Exposure to pesticides and oxidative stress in Brazilian agricultural communities

Abstract

Background:

Oxidative stress may be an important mechanism linking pesticide exposure to several diseases. We evaluated the association between pesticide exposures and oxidative stress biomarkers in Brazilian agricultural workers.

Methods:

A cross-sectional study was conducted in healthy agricultural (N=52) and non-agricultural workers (N=68) in a rural community in Rio de Janeiro State, Brazil. Regular pesticide use, sociodemographic, and lifestyle information was obtained by questionnaire. Oxidative stress biomarkers (N=7) were measured in serum and plasma. We calculated percent differences and 95% confidence intervals (CI) in oxidative stress biomarkers for use of pesticides adjusted for sex, age, education, smoking, and alcohol consumption, using multiple linear regression.

Results:

Living close to or in the fields was associated with glutathione peroxidase activity but not with any other markers. We observed significant positive associations between 8-isoprostane levels and activities of glutathione reductase, glutathione peroxidase, and glutathione-S-transferase with reported use of certain insecticides, fungicides, and herbicides. Our results also indicate a significant and negative association between glutathione-S-transferase activity and insecticide use.

Conclusions:

These findings suggest that use of pesticides may be associated with differences in oxidative stress biomarkers.

Introduction

A growing body of epidemiological evidence has suggested positive associations between occupational exposure to pesticides and different cancers (Alavanja et al. 2012; Latifovic et al. 2020), autoimmune diseases (Meyer et al. 2018; Parks et al. 2019) and neurodegenerative disorders (Tangamornsuksan et al. 2018; Huang et al. 2016). Oxidative stress, an imbalance between reactive oxygen species (ROS) and antioxidant defenses in biologic systems (Sies et al. 2017), may play an important role in these relationships (Abdollahi et al. 2004; Hitchon et al. 2004; Perl 2013). Excessive ROS such as superoxide anion, hydroxyl radical and hydrogen peroxide promotes oxidation of lipids, proteins, and DNA, leading to genetic or epigenetic alterations, inflammation, cellular damage, mutagenesis, and carcinogenesis (Hancock et al. 2001; Ayala et al. 2014; Biswas et al. 2016; Kreuz and Fischle 2016).

Antioxidant defenses include enzymatic and non-enzymatic components measured in serum, erythrocytes, or plasma for use as indirect markers of cellular injury promoted by oxidative stress in humans (Monastero and Pentyala 2017; Møller and Loft 2010). Oxidized biological molecules from proteins, lipids and DNA are also used as oxidative stress markers (Birben et al. 2012). Pesticides may induce oxidative stress directly or during their biotransformation leading to reactive metabolites that generate ROS (Bus and Gibson, 1984; Abdollahi et al. 2004). Figure 1 hypothetically illustrates the ROS generation by pesticides, antioxidant system action, and biological modifications that can be tracked as oxidative stress biomarkers.

Figure 1.

Oxidative stress produced during pesticide metabolism and possible formation pathways of selected biomarkers. ROS induce lipid oxidation produces 8-isoprostane (8-iso) (Dalle-Donne et al. 2006). The ROS superoxide anion (O₂^•−) is dismutation by superoxide dismutase enzyme (SOD) and converted to ROS hydrogen peroxide (H₂O₂), which is neutralized by catalase enzyme (CAT) (Ighodaro and Akinloye, 2017). The ROS alkyl peroxide (RO₂H) is neutralized by glutathione peroxidases enzyme (GPx). GPx oxidize the reduced glutathione (GSH) to reduce ROS (Dalle-Donne et al. 2006). The enzyme glutathione reductase (GR) reduces oxidized glutathione (GSSG) to GSH (Dalle-Donne et al. 2006). Pesticide metabolites also can be conjugated with glutathione by glutathione-S-transferase (GST) (Fabrini et al. 2012). The thiol group (Ths) is considered a marker of reducing power of cysteine present in proteins (Ptn-R−SH) and GSH (Mungli et al. 2009). Note. Docosahexaenoic acid (DHA) by Minutemen, Wikimedia Commons (https://commons.wikimedia.org/wiki/File:DHA.svg). In the public domain.

Pesticide poisonings have been associated with altered antioxidant enzyme activities and increased lipid peroxidation. Vidyasagar et al. (2004) found increased MDA levels and SOD activity, and decreased acetylcholinesterase activity, in farmers poisoned by organophosphate insecticides, further noting increased acetylcholinesterase activity and decreased MDA levels after treatment with a specific antidote. Reduced acetylcholinesterase activity among farmers poisoned by organophosphates was associated with increased MDA levels and SOD, CAT and GPx activities compared to controls (Hundekari et al. 2013).

