Determining whether environmental estrogens are associated with the risk of prostate cancer may have important implications for our general understanding of this disease. The estrogenic insecticide chlordecone was used extensively in the French West Indies, contaminating the population for more than 30 years. We analyzed the relationship between exposure to chlordecone and the risk of prostate cancer.

We investigated 623 men with prostate cancer and 671 controls. Exposure was analyzed according to case-control status, using either current plasma concentration or a cumulative exposure index based on years of exposure. We genotyped two single-nucleotide polymorphisms (rs3829125 and rs17134592) in the gene encoding chlordecone reductase.

We found a significant increase in the risk of prostate cancer with increasing plasma chlordecone concentration (odds ratio [OR], 1.77; 95% CI, 1.21 to 2.58 for the highest tertile of values above the limit of detection [LD]; P trend = .002) and for cumulative exposure index (OR, 1.73; 95% CI, 1.04 to 2.88 for the highest quartile; P trend = .004). Stronger associations were observed among those with a positive family history of prostate cancer and among those who had lived in a Western country. The rs3829125 and rs17134592 allele variants were in complete linkage disequilibrium and were found at low frequency (0.04). Among subjects with plasma chlordecone concentrations above the LD, carriers of the allele variants had a higher risk of prostate cancer (OR, 5.23; 95% CI, 0.82 to 33.32).

These findings support the hypothesis that exposure to environmental estrogens increases the risk of prostate cancer.

Prostate cancer is the most commonly diagnosed cancer among men in developed countries. However, little is known about the risk factors associated with this cancer. Advancing age, ethnic origins, and a family history of prostate cancer are the only established risk factors.1,2 Many lifestyle-related risk factors have been implicated, but their roles in prostate cancer etiology remain unclear.1,2 The effect of exposure to synthetic chemicals in the environment on prostate cancer development remains a matter of debate with implications for public health. Chemicals with hormonal properties, also called endocrine disruptors, are thought to be involved.3,4 However, no epidemiologic evidence of a positive link between environmental exposure to endocrine chemicals and prostate cancer has yet been established.

Chlordecone (also known as Kepone) is an organochlorine insecticide with well defined estrogenic properties.5,6 It was extensively used from 1973 to 1993 in the French West Indies, to control the banana root borer. A few years after the introduction of chlordecone, the widespread pollution of soils, river water, wild animals, and vegetables growing in polluted soils was reported.7,8 This pesticide undergoes no significant biotic or abiotic degradation in the environment, so permanently polluted soils and waters have remained the primary source of foodstuffs contamination, and human beings continue to be exposed to this chemical.913

Chlordecone is a potential carcinogen and has been shown to cause hepatic tumors in laboratory rats and mice.14 The carcinogenic and hormonal properties of chlordecone and its long biologic half-life raise the possibility of long-term effects, such as cancer. We tested the hypothesis that chlordecone exposure in adulthood, over a 30-year period, favors the development of prostate cancer in the French West Indies, where the incidence of this disease is particularly high.15 Chlordecone reductase catalyzes the reduction of chlordecone to chlordecone alcohol, increasing the biliary excretion of chlordecone and decreasing its toxicity.16 We therefore hypothesized that the prostate cancer-causing effects of chlordecone would be stronger in individuals with functional variants of the chlordecone reductase gene, rs3829125 and rs17134592, resulting in the production of a protein with low levels of enzymatic activity.

We report the results of a case-control study in which chlordecone exposure was evaluated by measuring its concentration in the blood.

Study Population

This study was conducted in Guadeloupe, a French archipelago in the Caribbean, most of the 405,000 inhabitants of which are of African descent. This study was carried out on 709 consecutive incident cases of histologically confirmed prostate cancer (June 2004 to December 2007) and 723 controls without prostate cancer (January 2005 to December 2006). For details of the selection of cases and controls, see the Appendix (online only).

Trained nurses obtained information for both patients and controls, concerning demographic characteristics, anthropometrics, periods and places of residence since birth, lifestyle, occupational data, family history of prostate cancer, medical background, use of medication, and prostate cancer screening history. Weight, height, and the circumferences of the waist and hips were measured at the end of the face-to-face interview. Participants were also asked to provide a blood sample. The study was approved by the Guadeloupean ethics committee for studies involving human subjects. Each participant provided written informed consent.

Laboratory Procedures and Data Management and Statistical Analyses

Chlordecone analysis, total lipid determination, genotyping methods and linkage analysis are described in the Appendix (online only).

A reference date was assigned to each participant, corresponding to the date of histopathologic diagnosis of prostate cancer for patients and the last date of prostate-specific antigen (PSA) determination for controls. Plasma chlordecone concentrations were classified according to their distribution among control samples. Values equal to or below the limit of detection (LD) were used as the reference group and values above the LD were grouped into tertiles. Based on the realistic assumption that the Guadeloupean population has been continuously exposed from 1973 to the present day, we obtained a cumulative exposure index score for each subject. This score was calculated as the product of current plasma chlordecone concentration and the number of years of residence in the French West Indies between 1973 and the year of the reference date. We calculated such scores for subjects with plasma chlordecone concentrations above LD. The cumulative exposure scores were grouped into quartiles, based on their distribution among controls, the lowest quartile being the reference group.

