Serum Molecular Fingerprints of Oxidative Stress Underlying Neurocognitive and Motor Dysfunction in Pesticide‐Exposed Populations

ABSTRACT

We investigated the relationship between chronic pesticide exposure, oxidative stress, and neurocognitive alterations in rural populations of Celendín, Cajamarca, Peru. A total of 101 adults (18–86 years) were evaluated through neuropsychological (NEUROPSI) and neuromotor assessments, serum oxidative stress biomarkers, and Fourier‐transform infrared spectroscopy (FT‐IR) of serum samples. Biomarkers analyzed included superoxide dismutase (SOD), catalase, glutathione peroxidase (GPx), total glutathione, lipid peroxidation (malondialdehyde), biopyrrins (BPn), and 8‐hydroxy‐2′‐deoxyguanosine (8‐OHdG). Statistical comparisons (Mann–Whitney U) and multivariate analyses (PCA, PLS‐DA) were performed to identify biochemical and spectral alterations. Pesticide users exhibited a significantly higher frequency of attention deficit (OR = 4.4, 95% CI = 1.2–15.9) and elevated serum 8‐OHdG levels (p = 0.042), indicating increased oxidative DNA damage. While antioxidant enzyme activities did not differ significantly, higher malondialdehyde and BPn levels suggested enhanced lipid and bilirubin oxidation. FT‐IR analysis revealed distinct spectral signatures in individuals with moderate cognitive impairment, impaired left‐hand motor control, and executive dysfunction (calculation). Significant bands appeared in the 440–493 cm⁻¹ region (disulfide/polysulfide bonds), 2573 cm⁻¹ (thiols), and 3630–3840 cm⁻¹ (free hydroxyl groups), consistent with oxidative protein and glycan modifications. Additional signals between 2089 and 2255 cm⁻¹ suggested carbamylation‐related isothiocyanate and cyanate groups. These findings suggest that oxidative stress is a key mechanism underlying pesticide‐related neurocognitive dysfunction. Combining oxidative stress biomarkers with FT‐IR provides a rapid, minimally invasive approach for identifying serum molecular fingerprints with potential applications in screening and monitoring pesticide‐induced neurotoxicity.

Chronic occupational pesticide exposure is associated with oxidative stress and neurocognitive dysfunction. Serum FT‐IR spectroscopy identified characteristic molecular fingerprints linked to oxidative damage, including alterations in thiol/disulfide balance, cyanate, and isothiocyanate‐related spectral regions. These molecular signatures correlated with oxidative stress biomarkers and cognitive impairment, supporting FT‐IR as a complementary, non‐invasive approach for detecting oxidative stress‐associated neurotoxicity in pesticide‐exposed populations.

Abbreviations

ADHD

Attention‐deficit/hyperactivity disorder

BPn

Biopyrrins

CIs

Confidence intervals

CSF

Cerebrospinal fluid

SD

Standard deviation

SEM

Standard error of the mean

FDR

False discovery rate

FT‐IR

Fourier‐transform infrared spectroscopy

GFAP

Glial fibrillary acidic protein

GPx

Glutathione peroxidase

GSH

Reduced glutathione

HCA

Hierarchical cluster analysis

MDA

Malondialdehyde

MCI

Mild cognitive impairment

ModCI

Moderate cognitive impairment

NEUROPSI

Brief Neuropsychological Assessment in Spanish

NF‐L

Neurofilament light chain

OR

Odds ratio

PLS‐DA

Partial least squares discriminant analysis

PCA

Principal component analysis

PKA

Protein kinase A

ROS

Reactive oxygen species

RMSEP

Root mean squared error of prediction

SOD

Superoxide dismutase

SS/TT

Disulfide‐to‐thiol ratio

VIP

Variable importance in projection

WHO

World Health Organization

¹H NMR

Proton nuclear magnetic resonance

8‐OHdG

8‐Hydroxy‐2′‐deoxyguanosine

1. Introduction

Pesticides are chemical substances extensively used for crop protection and pest control, with an estimated global annual usage of approximately two million tons. The World Health Organization (WHO) has classified several of these compounds as highly hazardous to human health (Shekhar et al. 2024). Chronic exposure to pesticides has been linked to a wide range of adverse health outcomes, including cancer, peripheral neuropathies, neurobehavioral and neurocognitive disorders, and immune system dysfunction (Tudi et al. 2022). Cognitive impairment related to long‐term pesticide exposure appears to result from multiple interacting factors, such as duration of exposure, age, and pesticide type. Mechanistically, oxidative stress, mitochondrial dysfunction, epigenetic alterations, accumulation of neurodegenerative proteins associated with Alzheimer's disease, and inflammatory processes have been implicated in the development of neurocognitive deficits among exposed individuals (Aloizou et al. 2020) (Botnaru et al. 2025) (Wen et al. 2023). Epidemiological evidence supports this association: a study conducted in Greece reported that individuals living near pesticide‐sprayed agricultural fields exhibited poorer cognitive performance than those who had never lived in such areas (Dardiotis et al. 2019), while another study among military personnel exposed to pesticides and pyridostigmine bromide demonstrated cognitive decline (Sullivan et al. 2018). Similarly, exposure to organophosphate pesticides has been associated with significant reductions in cognitive function (Paul et al. 2018).

Evidence indicates that the accumulation of acetylcholine resulting from acetylcholinesterase inhibition leads to overstimulation of muscarinic and nicotinic receptors, thereby impairing memory and learning processes (Finhler et al. 2023). Consistently, a previous study reported a significant correlation between decreased blood acetylcholinesterase activity and poorer cognitive performance among individuals exposed to organophosphate pesticides (Halim et al. 2018). In addition, a cohort study identified plasma biomarkers linked to the progression from cognitive decline to dementia, including molecules associated with inflammation, extracellular matrix remodeling, endothelial injury, lipid metabolism, and neurodegeneration—particularly the neurofilament light chain (NF‐L) (Kivisäkk et al. 2022).

Similarly, higher serum concentrations of tau protein, total cholesterol, LDL cholesterol, and triglycerides have been observed in individuals with moderate to severe cognitive impairment, whereas amyloid beta‐42 and HDL cholesterol levels were markedly reduced (Iqbal et al. 2020). Glial fibrillary acidic protein (GFAP) and tau protein have also been proposed as promising biomarkers of cognitive deterioration (Bivona et al. 2023). Supporting this, alterations in plasma levels of phosphorylated tau (p‐tau181), NF‐L, and GFAP have been associated with the transition from normal cognition to mild cognitive impairment (MCI) (Soldan et al. 2024).

Metabolomic approaches have further contributed to this understanding. Proton nuclear magnetic resonance (¹H NMR) spectroscopy of serum from patients with Alzheimer's disease and MCI revealed metabolites with diagnostic and prognostic potential, including phenylalanine, lysine, pyruvate, choline, lactate, and N‐acetylglucosamine (Botello‐Marabotto et al. 2023). Moreover, cerebrospinal fluid (CSF) analysis of Alzheimer's patients demonstrated altered levels of Aβ42 and tau proteins years before the onset of clinically detectable cognitive symptoms (Kyle et al. 2014).

Beyond these molecular biomarkers, Fourier transform infrared spectroscopy (FT‐IR) has emerged as a valuable analytical technique for identifying molecular fingerprints associated with diverse pathological conditions, including cancer. Its capacity to detect biochemical alterations at early stages highlights its potential for the diagnosis and prognosis of neurodegenerative and neurocognitive disorders (Ollesch et al. 2013)(Al‐Kelani and Buthelezi 2024). In this context, our study aimed to identify specific molecular signatures related to neurological and cognitive dysfunctions in pesticide‐exposed populations from northern Peru using FT‐IR.