Oxidative stress markers have been largely investigated in agricultural workers chronically exposed to pesticides (Sharma et al. 2013; Alves et al. 2016; Madani et al. 2016; Taghavian et al. 2015). However, few studies have examined alterations in these oxidative stress markers associated with different types of pesticides (Zepeda-Arce et al. 2017; Lerro et al. 2017; Wang et al. 2016). Although the oxidative stress markers have no specificity, some studies have suggested differential associations depending on the type of pesticide (Zepeda-Arce et al. 2017; Lerro et al. 2017). Thus, we conducted a study in one of the most important agricultural areas in the State of Rio de Janeiro to examine oxidative stress biomarkers in farmers chronically exposed to pesticides.

Methods

Study design and subject recruitment

In this cross-sectional study conducted between July and August 2016, healthy volunteers were recruited from five rural areas in the outskirts of Teresopolis County, one of the largest producers of greens and vegetables in the state of Rio de Janeiro. Family health clinics in five Teresopolis districts were visited by the study researchers, who provided information about the study to clinic professionals and community health workers. Each clinic was asked to invite up to 25 individuals (agricultural or non-agricultural workers) to participate in the study; eligible participants were enrolled during visits to the clinic regardless of their occupational status or reason for visiting the clinic. Individuals were eligible to join the study if they resided for at least one year in one of the studied districts. Exclusion criteria were current smoking, diagnosis of diabetes, hypertension, cancer, neurodegenerative, autoimmune, and endocrine disease, or the use of antioxidants or vitamins. The final sample included 52 agricultural workers (working at least one year in farming) and 68 non-agricultural workers ages 17–69 years.

Data collection and biological assays

The questionnaire was previously designed to assess agricultural pesticide use in the adjacent County of Nova Friburgo (Moreira et al. 2002, De Araújo et al. 2007). Data included sociodemographic characteristics, general farming information (types of crops grown and tasks), farm residence (whether they lived within 50 meters of crops or fields), and lifestyle (e.g., never vs. ever smoked, and any vs. no current alcohol use). Questions on pesticide use included structured questions on regular use of specific chemicals and practices (i.e., Do you come into contact with agricultural pesticides regularly?). If yes, what kind of contact, (i.e., mix, apply, both, pulling the pesticide discharge hose, cleaning spraying equipment, cleaning work clothes contaminated with pesticides) and use of protective equipment (i.e., Do you use any personal protective equipment (PPE) when you mix pesticides? What PPE do you use when you mix pesticides? Do you use PPE regularly? What PPE do you regularly use?). Questions were then asked about frequency of pesticide use, i.e., hours per day and days per month pesticides applied and which specific pesticides were regularly used by the participant.

Venous blood samples were collected at the family health clinics in 5 ml polypropylene tubes, without anticoagulants for serum-requiring assays (SOD, Thiol group, GST, CAT), and with EDTA for those performed on plasma (8-iso, GPx, GR). Samples were kept in a box with ice and immediately transported to the lab, stored for 2 hours at 2°−8°C until centrifugation and aliquoting, stored at −80°C until analysis.

Catalase (CAT) Activity

CAT activity was assayed according to the method developed by Goth (1991). Briefly, 100 μL serum samples were incubated with 65 mM hydrogen peroxide diluted in a 60mM Phosphate buffer (pH 7.4) at 37° C for 1 min. The reaction was stopped by adding 500 μL of 32.4mM ammonium molybdate. The formation of a stable color complex was measured colorimetrically at 405 nm. Results were reported as kilo units per liter (kU/L). Intra-assay coefficient of variation (CV) ranged from 1.6 to 5.1%.

Superoxide dismutase (SOD) Activity

Cayman Superoxide dismutase kit (706002) was used to evaluate SOD activity. Briefly, 10 μL of the serum sample was diluted in 50 mM Tris-HCl, pH 8.0 buffer and then 200 μL of tetrazolium salt was added for the detection of superoxide radicals generated by the addition of 20 μL xanthine oxidase. A calibration curve was obtained from the superoxide dismutase standard. The assay measures the sample ability to inhibit the reaction of the superoxide anion with tetrazolium salt that forms a compound called formazan whose absorbance was read at 440–460 nm. Results were reported in units per milliliter (U/mL). Intra-assay CV ranged from 3.4 to 8.0%.

