Is exposure to pesticides associated with biological aging? A systematic review and meta-analysis

Objective: Exposure to pesticides is a risk factor for various diseases, yet its association with biological aging remains unclear. We aimed to systematically investigate the relationship between pesticide exposure and biological aging. Methods: PubMed, Embase and Web of Science were searched from inception to August 2023. Observational studies investigating the association between pesticide exposure and biomarkers of biological aging were included. Three-level random-effect meta-analysis was used to synthesize the data. Risk of bias was assessed by the Newcastle-Ottawa Scale. Results: Twenty studies evaluating the associations between pesticide exposure and biomarkers of biological aging in 10,368 individuals were included. Sixteen reported telomere length and four reported epigenetic clocks. Meta-analysis showed no statistically significant associations between pesticide exposure and the Hannum clock (pooled β = 0.27; 95 %CI: (cid:0) 0.25, 0.79), or telomere length (pooled Hedges ’ g = (cid:0) 0.46; 95 %CI: (cid:0) 1.10, 0.19). However, the opposite direction of effects for the two outcomes showed an indication of possible accelerated biological aging. After removal of influential effect sizes or low-quality studies, shorter telomere length was found in higher-exposed populations. Conclusion: The existing evidence for associations between pesticide exposure and biological aging is limited due to the scarcity of studies on epigenetic clocks and the substantial heterogeneity across studies on telomere length. High-quality studies incorporating more biomarkers of biological aging, focusing more on active chemical ingredients of pesticides and accounting for potential confounders are needed to enhance our understanding of the impact of pesticides on biological aging.


