Increased Sister Chromatid Exchange in Peripheral Blood Lymphocytes from Humans Exposed to Pesticide : Evidence Based on a Meta-analysis

Pesticides are a group of natural or synthetic chemical substances including insecticide, herbicide and fungicide, being designated to combat plagues that generally attack, harm or transmit illness to living organisms including humans. Since the mid-1940s, a large number of synthetic pesticides have been introduced in the market. At present, the pesticide manual includes 900 main entries and lists over 2600 products (Bolognesi et al., 2011). Each year, large amounts of pesticides are set free into the environment and many of them are known to have adverse biological effects on non-target organisms including humans. Human exposure to pesticides can occur via dermal contact, inhalation, ingestion, or across the placenta (Gilden et al., 2010). Most pesticides are acutely and chronically toxic to humans. Chronic health effects associated to pesticide exposures included neurological effects, reproductive or developmental problems and carcinoma. Epidemiological studies have shown that there was an association between pesticide exposure and increased risk of several human cancers (Alavanja et al., 2004; Mink et al., 2008; Shim et al., 2009; Shakeel et al., 2010; Balasubramaniam et al.,


Introduction
Pesticides are a group of natural or synthetic chemical substances including insecticide, herbicide and fungicide, being designated to combat plagues that generally attack, harm or transmit illness to living organisms including humans.Since the mid-1940s, a large number of synthetic pesticides have been introduced in the market.At present, the pesticide manual includes 900 main entries and lists over 2600 products (Bolognesi et al., 2011).Each year, large amounts of pesticides are set free into the environment and many of them are known to have adverse biological effects on non-target organisms including humans.Human exposure to pesticides can occur via dermal contact, inhalation, ingestion, or across the placenta (Gilden et al., 2010).

Literature source and analytical methods
We conducted a systematic search through the database of Medline/PubMed, Embase and web of science, the search terms utilized were: "sister chromatid exchange", "pesticide", "insecticide", "fungicide", "herbicide" and "lymphocyte".The ending date of search was up to September 30, 2013.Additional relevant references quoted in the searched articles were also selected.
Criteria of literature inclusion: (a) The papers should be published in English; (b) The papers should include pesticide exposure and the frequency of SCE in human peripheral blood lymphocytes; (c) The paper must contain the exposed groups and control groups; (d) The paper must have the sample size, arithmetic means and standard deviations (SD) or the information that can be used to infer the results.Accordingly, reviews and overlapping or repeated literatures were excluded.For repeated studies or overlapping studies, the publication with more information was selected.
In total, 33 published studies were identified with the frequency of SCE in human peripheral blood lymphocytes of pesticide-exposed population.We reviewed all papers in accordance with the criteria defined above and excluded four overlapping articles and two papers without offering full information.Therefore, twenty-seven studies were determined to enter our study.Among them, data from two studies (Rupa et al., 1989;Carbonell et al., 1990) were included in the stratified analysis by smoking status only, owing to that they had the same population as Rupa et al's study (Rupa et al., 1991) and Carbonell et al's study (Carbonell et al., 1993), respectively.

Data extraction
For each study, characteristics such as first author, year of publication, country of studied population, duration, arithmetic means and standard deviations or standard errors, sample size of exposed groups and control groups and covariates accounted for were noted.Two of the authors independently tabulated the data and inputted them from these eligible studies to an electronic database.We estimated the summary arithmetic means and standard deviations, if the study provided stratum information, the data coming from similar stratum were combined to make a full use of them.Characteristics of individual study were summarized in Table 1.