Thiols

The colorimetric assay described by Hu (1994) was chosen to quantify the thiol group in the serum sample. Serum samples (25 μL) were mixed in 950 μL of 250M Tris, 20mM EDTA buffer, and 25 μL of 5,5-dithio-bis-2-nitrobenzoic acid (DTNB). A calibration curve was obtained from reduced glutathione (GSH), used as a standard of sulfhydryl group. The absorbance was read at 412nm and the results were reported in millimoles per liter (mmol/L). The limit of detection of the method was 0.09 mmol/L (mM) and intra-assay CV ranged from 0.7 to 4.4%.

Glutathione S transferase (GST) Activity

GST activity was determined spectrophotometrically using the method described by Habig et al. (1974) and adapted by Habdous et al. (2002). The enzyme assay was performed by adding 100 μL of serum sample to 700 μL of 100 mM Na-phosphate buffer (pH 5.5), 100 μL of 25 mM 1-chloro-2,4-dinitrobenzene (CDNB) and 100 μL of 50 mM reduced glutathione (GSH). Results were monitored at 340 nm, as units per liter (U/L). Intra-assay CV ranged from 1.4 to 8.8%.

Glutathione Peroxidase (GPx) Activity

GPx activity was measured using the Glutathione Peroxidase Assay Kit (Cayman Chemical, 703102). In a test tube, 100 μl of Assay Buffer, 50 μl of Co-Substrate Mixture (NADPH, GSH, and GR) and 20 μl of plasma sample were added. GPx activity is measured by monitoring the decrease of NADPH levels after adding 20 μl of cumene hydroperoxide. GPx catalyzes the cumene hydroperoxide reduction to non-reactive molecules using GSH oxidized to GSSG form. GSSG was recycled to its reduced state by GR, which oxidizes the NADPH. The decrease in NADPH was followed by a decrease in absorbance at 340 nm, and it is proportional to the activity of GPx. Results were presented as nmol/min/ml. Intra-assay CV ranged from 2.7 to 3.4%.

Glutathione Reductase (GR) Activity

The Glutathione Reductase Assay Kit (Cayman Chemical, 703202) was used to assess the GR activity by measuring the NADPH oxidation rate. Briefly, 100 μl of the assay buffer (50mM potassium phosphate, pH 7.5, containing 1mM EDTA and 1mg/ml bovine serum albumin), 20 μl of GSSG, and 20 μl of plasma sample were added to a test tube. The reaction was initiated by adding 50 μl of NADPH. Then, the decrease in absorbance that followed the oxidation of NADPH to NADP⁺, was measured at 340 nm. GR activity results were then presented as nmol/min/ml. Intra-assay CV ranged from 1.1 to 6.1%.

8-Isoprostane (8-ISO)

8-ISO was measured by a commercial competitive enzyme immunoassay (Cayman Chemical, 516351). Aliquots of plasma (200 μl) received 0.005% butylated hydroxytoluene (Sigma-Aldrich Corp., Germany) before freezing. Samples were purified using C-18 solid-phase extraction (SPE). The samples were dried under liquid nitrogen and then resuspended in ELISA buffer. The immunoenzymatic reaction product was measured at 412 nm and expressed as picograms per milliliter of plasma. Intra-assay CV ranged from 2.5 to 9.7%.

Statistical Analyses

In the full sample, we looked at residential proximity to crops or fields (within 50 meters, as a proxy for potential residential exposure to pesticides), and potential pesticide exposures among agricultural workers who reported regularly having mixed and applied pesticides, including use of PPE when mixing pesticides (don’t mix, mix with PPE, mix without PPE). Frequency of pesticide use in these individuals was modeled as categorical variable 0 = none; 1 to 2 times per month; and 3 or more times per month. Duration of application, collected in hours and minutes per application, was categorized as follows: 0 = none; 1 = apply pesticides up to 1 hour; 2 = apply pesticides more than 1 hour. For each pesticide-related variable, including use of specific pesticides, the exposed group included individuals who reported the specific exposure and unexposed group included non-exposed individuals regardless of their agricultural work status.