Baseline covariates were compared between patients and controls (Table 1). Odds ratios (OR) and 95% CIs for the risk of prostate cancer were estimated by unconditional logistic regression. We investigated whether covariates were confounding factors or effect modifiers, by looking at the association between chlordecone exposure and prostate cancer separately for each of the following covariates: age, Caribbean origin, education, body mass index, waist to hip ratio, alcohol use, smoking, past urogenital infections, diabetes (type 2), viral infections, prostate cancer screening history (during the last 5 years), family history of prostate cancer (first degree), banana farming, past residence in Western countries, year of reference date, total plasma lipid concentration, and the series and batch number of chlordecone analysis. Missing data for covariates varied from none to lower than 5%, except for waist to hip ratio and genotype, for which 17% and 8.6%, respectively, of the data were missing. Categorical covariates, including missing data as an additional response, were modeled as dummy variables representing the different levels. Confounding covariates were included into the logistic model if they modified the OR estimates by more than 10%. Tests for linear trend across exposure categories were performed, with the natural log-transformed chlordecone concentration treated as a continuous variable. For values equal to or below the LD, we assigned a value of the LD divided by 2. We tested for a modifier effect, by creating an interaction term based on the cross product of these variables and exposure categories (treated as a continuous variable). Interactions were examined by entering the interaction terms into the logistic regression model. We quantified the significance of the interaction term using the Wald test. We used polytomous logistic regression to estimate risk simultaneously among controls and non ordered subgroups of cases, as a function of age at onset (< 60 v > 60) and prostate cancer stage (local v regional/distant stage). We also used a composite index of prostate cancer aggressiveness at diagnosis: high aggressive potential (PSA > 30 ng/mL or regional/distant stage or Gleason score of 4 + 3 and higher) and low aggressive potential (PSA < 30 ng/mL, local stage, and Gleason score of 3 + 4 and lower).17 P values lower than .05 were considered significant. All statistical tests were two tailed and were carried out with SAS version 9.1 (SAS Institute Inc, Cary, NC).


Table 1. Baseline Characteristics of Patients and Controls

Table 1. Baseline Characteristics of Patients and Controls

Characteristic Patients (n = 623)
Controls (n = 671)
No./Total No. % No./Total No. %
Median age, years 66.2 60.6 < .01
    Interquartile range 60.5-71.5 54.0-67.1
Caribbean origin < .01
    French West Indies (Guadeloupe or Martinique) 96.5 91.4
    Haiti or Dominica 3.5 8.6
Education .03
    High school and higher 13.3 10.7
    Secondary 25.4 31.9
    Primary 61.4 57.4
Body mass index .49
    Normal (< 25) 44.0 47.0
    Overweight (25-29.9) 44.3 41.0
    Obese (≥ 30) 11.7 12.0
Waist-to-hip ratio > 0.95 45.4 30.1 < .01
Current or past alcohol use 86.8 82.9 .05
Current or past smoking 38.3 37.1 .69
Past urogenital infection 16.1 17.2 .65
Type 2 diabetes 18.1 12.3 < .01
Viral infection 23.5 24.1 .66
Prostate cancer screening history 50.7 13.4 < .01
Family history of prostate cancer
    No 55.9 78.0 < .01
    Yes 24.4 10.2
    Do not know 19.7 11.8
Banana farming 11.9 10.0 .27
Past residence in Western countries, > 1 year§ 29.9 23.9 .02
Median duration of residence in Western countries§ 15.1 13.2 .04
    Interquartile range 7.0-26.3 6.5-24.8
Median years of residence in French West Indies from 1973 to reference date 32.8 32.7 .43
    Interquartile range 31.5-33.8 30.0-33.3
Median plasma chlordecone, μg/L 0.44 0.40 .39
    Interquartile range < 0.25-1.00 < 0.25-0.86
Chlordecone reductase SNP (No./total No. with data)
        Wild-type C 1,137/1,182 96.2 1,161/1,208 96.1 .91
        Variant allele G 45/1,182 3.8 47/1,208 3.9
        Wild-type C 1,137/1,182 96.2 1,161/1,208 96.1 .91
        Variant allele G 45/1,182 3.8 47/1,208 3.9
Median PSA levels, ng/mL 8.75
    Interquartile range 6.00-14.30
Gleason score
    ≤ 7 (3 + 4) 82.0
    4 + 3 and > 7 18.0
    T1c or T2 and N0 and M0 13.8
    T3 or T4, or N+ or M+ 86.2

Abbreviations: SNP, single nucleotide polymorphism; PSA, prostate-specific antigen.

*P values for continuous variables are those for non parametric Mann-Whitney rank tests; for plasma chlordecone concentration values equal to or below the limit of detection, we assigned rank values corresponding to the value of the limit of detection divided by 2; for categorical variables, P values were obtained in tests for heterogeneity across levels and for SNPs, in tests for Hardy–Weinberg equilibrium among controls.

†The body mass index is the weight of the subject in kilograms divided by the square of the height in meters.

‡Within the 5-year period before the PSA test preceding the histological diagnosis of cases or allowing the selection of controls.

§All but four of the migrants had lived in mainland France. The remaining four migrants had lived in Germany.

The results presented here were obtained from a study population comprising 623 of the 709 eligible prostate cancer cases and 671 of the eligible 722 controls, from whom we were able to obtain blood samples and determinations of plasma chlordecone concentration. The baseline characteristics of the study participants are summarized in Table 1. Chlordecone was detected in the plasma of 68.7% of the cases and 66.8% of the controls.

Age-adjusted (5-year intervals) and multivariable-adjusted ORs and 95% CI for the association between prostate cancer and plasma chlordecone concentrations are presented in Table 2. Using plasma chlordecone concentration as a means of measuring exposure, we found that the highest exposure categories were associated with a significantly higher risk. Moreover, the relationship between exposure and response was significant (P = .002). When we used the cumulative exposure index as a means of measuring exposure, we found that the highest quartile was associated with a significantly higher risk of prostate cancer and that the exposure-response relationship was significant (P = .004).