Introduction
Pesticides are defined as substances or mixtures of substances released deliberately into the environment to manage and control insects, weeds and rodents (Kim et al., 2017).Despite being primarily developed to protect crops and enhance agricultural productivity, these chemicals are also used in greenspaces (gardens, parks, sports fields), public infrastructure (sidewalks, graveyards, construction sites), transportation networks (roads, railways, airports) and veterinary practices (de Graaf et al., 2024;Nowell et al., 2021).Approximately 3 billion kilograms of pesticides are globally used annually, with a budget of around $40 billion (Sharma et al., 2020).Based on their target, pesticides are categorized mainly into herbicides, insecticides, fungicides and rodenticides.Most of the pesticides persist for many years, affecting non-target organisms and thereby being ubiquitously present in the environment (Sun et al., 2018).People may be occupationally exposed to pesticides during farming, manufacturing or application (Figueiredo et al., 2022), and/or through inhalation of residual gas.Likewise, exposure can occur via the consumption of pesticide-contaminated food or beverages, domestic use of pesticide-containing products, or residing in proximity to agricultural areas (Damalas and Eleftherohorinos, 2011;Ottenbros et al., 2023;Simões et al., 2022).
Aging is a multifaceted and ever-degrading phenomenon that begins with alterations at the cellular level and gradually erodes the resilience and integrity of tissues and organs throughout time, resulting in impaired function and increased susceptibility to death (López-Otín et al., 2013).Due to growing life expectancy and declining fertility rates, the population is aging rapidly across the world.It is estimated that the number of people aged over 65 years worldwide will rise from 761 million in 2021-1.6 billion in 2050 (Affairs et al., 2023).Importantly, there is considerable between-person variation in the rate of aging and this indicates that chronological age alone does not sufficiently capture individual differences in the aging process (Zhou et al., 2022).In recent years, the concept of biological age, defined by the level of age-dependent biological changes such as molecular and cellular damage accumulation, has emerged as a promising measure for capturing the diversity underlying aging (Moqri et al., 2023).
To evaluate the actual state and differences in the rate of aging for each individual, a series of biomarkers of aging have been developed.These include epigenetic clocks (Belsky et al., 2022;Horvath and Raj, 2018), telomere length (Zglinicki, 2002), transcriptomic predictors, proteomic predictors, metabolomics-based predictors and composite (clinical) biomarker predictors (Jylhävä et al., 2017;Prattichizzo et al., 2024).Among these biomarkers, the most widely utilized are epigenetic clocks and telomere length.Epigenetic clocks, also referred to as DNA methylation based-clocks, capture information from ten to thousands of cytosine-phosphate-guanine (CpG) sites using machine learning techniques (Williams et al., 2023).The first generation clocks, represented by Horvath, Skin and Blood clock and Hannum, exhibited remarkable accuracy in predicting chronological age (Hannum et al., 2013;Horvath, 2013;Horvath et al., 2018).The second generation clocks further incorporated clinical phenotypic age markers linked to DNA methylation measures and were trained to predict mortality risk and age-related diseases (Levine et al., 2018;Lu, Quach, et al., 2019).The most recent third generation clock DunedinPACE, estimated the pace of aging and reflects the overall integrity of several organ systems based on mixed-effects growth modeling of longitudinal change in 19 biomarkers (Belsky et al., 2022).Accelerated epigenetic age /pace of aging estimated by these clocks is associated with an elevated risk of mortality from all-natural causes, even after accounting for known risk factors (Duan et al., 2022;Fransquet et al., 2019).Faul et.al found that second and third generation clocks were significant predictors of later life health outcomes including cognitive dysfunction, functional limitations and chronic conditions (Faul et al., 2023).A meta-analysis demonstrated a significant correlation between first and second generation clocks and incidence of cancer (Oblak et al., 2021).Nonetheless, different epigenetic clocks were found to exhibit varying degrees of associations with risk factors and outcomes, indicating their potential to capture distinct aspects of the aging process (Horvath and Raj, 2018).
Another biomarker of aging is the length of telomeres, which are repetitive DNA repeats at the end of chromosomes that protect them from degradation, end to end fusion and recombination.Telomeres progressively shorten with every cycle of cell division, leading to the initiation of cellular senescence.It is commonly believed that constant accumulation of senescent cells may lead to dysfunctional tissues and organs, thereby accelerating the aging process and subsequent occurrence of age-related diseases (Fasching, 2018;Jylhävä et al., 2017;Moix et al., 2024).Telomere length has been employed as a reliable marker of biological aging and to predict an individual's health condition.However, the variety of approaches used to quantify telomere length sometimes led to varying outcomes, raising concerns about its accuracy (Lin and Yan, 2005).
Besides epigenetic clocks and telomere length, other aging indicators such as Klemera-Doubal method (KDM) biological age and methylation estimator of telomere length (DNAmTL) have also been extensively used in epidemiological studies (Halabicky et al., 2024;Van der Laan et al., 2022;Wang et al., 2024).KDM biological age integrates information on the integrity of multiple organ systems in the body and has good reliability in predicting mortality (Levine, 2013).DNAmTL, based on 140 CpGs and validated in leukocytes from several cohorts, showed a strong association with age and outperformed several limitations of the classical techniques (Lu, Seeboth, et al., 2019).
So far, a growing body of research has shown that exposure to environmental pollutants such as air pollution, heavy metals or chemicals is associated with accelerated biological aging (Khodasevich et al., 2023;Liang et al., 2024;Wang et al., 2024).As a common class of environmental chemicals, exposure to pesticides is also considered a risk factor for aging.While several studies have evaluated the links between pesticide exposure and epigenetic clocks, the evidence remains inconclusive.A prior systematic review and meta-analysis revealed shorter telomere length in occupationally exposed subjects, yet the number of studies included was small and exclusively among occupational populations (Passos et al., 2022).To date, there is no systematic review or meta-analysis on the association between pesticide exposure and biological aging using different biomarkers.To address this gap and better understand the relationship between pesticides and biological aging, we conducted a systematic review and meta-analysis.

Methods
This systematic review and meta-analysis was reported in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines (Page et al., 2021).The protocol was registered in PROSPERO (CRD42023392845).

Eligibility criteria
Inclusion criteria were applied according to PECO (Population, Exposure, Comparator, Outcome) statement (Table 1).Epidemiological studies using cohort, case-control, and cross-sectional designs and published in English were included.Related systematic reviews were not included but used to scan for references.Ecological studies, methodological papers, and conference abstracts were excluded.

Search strategy
PubMed, EMBASE and Web of Science were searched for studies published from inception to August 2023.The search strategy was developed for each database, utilizing Emtree in Embase, and combining free text and MeSH terms in PubMed for retrieval.All pesticides, categorized by their targets (insecticides, herbicides, fungicides, fumigant, bactericides, and rodenticides), were searched across the databases.Outcomes comprised epigenetic clocks, telomere length, KDM biological age, homeostatic dysregulation, allostatic load and biological health score, along with their broad categories including biological aging, accelerated aging and epigenetic age.Detailed search terms were shown in Table A. 2. Reference lists of included papers and relevant reviews were manually scanned for additional publications.