2. Materials and Methods

2.1. Population and Sample

The study was carried out in rural communities of the Celendín province, Cajamarca region, northern Peru. The surveyed localities included Celendín, Huacapampa–José Gálvez, Fraylecocha, and Sucre, situated at an altitude of approximately 2620 m above sea level and comprising a total population of around 21,000 inhabitants (Instituto Nacional de Estadistica e Informatica 2017). The economy of these communities relies primarily on agriculture and livestock farming. The research protocol was approved by the Ethics Committee of the National Institute of Health (Resolution No. 062‐2024‐DIIS/INS) and adhered to national and international ethical standards concerning confidentiality and informed consent. A total of 101 adults aged 18 to 86 years who met the inclusion criteria were enrolled in the study. Individuals with metabolic, neurological, or chronic diseases; severe visual impairment; harmful use of alcohol or tobacco; illiteracy; or pregnancy were excluded. Data collection was conducted during the first 2 weeks of December 2024.

2.2. Biological Sample Collection

Coordination was established with the Celendín Health Network to obtain authorization for access to local health centers and to facilitate participant recruitment. Sampling stations were set up in each facility for biological collection and survey administration under standardized quality control conditions. After obtaining written informed consent, venous blood samples were collected into vacuum‐sealed tubes and centrifuged at 3500 rpm for 5 min. The resulting serum was stored at −20 °C and subsequently transported to the laboratory for biochemical analysis.

2.3. Neurological and Neuropsychological Assessment

All participants meeting the inclusion criteria underwent clinical evaluations using a comprehensive neurocognitive and neuromotor test battery administered by certified specialists. Cognitive performance was assessed with the Brief Neuropsychological Assessment in Spanish (NEUROPSI) (Ostrosky‐Solís et al. 1999), a tool validated for Latin American populations and specifically for Peru (Cronbach's α = 0.863) (Marreros‐Tananta and Guerrero‐Alcedo 2022). This test, which lasts approximately 25 min, evaluates eight cognitive domains: orientation, attention/concentration, memory, language, reading, writing, executive functions, and recall. The total score ranges from 0 to 130 points, classifying cognitive performance as normal, mild, moderate, or severe according to age and educational level.

Five fundamental motor functions were also evaluated: tremor, simple reaction time, and visuomotor coordination, postural control (on stable and unstable surfaces), and cranial nerve motor function according to standardized protocols for occupational exposure. The NEUROPSI protocol provided specific sub‐scores for orientation (time, place, person), attention and concentration (digit span regression, visual detection, subtraction), memory (word and picture encoding), language (naming, repetition, comprehension, semantic and phonological fluency), reading and writing, as well as conceptual executive functions (similarities and calculation). Motor functions included right‐hand, left‐hand, alternating, and opposite movements.

In addition, single‐leg balance, tandem gait, and Romberg tests were performed, along with tremor evaluation and screening for cerebellar syndrome, sensory ataxia, and other neurological sequelae. A global cognitive impairment index was calculated for each participant based on combined neuropsychological and neuromotor outcomes.

2.4. Oxidative Stress Assessment

Seven oxidative stress biomarkers were quantified in serum samples: superoxide dismutase (SOD) activity, catalase activity, lipid peroxidation (malondialdehyde), glutathione peroxidase (GPx) activity, total glutathione (reduced and oxidized glutathione), biopyrine (BPn), and 8‐hydroxy‐2′‐deoxyguanosine (8‐OHdG). All measurements were performed using commercial kits in accordance with manufacturers' protocols.

SOD activity was determined using the Superoxide Dismutase (SOD) Assay Kit (Sigma‐Aldrich) via a kinetic method; catalase activity with the Catalase Activity Assay Kit (ab83464, Abcam) using an endpoint colorimetric method; malondialdehyde (MDA) with the Lipid Peroxidation Assay Kit (Sigma‐Aldrich) using an endpoint colorimetric method; GPx activity with the Glutathione Peroxidase Assay Kit (Sigma‐Aldrich) via a kinetic method; total glutathione with the Glutathione Assay Kit (Sigma‐Aldrich) via a kinetic method; biopyrine with the Human Biopyrin (BPn) ELISA Kit (MyBioSource) using an ELISA method; and 8‐OHdG with the 8‐OHdG ELISA Kit (ab285254, Abcam), also by ELISA. Absorbance readings were obtained using a Synergy HTX Multi‐Mode Microplate Reader (BioTek).

2.5. FT‐IR Spectroscopic Analysis

Fifty microliters of each serum sample were deposited onto high‐infrared‐transmittance glass slides (MirrIR Low‐E, Kevley Technologies) and air‐dried for approximately 30 min at room temperature (22 ± 2 °C) prior to analysis. FT‐IR transmission spectra were acquired using a Thermo Fisher Nicolet iS50 spectrometer equipped with a DTGS detector, operating in transmission mode. Each spectrum represented an average of 32 scans collected at a spectral resolution of 0.4 cm⁻¹ across the 400–4000 cm⁻¹ range.

Spectral data were expressed as percent transmittance (%) versus wavenumber (cm⁻¹). Prior to analysis, spectra were baseline‐corrected, smoothed using a Savitzky–Golay filter (9 points), and normalized with OMNIC software (Thermo Fisher Scientific) to reduce instrumental noise and enhance comparability among samples.

2.6. Statistical Analysis

To investigate the association between pesticide exposure and the occurrence of neurological or cognitive conditions, a cross‐sectional analytical design was applied. Exposure status was determined through self‐reported pesticide use obtained during structured interviews, while clinical diagnoses were classified into predefined neurological and cognitive categories. Odds ratios (ORs) and their 95% confidence intervals (CIs) were computed. Descriptive statistics were computed for each oxidative stress biomarker, including the mean, standard deviation (SD), standard error of the mean (SEM), median, and range (minimum–maximum). Comparisons between participants who reported pesticide use and those who did not were conducted using the Mann–Whitney U test. Similarly, independent comparisons were performed between the control group (participants without neurological or neurocognitive disorders) and individuals diagnosed with specific conditions, including mild and moderate cognitive impairment, attention deficit, cerebellar hemispheric syndrome, polyneuropathy, and executive function deficits (manual and conceptual/calculative).

FT‐IR data were analyzed using multivariate statistical approaches. Principal component analysis (PCA) was first applied to scaled data to examine overall variability and visualize potential natural clustering among groups. Subsequently, partial least squares discriminant analysis (PLS‐DA) was employed using three principal components to assess the supervised classification capability of the spectra according to each evaluated condition. Variable importance in projection (VIP) scores were calculated to identify the wavenumbers contributing most significantly to group discrimination.

Additionally, independent Student's t‐tests were performed for each spectral frequency, with p value adjustment using the false discovery rate (FDR) correction method to control for multiple comparisons. Group mean intensity profiles and standard deviations were then plotted as a function of wavenumber, highlighting characteristic spectral features associated with the analyzed conditions.

On the other hand, the original dataset was transposed to obtain a sample‐by‐variable matrix, and all stress oxidative variables, were converted to numeric format following data cleaning. FT‐IR transmittance values were extracted as predictor variables, whereas oxidative stress biomarkers were used as response variables. To facilitate supervised classification, each oxidative stress biomarker was dichotomized into low and high categories according to its quartile distribution, with individuals in the lowest quartile classified as the low group and those in the highest quartile classified as the high group; samples within the two intermediate quartiles were excluded from the classification analyses. Supervised discrimination between groups was performed using PLS‐DA with three latent components and autoscaling. VIP scores were calculated to identify the most discriminative FT‐IR wavenumbers. In parallel, each FT‐IR wavenumber was compared between groups using independent two‐sample t‐tests followed by FDR correction for multiple testing. Pearson correlation coefficients were subsequently calculated between each oxidative stress biomarker and all FT‐IR bands to identify spectral regions exhibiting the strongest linear associations. Mean FT‐IR ± standard deviation was calculated for each biomarker category and displayed as multipanel spectral profiles using the conventional reversed wavenumber axis.