Table 2. Plasma Chlordecone Concentration and Risk of Prostate Cancer

Table 2. Plasma Chlordecone Concentration and Risk of Prostate Cancer

Chlordecone No. of Patients No. of Controls Age Adjusted
OR 95% CI OR 95% CI
Plasma concentration, μg/L
    ≤ 0.25 (LD) 195 223 1.00 1.00
     > 0.25-0.47 128 150 0.95 0.69 to 1.31 1.11 0.75 to 1.65
     > 0.47-0.96 139 149 1.16 0.84 to 1.59 1.22 0.82 to 1.83
     > 0.96 161 149 1.27 0.93 to 1.72 1.77 1.21 to 2.58
Cumulative exposure index by quartile, μg/L × No. of years†
    1 (lowest) 88 112 1.00 1.00
    2 101 112 1.05 0.69 to 1.58 1.06 0.62 to 1.82
    3 101 112 1.15 0.76 to 1.74 1.23 0.72 to 2.11
    4 134 112 1.33 0.89 to 1.99 1.73 1.04 to 2.88

Abbreviations: OR, odds ratio; LD, limit of detection.

*The multivariable logistic model includes age (5-year intervals), total plasma lipid concentration (continuous), waist-to-hip ratio (≤ 0.95, > 0.95) and history of prostate cancer screening (no, yes).

†For subjects with values above LD.

When we looked for potential interactions, no modifier effects were observed for any covariates, other than a history of prostate cancer in first-degree relatives and previous residence in a Western country. Results for analyses stratified for these two covariates are presented in Table 3. Associations between chlordecone exposure and prostate cancer were stronger in men with a family history of prostate cancer and in men who had previously lived in Western countries, with significant linear relationships (P = .03 and P = .005, respectively). The interaction terms were found to be significant for family history of prostate cancer and past residence in a Western country (Table 3). Using chlordecone concentrations below the LD as the reference category and concentrations above LD as the exposure category, we carried out a double stratification, according to family history of prostate cancer and past residence in Western countries. Subjects with both a family history of prostate cancer and previous residence in a Western country had a higher risk of prostate cancer (OR, 4.94; 95% CI, 1.15 to 21.23). The risk did not differ from one for those with neither or only one of these factors (data not shown).


Table 3. Plasma Chlordecone Concentration and Interaction With Family History of Prostate Cancer, and Past Residence in Western Countries

Table 3. Plasma Chlordecone Concentration and Interaction With Family History of Prostate Cancer, and Past Residence in Western Countries

Chlordecone Plasma Concentration (μg/L) Without Family History of Prostate Cancer*
With Family History of Prostate Cancer*
Interaction P Without Past Residence in Western Countries*
With Past Residence in Western Countries*
Interaction P*
No. of Patients No. of Controls OR 95% CI No. of Patients No. of Controls OR 95% CI No. of Patients No. of Controls OR 95% CI No. of Patients No. of Controls OR 95% CI
≤ 0.25 (LD) 116 161 1.00 45 26 1.00 137 165 1.00 58 56 1.00
> 0.25 to 0.47 78 111 1.35 0.80 to 2 0.26 26 19 0.97 0.33 to 2.83 87 116 1.09 0.68 to 1.74 41 34 1.15 0.53 to 2.48
> 0.47 to 0.96 81 115 1.13 0.66 to 1.95 34 8 3.22 1.03 to 10.05 103 110 1.12 0.69 to 1.82 36 39 1.33 0.62 to 2.86
> 0.96 68 123 1.27 0.76 to 2.13 45 14 3.00 1.12 to 8.07 < .001 110 118 1.53 0.98 to 2.39 51 31 2.71 1.26 to 5.83 < .001

Abbreviations: OR, odds ratio; LD, limit of detection.

*The multivariable logistic model includes age (five-year intervals), total plasma lipid concentration (continuous), waist-to-hip ratio (≤ 0.95, > 0.95) and history of prostate cancer screening (no, yes).

We determined whether the chlordecone-prostate cancer association depended on clinical characteristics. We found a significantly higher risk for patients age 60 years or older (for the highest tertiles of detectable values, OR, 1.91; 95% CI, 1.24 to 2.94; P for trend < .001) whereas, for patients younger than 60 years, the OR was 1.22 (95% CI, 0.58 to 2.57; P for trend = .94). We found that the risk was significantly higher for both local (OR, 1.58; 95% CI, 1.07 to 2.35; P for trend = .021) and regional/distant stage (OR, 2.25; 95% CI, 1.16 to 4.34; P for trend = .018). However, using the composite index of aggressiveness, a significantly higher risk was found only for high aggressive forms (OR, 2.16; 95% CI, 1.33 to 3.51; P for trend = .004), with the OR for low aggressive forms being only 1.45 (95% CI, 0.96 to 2.19; P for trend = .04).

Both SNPs, rs3829125 and rs17134592, were in Hardy-Weinberg equilibrium in the controls (P > .05). These variants were in complete linkage disequilibrium with D' and r2 of 1, and were present at identical frequencies in cases and controls. The frequencies of the allele variants were low (0.04). We therefore stratified the analysis according to genotype, using chlordecone concentrations at or below the LD as the reference category and concentrations above the LD as the exposure category. For carriers of variant alleles, a non significant increase in prostate cancer risk was observed (OR, 5.23; 95% CI, 0.82 to 33.32), whereas the OR for wild-type allele homozygotes was 1.30 (95% CI, 0.91 to 1.85).