Study selection
References were de-duplicated in Endnote software X9.All titles and abstracts were screened for relevance then full-text examination was conducted independently by two reviewers (SZ and VG).Any discrepancies were resolved by discussion.The following information from the included articles was extracted: first author, year of publication, country, study design, sample size, age and gender of participants, type of pesticide(s), exposure setting and assessment, biomarker(s) of biological aging, statistical model, covariates adjustment, main results including maximally adjusted effect sizes along with standard error (SE) or 95 % confidence intervals (CI).
For studies reporting values of aging biomarkers between exposed and non-exposed groups, sample size, mean and standard deviation (SD) were retrieved from each group.Data presented as median and interquartile range (IQR) were converted to mean and SD (Wan et al., 2014).When results of multiple independent subgroups were reported in one study, they were merged into a single group (Higgins et al., 2019).
If data were only presented in graphs, WebPlotDigitizer (https://a utomeris.io/WebPlotDigitizer/,Version 4.7) was used to extract numerical values.When studies had overlapping populations and information, the one with larger sample size or provided more information (SD, 95 %CI or p-value) was selected.Corresponding authors were contacted for missing data.

Risk of bias assessment
Two reviewers (SZ and VS) independently assessed the risk of bias by using the Newcastle-Ottawa Scale (NOS) for cohort, case-control studies (https://www.ohri.ca/programs/clinical_epidemiology/oxford.asp), and the adapted version for cross-sectional studies (Herzog et al., 2013).A score of more than 7 was considered high quality, 6-7 was considered moderate quality and a less than 6 was considered low quality.The higher the score, the lower the risk of bias.Any disagreements in scoring between the two reviewers were resolved by the third reviewer (VG).

Synthesis of results
From the initial search, telomere length and epigenetic clocks were the two biomarkers of biological aging identified.Synthesis of results encompassed meta-analysis, sensitivity analysis and moderator analysis.Studies that could not be quantitatively analysed in meta-analysis were summarized in a narrative form.

Meta-analysis
Meta-analysis was performed where data were available in at least two studies.To account for the multiple dependent effect sizes due to multiple pesticides reported within one population, a three-level random-effects model was employed (Harrer et al., 2021).Contrary to the traditional univariate meta-analytic approach, which only considers sampling variance of the extracted effect sizes and variance between studies, three-level meta-analysis additionally models the variation of effect sizes within studies (Assink and Wibbelink, 2016;Cheung, 2014).
All studies using epigenetic clocks as an outcome reported linear relationship between pesticide exposure as a ln-transformed continuous variable and different clocks.In studies which used telomere length as an outcome, various exposure assessment methods led to reporting of different types of effect sizes (i.e., β coefficient, mean difference).The corrected standardized mean difference (Hedges' g) on the raw scale was chosen as the effect size for the meta-analysis, representing the mean difference in telomere length between individuals with higher and lower exposure to pesticides.Positive values indicated longer telomere length in individuals with higher exposure compared to those with lower exposure, while negative values suggested shorter telomere length.The application of Hedges' g maximized the inclusion of the results in the meta-analysis.Since the majority of studies reported telomere length in exposed and non-exposed groups, Hedges' g could be directly calculated.Moreover, effect size transformation methods were available to convert β coefficients from linear regression models with binary and continuous exposure variables into Hedges' g.Formulas for calculation and transformation were presented in Appendix B.
Heterogeneity was estimated by the I 2 statistic using the var.comp function in demeta (Cheung, 2014).The total amount of heterogeneity was decomposed into: a) sampling variance at level 1, b) within-study heterogeneity at level 2 (I 2 within ), and c) between-study heterogeneity at level 3 (I 2 between ).Statistical significance of heterogeneity was tested using Cochran's Q statistic.

Sensitivity analysis
Outlier detection and influence diagnostics were performed to elucidate the source of heterogeneity and test the robustness of the pooled effect estimate.Effect sizes falling below the first quartile or above the third quartile of the interquartile range (IQR) were considered outliers (Gómez Penedo and Flückiger, 2023).Effect sizes were considered influential if their Cook's distance exceeded three times the mean Cook's distance (Cook, 1977;Ellis et al., 2023).Sensitivity analysis was carried out by removing outlier effect sizes, influential effect sizes, low-quality studies, effect sizes with transformations and with no adjustment of confounders.