To further evaluate the predictive performance of FT‐IR, partial least squares regression (PLS‐R) models were independently developed for each oxidative stress biomarker using leave‐one‐out cross‐validation. Five latent components were initially fitted, and model performance was assessed using the cross‐validated coefficient of determination and the root mean squared error of prediction (RMSEP). To investigate the global relationships between serum molecular fingerprints and oxidative stress biomarkers, hierarchical cluster analysis (HCA) was performed using Pearson correlation coefficients. The analysis was restricted to the 900–1800 cm⁻¹ spectral region, commonly referred to as the biochemical fingerprint region, because it encompasses the major vibrational bands associated with proteins (amide I and II), lipids, carbohydrates, phospholipids, and nucleic acids, thereby providing the most biologically informative molecular signatures of serum composition. To reduce dimensionality while preserving the most informative spectral information, the 100 wavenumbers exhibiting the highest variance within this spectral region were selected to construct the correlation matrix, which was subsequently visualized as a heatmap with hierarchical clustering applied to both rows (wavenumbers) and columns (oxidative stress biomarkers).

All statistical computations and data visualizations were performed using the R programming language (version 4.4.2) within the RStudio integrated development environment.

3. Results

3.1. Pesticide Exposure and Neurocognitive Alterations

The findings revealed a clear association between pesticide exposure and several neurological and cognitive alterations (Table 1). Attention deficit disorder was significantly more prevalent among pesticide users (OR = 4.4, 95% CI = 1.2–15.9), indicating that individuals exposed to pesticides had approximately fourfold higher odds of developing this condition compared to non‐users. Although mild (OR = 3.2) and moderate (OR = 3.9) cognitive impairment also exhibited a trend toward greater frequency among pesticide users, these associations did not reach statistical significance. Conversely, cerebellar hemispheric syndrome (OR = 0.18), polyneuropathy, and right central facial paralysis showed no significant differences between exposure groups. Overall, these results suggest a potential relationship between chronic pesticide exposure and cognitive alterations, particularly attention deficit disorder.

TABLE 1.

Association between pesticide use and neurological or cognitive conditions.

Condition	Odds ratio (OR)	ρ‐value	95% Confidence interval
Attention deficit	4.4 a	0.02	1.2–15.9
Mild cognitive impairment	3.2	0.13	0.71–14.9
Cerebellar hemispheric syndrome	0.18	0.2	0.009–3.3
Polyneuropathy	1.95	0.52	0.26–14.8
Moderate cognitive impairment	3.9	0.28	0.34–45.5
Right central facial paralysis	0.64	0.79	0.03–16.47
Severe cognitive impairment	5.8	0.29	0.23–148.23
Ocular myasthenia gravis	5.8	0.29	0.23–148.23

^a Statistically significant association (p < 0.05).

Serum oxidative stress biomarkers showed variable patterns between pesticide users and non‐users (Table 2). The average activities of antioxidant enzymes (SOD, catalase, and GPx) were comparable between groups, suggesting that overall enzymatic antioxidant capacity was not markedly altered by pesticide use. However, biomarkers of oxidative damage tended to be higher in pesticide users. Mean MDA levels were elevated (14.94 vs. 12.33 nmol/mL), indicating increased lipid membrane damage, while 8‐OHdG concentrations were also higher in the exposed group (631.1 vs. 426.5 ng/mL). Additionally, the substantial increase in BPn (21.99 vs. 11.55 pg./mL) suggests an enhanced generation of reactive oxygen species (ROS). Conversely, total glutathione levels were slightly reduced among pesticide users.

TABLE 2.

Descriptive statistics and comparative analysis of oxidative stress biomarkers between participants with and without pesticide use.

Parameter	Group	Mean	Median	Standard deviation (SD)	Minimum	Maximum	p (Mann–Whitney U test, asymptotic, 2‐tailed)
SOD (U/ml)	No pesticide use	0.678	0.451	0.741	0.139	4.371	0.272
	Pesticide use	0.772	0.585	0.642	0.162	3.303
Catalase activity (mU/mL)	No pesticide use	7.428	7.459	3.902	1.322	15.688	0.257
	Pesticide use	7.405	3.498	7.396	0.577	37.806
Lipid peroxidation (MDA, nmol/mL)	No pesticide use	12.33	10.249	7.75	2.101	35.978	0.562
	Pesticide use	14.936	11.921	13.008	3.388	66.424
Glutathione peroxidase (U/L)	No pesticide use	272.774	265.33	100.206	80.948	734.53	0.329
	Pesticide use	273.155	233.851	151.088	134.914	929.406
Total glutathione (nmol/mL)	No pesticide use	6.902	6.031	3.832	1.885	27.139	0.68
	Pesticide use	6.39	5.277	2.826	1.885	13.193
Biopyrrins (pg/mL)	No pesticide use	11.553	8.428	11.236	1.411	60.247	0.369
	Pesticide use	21.995	9.007	56.879	1.785	369.151
8‐OHdG (ng/mL)	No pesticide use	426.546	131.812	777.306	42.123	4199.746	0.042 a
	Pesticide use	631.125	335.064	983.455	25.416	5141.466

^a Statistically significant association (p < 0.05).

Non‐parametric analysis using the Mann–Whitney U test indicated that, overall, differences between exposed and unexposed groups were not statistically significant (Table 2) for most oxidative stress and antioxidant defense biomarkers (SOD, catalase, MDA, GPx, total glutathione, and BPn; ρ value > 0.05 in all cases). However, a significant difference was observed in serum 8‐OHdG levels (p = 0.042), a biomarker of oxidative DNA damage, suggesting that chronic pesticide exposure may be associated with increased genotoxic oxidative stress.

3.2. Oxidative Damage in Neurocognitive and Neuromotor Impairments

Serum oxidative stress biomarkers were analyzed in participants grouped according to their neurological or neurocognitive status (Table 3). In the control group (without neurological or neurocognitive disorders), the mean SOD activity was 0.64 ± 0.09 U/mL, catalase activity was 6.99 ± 0.50 mU/mL, and MDA reached 13.66 ± 1.28 nmol/mL. The mean GPx activity was 254.06 ± 10.13 U/L, while total glutathione averaged 6.85 ± 0.50 nmol/mL. The mean concentrations of BPn and 8‐OHdG were 12.52 ± 1.55 pg./mL and 452.58 ± 92.71 ng/mL, respectively.

TABLE 3.

Descriptive statistics of oxidative stress biomarkers and comparative analysis of oxidative stress biomarkers between participants with neurological/neurocognitive disorders and controls without cognitive or neurological impairment.