This study reveals that chlordecone exposure, evaluated by determining plasma chlordecone concentration, is consistently associated with an increase in the risk of prostate cancer, and with a significant exposure-response relationship. This is the largest study to have investigated the effects of organochlorine compounds on prostate cancer risk through the evaluation of exposure by biologic measurement. Previous studies have been exploratory and based on less than 80 incident or prevalent prostate cancer cases.1821

A significant increase in risk was observed after adjustment for covariates. This unusual finding is explained by the contribution of waist-to-hip ratio (abdominal obesity) and prostate cancer screening, which act as negative confounding factors. In our population study, both abdominal obesity and prostate cancer screening were associated with a significant increase in the risk of prostate cancer (data not shown), as reported for abdominal obesity in other populations,22,23 and for prostate cancer screening. We also found that these two covariates were associated with lower plasma chlordecone concentrations in our study. In the case of abdominal obesity, peripheral fat may sequester organochlorine lipophilic compounds, leading to an inverse association between peripheral fat levels and circulating concentrations of organochlorine compounds.24 Many factors may account for the observed inverse relationship between prostate cancer screening and plasma chlordecone concentration. For example, we found that individuals with a history of prostate cancer screening were more likely to have lived in Western countries, and were less likely to have worked on banana plantations, these factors in turn being associated with lower plasma chlordecone concentration. In addition, screening procedures may introduce distortions in the associations between exposures of interest and cancer outcomes if the study would have included fewer cases in the absence of screening.25 This is particularly true for Guadeloupe, where the recorded incidence of prostate cancer increased dramatically during the study period,15 mostly due to the development of PSA testing in the population. Consistent with this, it should be borne in mind that one of the potential negative consequences of prostate cancer screening is an increase in the diagnosis of indolent or latent tumors that will never be a problem to the patient and would not have been detected without screening. Such tumors, like those found in autopsy series,26 may reflect the natural aging process rather than the consequences of exposure to chemicals, such as chlordecone, in the environment.

Cancer is generally thought to occur a long time after the first exposure and after long periods of continuous exposure. Our study population fits this requirement, with the median exposure period being almost 33 years. Moreover, given that chlordecone exposure in our study population began in adulthood, at a median age of 30.2 years (the youngest subject being 13 years old at the start of exposure), our study did not include cases of exposure during critical periods for carcinogenesis, such as the fetal or neonatal periods.

The prostate cancer risk associated with chlordecone exposure was higher in subjects with a family history of prostate cancer. Similar findings were reported for pesticide exposure in the Agricultural Health Study.2730 Study subjects and their first-degree relatives may have similar patterns of exposure, and this might lead to a statistical interaction between chlordecone exposure and family history of prostate cancer; alternatively, the observed interaction may be indicative of an inherited genetic trait, such as a polymorphism in a metabolic enzyme that alters the balance between bioactivation and detoxification in the body.

The prostate cancer risk associated with chlordecone exposure was particularly marked in subjects who had spent some time living in a Western country. Most such subjects in our study migrated to a Western country at the beginning of adulthood, remaining there for a mean duration of 14 years. Most of these subjects returned to the French West Indies at the time at which chlordecone use began. Many intervening factors may explain the interaction observed and any potential causal relationship should be interpreted with caution. Migration constitutes a period of exposure to specific environmental risk factors, including hazardous chemicals or nutritional agents. Residing in Western countries may induce significant changes in an individual, due, for example, to the adoption of a Western lifestyle, including, in particular, eating habits that may be risk factors for prostate cancer.31 Such changes in lifestyle may then be maintained by these individuals after their return to the French West Indies.

Previous studies have shown differences in the interindividual activity of chlordecone reductase in the liver between white and Japanese individuals.16,32 Such variability has been associated with a protein variant that arises from two nucleotide (C to G) substitutions at positions 434 and 931 of the cDNA.32 We show here that these two mutations are in complete linkage disequilibrium and that the frequency of the variant allele (0.04) is lower than that previously reported in whites (0.10).33 Although limited by the small number of subjects, the observed high risk associated with carrying at least one variant allele, for individuals with chlordecone concentrations above the LD, is consistent with the mutated enzyme being less efficient. This may lead to lower levels of biliary excretion and of chlordecone clearance from the circulation.

Chlordecone may act as a tumor promoter, through hormone-mediated effects. Chlordecone binds the estrogen receptors α (ERα) and β (ERβ), acting as an agonist of ERα and an antagonist of ERβ.6,34,35 ERα mediates the adverse effects of estrogen on the prostate, such as aberrant proliferation, inflammation, and malignancy. ERβ exerts opposite and beneficial effects, such as antiproliferative, anti-inflammatory, and, potentially, anticarcinogenic effects.36 The human prostate expresses both ERα and ERβ, with ERα expressed primarily on stromal cells and ERβ found in the differentiated epithelium.37,38 The interplay between the agonistic effects on ERα and the antagonistic effects on ERβ of chlordecone may increase proliferation of estrogen-sensitive tissues, increasing the risk of cancer. In addition to its interaction with nuclear ER, chlordecone may activate alternative estrogen signaling pathways or other enzymes and receptors involved in steroidal homeostasis.3943 It remains unclear whether such mechanisms actually take place in human prostate and whether they trigger prostate cancer.

We are aware of the inherent limitations of patient-control studies. Several factors potentially generating bias must be considered, particularly those relating to differential errors in the measurement of disease or exposure. Patient identification was based on unequivocal histologic criteria and controls were selected on the basis of strict criteria, such as normal findings on digital rectal examination and PSA in the normal range for age, taking into account the ethnic origin of the population. We recruited incident rather than prevalent patients, and controls were selected from a representative sample of the male Guadeloupean population during the study period. Exposure was evaluated on the basis of objective determinations of plasma chlordecone concentration, rather than on questionnaires, which might generate recall bias. Single determinations of plasma chlordecone concentration have been shown to provide an accurate reflection of the load of this compound in the body.44,45 Classical organochlorine compounds (determined in blood or fat tissues) have been found at lower concentrations in French West Indies populations than in most other populations and are not correlated with chlordecone levels.13 The quantitative evaluation of chlordecone exposure-response is therefore unlikely to have been confounded by the presence of other organochlorine compounds in our study. However, we cannot exclude the possibility that some unknown confounding factors remain that may account for the associations observed.

In conclusion, our results suggest that there is a causal relationship between chlordecone exposure and prostate cancer risk. Our study also suggests that this association may be affected by genetic background, together with environmental agents related to diet or lifestyle. These findings require further investigation.