Moderator analysis
Additional moderator analysis was performed using omnibus tests to explore potential sources of heterogeneity across the following moderators: chemical class of pesticide (organophosphate or organochlorine), gender (male or female), exposure setting (occupational or environmental) and expression of telomere length (T/S ratio, bp or kb/genome).

Publication bias
Publication bias was assessed through visual inspection of funnel plots in combination with the Egger's test.All statistical analyses were conducted using the metafor and dmeta packages of R statistical software (version 4.3.0).

Study selection
The initial search retrieved 2079 records, and 2 additional records were identified through reference section of the selected systematic reviews.After removing 540 duplicates, 1541 were screened by title and S. Zuo et al.
Despite the overlapping population between (Hou et al., 2013) and (Andreotti et al., 2015), they were treated as two separate studies and included in the systematic review, since the size of overlap was small (40/1234 and 40/568).

Characteristics of the studies
Summary characteristics of the studies were described in Table 2 and  Table 3. Twenty studies evaluating the associations between pesticide exposure and two biomarkers of biological aging in 10368 individuals aged between 7 and 92 years (6647 females and 3721 males) were included.Sixteen studies reported telomere length and four studies reported epigenetic clocks.The vast majority were cross-sectional studies, with only three studies using cohort design.The studies were carried out in South America (n = 6), North America (n = 5), Europe (n = 4), Asia (n = 4) and Africa (n = 1).More studies assessed occupational exposure (55 %) compared to environmental exposure (40 %), while the sample in one study was composed of occupationally exposed and nonoccupationally exposed populations (Ali et al., 2023).Studies primarily assessed exposure through concentration of pesticide metabolites in biological samples (blood/urine).Other approaches included using occupation as a surrogate for exposure and collecting self-reported data via questionnaires.

Risk of bias
The quality scores of the studies were in the range of 5-9 points (Table A. 5 and Table A. 6).A total of 9 studies were rated high quality, considering the comparability of subjects in different groups, adjustments for important confounding variables, and assessing exposure in biological samples, through validated or well-described tools.Eight studies were considered as moderate quality, and three were categorized as low quality.The reasons for the low quality were related to the lack of description of exposure tool and insufficient control of confounding factors.