Neurological/neurocognitive condition	Parameter	Mean	Median	Standard deviation (SD)	Minimum	Maximum	Asymptotic sig. (2‐tailed)*	Exact sig. (2‐tailed)*
No neurological and/or neurocognitive disorder	SOD (U/mL)	0.6368	0.4351	0.7064	0.1385	4.3708	—	—
	Catalase activity (mU/mL)	6.9941	7.1706	4.0114	0.577	13.6079	—	—
	Lipid peroxidation (MDA, nmol/mL)	13.6612	11.3208	10.2755	2.1012	66.4237	—	—
	Glutathione peroxidase (U/L)	254.0641	258.5847	81.0812	80.9483	458.7068	—	—
	Total glutathione (nmol/mL)	6.8496	6.031	4.0069	1.8847	27.1394	—	—
	Biopyrrins (pg/mL)	12.5241	8.3226	12.4022	1.4602	60.2471	—	—
	8‐OHdG (ng/mL)	452.5829	199.0333	741.6749	33.8881	4199.7456	—	—
Mild cognitive impairment (MCI) (language, reading, executive functions)	SOD (U/mL)	0.919	0.837	0.559	0.312	1.992	0.047**	—
	Catalase activity (mU/mL)	6.437	3.035	4.641	2.368	11.576	0.758	—
	Lipid peroxidation (MDA, nmol/mL)	16.24	12.393	11.839	5.103	38.551	0.589	—
	Glutathione peroxidase (U/L)	243.273	230.852	72.878	158.898	386.753	0.664	—
	Total glutathione (nmol/mL)	7.646	7.162	2.321	5.277	12.439	0.203	—
	Biopyrrins (pg/mL)	16.762	12.732	13.017	3.906	38.999	0.239	—
	8‐OHdG (ng/mL)	397.077	261.121	369.768	113.533	1032.676	0.643	—
Attention deficit	SOD (U/mL)	0.797	0.585	0.813	0.162	3.303	0.301	—
	Catalase activity (mU/mL)	8.098	6.491	7.955	1.899	31.435	0.38	—
	Lipid peroxidation (MDA, nmol/mL)	9.253	8.533	4.884	3.816	19.254	0.114	—
	Glutathione peroxidase (U/L)	304.305	292.313	160.979	178.386	764.511	0.289	—
	Total glutathione (nmol/mL)	6.872	6.031	1.898	4.9	10.554	0.57	—
	Biopyrrins (pg/mL)	39.283	8.676	99.79	1.46	369.151	0.886	—
	8‐OHdG (ng/mL)	762.172	373.89	1351.144	84.781	5141.466	0.265	—
Cerebellar hemispheric syndrome	SOD (U/mL)	0.745	0.528	0.564	0.21	1.622	0.502	0.521
	Catalase activity (mU/mL)	10.041	9.106	3.326	7.153	15.688	0.277	0.291
	Lipid peroxidation (MDA, nmol/mL)	8.877	9.391	1.403	7.247	10.678	0.276	0.291
	Glutathione peroxidase (U/L)	369.064	274.325	206.815	242.845	734.53	0.168	0.174
	Total glutathione (nmol/mL)	5.428	5.277	1.656	3.015	7.162	0.339	0.359
	Biopyrrins (pg/mL)	14.266	15.179	4.364	9.172	20.334	0.085	0.085
	8‐OHdG (ng/mL)	143.574	87.622	109.846	83.155	337.811	0.061	0.061
Polyneuropathy	SOD (U/mL)	0.897	0.568	0.929	0.186	2.263	0.602	0.624
	Catalase activity (mU/mL)	6.401	5.794	2.982	3.486	10.53	0.855	0.87
	Lipid peroxidation (MDA, nmol/mL)	21.494	20.969	10.417	9.348	34.691	0.078	0.078
	Glutathione peroxidase (U/L)	286.692	305.805	74.021	188.879	346.279	0.368	0.383
	Total glutathione (nmol/mL)	5.371	5.277	0.474	4.9	6.031	0.281	0.304
	Biopyrrins (pg/mL)	7.537	5.527	7.115	1.411	17.682	0.236	0.246
	8‐OHdG (ng/mL)	908.622	287.256	1368.096	107.731	2952.246	0.62	0.642
Moderate cognitive impairment (encoding, reading, recall functions)	SOD (U/mL)	1.113	1.041	0.633	0.519	1.779	0.056	0.056
	Catalase activity (mU/mL)	17.186	11.234	18.382	2.518	37.806	0.458	0.477
	Lipid peroxidation (MDA, nmol/mL)	8.09	7.676	3.641	4.674	11.921	0.262	0.283
	Glutathione peroxidase (U/L)	501.679	287.816	370.422	287.816	929.406	0.089	0.095
	Total glutathione (nmol/mL)	4.775	5.277	2.674	1.885	7.162	0.344	0.372
	Biopyrrins (pg/mL)	9.826	11.745	6.957	2.111	15.623	0.952	0.966
	8‐OHdG (ng/mL)	1191.339	141.71	1873.976	77.404	3354.901	1	1

*From Mann–Whitney U test.

**Statistically significant association (p < 0.05).

Subjects with MCI showed a slight increase in SOD activity (0.92 ± 0.21 U/mL) and lipid peroxidation (16.24 ± 4.48 nmol/mL), along with a reduction in catalase and GPx activity compared to controls. Attention deficit cases exhibited higher GPx activity (304.31 ± 44.65 U/L) and markedly elevated BPn (39.28 ± 27.68 pg./mL) and 8‐OHdG (762.17 ± 374.74 ng/mL) levels, suggesting increased oxidative stress. In the cerebellar hemispheric syndrome group, catalase activity and GPx levels were higher (10.04 ± 1.49 mU/mL and 369.06 ± 92.49 U/L, respectively), whereas total glutathione was comparatively lower. Participants with polyneuropathy displayed the highest mean MDA concentration (21.49 ± 5.21 nmol/mL) and 8‐OHdG (908.62 ± 684.05 ng/mL). Finally, the moderate cognitive impairment (ModCI) group showed the highest mean GPx activity (501.68 ± 213.86 U/L) and 8‐OHdG concentration (1191.34 ± 1081.94 ng/mL).

Non‐parametric comparisons (Mann–Whitney U test) revealed significant differences (Table 3) between the control group and subjects with MCI for SOD activity (U = 121, Z = −1.987, p = 0.047), indicating increased enzymatic antioxidant response in MCI. No other comparisons showed statistically significant differences at p < 0.05, although trends were observed for higher oxidative damage (8‐OHdG and MDA) in cerebellar hemispheric syndrome, polyneuropathy, and ModCI groups.

3.3. Chemical Fingerprints Associated With Cognitive Impairment

The VIP analysis of FT‐IR revealed distinct vibrational bands associated with pesticide exposure and neurological or neurocognitive conditions (Table 4). For pesticide use, the most discriminant wavenumbers were located in the regions of 400–401 cm⁻¹, 1870–1871 cm⁻¹, 2065–2236 cm⁻¹, and 3745 cm⁻¹ (VIP > 4.7; p < 0.05), corresponding to vibrational modes of disulfide bonds (–S–S–), metal–carbonyl complexes, isothiocyanates (–N=C=S), cyanates (–O–C≡N), and free hydroxyl groups, respectively (Figure 1). Subjects with MCI showed significant spectral differences (Figure 2) in the 1933–2167 cm⁻¹ region (VIP > 5.0; p < 0.05), associated with carbonyl and isocyanate stretching modes. Similarly, individuals with attention deficit exhibited relevant absorptions (Figure 2) around 1734–1943 cm⁻¹, typically attributed to C=O stretching of lipids and proteins. In contrast, subjects with hemispheric cerebellar syndrome and polyneuropathy displayed less pronounced (Figure 2), yet consistent, variations within the 1864–2206 cm⁻¹ range (VIP 2.4–4.4), possibly reflecting cumulative oxidative and metabolic disturbances. However, these did not have statistical significance.

TABLE 4.

Variable importance in projection (VIP) scores, wavenumbers, and statistical significance of FT‐IR bands associated with pesticide use and neurological/neurocognitive conditions.