© 2010 by American Society of Clinical Oncology

Supported by the Programme Hospitalier de Recherche Clinique, the Plan Pluri-Formation from the French Ministry of Education and Research, the French Ministry of Overseas Territories, the French Agency for Environmental and Occupational Health Safety, the Guadeloupean Division of Social, Health, and Welfare Affairs Directorate, the National Health Directorate, the Association pour la Recherche sur le Cancer, the Comité Guadeloupe Ligue contre le Cancer; Grant No. 212844 from the European Union (Developmental Effects of Environment on Reproductive Health).

Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

The author(s) indicated no potential conflicts of interest.

Conception and design: Luc Multigner, Pascal Blanchet

Financial support: Luc Multigner, Marc Romana, Bernard Jégou, Pascal Blanchet

Provision of study materials or patients: Helene Delacroix-Maillard, Pascal Blanchet

Collection and assembly of data: Luc Multigner, Jean Rodrigue Ndong, Arnaud Giusti, Marc Romana, Helene Delacroix-Maillard, Pascal Blanchet

Data analysis and interpretation: Luc Multigner, Jean Rodrigue Ndong, Arnaud Giusti, Marc Romana, Sylvaine Cordier, Bernard Jégou, Jean Pierre Thome, Pascal Blanchet

Manuscript writing: Luc Multigner, Jean Rodrigue Ndong, Arnaud Giusti, Marc Romana, Helene Delacroix-Maillard, Sylvaine Cordier, Bernard Jégou, Jean Pierre Thome, Pascal Blanchet

Final approval of manuscript: Luc Multigner, Jean Rodrigue Ndong, Arnaud Giusti, Marc Romana, Helene Delacroix-Maillard, Sylvaine Cordier, Bernard Jégou, Jean Pierre Thome, Pascal Blanchet