Associations between pesticide exposure and epigenetic clocks
Four studies assessed the association between pesticide exposure and epigenetic clocks (Table 2 and Table A. 3      cross-sectional components to investigate the prenatal (cohort) and childhood (cross-sectional) pesticide exposure in relation to children's epigenetic age (de Prado-Bert et al., 2021).The total sample size of the included studies was 3672, comprising 2267 males and 1405 females.Three studies focused on middle-aged and elderly individuals, with ages ranging from 45 to 70 years, while one study was conducted among children, with an average age of 7 years.A total of 25 metabolites from 6 chemical categories were reported, with organochlorine and organophosphate pesticides being the most analyzed.One study included a wide range of pesticides including chlorophenoxy (Dicamba, 2,4-D), chloroacetanilides (Acetochlor, Metolachlor), Pyridine (Picloram) and Trazines (Atrazine) (Hoang et al., 2021).Most studies focused on more than one epigenetic clock, including Horvath (Hoang et al., 2021;Lucia et al., 2022), Hannum (Hoang et al., 2021;Lind et al., 2018;Lucia et al., 2022), PhenoAge (Hoang et al., 2021;Lucia et al., 2022), DunedinPoAm (Lucia et al., 2022), GrimAge (Hoang et al., 2021), and Horvath Skin and Blood (de Prado-Bert et al., 2021;Hoang et al., 2021).Among these, the Hannum clock was the most frequently utilized clock, being employed in three studies.
Three studies showed positive significant associations between metabolites of four pesticides (p,p'-DDE, DDT, TNC, and AMPA) and accelerated epigenetic aging (Hoang et al., 2021;Lind et al., 2018;Lucia et al., 2022).On the contrary, one study reported a negative association between dimethyldithiophosphate (DMDTP) and epigenetic age; the researchers interpreted this as the result of higher intake of fruits and vegetables.However, the results did not change substantially after additionally adjusting for fruit intake, vegetable intake or urinary hippurate, a metabolite marker of fruits and vegetables (de Prado-Bert et al., 2021).
Two studies reporting five effect sizes of the association between concentration of pesticide metabolites and the Hannum clock were included in the meta-analysis.Results were presented in the forest plot (Fig. 2).Overall, a positive but non-significant association was found between pesticide exposure and epigenetic age using the Hannum clock (pooled β= 0.27; 95 %CI: − 0.25, 0.79).23.83 % of the overall heterogeneity was attributable to between-study differences (I 2 between ), while 42.57% was attributable to within-study differences (I 2 within ).
Nine studies reported results on a total of 72 active chemical ingredients of pesticides, mainly belonging to the organochlorine and organophosphate classes, followed by carbamate.The remaining seven studies reported on pesticide mixtures.In all studies, telomere length was assessed using the quantitative polymerase chain reaction (qPCR) method.Specifically, telomere length was expressed as the ratio of telomere repeat copy to the relative number of a single copy gene (T/S ratio) in ten studies, as base pairs (bp) in four studies, and as kb/diploid genome in two studies.Overall, the association between pesticide exposure and telomere length was not consistent (Table A . 4).Ten studies reported shorter telomeres, while two studies reported longer telomeres (Cosemans et al., 2022;Duan et al., 2017).One study found shorter telomeres with diethylphosphate (DE) derived metabolites but not with dimethylphosphate (DM) derived metabolites (Ali et al., 2023).In one study, a negative association was found between 3,5,6-trichloro-2-pyridinol (TCPY) and telomere length, while a positive association was found for diethyl thiophosphate (DETP) (Ock et al., 2020).Furthermore, one study indicated an inverted U-shaped change in telomere length with increasing pesticide concentration (Shin et al., Fig. 2. Forest plot for meta-analysis of association between pesticide exposure (ln-transformed) and Hannum clock.2010), and two studies found no overall association between pesticide exposure and telomere length (de Oliveira et al., 2019;dos Santos et al., 2022).
One study was excluded from the meta-analysis because it reported neither standard error, confidence intervals, nor p-values (Shin et al., 2010).Of the 15 studies included in the primary meta-analysis, 11 studies presented means of telomere length in exposed and non-exposed groups, allowing for the calculation of mean difference and subsequently Hedges' g.In the remaining studies, results were converted from β coefficients from linear regression models with binary and continuous exposure variables into Hedges' g (Cosemans et al., 2022;Guzzardi et al., 2016;Karimi et al., 2020;Ock et al., 2020).
Results of the meta-analysis on association between pesticide exposure and telomere length were shown in the forest plot (Fig. 3).Overall, telomere length in higher-exposed populations was shorter compared to the lower exposed population, but the association fell short of statistical significance (pooled g= − 0.46; 95 %CI: − 1.10, 0.19).99.15 % of the heterogeneity was explained by the between-study, with 0.39 % explained by the within-study.

Sensitivity analysis
Sensitivity analysis showed that excluding effect sizes identified as outliers, through transformation, or those without confounder adjustments did not have a strong influence on the results, with the pooled effects consistently remaining negative.

Moderator analysis
A summary of moderator analysis results was presented in Table 5.Chemical classes of pesticides, gender, exposure settings, and expression of telomere length did not moderate the effect of pesticide exposure on telomere length.

Publication bias
Funnel plot showed an asymmetrical distribution (Fig. A.1), and Egger's test was statistically significant (P = 0.0051), suggesting publication bias favoring studies with null-effect sizes.

Discussion
To our knowledge, this is the first systematic review summarizing the current evidence on the association between pesticide exposure and biological aging.In this systematic review, epigenetic clocks and telomere length were the two biomarkers of biological aging captured from existing research in relation to exposure to pesticides.Using a three-level meta-analysis to quantitatively synthesize data, we found that pesticide exposure was not statistically associated with either epigenetic clock or with telomere length.However, the opposite direction of the pooled effects for both outcomesthe positive association with Hannum clock and the negative with telomere length, showed a possible indication of biological aging acceleration in relation to pesticide exposure.Importantly, after removal of influential effect sizes or low-quality studies, pesticide exposure was significantly associated with shorter telomere length.