We calculated frequencies for categorical variables and conducted chi-square tests to compare categorical variables. The distribution of continuous variables was tested for normality using the Kolmogorov–Smirnov test, and examination of their histograms and differences between means and medians. Enzyme activities were expressed as the median and interquartile range (IQR) for skewed ones. We utilized non-parametric Mann Whitney U test to examine the difference of oxidative stress markers levels between agricultural and non-agricultural workers. Biomarkers of oxidative stress were log (10)-transformed to satisfy the normality condition. Multiple linear regression was performed to examine the adjusted percent differences and 95% confidence intervals (CI) in oxidative stress markers in association with pesticide exposure variables. The adjusted differences were calculated using the formula $\widehat{\Delta M}\%=100\left(e^{\widehat{\beta }c}-1\right)\%$ that indicates the relative change in the median (geometric mean) of Y when adding c units to X if X is binary (Barrera-Gómez and Basagaña, 2015). Potential confounding factors such as sex, age, smoking, education, and current use of alcohol were included in the regression models. Overall, agricultural workers reported the use of 18 different pesticides, but associations were estimated for only ten pesticides used by four or more participants. All statistical analyses were conducted using SPSS (IBM, SPSS Statistics Version 21).

Ethical Considerations

The study was approved by the Institute of Public Health’s Ethics Committee at the Federal University of Rio de Janeiro (approval number 1.537.073). Each participant was informed about all relevant aspects of the study and voluntarily signed a consent form.

Results

Agricultural workers were older than non-agricultural workers (median = 43 vs. 38.5; p<0.001) (Table 1). The proportion of male was higher among agricultural workers than non-agricultural workers (n=28 (53.8%) vs. n=18 (26.5%); p = 0.002). Non-agricultural workers reported higher educational levels (p=0.002). Smoking history and alcohol consumption were similar in both groups. Agricultural workers were more likely to live close to or within the fields (80.8% vs. 36.8%; p<0.001). None of the non-agricultural workers reported occupational pesticide use while 94% of agricultural workers reported using pesticides. The levels of oxidative stress biomarkers in both groups did not differ significantly when comparing geometric means, though agricultural workers had somewhat higher levels of GP and GPx (p>0.10; Table 2).

Table 1.

Main characteristics of the study sample.

	Total N=120	Agricultural workers N=52	Non-agricultural workers N=68	cp value
Agea	41 (33.3, 49.0)	43 (40.0, 53.0)	38.5 (30.0, 46.0)	<0.001
Sexb
Male	46 (38.3)	28 (53.8)	18 (26.5)	0.002
Female	74 (61.7)	24 (46.2)	50 (73.5)
Educationb
High school or less	109 (90.8)	52 (100)	57 (83.8)	0.002
More than high school	11 (9.2)	0 (0)	11 (16.2)
Smoking historyb
Never	109 (90.8)	46 (88.5)	63 (92.6)	0.431
Ever	11 (9.2)	6 (11.5)	5 (7.4)
Current alcohol useb
No	72 (60.0)	32 (61.5)	40 (58.8)	0.940
yes	48 (40.0)	20 (38.5)	28 (41.2)
Living in house nearby or in fieldsb
No	53 (44.2)	10 (19.2)	43 (63.2)	<0.001
Yes	67 (55.8)	42 (80.8)	25 (36.8)
Occupational use of pesticidesb
Never	71 (59.2)	3 (5.8)	68 (100)	<0.001
Ever	49 (40.8)	49 (94.2)	0 (0)

^aMedian (Q1, Q3), Mann Whitney non-parametric test.

^bN (%),

^cChi-square test.

Table 2.

Levels of oxidative stress biomarkers in the study population (n=120).

		Geometric Mean (GSD)
Biomarkers	N (%)	Agricultural workers	Non-agricultural workers	a p Value
8-ISO (pg/mL)b	117 (97.5)	118.3 (±2.2)	117.9 (±2.4)	0.998
Thiol (mmol/L)	120 (100.0)	0.4 (±1.6)	0.3 (±1.7)	0.336
SOD (U/mL)b	119 (99.2)	1.1 (±1.4)	1.1 (±1.5)	0.722
CAT (U/L)	120 (100.0)	37.0 (±2.4)	41.3 (±2.1)	0.433
GR (nmol/mL)b	111 (92.5)	32.0 (±1.8)	27.6 (±1.9)	0.118
GPx (nmol/mL)b	118 (98.3)	100.0 (±1.4)	90.7 (±1.3)	0.120
GST (U/L)b	119 (99.2)	3.4 (±3.2)	3.9 (±3.2)	0.338

^aMann-Whitney non-parametric test.