1. AW Hsing, AP Chokkalingam : Prostate cancer epidemiology Front Biosci 11: 13881413,2006 Crossref, MedlineGoogle Scholar
2. JE Damber, G Aus : Prostate cancer Lancet 371: 17101721,2008 Crossref, MedlineGoogle Scholar
3. E Diamanti-Kandarakis, JP Bourguignon, LC Giudice , etal : Endocrine-disrupting chemicals: An Endocrine Society scientific statement Endocr Rev 30: 293342,2009 Crossref, MedlineGoogle Scholar
4. GS Prins : Endocrine disruptors and prostate cancer risk: Endocr Relat Cancer 15: 649656,2008 Google Scholar
5. B Hammond, Katzzenellenbogen, N Krauthammer , etal : Estrogenic activity of the insecticide chlordecone (Kepone) and interaction with uterine estrogen receptors Proc Natl Acad Sci U S A 76: 66416659,1979 Crossref, MedlineGoogle Scholar
6. VP Eroschenko : Estrogenic activity of the insecticide chlordecone in the reproductive tract of birds and mammals J Toxicol Environ Health 8: 731742,1981 Crossref, MedlineGoogle Scholar
7. J Snegaroff : Organochlorinated insecticide residues in the grounds and the rivers of the bananiere area of Guadeloupe [in French] Phytiatrie-Phytopharmacie 26: 251268,1977 Google Scholar
8. N Cavelier : A Kermarrec Contamination of fauna by the organochlorinated pesticides Current level of the biological chains in Guadeloupe: Pesticides and heavy metals 1979-1980, INRA, French Ministry of the Environment [in French] 113128,1980 Google Scholar
9. YM Cabidoche, R Achard, P Cattan , etal : Long-term pollution by chlordecone of tropical volcanic soils in the French West Indies: A simple leaching model accounts for current residue Environ Pollut 157: 16971705,2009 Crossref, MedlineGoogle Scholar
10. G Bocquené, A Franco : Pesticide contamination of the coastline of Martinique Mar Pollut Bull 51: 612619,2005 Crossref, MedlineGoogle Scholar
11. C Dubuisson, F Héraud, JC Leblanc , etal : Impact of subsistence production on the management options to reduce the food exposure of the Martinican population to Chlordecone Regul Toxicol Pharmacol 49: 516,2007 Crossref, MedlineGoogle Scholar
12. L Multigner, P Kadhel, F Huc-Terki , etal : Exposure to chlordecone and male fertility in Guadeloupe (French West Indies) Epidemiology 17: 6,2006 suppl abstr S372 Crossref, MedlineGoogle Scholar
13. P Kadhel : Pesticides in the Antilles, impact on the function of reproduction [in French] PhD thesis 2008 Guadeloupe, French West Indies Université des Antilles et de la Guyane Google Scholar
14. AE Sirica, CS Wilkerson, LL Wu , etal : Evaluation of chlordecone in a two-stage model of hepatocarcinogenesis: A significant sex difference in the hepatocellular carcinoma incidence Carcinogenesis 10: 10471054,1989 Crossref, MedlineGoogle Scholar
15. S Mallick, P Blanchet, L Multigner : Prostate cancer incidence in Guadeloupe, a French Caribbean Archipelago Eur Urol 47: 769772,2005 Crossref, MedlineGoogle Scholar
16. DT Molowa, AG Shayne, PS Guzelian : Purification and characterization of chlordecone reductase from human liver J Biol Chem 261: 1262412627,1986 MedlineGoogle Scholar
17. O Cussenot, AR Azzouzi, N Nicolaiew , etal : Combination of polymorphisms from genes related to estrogen metabolism and risk of prostate cancers: The hidden face of estrogens J Clin Oncol 25: 35963602,2007 LinkGoogle Scholar
18. JM Ritchie, SL Vial, LJ Fuortes , etal : Comparison of proposed frameworks for grouping polychlorinated biphenyl congener data applied to a case-control pilot study of prostate cancer Environ Res 98: 104113,2005 Crossref, MedlineGoogle Scholar
19. L Hardell, SO Andersson, M Carlberg , etal : Adipose tissue concentrations of persistent organic pollutants and the risk of prostate cancer J Occup Environ Med 48: 700707,2006 Crossref, MedlineGoogle Scholar
20. KJ Aronson, JW Wilson, M Hamel , etal : Plasma organochlorine levels and prostate cancer risk J Expo Sci Environ Epidemiol [epub ahead of print on June 10, 2009] Google Scholar
21. X Xu, AB Dailey, EO Talbott , etal : Associations of Serum Concentrations of Organochlorine (OC) pesticides with breast cancer and prostate cancer in U.S. adults Env Health Perspect 118: 6066,2010 Crossref, MedlineGoogle Scholar
22. AW Hsing, J Deng, IA Sesterhenn , etal : Body size and prostate cancer: A population-based case-control study in China Cancer Epidemiol Biomarkers Prev 9: 13351341,2000 MedlineGoogle Scholar
23. MD Jackson, SP Walker, CM Simpson , etal : Body size and risk of prostate cancer in Jamaican men Cancer Causes Control 21: 909917,2010 Crossref, MedlineGoogle Scholar
24. A Agudo, F Goñi, A Etxeandia , etal : Polychlorinated biphenyls in Spanish adults: Determinants of serum concentrations Environ Res 109: 620628,2009 Crossref, MedlineGoogle Scholar
25. NS Weiss : Adjusting for screening history in epidemiologic studies of cancer: Why, when, and how to do it Am J Epidemiol 157: 957961,2003 Crossref, MedlineGoogle Scholar
26. J Adolfsson : Commentary: On the incidence of histological prostate cancer and the probable diagnosis of cases with tumors too small to produce symptoms or to attract attention on physical examination: The findings of Dr Arnold Rice Rich Int J Epidemiol 36: 285287,2007 Crossref, MedlineGoogle Scholar
27. MC Alavanja, C Samanic, M Dosemeci , etal : Use of agricultural pesticides and prostate cancer risk in the Agricultural Health Study cohort Am J Epidemiol 157: 800814,2003 Crossref, MedlineGoogle Scholar
28. R Mahajan, MR Bonner, JA Hoppin , etal : Phorate exposure and incidence of cancer in the Agricultural Health Study Environ Health Perspect 114: 12051209,2006 Crossref, MedlineGoogle Scholar
29. SM Lynch, R Mahajan, LE Beane Freeman , etal : Cancer incidence among pesticide applicators exposed to butylate in the Agricultural Health Study (AHS) Environ Res 109: 860868,2009 Crossref, MedlineGoogle Scholar
30. CH Christensen, EA Platz, G Andreotti , etal : Coumaphos exposure and incident cancer among male participants in the Agricultural Health Study (AHS) Environ Health Perspect 118: 9296,2010 Crossref, MedlineGoogle Scholar
31. JM Chan, PH Gann, EL Giovannucci : Role of diet in prostate cancer development and progression J Clin Oncol 23: 81528160,2005 LinkGoogle Scholar
32. T Kume, H Iwasa, H Shiraishi , etal : Characterization of a novel variant (S145C/L311V) of 3alpha-hydroxysteroid/dihydrodiol dehydrogenase in human liver Pharmacogenetics 9: 763771,1999 Crossref, MedlineGoogle Scholar
33. National Center for Biotechnology Information Google Scholar
34. GG Kuiper, JG Lemmen, B Carlsson , etal : Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor beta Endocrinology 139: 42524263,1998 Crossref, MedlineGoogle Scholar
35. G Lemaire, W Mnif, P Mauvais , etal : Activation of alpha- and beta-estrogen receptors by persistent pesticides in reporter cell lines Life Sci 79: 11601169,2006 Crossref, MedlineGoogle Scholar
36. SJ Ellem, GP Risbridger : The dual, opposing roles of estrogen in the prostate Ann N Y Acad Sci 1155: 174186,2009 Crossref, MedlineGoogle Scholar
37. H Schulze, S Claus : Histological localization of estrogen receptors in normal and diseased human prostates by immunocytochemistry Prostate 16: 331343,1990 Crossref, MedlineGoogle Scholar
38. E Enmark, M Pelto-Huikko, K Grandien , etal : Human estrogen receptor β-gene structure, chromosomal localization, and expression pattern J Clin Endocrinol Metabol 82: 42584265,1997 MedlineGoogle Scholar
39. P Thomas, J Dong : Binding and activation of the seven-transmembrane estrogen receptor GPR30 by environmental estrogens: A potential novel mechanism of endocrine disruption J Steroid Biochem Mol Biol 102: 175179,2006 Crossref, MedlineGoogle Scholar
40. S Ray, F Xu, P Li , etal : Increased level of cellular bip critically determines estrogenic potency for a xenoestrogen kepone in the mouse uterus Endocrinology 148: 47744785,2007 Crossref, MedlineGoogle Scholar
41. ML Scippo, C Argiris, C Van De Weerdt , etal : Recombinant human estrogen, androgen and progesterone receptors for detection of potential endocrine disruptors Anal Bioanal Chem 378: 664669,2004 Crossref, MedlineGoogle Scholar
42. N Benachour, S Moslemi, H Sipahutar , etal : Cytotoxic effects and aromatase inhibition by xenobiotic endocrine disrupters alone and in combination Toxicol Appl Pharmacol 222: 129140,2007 Crossref, MedlineGoogle Scholar
43. J Lee, RC Scheri, Y Zhang , etal : Chlordecone, a mixed pregnane X receptor (PXR) and estrogen reeptor alpha (ERalpha) agonist, alters cholesterol homeostasis and lipoprotein metabolism in C57BL/6 mice Toxicol Appl Pharmacol 233: 193202,2008 Crossref, MedlineGoogle Scholar
44. SB Cannon, KM Veazey, RS Jackson , etal : Epidemic Kepone poisoning in chemical workers Am J Epidemiol 107: 529537,1978 Crossref, MedlineGoogle Scholar
45. WJ Cohn, JJ Boylan, RV Blanke , etal : Treatment of chlordecone (Kepone) toxicity with cholestyramine resultas of a controlled trial N Engl J Med 298: 243248,1979 CrossrefGoogle Scholar


We thank the staff from the Department of Urology of the University Hospital of Pointe à Pitre, the Urology Clinics of Saint Pierre Hospital, the Sainte Genevieve Health Prevention Center, and the Guadeloupean Medical Analysis Laboratories for their participation in this project; and Cedric Broquere and Murielle Louvet for their technical assistance with this work.