Pesticides and epigenetic clocks
A minority of the associations tested in the included studies showed a significant association between pesticides and epigenetic clock, mostly but not always in the direction of accelerated aging.Of the eight statistically significant associations with pesticides, six were positive, while the remaining two were negative.The inconsistent results may be attributed to variations in study populations.In three of the four studies, one focused solely on males, one on females, one on children.Generally, the meta-analysis showed a positive but non-significant association between pesticide exposure and the Hannum clock.However, the results should be interpreted with caution due to the low number of studies included in the meta-analysis, and the inability to analyze other epigenetic clocks.On one hand, in contrast to the Hannum clock, which was developed to estimate chronological age, the second-generation clocks demonstrated superior performance in predicting age-related clinical phenotypes and all-cause mortality (McCrory et al., 2020).On the other hand, in comparison to the Horvath clock algorithm, designed as a robust multi-tissue age predictor based on DNA methylation at 353 CpG sites, the Hannum clock serves as a blood-based estimator, characterized by DNA methylation at only 71 CpG sites (Fransquet et al., 2019).Across the results reported in the studies included in our review, the effects of pesticides were predominantly found on first generation clocks and not on second or third generation clocks.These pesticides included p, p'-DDE, DDT, TNC, and AMPA.In prior literature, second and third generation clocks were more frequently reported in significant association with environmental pollutants (Blechter et al., 2023;Boyer et al., 2023;Wang et al., 2024).In our review, only two studies evaluated the associations between 17 active ingredients of pesticides and more than one generation of clocks, limiting our ability to investigate the differential associations with pesticides between the different clocks (Hoang et al., 2021;Lucia et al., 2022).However, both studies consistently found only significant associations of pesticide exposure with first generation epigenetic clocks and not with second or third generation clocks.The explanation for these differences may be found in the specific phenotypes that these clocks are trained on: the second generation uses specific inflammatory biomarkers in blood that are associated with mortality and the third generation uses longitudinal changes in 19 indicators of organ-system integrity.Maybe these clocks are too specific to capture the different mechanisms by which pesticide exposure may exert an effect on biological aging including neurotoxicity and endocrine disruption.More studies are needed in the future to investigate the differential effect of pesticide exposure on the different clocks.
DNA methylation patterns change dynamically, with alternations in some DNA methylation sites linearly correlated with age (Duan et al., 2022;Horvath et al., 2012).Existing evidence from both adults and children suggests an association between organophosphate exposure, memory impairment, neuropsychiatric issues, and neurodegenerative diseases (Sarailoo et al., 2022).A recent systematic review revealed that exposure to pesticides was associated with cardiovascular diseases (Zago et al., 2022).Furthermore, some epigenome-wide association studies (EWAS) identified differential methylation in relation to pesticides.A case-control study found correlations between organophosphate exposure and methylation levels at 70 significant CpGs, with overrepresentation of genes related to GABA-B receptor II and endothelin signaling pathways, which had impact on neurotransmitter release and inflammation, respectively (Paul et al., 2018).Another study in the Netherlands also reported differential methylation related to occupational pesticide exposure, and further found that differential methylation at specific genomic locations induced by pesticides may play a role in airway disease pathogenesis (Plaat et al., 2018).One recent study by (Hoang et al., 2021) found that glyphosate was linked to 24 differentially methylated CpG sites, while 162 CpG sites were discovered in association with eight pesticides (acetochlor, atrazine, dicamba, malathion, metolachlor, mesotrione, picloram, and heptachlor).