^bMissing values for up to 3 specimens per assay, except for GR with 9 missings. GSD: Geometric Standard Deviation.

Living close to or within crop fields was positively associated with GPx activity (Table 3). Compared to non-agricultural workers and workers who did not personally mix or apply pesticides, pesticide mixing and application, use of personal protective equipment (PPE) when mixing pesticides, and frequency and duration of pesticide applications were not significantly associated with oxidative stress biomarkers.

Table 3.

Adjusted percent difference^a (95% CI) in oxidative stress biomarkers associated with potential occupational and environmental pesticide exposure (n=120).

		Adjusted differences (95% CI)
	N (%)	8-ISO	THIOL	SOD	CAT	GR	GPx	GST
Living in house nearby or in fields	67 (55.8)	31.8 (−35.4, 175.4)	41.3 (−8.8, 118.8)	9.6 (−20.6, 54.9)	9.6 (−46.3, 123.9)	14.8 (−35.4, 108.9)	31.8 (2.3, 77.8)	129.1 (−16.8, 531.0)
Pesticide useb
Regularly mix and apply	37 (30.8)	−4.5 (−59.3, 123.9)	41.3 (−14.9, 139.9)	0.0 (−30.8, 47,9)	17.5 (−48.7, 175.4)	54.9 (−22.4, 209.0)	12.2 (−18.7, 55)	0 (−70.5, 238.8)
When mixing pesticidesc
No PPE use	30 (25.0)	−27.6 (−71.8, 86.2)	31.8 (−25.9, 336.5)	0.0 (−33.9, 54.9)	51.4 (−41.1, 280.2)	47.9 (−30.8, 216.2)	99.5 (−24.1, 54.9)	−14.9 (−77.6, 231.1)
Yes PPE use	7 (5.8)	123.9 (−49.9, 877.2)	−42.5 (−30.8, 336.5)	0.0 (−48.7, 90.5)	−43.8 (−87.1, 145.5)	−43.8 (−43.8, 475.4)	34.9 (−22.4, 139.9)	62.2 (−80.9, 1249.0)
Frequency appliedd	37 (30.8)	0.0 (−35.4, 58.5)	14.8 (−12.9, 51.4)	0.0 (−18.7, 20.2)	17.5 (−24.1, 86.2)	17.5 (−18.7, 69.8)	9.6 (−6.7,31.8)	−4.5 (−49.9, 82.0)
Duration of applicationd	37 (30.8)	0.0 (−42.5, 69.8)	12.2 (−20.6, 54.9)	0.0 (−22.4, 25.9)	25.9 (−25.9, 113.8)	23.0 (−22.4, 90.5)	23.0 (0, 47.9)	25.9 (−42.5, 169.2)

^aIn terms of percent change, the pesticide exposure measured by proxy variables compared to no or lower exposure can be associated with an % increase in the geometric mean of oxidative stress biomarkers. The models were adjusted for age, sex, education levels, past smoke and alcohol consumption.

^bMix and apply (Of 49 participants who reported pesticide use, 12 reported only pulling the discharge hose but not personally mixing and applying – these are included in the referent category along with non-agricultural workers).

^cUse of dummy variables. Reference group included all subjects except those that didn’t use and used PPE when mixing.

^dFrequency of pesticide use (0 - none; 1 – up to twice per month; 2 - three or more times per month) and duration per application (0 = no; 1 = up to 1 hour; 2 = more than 1 hour) were considered as ordinal variables. CI: confidence interval.

The results suggest a positive association between 8-ISO levels and use of the herbicide glyphosate (182%, 95% CI= 12, 608%) (Table 4). In addition, positive associations between GPx activity and use of the fungicides difenoconazole (23%, 95% CI= 2, 49%) and copper sulfate (35%, 95% CI= 2, 79%) and the insecticide methomyl (38%, 95% CI= 11, 70%) were observed. GST activity can be positively associated with use of the fungicide copper sulfate (426%, 95% CI= 82, 1418%) and the herbicide linuron (197%, 95% CI= 13, 677%), and negatively associated with use of the insecticide deltamethrin (−147%, 95% CI= −125, −187%).

Table 4.

Adjusted percent difference^a (95% CI) in oxidative stress biomarkers in association with specific pesticide use (n = 120).