Quantitative data synthesis
To assess the relationship between SCE frequency and exposure to pesticide, we conducted a meta-analysis of identified studies.Data were combined using either a fixed-effects model or a random-effects model Bauchinger (Bauchinger et al., 1982) 1982 Workers in plant 11.41±6.37years Germany Smoking a Carbonell (Carbonell et al., 1990(Carbonell et al., ) 1990 Agricultural workers More than 10 years Spain Smoking Carbonell (Carbonell et al., 1993) 1993 Agricultural workers 170.7 ± 179.1 hours per year Spain Agricultural activity and duration Costa (Costa et al., 2006) 2006 Agricultural explorations and Range from 0.5 to 48 years Portugal Sex and smoking greenhouses c Ergene (Ergene et al., 2007) 2007 Subjects living in region 34.56±10.47years Turkey Smoking contaminated with pesticides De Ferrari (De Ferrari et al., 1991) 1991 Floriculturists Unknown Italy Age and smoking Gomez-Arroyo 1992 Agricultural workers Range from 1 to 35 years Mexico Smoking, drinking (Gomez-Arroyo et al., 1992) and duration Gomez-Arroyo (Gomez-Arroyo et al., 2000) 2000 Floriculturists in greenhouses 7.7±3.8years Mexico Sex c Hatjian (Hatjian et al., 2000) 2000 Agricultural college Short exposure Unite Kingdom students dipping sheep Hoyos (Hoyos et al., 1996) 1996 Farm workers 16.5±8.8years Colombia Age, cigarette smoking and drinking Jablonicka (Jablonicka et al., 1989) 1989 Pesticides production Unknown Czechoslovakia Unknown Joksic (Joksic et al., 1997) 1997 Agricultural workers 12.1± 6.02 years Yugoslavia Unknown Kourakis (Kourakis et al., 1996) 1996 Pesticides sprayers Unknown Greece Unknown Lander (Lander et al., 1995) 1995 Greenhouse sprayers An average of 17 (range 1-50) years Denmark Smoking and age Linnaimaa (Linnainmaa, 1983) 1983  (Padmavathi et al., 2000) 2000 Pesticide industry Range from 1 to 24 years India Smoking and duration Pasquini (Pasquini et al., 1996) 1996 Farmers 18.35±12.42years Italy Age, smoking and duration c Rowland (Rowland et al., 2007) 2007 War veterans Unknown New Zealand Age, smoking and drinking Rupa (Rupa et al., 1988) 1988 Vegetable gardens 20.24±7.77years India Smoking and duration b Rupa (Rupa et al., 1989) 1989 Applicators in cotton fields 8 hours per day in the spring and winter India Duration and smoking seasons (range 1-25 years) Rupa (Rupa et al., 1991) 1991 Applicators in cotton fields 8 hours per day and 9 months per year India Duration Scarpato (Scarpato et al., 1996) 1996 Greenhouse floriculturists Unknown Italy Smoking and sex Shaham (Shaham et al., 2001) 2001 Greenhouse farmers Mean exposure period: 28.3 years Israel Age, smoking, (range 2.5-55.5 years) education and origin Steenland (Steenland et al., 1997) 1997 Applicators Unknown Mexico Age and smoking c Suarez (Suarez et al., 2003) 2003 Workers exposed pesticides Unknown Spain Unknown by tap water Zeljezic (Zeljezic et al., 2002) 2002  (DerSimonian et al., 1986).The Cochrane Q statistics test was performed for the assessment of heterogeneity among studies.A fixed-effects model is employed when the effects are assumed to be homogenous, while a randomeffects model is employed when they are heterogeneous.We computed the weighted mean difference and 95% confidence intervals (95%CI) for each study.Egger's test was used to test publication bias (Egger et al., 1997).Begg's rank correlation test was also used to check the publication bias (Begg et al., 1994).
The meta-analysis was performed with Review Manager (Version 5.0.24,The Cochrane Collaboration) and STATA 10.0 software package (Stata Corporation, College Station, Texas).All the tests were two-sided, and a P value of <0.05 for any test or model was considered to be statistically significant difference.

Meta-analysis database
We established a database according to the extracted information from each article.Some general information was listed in Table 1.It indicates first author, year of publication, exposure to pesticide, duration, country of studied population and covariates accounted for.There were a total of 25 studies with 1214 exposures and 1068 controls.

Test of heterogeneity
Table 2 shows the difference in SCE frequency between pesticide-exposure groups and control groups.The heterogeneity of the 25 studies was analyzed.Our results showed that all meta-analyses on SCE frequency had heterogeneity with P value<0.05.Therefore, we estimated the summary weighted mean difference for them with a random-effects model.

Bias diagnosis
Publication bias was assessed in this study.The result for SCE in total population was an asymmetric funnel plot (Figure 2).Our results from both Egger's test and Begg's test suggested that publication bias might not have a significant influence on the summary estimate of SCE among total population, smokers, pesticide applicator, pesticide producer, other exposure population,

Figure 1. Meta-analysis is Conducted on Sister Chromatid Exchange (SCE) Among Total Population.
Each estimate of weighted mean difference on SCE per cell is designated by a solid square, and the 95% confidence intervals (95%CI) of each subgroup is shown by transverse line.The blank rhombus at the bottom is the pooled estimate of weighted mean difference by a random-effects model.Asian population, European population and American population.Maybe, there was publication bias in the meta-analysis for nonsmokers because there was some uncertainty with a P value being equal to 0.05 in Egger's test.

