [50]. The azole moiety is known to inhibit several CYP450s isoforms, yet, its effect is diverse, depending on the insecticide with which it interacts. For example, with pyrethroids, the toxicity is often enhanced, yet with compounds (such as organophosphorothioates) whose oxon derivatives are the toxic counterparts, the triazoles, through the inhibition of CYP450 enzymes, block transformation to oxon derivatives, leading to a decrease in toxicity [50].

Conclusions
A developed and validated LC-ESI-MS/MS analytical method was applied for the detection and quantification of suspect pesticides in biological fluids collected from a documented pesticide intoxication incident from a Greek hospital. Significant levels of chlorpyrifos and myclobutanil were determined in gastric fluid samples, and in the serum sample, verifying the severity of the incident. UHPLC-HRMS unveiled the presence of key metabolites of both suspect active substances, which were more prolific in the gastric fluids, where the elevated concentrations were observed. The neonicotinoid imidacloprid was also identified in the gastric fluids. Lastly, GC-MS analysis unveiled the presence of several co-formulants in gastric fluids. Next steps of the herein presented analytical approaches include the incorporation in the LC-ESI-MS/ MS, and UHPLC-HRMS methods of dialkylphosphates, and other metabolites not currently enrolled in the analytical scope, and their subsequent quantification, which is further important from a risk assessment perspective.