		Adjusted differences (95% CI)
	N (%)	8-ISO	THIOL	SOD	CAT	GR	GPx	GST
FUNGICIDES
Difenoconazole	12 (10.0)	69.8 (−49.9, 488.8)	14.8 (−43.8, 134.4)	−16.8 (−49.9, 38.0)	14.8 (−63.7, 263.1)	43.3 (−4.9, 113.8)	23.4 (2.0, 49.2)	9.4 (−46.7, 127.0)
Azoxystrobin	7 (5.8)	181.8 (−35.4, 1130.3)	7.2 (−57.3, 169.2)	−6.7 (−51.0, 82.0)	223.6 (−24.1, 1312.5)	29.7 (−20.5, 113.8)	22.1 (−25.9, 103.4)	−57.7 (−83.0, 5.1)
Copper sulfate	5 (4.2)	104.2 (−66.9, 1158.9)	−8.8 (−69.8, 175.4)	−10.9 (−59.3, 95.1)	86.2 (−67.6, 996.5)	18.5 (−35.6, 118.1)	35 (2.0, 78.6)	425.9 (82.2, 1418.0)
Mancozeb	5 (4.2)	−82.6 (−97.5, 20.2)	14.8 (−60.2, 231.1)	0 (−54.3, 113.8)	134.4 (−57.3, 1218.3)	17.4 (−39.3, 127.0)	29.7 (−3.0, 71.6)	−53.7 (−83.5, 40.5)
HERBICIDES
Paraquat	18 (15.0)	−41.1 (−78.6, 62.2)	66 (−8.8, 209.0)	−4.5 (−39.7, 47.9)	66 (−36.9, 336.5)	3 (−26.7, 44.8)	9.4 (−6.8, 28.4)	−40.5 (−67.4, 8.3)
Glyphosate	24 (20.0)	181.8 (12.2, 607.9)	69.8 (−4.5, 202.0)	51.4 (−2.3, 123.9)	−16.8 (−66.9, 108.9)	18.5 (−13.9, 64.9)	10.5 (−4.9, 28.4)	25.9 (−29.5, 124.8)
Fluazifop	14 (11.7)	−35.4 (−78.6, 95)	17.5 (−39.7, 129.1)	−4.5 (−42.5, 58.5)	47.9 (−49.9, 346.7)	4.1 (−28.8, 50.7)	−5.8 (−21.3, 12.7)	−47.8 (−73.3, 1.0)
Linuron	7 (5.8)	246.7 (−22.4, 1484.9)	66 (−33.9–316.9)	−12.9 (−55.3, 69.8)	−65.3 (−92.1, 47.9)	50.7 (−9.5, 148.4)	12.7 (−12.2, 44.8)	197.4 (12.7, 676.8)
INSECTICIDES
Deltamethrin	16 (13.3)	58.5 (−35.4, 280.2)	34.9 (−29.2, 151.2)	−8.8 (−42.5, 41.3)	108.9 (−24.1, 475.4)	12.7 (−20.5, 60.0)	−3.9 (−18.9, 13.9)	−53.2 (−74.8, −13.1)
Methomyl	9 (7.5)	51.4 (−58.3, 449.5)	−2.3 (−58.3, 123.9)	−36.9 (−66.1, 17.5)	−57.3 (−88.8, 62.2)	66.5 (2.0, 171.8)	37.7 (10.5, 69.9)	101.4 (−16.5, 380.7)

^aIn terms of percent change, the use of a specific pesticide compared to no use can be associated with an % increase in the geometric mean of oxidative stress biomarkers. The models were adjusted for age, sex, education levels, past smoke, and alcohol consumption. 95% CI: 95% confidence interval.

Discussion

In this rural Brazilian population, oxidative stress biomarkers were not associated with overall pesticide use. However, our results suggest some possible associations between regular use of pesticides and oxidative stress. In particular, a positive association was seen between levels of 8-ISO and the commonly used herbicide glyphosate. Use of two fungicides, difenoconazole and copper sulfate, and the insecticide methomyl were positively associated with GPx activity, an enzyme that specifically scavenges organic peroxides. Living in or close to agricultural fields was also positively associated with GPx activity. Use of the fungicide copper sulfate and the herbicide linuron was positively associated with GST activity, an enzyme that neutralizes reactive metabolites during xenobiotic metabolism, while the insecticide deltamethrin was negatively associated with the activity of this enzyme.