Category	Wavenumber (cm−1)	VIP	p	Adjusted ρ‐value
Pesticide use	400.16	6.23	0.01	1
	2216.81	6.04	0.01	1
	2216.33	5.88	0.01	1
	2217.29	5.78	0.01	1
	400.65	5.73	0.01	1
	2154.13	5.28	0.02	1
	2215.84	5.25	0.02	1
	2153.65	5.24	0.02	1
	2217.77	5.21	0.02	1
	1870.16	5.16	0.02	1
	401.13	5.13	0.03	1
	1870.64	5.09	0.02	1
	2154.62	5	0.03	1
	2064.94	4.96	0.03	1
	2153.17	4.9	0.03	1
	2065.42	4.88	0.03	1
	2236.09	4.83	0.03	1
	2235.61	4.81	0.04	1
	3745.14	4.77	0.03	1
	3745.63	4.75	0.03	1
Attention deficit	1907.77	3.27	0	0.42
	1908.25	3.25	0	0.42
	1942.48	3.22	0.01	0.42
	1942.96	3.15	0.02	0.42
	1907.29	3.07	0.01	0.42
	1908.73	3	0	0.42
	1749.63	2.9	0	0.42
	1749.15	2.89	0	0.42
	1942	2.88	0.01	0.42
	1750.11	2.87	0	0.42
	1748.67	2.85	0	0.42
	1735.17	2.84	0	0.42
	1735.65	2.83	0	0.42
	1736.13	2.82	0	0.42
	1734.68	2.81	0	0.42
	1736.61	2.81	0	0.42
	1739.02	2.8	0	0.42
	1739.51	2.8	0	0.42
	1748.18	2.8	0	0.42
	1738.54	2.8	0	0.42
Mild cognitive impairment	2061.57	7.12	0.03	1
	2062.05	6.9	0.04	1
	2061.08	6.85	0.03	1
	2060.6	6.27	0.04	1
	2062.53	6.15	0.08	1
	2166.19	6.12	0.01	1
	2166.67	6.11	0	1
	2235.13	5.6	0.08	1
	2165.7	5.59	0.02	1
	2011.91	5.57	0.01	1
	2060.12	5.55	0.07	1
	2234.65	5.53	0.06	1
	2167.15	5.51	0	1
	2011.42	5.44	0.01	1
	1932.84	5.42	0	1
	2235.61	5.4	0.12	1
	2012.39	5.37	0.02	1
	1932.36	5.37	0.01	1
	2234.17	5.24	0.05	1
	1933.32	5.22	0.01	1
Hemispheric cerebellar syndrome	2063.98	4.42	0.1	0.97
	2064.46	4.31	0.11	0.97
	2063.49	4.29	0.1	0.97
	2064.94	4.02	0.14	0.97
	2164.74	3.94	0.04	0.97
	2063.01	3.92	0.12	0.97
	2164.26	3.7	0.05	0.97
	2165.22	3.7	0.04	0.97
	1983.94	3.68	0	0.97
	1983.46	3.67	0.01	0.97
	2065.42	3.6	0.18	0.97
	1984.43	3.59	0	0.97
	1982.98	3.54	0.01	0.97
	2205.72	3.47	0.11	0.97
	1984.91	3.45	0	0.97
	2206.2	3.45	0.1	0.97
	2160.88	3.39	0.02	0.97
	2062.53	3.36	0.17	0.97
	2206.68	3.31	0.11	0.97
	2205.24	3.3	0.14	0.97
Polyneuropathy	2198.97	3.51	0.01	0.4
	2199.45	3.48	0.01	0.4
	2063.49	3.28	0.02	0.4
	2198.49	3.27	0.01	0.4
	2199.93	3.2	0	0.4
	2063.97	3.13	0.02	0.4
	2063.01	3.11	0.05	0.4
	2198.01	2.84	0.01	0.4
	2064.46	2.72	0.03	0.4
	2200.42	2.72	0	0.4
	2062.53	2.62	0.12	0.4
	1868.23	2.57	0.06	0.4
	1868.71	2.57	0.07	0.4
	1867.75	2.53	0.05	0.4
	1869.2	2.5	0.07	0.4
	1864.86	2.5	0.01	0.4
	1864.38	2.5	0.01	0.4
	1867.27	2.47	0.03	0.4
	1865.34	2.46	0.01	0.4
	1863.89	2.42	0.02	0.4

FIGURE 1.

Fourier Transform infrared spectral comparison of serum samples from pesticide users and non‐users.

FIGURE 2.

Fourier Transform infrared spectral comparison of serum samples from individuals with and without neurological or cognitive disorders.

FT‐IR analysis of serum samples revealed statistically significant spectral differences (Table 5) exclusively in participants with ModCI, left‐hand motor dysfunction, and conceptual executive calculation deficit (Figures 3, 4, and 5), while no significant associations were observed in other variables evaluated using NEUROPSI. Characteristic signals in the 440–493 cm⁻¹ region corresponded to disulfide (–S–S–) and polysulfide linkages, indicative of oxidative modifications in serum proteins. Absorption bands between 3630 and 3840 cm⁻¹ were attributed to free hydroxyl (O–H) groups of unbound sugars and phenolic metabolites, suggesting alterations in redox homeostasis or serum carbohydrate composition. In the 2089–2255 cm⁻¹ region, signals associated with isothiocyanate (–N=C=S) and cyanate (–O–C≡N) functional groups were detected, possibly arising from protein carbamylation or cyanate metabolism. A distinct band at 2573.58 cm⁻¹ indicated the presence of thiol (–S–H) groups derived from cysteine or reduced glutathione (GSH), whereas absorptions between 2330 and 2340 cm⁻¹ corresponded to endogenous carbonates and bicarbonates. Finally, the signal observed between 1860 and 1900 cm⁻¹ suggested the presence of metal–carbonyl complexes, potentially reflecting interactions involving serum trace metals. Collectively, these spectral features highlight biochemical alterations related to oxidative processes and metabolic dysfunction in individuals with cognitive and motor impairments (Table 6).

TABLE 5.

Variable importance in projection (VIP) scores, wavenumbers, and statistical significance of FT‐IR bands associated with moderate cognitive impairment, motor executive dysfunction (left hand), and conceptual executive dysfunction (calculation).

Cognitive domain/function	Wavenumber (cm−1)	VIP	p	Adjusted ρ‐value
Moderate cognitive impairment	3838.19	2.39	0.01	0.03 a
	3838.68	2.35	0.01	0.03 a
	3837.71	2.33	0.02	0.05
	2089.05	2.3	0.03	0.07
	2088.56	2.24	0.06	0.1
	3839.16	2.24	0.02	0.04 a
	2089.53	2.23	0.02	0.04 a
	3837.23	2.16	0.04	0.08
	2088.08	2.1	0.09	0.14
	2190.29	2.08	0.14	0.2
	2255.38	2.08	0.04	0.07
	2254.9	2.08	0.01	0.03 a
	3839.64	2.07	0.03	0.06
	2189.81	2.04	0.11	0.17
	2255.86	2.04	0.07	0.12
	2190.77	2.03	0.16	0.22
	2090.01	2.02	0.02	0.04 a
	2254.42	2	0.02	0.05
	2256.34	1.98	0.11	0.17
	2573.58	1.97	0.01	0.03 a
Motor executive functions—left hand	1983.94	1.93	0.01	0.05
	1983.46	1.91	0.01	0.05
	491.77	1.91	0	0.03 a
	492.25	1.91	0	0.03 a
	491.29	1.9	0	0.03 a
	492.73	1.88	0	0.03 a
	1893.79	1.88	0	0.03 a
	443.07	1.88	0	0.03 a
	1984.43	1.88	0.02	0.06
	442.59	1.87	0	0.03 a
	1894.27	1.86	0	0.03 a
	490.8	1.86	0	0.03 a
	443.56	1.86	0	0.03 a
	3680.54	1.85	0	0.03 a
	446.93	1.85	0	0.03 a
	447.41	1.85	0	0.03 a
	493.21	1.85	0	0.03 a
	1982.98	1.85	0.01	0.05
	3681.02	1.84	0	0.03 a
3680.06	1.84	0	0.03 a
Conceptual executive functions—calculation	2331.55	1.8	0	0.02 a
	2331.07	1.79	0	0.02 a
	443.56	1.75	0	0.02 a
	2332.04	1.74	0	0.02 a
	443.07	1.73	0	0.02 a
	2330.59	1.72	0	0.02 a
	444.04	1.71	0	0.02 a
	3680.54	1.67	0	0.02 a
	3633.29	1.66	0	0.02 a
	3632.81	1.66	0	0.02 a
	1864.38	1.66	0	0.02 a
	3681.02	1.66	0	0.02 a
	3633.77	1.66	0	0.02 a
	3632.33	1.66	0	0.02 a
	1864.86	1.65	0	0.02 a
	442.59	1.65	0	0.02 a
	3680.06	1.65	0	0.02 a
	3634.26	1.65	0	0.02 a
	3631.84	1.65	0	0.02 a
	3631.36	1.64	0	0.02 a

^a Statistically significant association (p < 0.05).