The study sponsors had no role in the design of the study; in the collection, analysis, or interpretation of the data; or in the writing of the report. The corresponding author had full access to all of the data in the study and had final responsibility for the decision to submit the article for publication.

Selection of Patients and Controls

We invited 850 consecutive incident patients with histologically confirmed prostate cancer attending public and private urology clinics (June 2004 to December 2007) to participate in this study. Participating centers diagnosed most (approximately 80%) of the new prostate cancer patients, with a recruitment area covering the total territory of the Guadeloupe Archipelago. Ten (1.2%) of the 850 cases refused to participate, 28 (3.3%) did not fulfill the inclusion criteria (see below), and 103 (12.2%) had not fully completed a questionnaire before the end of the study period, were too ill to participate, or underwent early treatment. We therefore included a total of 709 patients.

Potential controls were recruited from men participating in a free systematic health screening program open to the general population (January 2005 to December 2006). In France, these preventive visits are provided by the French National Health Insurance to all subjects older than 6 years. Each year, a random population sample selected in accordance with the sex and age distribution of the general population was invited to participate. The participation rate in Guadeloupe was about 90%. Between January 1, 2005, and December 31, 2006, we aimed to invite the first 900 consecutive men age 45 years or older to participate, with selection according to an approximate age distribution of prostate cancer incidence in Guadeloupe,15 as follows: younger than 50 years (n = 70), 50 to 59 years (n = 280), 60 to 69 years (n = 350), 70 years or older (n = 200). In total, 24 subjects (2.7%) could not be approached, 12 (1.3%) refused to participate, 128 (14.2%) did not fulfill the inclusion criteria (see above), and 14 (1.6%) had not completed a questionnaire before the end of the study period. This resulted in the overall inclusion of 723 controls.

Inclusion criteria for both patients and controls were current residence in Guadeloupe and no history of past or current hormone treatment, including treatment with 5-α reductase inhibitors. In line with French law, ethnic origin could not be recorded. We limited the study to subjects with at least one parent born on a Caribbean island with a population recognized to be mostly of African descent. These criteria excluded temporary residents (mostly whites) and the most recent migrant population from the Middle East. The study population was, thus, as homogeneous as possible in terms of chlordecone pharmacokinetics and susceptibility genes—factors that may differ as a function of ethnic or geographic origin. Additional inclusion criteria for controls were normal findings upon digital rectal examination and total plasma PSA concentration no higher than the 75th percentile when plotted against age group for African American men without clinical evidence of prostate cancer (Morgan TO, Jacobsen SJ, McCarthy WF, et al: N Engl J Med 335:304-310, 1996). The cutoff points used were: < 1.1 ng/ml (< 50 years of age), < 2.3 ng/ml (50 to 59 years), < 3.5 ng/ml (60 to 69 years), and < 4.9 ng/ml (> 70 years).

Plasma Chlordecone Analysis

Blood samples were drawn into EDTA tubes. Plasma was separated by centrifugation, transferred immediately to propylene vials with a Teflon stopper, and frozen at −30°. Samples were identified only by a unique sample code and were transferred, on dry ice, to the Center for Analytical and Research Technology at Liege University for analysis. All laboratory personnel were blind to the identities of patients and controls.

Plasma samples were thawed, weighed, and prepared as previously described (Janak K, Jensen E: , Becher G: J Chromat B 734:219-227, 1999; Pauwels A, Wells DA, Covaci A, et al: J Chromat B 723:117-125, 1999; Sundberg SE, Ellington JJ, Evans, JJ: J Chromat B 831:99-104, 2006; Debier C, Pomeroy PP, Dupont C, et al: Mar Ecol Prog Ser 247:237-248, 2003). Proteins were denatured in 2 ml of plasma, by adding 100 μl triethylamine and 5 ml formic acid. We added 50 μl of polychlorinated biphenyl (PCB) 112 (100 pg/μl in acetone) as a surrogate marker. The mixture was subjected to solid-phase extraction in 6 ml (1 g) Supelco Supelclean™ Envi-C18 SPE Tubes (Supelco, Bellefonte, PA). The column was conditioned with 10 ml of methanol, followed by 10 ml of distilled water, using an elution device (Visiprep DL™ SPE Supelco; Supelco). Samples were then added run through the column. The vial containing the samples was washed with 3 ml of distilled water added to the cartridge. The column was then rinsed with 2 ml of distilled water and dried for 20 min under a 200 mbar vacuum. Chlordecone was eluted with 5 ml diethylether/n-hexane (85/15, v/v). The solvent was evaporated off to give a final volume of 50 μl. We then added 3 ml of n-hexane to the sample, which was subjected to acidic washing in 98-100% w/v sulfuric acid with evaporation to dryness. We added 50 μl n-hexane and 50 μl PCB 209 (100 pg/μl), used as an internal standard, to each sample. The samples were analyzed by high-resolution gas chromatography, with a Thermo Quest Trace 2000 gas chromatograph equipped with a Ni63 ECD detector (Thermo Quest, Milan, Italy). Chlordecone was separated on a 30 m × 0.25 mm DB-XLB capillary column (J&W Scientific, Santa Clara, CA). The carrier gas was hydrogen and the constituent gas was Ar/CH4 (95/5 v/v). The injector was at room temperature and the detector was kept at 300°C. The temperature program was as follows: 2 min at 60°C, gradual heating from 60 to 180°C at the rate of 30°C/min, 3 min at 180°C, gradual heating from 180 to 270°C at the rate of 2.5°C/min, and 12 min at 270°C.