Two previous cross-sectional studies have reported that fungicide use was negatively associated with GST and GR activities and thiol levels (Arnal et al. 2011; Mecdad et al. 2011). In our study, there were no associations seen for fungicides with GR activity or thiols levels. Instead, positive associations were observed between two fungicides (difenoconazole and copper sulfate) and GPx activity and between copper sulfate and GST activity. Results of experimental studies are inconsistent but provide some support for our findings. Rats treated with subacute doses of difenoconazole showed a significant decrease in GPx activity of hepatic tissues (Kasmi et al. 2018). Subchronic exposure to copper sulfate increased the activity of GPx and GST in rats’ blood (Quamar et al. 2019).

Herbicide use has been inconsistently associated with oxidative stress in farmers. For instance, in a longitudinal study conducted by Lerro et al. (2017), higher levels of urinary 2,4-dichlorophenoxyacetic acid (2,4-D) were associated with 8-ISO levels in farmers’ urine. However, in another longitudinal study conducted by the same author (Lerro et al. 2020), elevated 8-ISO levels were associated for the 3^rd quartile of 2,4-D use, but no exposure-response trend was observed. In our study, agricultural workers did not report the use of 2,4-D. Instead, the use of the herbicide glyphosate was positively associated with 8- ISO in farmworkers’ blood. There is no evidence in animals on glyphosate and 8- ISO, a product of lipid peroxidation. However, rats treated with glyphosate showed elevated MDA levels (another lipid peroxidation product) in plasma and liver (El-Shenawy, 2009; Milić et al. 2018).

Exposure to insecticides has been positively associated with CAT and SOD activities, and negatively associated with GST and GR activities and thiol levels in cross-sectional studies (Mecdad et al. 2011; Ogut et al. 2011). Zepeda-Arce et al. (2017) observed that the use of bendiocarb, a carbamate insecticide, and deltamethrin, a pyrethroid insecticide, were positively associated with GPx activity. In our study, deltamethrin use was negatively associated with GST activity, while the carbamate insecticide methomyl was positively associated with GR and GPx activities. Although speculative, these last results suggest that methomyl might be associated with an increased RO₂H (Figure 1). Rats orally exposed to deltamethrin showed suppression of GPx, GST, and CAT activities in liver and kidney tissues (Rehman et al. 2006; Sharma et al. 2014). In animal models, subacute exposure to methomyl was associated with decreased GPx and GR activity in rat serum (Mansur et al. 2017).

Environmental pesticide exposure may occur from contaminated food, water, and soil. Living close to the fields is also related to environmental exposure due to pesticide spraying to the planting areas. On the other hand, occupational exposure to pesticides may occur while transporting, mixing, or applying chemicals. Further, many agricultural workers do not use suitable PPE or use cloth masks that increase dermal and respiratory exposure. In this study, living in a house nearby or in the fields was positively related to GPx activity, but none of the other oxidative stress biomarkers. Duration and frequency of pesticide use were not associated with oxidative stress biomarkers, but lacked times reported for the specific pesticides, so a lack of effect may reflect the types most frequently applied.

Pesticide-induced oxidative stress can result from excessive ROS mediated by metabolizing enzymes. For example, bipyridyl herbicides produce ROS through redox cycling mediated by cytochrome P450 reductase (Blanco-Ayala et al. 2014). Some organochlorine insecticides induce cytochrome P450 enzymes, leading to the rate of O₂ production in cells increasing (Banerjee et al. 2001). In our study, difenoconazole, glyphosate, and methomyl were associated with oxidative stress. Difenoconazole seems to inhibit lanosterol-14a- demethylase activity (CYP51) and increase ROS generation, but it is unclear how this occurs (Kasmi et al. 2018). Some studies have suggested that glyphosate might activate glutamate receptors, promoting Ca₂⁺ influx, mitochondrial dysfunction, and oxidative stress (Olorunsogo, 1990; Cattani et al. 2017). The molecular mechanisms that link methomyl to oxidative stress are not yet understood. However, the acetylcholinesterase inhibition drive-by carbamates cause depletion of high-energy phosphates and affect cell mitochondrial function, elevating ROS (Milatovic et al. 2006).

The antioxidant defense is responsible for regulating the intracellular reduction and oxidation processes (redox). In pathological conditions promoted by oxidative stress, activation or silencing of genes encoding antioxidant enzymes with alteration in the levels of these structural proteins may occur (Dalton et al. 1999; Scandalios 2006). Thus, increased activity of antioxidant enzymes may be associated with ROS generation and oxidative stress. However, when oxidative stress is persistent, the antioxidant systems can be overwhelmed and attacked by reactive molecules and lose their function leading to irreversible biological damage (Birben et al. 2012). In this way, the results of epidemiological and experimental studies may sometimes appear inconsistent. In animal studies or human poisoning, high doses of pesticides have been related to toxic effects and may indicate a decreased activity of antioxidant enzymes due to damage to the antioxidant system. By contrast, exposure to long-term low doses may be related to increased activity of these enzymes since these doses can be associated with ROS production.