FIGURE 3.

Fourier Transform infrared spectral comparison of serum samples from individuals with moderate cognitive impairment and those without neurological or cognitive disorders.

FIGURE 4.

Fourier Transform infrared spectral comparison of serum samples from individuals with impaired executive motor function in the left hand and those without motor dysfunction.

FIGURE 5.

Fourier Transform infrared spectral comparison of serum samples from individuals with impaired executive cognitive function in calculation and those without cognitive dysfunction.

TABLE 6.

Characteristic infrared absorption bands and functional group assignments in human serum samples associated with moderate cognitive impairment, left‐hand motor dysfunction, and executive conceptual calculation deficit.

Wavenumber (cm−1)	Wavenumber (cm−1)	Definition	Reference
440–450	Polysulfides (S‐S stretch)	Disulfide bonds in proteins (–S–S–), polysulfide species from oxidized thiols	(Bayu et al. 2023)
490–493	Aryl disulfides (S‐S stretch)	Aromatic disulfide‐containing metabolites or modified proteins with aryl–S–S–aryl linkages	(Bayu et al. 2023)
3630–3690	Primary alcohol, OH stretch	Free sugar hydroxyls (mono‐ and disaccharides), alcohol‐containing metabolites, phenolic OH (unbound)	(Bayu et al. 2023)
3830–3840	Free O–H (not bonded by H)	Primary alcohol (O–H stretch). Free hydroxyl groups of sugars (mono‐ and disaccharides), alcohol‐containing metabolites, or unbound phenolic hydroxyls.	(Bayu et al. 2023)
2089–2090	Isothiocyanate (‐NCS)	Isothiocyanate (–N=C=S stretch). May arise from thiocyanate metabolites	(Bayu et al. 2023)
2254.9	Cyanate (‐OCN stretch)	Cyanate (–O–C ≡ N stretch). Detected in protein carbamylation or cyanate‐derived compounds.	(Bayu et al. 2023)
2573.58	Thiols (S‐H stretch)	Thiols (S–H stretch). From cysteine residues or reduced glutathione (GSH).	(Bayu et al. 2023)
2330–2340	CO2 absorption or carbonate/bicarbonate vibrations	CO2 absorption band or carbonate/bicarbonate asymmetric stretching vibrations.	(Bayu et al. 2023)
1860–1900	Transition metal carbonyls	Metal–carbonyl complexes (organometallics) or possible metal–pesticide complexes	(Bayu et al. 2023)

3.4. Chemical Fingerprints Associated With Oxidative Stress

The VIP analysis consistently identified spectral features within the 1900–2200 cm⁻¹ region as the principal contributors to the discrimination of oxidative stress biomarker categories (Table 7). This spectral interval is dominated by overtone and combination bands that have been associated with cyanate and isothiocyanate‐related vibrations, suggesting that oxidative stress is accompanied by subtle changes in serum molecular composition rather than by alterations in a single biochemical component. The most discriminative regions were centered at approximately 2003–2005 cm⁻¹ for SOD activity, 1985–2135 cm⁻¹ for catalase activity, 2002–2004 cm⁻¹ for GPx, 1932–1934 cm⁻¹ for total glutathione, 1984–2017 cm⁻¹ for BPn, and 2001–2096 cm⁻¹ for 8‐OHdG, whereas MDA exhibited an additional prominent feature near 2156 cm⁻¹. Pearson correlation analysis revealed weak‐to‐moderate linear relationships between individual FT‐IR bands and oxidative stress biomarkers (|r| = 0.25–0.31), with the strongest positive correlations observed for SOD activity (2238.99 cm⁻¹, r = 0.293) and 8‐OHdG (3899.91 cm⁻¹, r = 0.310), while catalase activity, GPx, total glutathione, and BPn exhibited inverse correlations with their dominant spectral bands. These findings indicate that oxidative stress biomarkers are associated with coordinated modifications of the serum molecular fingerprint rather than isolated spectral alterations, highlighting the capacity of FT‐IR to detect subtle biochemical perturbations associated with oxidative damage.

TABLE 7.

Characteristic infrared absorption bands and functional group assignments in human serum samples associated with oxidative stress.

Biomarker	VIP‐based dominant FTIR region (cm−1)	Functional group assignment	Best linear association (cm−1, r)
Superoxide dismutase activity	2003–2005 (VIP ~ 2.66–3.04)	Aromatic combination bands/C ≡ N/NCS (2000–2260 cm−1) (Bayu et al. 2023)	2238.99 (r = 0.293)
Catalase activity	1985–2135 (VIP ~ 2.63–2.73)	Nitriles/isothiocyanates/aromatic combination bands (1990–2175 cm−1) (Bayu et al. 2023)	2004.19 (r = −0.298)
Lipid peroxidation	400–401 and ~2156 cm−1 (VIP up to 4.17)	Low‐frequency region likely instrumental artifacts; 2100–2200 cm−1: nitriles/SCN/C ≡ C (2000–2260 cm−1) (Bayu et al. 2023)	3619.31 (r = −0.258)
Glutathione peroxidase activity	2002–2004 cm−1 (VIP ~ 2.94–3.29)	Aromatic combination bands/nitriles/isothiocyanates (2000–2260 cm−1) (Bayu et al. 2023)	1919.34 (r = −0.310)
Total glutathione	1932–1934 cm−1 (VIP up to 4.78)	Carbonyl‐related combinations/aromatic bands (1660–2000 cm−1) (Bayu et al. 2023)	1958.39 (r = −0.251)
Biopyrrins	1984–2017 cm−1 (VIP ~ 3.0–3.1)	Aromatic combination bands/nitriles/SCN groups (2000–2260 cm−1) (Bayu et al. 2023)	2004.19 (r = −0.288)
8‐Hydroxy‐2‐deoxyguanosine	2001–2096 cm−1 (VIP up to 4.07)	Nitriles/isothiocyanates/aromatic combination bands (2000–2260 cm−1) (Bayu et al. 2023)	3899.91 (r = 0.310)

The average FT‐IR further supported these observations by demonstrating highly conserved global spectral profiles between the low‐ and high‐biomarker groups, with only subtle but reproducible differences in absorbance intensity across several spectral regions (Figure 6). Hierarchical cluster analysis of the correlation matrix reinforced this interpretation by revealing two major clusters of oxidative stress biomarkers (Figure 7). One cluster grouped SOD, catalase activity, total glutathione, BPn, and 8‐OHdG, which exhibited similar positive correlation patterns with the selected FT‐IR variables, whereas glutathione peroxidase formed a distinct cluster characterized predominantly by negative correlations. Furthermore, MDA occupied an intermediate position between these clusters, indicating a partially shared but distinct spectral signature. The clustering of biomarkers with complementary biological functions suggests that FT‐IR captures coordinated molecular responses to oxidative stress, reflecting integrated alterations in serum proteins, lipids, carbohydrates, and other metabolites rather than changes attributable to individual biochemical pathways. Collectively, these findings demonstrate that serum FT‐IR molecular fingerprints provide a systems‐level representation of oxidative stress status and support their potential application as complementary biomarkers for monitoring oxidative stress‐related neurotoxic effects in pesticide‐exposed populations.

FIGURE 6.

Fourier Transform infrared spectral comparison of stress oxidative biomarkers.

FIGURE 7.

Heatmap with hierarchical clustering of oxidative stress biomarkers.