Discussion
Sister chromatid exchange was first visualized by Taylor in plant cells using tritium and autoradiography, which provided poor spatial resolution.It was later discovered that incorporation of the DNA base analog 5'-bromodeoxyuridine, in combination with Hoechst dye 33258 staining, would differentiate sister chromatids and reveal SCE (Wilson et al., 2007).SCE results from S-dependent lesions which during S-phase are eventually transformed to double strand break (DSB) and these can be repaired by a homologous recombination mechanism that may give rise to SCE.It has been described as a sensitive convenient method for monitoring exposure to environmental genotoxic agents (Anwar, 1994).
Several studies have shown that pesticide produced genotoxic effects in assays performed in vitro.Nikoloff et al reported that herbicide flurochloridone induced a significant and equivalent increase in SCEs regardless of the concentration in Chinese hamster ovary cells treated for 24h within the 0.25-15μg/ml concentration range (Nikoloff et al., 2012).Afugan induced a significant increase in the frequency of SCE in cultured human lymphocytes at all concentrations (2.5, 5, 10 and 20 μg/ ml) at both 24h and 48h treatment periods, compared with the negative controls (Yuzbasioglu et al., 2006).Federico et al reported that a statistically significant increase of SCE was observed in epithelial liver cell lines by four phenylurea herbicides (fenuron, chlorotoluron, diuron and difenoxuron), compared with the negative controls (Federico et al., 2011).200.0μg/ml of Permethrin induced a significant increase in SCE frequency over control values in cultured human lymphocytes (Turkez et al., 2012).Fipronil, a phenylpyrazole pesticide induced a statistically significant increase in the SCE frequency in a dose-dependent manner in human peripheral blood lymphocytes, compared with a negative control (Celik et al., 2014).
Our meta-analysis showed that the frequency of SCE in human peripheral blood lymphocytes was significantly higher in the pesticide-exposed groups than control groups as evidenced by a random-effects model, where the summary estimate of weighted mean difference was 1.69 (95%CI: 1.01-2.38).Our findings indicated that exposure to pesticide could induce significantly increased levels of chromosome damage in human peripheral blood lymphocytes measured by SCE, which might be an indicator of early genetic effects for pesticide-exposed population.
The availability of reference values is important for laboratories and research teams to validate protocols and analytical procedures and for epidemiologists to estimate the statistical power of field studies and assess the quality of data (Neri et al., 2005).Baseline Micronuclei frequencies for the cytokinesis block assay in adults and children have been published (Bonassi et al., 2001;Neri et al., 2005).In this paper, we provided the SCE baseline values of referent population from 25 papers included in this meta-analysis at the same time, which might be used for the planning phase of a study, but not for a standard to select proper controls in the conduct of future studies.doi.org/10.7314/APJCP.2014.15.22.9725 Sister Chromatid Exchange and Human Exposure to Pesticide There are some limitations inherent in this metaanalysis.Firstly, only published literatures were included in this study, which might cause publication bias.Both Egger's test and Begg's test were carried out to address this issue.Our results showed that the likelihood of key publication bias might not be present in this metaanalysis except for nonsmokers since there was some uncertainty with the P value being equal to 0.05 in Egger's test.Secondly, several factors such as age, sex, cigarette smoking, duration, exposure type and levels of environmental exposure might affect the frequency of SCE in human peripheral blood lymphocytes (Bonassi et al., 1995;Bolognesi et al., 1997;Martinez-Valenzuela et al., 2009;Ben Salah et al., 2011).Cigarette smoking and exposure type were stratified in this meta-analysis for SCE frequency further, and we also observed significantly higher frequency of SCE in human peripheral blood lymphocytes among smokers or nonsmokers, applicator, producer and other exposure population in comparison with their corresponding control groups, respectively.However, other confounders were not stratified in this meta-analysis, since only a few investigators reported such results and stratified range was not uniform for some factors.Thirdly, since studies included in this meta-analysis were heterogeneous, we performed sensitivity analysis further, and found that the studies were homogenous for SCE among total population while some studies were excluded.
Considering that the origin of studied population might have an effect on the SCE frequency among subjects exposed to pesticide, country of studied population was stratified in this meta-analysis further.We only observed that there was significantly increased frequency of SCE in exposed groups among Asian population, compared with control groups.The association was not found in European population and American population, maybe due to the different level of environmental exposure to pesticide among the different origin of country.
In summary, the findings show a significant increase in frequency of SCE in human peripheral blood lymphocytes of pesticide-exposed population.However, our metaanalysis was conducted on population-based studies, meta-analysis based on individual data might provide more precise and reliable results.Therefore, it is necessary to construct an international database on genetic damage among population exposed to pesticide that may contain all raw data of studies examining pesticide-genetic toxicity.It is required for authors of all of the published papers to share their raw data.With sufficient individual data, it may be likely to resolve this problem of confounding factors including age, sex, cigarette smoking, duration, exposure type and levels of environmental pesticide exposure.

Figure 3 .
Figure 3. Sensitivity Analysis is Conducted on SCE.Each estimate of weighted mean difference on SCE per cell is designated by a solid square, and the 95% confidence intervals (95%CI) of each subgroup is shown by transverse line.The blank rhombus at the bottom is the pooled estimate of weighted mean difference by a fixed-effects model

Figure 2 .
Figure 2. Funnel Plot Analysis is Used to Detect Publication Bias on SCE.Each point represents a separate study for the indicated association.For each study, the mean difference (MD) is plotted on as size-effect against the precision (standard error, SE).The line in the centre indicates the summary diagnostic MD.If bias is absent, small studies will have MD that are widely scattered but still centered round the MD estimates provided by large, more precise studies