Brazil is one of the largest consumers of pesticides globally (Zaller, 2020). Brazilian studies have suggested environmental contamination by insecticides and herbicides in freshwater samples and residential soils in some Brazilian States (Albuquerque et al. 2016; Fernandes et al. 2020). The Brazilian Program on Pesticide Residue Analysis in Food (PARA) has found non-authorized pesticides or concentrations of residues above the MRL in about 20% of animal and vegetable products (Arisseto-Bragotto et al. 2017). There are no studies on contamination by pesticides in freshwater or soil in the studied area, but two studies in adjacent cities reported detectable insecticide levels in superficial water (Moreira et al. 2002; Veiga et al. 2006).

There are limitations in our study that must be noted. The number of individuals reporting certain pesticides was low, so statistical power was limited. Furthermore, the number of comparisons made was high, so some findings could be due to chance. Given the lack of prior evidence, the findings might be viewed as hypothesis-generating. Sample selection was based on invitation by the medical team, which may have introduced selection bias (Pahwa et al. 2012), as participants may have been identified based on personal relationships and easy access to their residences. Therefore, the results in this volunteer sample may not be generalizable to other populations. We evaluated the exposure to pesticides through a standardized questionnaire. However, self-reported use may be prone to exposure misclassification, and in our survey open-ended questions referred to regular use and lacked a specified time frame to guide recall of specific pesticides. Information on environmental exposures was limited to residential proximity to fields, and no data were collected on personal non-occupational uses. Other exposures could be correlated with certain agricultural tasks (e.g., fertilizer use). These methodological limitations emphasize the need to conduct prospective studies in a larger sample of the Brazilian agricultural population along with a valid and accurate exposure assessment to pesticides to obtain robust evidence of pesticides’ effects on workers and residents (Paumgartten et al. 2020). Future studies are needed using biomarkers of pesticide exposure, although for some exposures, biomarkers may be transient and reflect recent rather than chronic exposures. Similarly, oxidative stress markers were measured at one point in time, so observed associations may be due to recent exposures or other unmeasured factors. Most of the evaluated oxidative stress biomarkers in our study might reflect potential alterations in antioxidant enzymes and oxidative damage in lipids. Future studies should also include biomarkers of damage to proteins and DNA (Costa et al. 2020; Ramos et al. 2020) that are intracellular redox imbalance early products to evaluate the extent of pesticides’ toxic effects in different cellular structures. Finally, analytical difficulties in measuring the thiol group (Winther and Thorpe, 2013) can introduce error into the estimation of this biomarker, and enzymatic biomarkers tend to be more stable.

Strengths of our study include the evaluation of the relationship between pesticides and multiple oxidative stress biomarkers, using robust multivariate statistical analysis to adjust for potential confounding variables. In addition, the medical team collaboration was crucial to creating trust and promote population participation, also allowing subjects to be comfortable responding to potentially sensitive survey questions related to pesticide use. To the best of our knowledge, ours is one of the few studies to evaluate the relationship between different pesticides and oxidative stress using adjusted models (Arce et al. 2017; Lee et al. 2017; Lerro et al. 2017; Wang et al. 2016), while many others have investigated these associations in individuals exposed to pesticides in general or classes of pesticides (Alves et al. 2016; Kahl et al. 2016; Fareed et al. 2016; Venkata et al. 2016; Gaikwad et al. 2015; Mori et al. 2015; Ogut et al. 2015; Murussi et al. 2014; Bayrami et al. 2012). Our findings that oxidative stress markers were not associated with overall agricultural occupation and agricultural pesticide use suggest that observed associations for different types of pesticides were not due to common underlying differences between agricultural workers and other participants. Finally, we evaluated multiple markers, which may capture a broader range of potential molecular pathways relating pesticides to oxidative stress.

Conclusion

Our findings suggest that exposure to herbicides, fungicides, and insecticides may contribute to oxidative stress status in Brazilian agricultural workers. Future epidemiological studies are needed to confirm these findings in larger samples and other populations, also including biomarkers of pesticide exposure to confirm findings for specific pesticides.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author, A.S.E.S., upon reasonable request.