4. Discussion

The results of our study suggest an association between pesticide exposure and the presence of neurological and cognitive impairments, with particular emphasis on attention deficit, which was significantly more frequent among pesticide users (OR = 4.4, 95% CI = 1.2–15.9). These findings suggest that chronic pesticide exposure may disrupt neurobiological processes related to attention, potentially through oxidative stress mechanisms, as supported by previous studies on organophosphate (Tsai and Lein 2021) and carbamate neurotoxicity (Mudyanselage et al. 2023). Although other conditions, such as MCI andModCI, were observed in pesticide users, these associations did not reach statistical significance, potentially due to the limited sample size or interindividual variability in exposure and biological susceptibility. Prenatal or early‐life exposure to low levels of organophosphates has been linked to neurobehavioral disorders, including attention‐deficit/hyperactivity disorder (ADHD), as well as impairments in memory and motor skills. Evidence indicates that organophosphates induce the production of reactive oxygen and nitrogen species, causing progressive oxidative damage and DNA alterations in the brain (Chang et al. 2018). Occupational exposure to chlorpyrifos has also been associated with ADHD symptoms in adolescents (Rohlman et al. 2019).

Consistent with these observations, oxidative stress markers in our study support this interpretation. Serum levels of 8‐OHdG were significantly elevated in pesticide users (p = 0.042), indicating enhanced oxidative DNA damage. 8‐OHdG is widely recognized as a biomarker of oxidative stress on genetic material and is associated with mitochondrial dysfunction and neuronal degeneration (Graille et al. 2020). Previous studies have reported elevated urinary and serum 8‐OHdG levels in children and adults exposed to organophosphates (Ding et al. 2012; Pandiyan et al. 2021). Additional markers of oxidative stress, including MDA and BPn, were also higher in pesticide users, although not reaching statistical significance. Exposure to pesticides such as chlorpyrifos has been linked to increased ROS production, leading to mitochondrial membrane alterations and lipid peroxidation, as reflected by elevated MDA levels (Labitta et al. 2019; Montanarí et al. 2024). BPn, products of bilirubin oxidation by ROS, further suggest heightened oxidative stress (Bakry et al. 2016). In contrast, antioxidant enzyme activities including SOD, catalase, and GPx activity did not differ significantly between groups, potentially reflecting adaptive exhaustion of enzymatic defenses in response to sustained oxidative damage.

Similarly, we observed that serum SOD activity was significantly elevated in individuals with MCI, which may reflect oxidative stress induced by pesticide exposure. Previous studies have reported increased SOD activity in individuals exposed to pesticides, likely as a compensatory response to elevated ROS levels (Vidyasagar et al. 2004; Hundekari et al. 2013; Sule et al. 2022). Serum levels of 8‐OHdG, BPn, and MAD were also higher in participants with MCI, although these differences were not statistically significant. In contrast, catalase, glutathione peroxidase, and total glutathione levels were lower in the MCI group, consistent with previous reports showing decreased catalase activity and increased MDA levels in individuals with MCI (Ioannidou et al. 2025; Aly et al. 2024).

In participants with ADHD, levels of 8‐OHdG, BPn, and SOD activity were elevated, though not statistically significant, aligning with evidence of increased oxidative stress markers in ADHD patients (Predescu et al. 2024). Similarly, in participants with polyneuropathy, 8‐OHdG, SOD activity, and MDA levels were higher, reflecting the elevated ROS and reduced endogenous antioxidant defenses reported in peripheral polyneuropathy (Mallet et al. 2020). In the ModCI group, SOD, catalase activity, and 8‐OHdG levels were elevated, consistent with prior findings linking cognitive decline to increased oxidative stress and decreased glutathione levels (Hajjar et al. 2018). Participants with cerebellar hemispheric syndrome also exhibited higher levels of SOD and catalase activities and BPn, likely reflecting oxidative stress in brain regions critical for memory, learning, and emotional regulation, including the hippocampus, amygdala, and prefrontal cortex. Vulnerability of hippocampal subregions (CA1, CA3, and dentate gyrus) to ROS can impair synaptic plasticity and neurogenesis, ultimately compromising neuronal connectivity and cognitive function (Salim 2017).

FT‐IR analysis of dried serum revealed several bands indicative of biochemical alterations associated with oxidative stress, serving as potential biomarkers of ModCI. Signals in the 3630–3690 cm⁻¹ region were attributed to O–H stretching of primary alcohols, corresponding to free hydroxyl groups in mono‐ and disaccharide sugars, metabolites, or unlinked phenols (Bayu et al. 2023). Individuals with MCI, left‐hand motor impairment, or calculation deficits exhibited higher levels of this chemical group than controls. This variation may reflect increased glycosylation of proteins, lipids, or glycans, which has been associated with neurological disorders (Pradeep et al. 2023). Moreover, ROS generated by oxidative stress can directly modify glycans or disrupt glycosyltransferase and glycosidase activity (Khoder‐Agha and Kietzmann 2021). Additional bands at 3830–3840 cm⁻¹, corresponding to free O–H groups not involved in hydrogen bonding, were also elevated in individuals with cognitive or motor impairments, supporting the hypothesis of oxidative stress–related glycosylation modifications and indicating the presence of free sugars or phenolic groups.

In the mid‐to‐high wavenumber region, notable differences were observed at 2573.58 cm⁻¹, corresponding to S–H stretching vibrations of thiols (Bayu et al. 2023) in individuals with ModCI. The increased S–H signal in the FT‐IR may reflect elevated levels of free or reduced thiols, potentially indicating an activation of the antioxidant system in response to a prior redox imbalance. In the context of MCI, this could represent an early compensatory mechanism against increased ROS.

Additionally, signals at 2089–2090 cm⁻¹ (assigned to isothiocyanates, –NCS) and 2254.9 cm⁻¹ (assigned to cyanates, –OCN) were also elevated in ModCI participants. These species are not typically present in human serum and may indicate protein carbamylation processes. Cyanate ions can be generated through myeloperoxidase‐catalyzed oxidation of pseudohalide thiocyanate ions or via urea decomposition into hydrocyanic acid and its conjugate base, cyanate. Hydrocyanic acid exists in equilibrium with its reactive form, isocyanic acid, which reacts with nucleophilic protein groups, such as amino groups. Carbamylation is thus part of cellular damage mechanisms associated with metabolic, inflammatory, and oxidative disorders and has been linked to atherosclerosis, as well as potential modifications of tau protein in Alzheimer's disease (Verbrugge et al. 2015).

Furthermore, participants with impairments in left‐hand executive function or calculation exhibited significantly different bands in the 440–450 cm⁻¹ and 490–493 cm⁻¹ regions, corresponding to disulfide and polysulfide bond vibrations (Bayu et al. 2023). This indicates elevated serum levels of disulfides and polysulfides compared to controls. In various pathologies, increased disulfide levels or higher disulfide‐to‐thiol ratios (SS/TT) signify a shift toward an oxidative state, reflecting oxidative stress or cellular damage. For instance, in Alzheimer's disease, SS/TT ratios are elevated, indicating oxidative damage (Erel and Erdoğan 2020). Collectively, these chemical groups identified in individuals with MCI, left‐hand executive impairment, or calculation deficits appear to be directly associated with oxidative stress and may serve as potential biomarkers of neurocognitive dysfunction.

Importantly, many of the discriminative FT‐IR frequencies identified in relation to oxidative stress biomarkers were located within the same spectral regions previously associated with neurocognitive and motor dysfunction in this study. In particular, alterations within regions assigned to isothiocyanates and cyanates showed alterations both in participants with elevated oxidative stress biomarkers and in those exhibiting cognitive or motor deficits. The recurrence of these spectral features across independent analyses suggests that oxidative stress may be a major driver of the serum molecular fingerprints associated with neurological impairment. This interpretation is consistent with the well‐established role of reactive oxygen species in promoting lipid peroxidation, protein oxidation, DNA damage, mitochondrial dysfunction, and neuroinflammation, processes that collectively contribute to neuronal injury and cognitive decline (Hassan et al. 2022).