Data were recorded with Chromcard 2.2 (Fisons Instruments, Milano, Italy) software for Windows. Chlordecone was identified on the basis of its retention time relative to that of the internal standard used for quantification (ie, PCB 209). The linear calibration curve (5 to 120 pg/μl) was established with a certified chlordecone solution (Riedel-de Haën, Seelze, Germany) and a good correlation (r > 0.99) was achieved. A procedural blank, consisting of 2 ml human serum (Cambrex Bio Science, Walkersville, MD), was run with each series of 10 samples, to control the clean-up procedure. Quality control (QC) was performed by regular analyses of procedural blanks, by the injection of standard and n-hexane blanks. The QC was human serum enriched with defined concentrations of chlordecone. The chlordecone concentration in each sample and in the QC was corrected for initial sample weight, and the percentage recovery of the surrogate PCB 112. Recovery rates were always between 70 and 130% (Debier C, Pomeroy PP, Dupont C, et al: Mar Ecol Prog Ser 247:237-248, 2003; JOCE 17/8/2002: Journal Officiel des Communautés Européennes L221: 8-36, 2002). The limit of detection (LD) was fixed at three times the background noise of the chromatogram, at 0.25 μg/l. The limit of quantification (LQ) was determined with spiked human serum samples and was established as 0.75 μg/l, corresponding to ten times the background noise of the chromatogram.

The mean recovery efficiencies and coefficients of variation (repeatability or interassay) at concentrations corresponding to the LD and the LQ were 107.8% and 17.4%, respectively, for the LD, and 102.8% and 11.9%, respectively, for the LQ. At concentrations above the LQ and up to 50 μg/l the mean recovery efficiency and coefficient of variation (repeatability or interassay) were 104.6% and 10.3%, respectively. During the plasma sample analysis sequence, a control chart was established for analyzing the variation of QC recovery efficiencies and coefficient of variation. The QC for testing the accuracy and precision of the analysis routinely used was human serum spiked with 12 μg/l chlordecone. These QCs were carried out by two different technicians at different times. The mean recovery efficiency and coefficient of variation (reproducibility or interassay/interoperator) were 91.4% and 11.3%, respectively.

Lipophilic molecule concentrations are routinely expressed on a per-unit serum-lipid basis (from total cholesterol and triglyceride concentrations). However, the suitability of this approach has been questioned (Porta M, Jariod M, López T, et al: Environ Int 35:1080-1085, 2009). Furthermore, as a substantial portion of the chlordecone in blood is associated with proteins (mainly albumin) and high-density lipoproteins (Soine PJ, Blanke RV, Guzelian PS, et al: J Toxicol Environ Health 9:107-118, 1982; rather than low-density lipoproteins and very high-density lipoproteins, as observed for most other persistent organic pollutants), expressing chlordecone concentration on a per-unit-serum-lipid basis may be misleading. Nevertheless, to take into account the chlordecone fraction transported by lipids we considered total lipids in adjustment models. Such adjustment appears to be justified, as chlordecone may alter lipid homeostasis.43

Total Lipids and Genotyping

Plasma total cholesterol and total triglyceride concentrations were determined enzymatically (DiaSys Diagnostic Systems GmbH; Holzheim, Germany) and total lipid concentration was calculated as described (Bernert JT, Turner WE, Patterson DG Jr, Needham LL: Chemosphere 68:824-831, 2007).

Genomic DNA was obtained from peripheral blood lymphocytes and used to genotype variants of chlordecone reductase (Alpha Keto Reductase 1C4).

From the five single-nucleotide polymorphisms (SNPs) previously identified in the AKR1C4 coding sequence, we selected the two functionally relevant SNPs, rs3829125 and rs17134592, located within exons 4 and 9 respectively (National Center for Biotechnology Information: SNPs were genotyped with the SNaPshot Genotyping System from Applied Biosystems (Foster City, CA). Briefly, the template fragments amplified from genomic DNA were obtained with the primer pairs TACTTCAgCCAggTgAgACg/ gAgAgATAAATgTCCATggAgC for rs3829125 and TTgAACAATTgACCCTgTgTAg/ACAgggCCTTCTgCTAgATg for rs17134592, under the following conditions: 94°C for 5 min, 35 cycles of 94°C for 30 s, 55°C for 30 s, 72°C for 45 s and, finally, 72°C for 4 min. The extended primers for the genotyping of rs3829125 and rs17134592 were gTC-CTC-CAA-gCA-CTC-ACC-TCC-CAT-gTg-gCA and AAA-AAA-AAA-AAA-AAA-AAA-TAT-TTA-TgT-ACT-ATC-CCT-TTC-TCT-TT-TCA-gT, respectively. Primer extension was carried out with the ABI SNaPshot multiplex kit (Applied Biosystems, Foster City, CA), according to the manufacturer's recommendations. Labeled extension products were treated with shrimp alkaline phosphatase and separated on an ABI Prism 310 Genetic Analyzer. For quality control purposes, 10% of the samples tested were blind duplicates, all of which gave consistent results.

Both SNPs were found to be in Hardy-Weinberg equilibrium among controls (P > .05). Linkage disequilibrium between the two SNPs, rs3829125 and rs17134592, was analyzed with LD2SNPing software, by determining D' and r2 values (Chang HW, Chuang LY, Chang YJ, et al: BMC Genet 10:26, 2009).


No companion articles


DOI: 10.1200/JCO.2009.27.2153 Journal of Clinical Oncology 28, no. 21 (July 20, 2010) 3457-3462.

Published online June 21, 2010.

PMID: 20566993

ASCO Career Center