Pesticides and telomere length
An existing systematic review on occupational pesticide exposure and telomere length included six studies that only reported data on exposed and non-exposed groups (Passos et al., 2022).Expanding on this, with sixteen studies included, we took into account both occupational and environmental exposure routes, and different exposure assessment methods, resulting in an enrichment of the meta-analysis.The findings showed a negative association between pesticide exposure and telomere length, which however did not reach statistical significance.Moderator analysis made no significant change but showed a concordant direction in the pooled results.Despite the fact that the potential mechanisms are not entirely understood, oxidative stress (Janoš et al., 2023;Mehta et al., 2008) and inflammatory response (Lopes-Ferreira et al., 2023) are two of the main pathways through which pesticides might eventually damage telomeres.Due to the high guanine content in certain telomere sequences and the deficiency in the repair of single strand breaks, telomeres are extremely sensitive to damage by oxidative stress (Honda et al., 2001;Hou et al., 2013).Additionally, it has been demonstrated that pesticides potentially induce inflammation in different experimental models, manifested through skin irritation, respiratory impairment, or systemic effects with effects on telomere length (Liu et al., 2023).Nonetheless, some studies reported associations of telomere lengthening with adverse health outcomes such as breast cancer (Li and Ma, 2022), gastric cancer (Wang et al., 2018), and lung cancer (Doherty et al., 2018;Tsatsakis et al., 2023), which most likely result from high telomerase activity and prolonged cell survival (Aviv et al., 2017).
Substantial heterogeneity between studies persisted, which could not be explained after moderator analysis.It may be attributed to systematic differences in characteristics of the study populations as well as to the variation in exposure assessments.Pesticide exposure methods included biomonitoring, questionnaires and use of occupation as a proxy for exposure.A wide range of active ingredients of pesticides were examined; with seven studies focusing on pesticide-mixtures.Pesticides are a heterogeneous group of chemically diverse active ingredients, even within their broad application group of insecticides, herbicides and fungicides.The mixtures might mix chemicals with no health effects and chemicals with a rather significant health effect (Karalexi et al., 2021).In sensitivity analysis, after removing the influential effect sizes or the low-quality studies, statistically significant associations of pesticide exposure with shorter telomere length were found, which highlighted the importance of longitudinal studies with larger sample size and proper adjustment for confounders in producing meaningful results.Regarding publication bias, the funnel plot suggested that studies showing a significant negative association between pesticide exposure and telomere length could be missing, and, consequently, our pooled effect size could be underestimated.

Strengths and limitations
This is the first comprehensive systematic review and meta-analysis examining the association between pesticide exposure and biological aging.Considering that the majority of studies reported effects of more than one pesticide within one study, a three-level meta-analytic model was employed to include all effect sizes in our analysis.This approach not only addresses the dependency of effect sizes but also ensures that all available information is preserved, allowing for the attainment of maximum statistical power (Assink and Wibbelink, 2016).Besides, existing effect size transformation methods were applied to homogenize the effect sizes and enhance the interpretability.Finally, we tested the robustness and explored the heterogeneity through sensitivity and moderator meta-analysis.
Nevertheless, several limitations should be noted.Substantial between-study heterogeneity persisted in the meta-analysis of pesticide exposure and telomere length, despite the implementation of moderator analysis, which may hinder producing reliable results.Secondly, the predominance of cross-sectional studies included weakened the pooled level of evidence.Thirdly, the meta-analysis on epigenetic clocks inlcuded only two studies, limiting the amount of usable evidence and the ability to conduct further analysis.Fourthly, a large portion of studies only reported broad categories of pesticide mixtures rather than the active chemical ingredients, making it difficult to evaluate individual substances.Lastly, only two biomarkers of biological aging were retrieved in this systematic review, restricting our ability to generalize the results to the overall aging process.

Conclusion
This systematic review and meta-analysis does not allow drawing a final conclusion on association between pesticide exposure and biological ageing measured as epigenetic clocks or as telomere length.However, both analyses are suggestive of a possible positive association between pesticide exposure and accelerated aging, although this is not always firmly supported by statistically significant pooled effect estimates.Inadequate study quality, high heterogeneity, publication bias in studies reporting on telomere length, and the small number of studies reporting on epigenetic clocks, are the main constraints limiting our inference.Future studies conducted on large samples coming from highquality cohort samples, integrating more biomarkers of biological aging, and assessing exposure of pesticides through biomonitoring with refined adjustments for potential confounders are warranted.
S.Zuo et al.

Fig. 3 .
Fig.3.Forest plot for meta-analysis of association between pesticide exposure and telomere length.
S.Zuo et al.

Table 2
Characteristics of studies on pesticide exposure and epigenetic clocks.
S.Zuo et al.

Table 3
Characteristics of studies on pesticide exposure and telomere length.

Table 3
(continued ) (continued on next page) S.Zuo et al.

Table 4
Sensitivity analysis on association between pesticide exposure and telomere length.

Table 5
Moderator analysis on the association between pesticide exposure and telomere length.