Interestingly, one of the spectral regions associated with neurological impairment in the present study (approximately 2089–2090 cm⁻¹) has been tentatively assigned to isothiocyanate‐related vibrations. Although FT‐IR cannot identify individual metabolites, this spectral assignment is noteworthy because isothiocyanates are potent activators of the Nrf2/Keap1 antioxidant pathway, one of the major cellular defense mechanisms against oxidative stress. Activation of Nrf2 promotes the transcription of Phase II detoxification enzymes, including glutathione S‐transferases and heme oxygenase‐1, while preserving intracellular glutathione homeostasis and suppressing NF‐κB‐mediated inflammation (Sita et al. 2016).

Another noteworthy finding was the identification of spectral bands around 2254.9 cm⁻¹, tentatively assigned to cyanate (–OCN) groups, which complements the alterations observed in the isothiocyanate‐related region. Cyanate is the precursor of protein carbamylation, a non‐enzymatic post‐translational modification generated through the reaction of isocyanic acid with free amino groups, particularly the ε‐amino group of lysine residues. Besides its formation from urea dissociation, isocyanic acid can also be generated during inflammatory responses through the myeloperoxidase‐mediated oxidation of thiocyanate, linking protein carbamylation to both oxidative stress and chronic inflammation. Carbamylation has been shown to modify protein charge, conformation, and biological activity while increasing susceptibility to oxidative damage, thereby contributing to the accumulation of dysfunctional proteins in several chronic disorders (Carracedo et al. 2018).

The primary brain regions involved in calculation include the left dorsolateral prefrontal cortex, the left angular gyrus, and the left anterior cingulate cortex (Figure 8) (Arsalidou et al. 2018). Within the dorsolateral prefrontal cortex, specific neuronal circuits rely on tightly regulated intracellular calcium and cAMP levels. During stress, a signaling cascade increases cAMP and intracellular Ca²⁺, primarily mediated by activation of dopaminergic D1 and adrenergic β₁ receptors. This activates protein kinase A (PKA), which phosphorylates calcium channels and NMDA receptors, facilitating excessive Ca²⁺ influx and release from the smooth endoplasmic reticulum via ryanodine receptors and IP₃ receptors. The resulting calcium overload disrupts cellular homeostasis and stimulates ROS production in mitochondria and the SER (Joyce et al. 2025).

FIGURE 8.

Areas of the brain involved in left‐hand executive processes and conceptual calculation.

In parallel, ROS‐induced demethylation of CpG sites occurs in the hippocampus and cingulate cortex, altering epigenetic marks essential for memory formation and consolidation. DNA oxidation, particularly as 8‐oxoguanine, can modify methylation patterns and induce epigenetic instability through repair mechanisms such as the base excision repair pathway, mediated by OGG1. This can affect the conversion of 5‐methylcytosine to cytosine, ultimately influencing gene expression in regions critical for synaptic plasticity (Kandlur et al. 2020). Thus, oxidative stress directly impacts the function of brain regions involved in calculation, suggesting that the chemical groups identified in FT‐IR may reflect these molecular alterations.

Regarding executive motor function of the left hand, the principal regions include the right primary motor cortex, right supplementary motor area, and right premotor cortex (Figure 8) (Grabowska et al. 2012; Hardwick et al. 2013). Oxidative stress disrupts the excitability of pyramidal neurons in the motor cortex. Experimental induction of oxidative stress using cumene hydroperoxide in cortical neurons has demonstrated alterations in both passive (e.g., resting membrane potential) and active (e.g., action potential firing rate) properties, leading to cortical hyperexcitability. This early phenomenon, observed in conditions such as ALS, manifests as decreased motor thresholds, fasciculations, and exaggerated reflexes. The hyperexcitability is linked to modifications in ion channels (e.g., Nav1.6, Kv1), impaired glutamate reuptake by glial cells, and intracellular calcium overload. Enhanced neuronal excitation further generates ROS, creating a self‐perpetuating cycle that exacerbates neurodegeneration. Consequently, oxidative stress and pyramidal neuron hyperexcitability represent a key pathological axis underlying motor cortex dysfunction in Amyotrophic lateral sclerosis and other neurodegenerative diseases (Pardillo‐Díaz et al. 2022). Nevertheless, this study has several limitations. First, its cross‐sectional design precludes establishing causal relationships between pesticide exposure, oxidative stress, serum molecular fingerprints, and neurocognitive impairment. Second, pesticide exposure was assessed through self‐reported occupational history rather than biological monitoring, preventing identification of the specific pesticides, exposure duration, intensity, or cumulative dose, and introducing the potential for recall bias, exposure misclassification, and social desirability bias. Third, although exposed and non‐exposed groups were compared, factors such as age, sex, diet, lifestyle, and other environmental exposures may have influenced oxidative stress biomarkers and serum FT‐IR molecular fingerprints. Finally, FT‐IR provides global molecular fingerprints rather than direct identification of individual compounds; therefore, the assignment of specific functional groups remains tentative and should be confirmed using complementary analytical techniques.

5. Conclusion

FT‐IR identified reproducible serum molecular fingerprints characterized by spectral changes associated with oxidative stress biomarkers and neurocognitive dysfunction, particularly within the 1900–2300 cm⁻¹ region, suggesting coordinated alterations in protein, lipid, and redox metabolism rather than isolated biochemical changes. The overlap between oxidative stress‐associated spectral signatures and those identified in individuals with cognitive and motor impairments further supports the hypothesis that oxidative damage is a major contributor to the molecular alterations underlying neurological dysfunction in pesticide‐exposed populations. Collectively, these findings demonstrate that integrating conventional oxidative stress biomarkers with FT‐IR provides a complementary systems‐level approach for identifying serum molecular signatures of pesticide‐induced neurotoxicity and highlights the potential of FT‐IR as a rapid, minimally invasive tool for screening and monitoring oxidative stress‐related neurocognitive impairment.

Author Contributions

Jhonatan Rabanal‐Sanchez: conceptualization, investigation, writing – original draft, writing – review and editing, methodology, formal analysis, data curation, visualization. Carmen Rosa Silva‐Correa: conceptualization, investigation, writing – original draft, writing – review and editing, methodology. Miguel Angel Burgos‐Flores: conceptualization, investigation, writing – original draft, writing – review and editing, methodology. Kevin Jesus Mayma‐Aguirre: conceptualization, investigation, writing – original draft, writing – review and editing. Jimmy Andreyvan Cainamarks‐Alejandro: writing – original draft, writing – review and editing, investigation, conceptualization. Jonh Maximiliano Astete‐Cornejo: conceptualization, investigation, writing – original draft, writing – review and editing, supervision, project administration, resources, funding acquisition. Jaime Rosales‐Rimache: conceptualization, investigation, writing – original draft, writing – review and editing, supervision, validation.

Funding

This work was funded by Programa Nacional de Investigación Científica y Estudios Avanzados (PROCIENCIA) of Consejo Nacional de Ciencia, Tecnologia e Innovacion Tecnologica (CONCYTEC), PROCIENCIA ‐ CONCYTEC, under the Basic Research Projects call E041‐2023‐01 [Contract No. PE501083171‐2023‐PROCIENCIA].

Disclosure

The work described has not been published previously. The article is not under consideration for publication elsewhere. The article's publication is approved by all authors and tacitly or explicitly by the responsible authorities where the work was carried out.

Ethics Statement

The study protocol was reviewed and approved by the Institutional Ethics Committee of the National Institute of Health, Lima, Peru (Director's Resolution No. 062‐2024‐DIIS/INS, dated February 22, 2024). Ethical guidelines were followed to ensure participant confidentiality and well‐being. Before sample collection, all participants provided written informed consent in accordance with ethical research standards.

Conflicts of Interest

The authors declare no conflicts of interest.

Rabanal‐Sanchez, J. , Silva‐Correa C. R., Burgos‐Flores M. A., et al. 2026. “Serum Molecular Fingerprints of Oxidative Stress Underlying Neurocognitive and Motor Dysfunction in Pesticide‐Exposed Populations.” European Journal of Neuroscience 64, no. 3: e70645. 10.1111/ejn.70645.

Data Availability Statement

Data are available from the corresponding author on reasonable request.