Persistent Pesticides in Human Breast Milk and Cryptorchidism

1Studies published during the last decades have indicated that the birth prevalence of cryptorchidism may have increased in some regions. Genetic factors may contribute to these findings. However, the short time interval of this increase suggests that environmental factors may be important.

2Several organochlorine pesticides can cause adverse effects in the male reproductive system in animals. An increased incidence of cryptorchidism in male panthers has been attributed to endocrine-disrupting chemicals in the environment, such as 1,1-dichloro-2,2-bis(4-chlorophenyl)ethane (p,p′-DDE). In utero exposure of male rats and rabbits to dichlorodiphenyldichloroethylene (DDT) resulted dose dependently in reduced anogenital distance, hypospadias, cryptorchidism, and epididymal agenesis.

3Increased rates of orchidopexy in areas with extensive pesticide use in agriculture have been reported. One study also found higher pesticide levels in fat tissue samples from boys operated on for cryptorchidism than in children who were operated on for other reasons.

4The aim of this study was to investigate, in a case–control study nested within a prospective cohort, whether individual in utero exposure to organochlorine pesticides estimated by measurements of maternal breast milk concentrations was associated with cryptorchidism among male children.

5We conducted a joint prospective, longitudinal birth cohort study in Finland (Turku University Hospital) and Denmark (National University Hospital at Copenhagen) from 1997 to 2001. In this study we aimed to describe regional prevalence rates and risk factors (lifestyle and exposure) for cryptorchidism by means of questionnaires and biologic samples (blood samples of mother and child, placentas, and one breast milk sample per child). The study was prospectively planned by both research groups as a joint venture in 1996. Recruitment and examinations were completely standardized. The cohort (antenatal recruitment and inclusion, inclusion criteria, and clinical examinations) has been described previously in detail. All boys in the cohort were examined by a group of specially trained doctors for signs of cryptorchidism. The examination technique and definition of cryptorchidism developed by Scorer were applied. Standardization of the clinical procedures was achieved by repetitive workshops, and borderline cases were examined by two researchers from the national study groups. All boys were examined shortly after birth and again at 3 months of age. Boys born prematurely were examined at the expected date of delivery.

6From the total biobank of breast milk samples, we included 65 samples from each country for organochlorine pesticide measurements. The number of samples was determined by funding. The samples represent 29 Danish and 33 Finnish cases defined as boys with cryptorchidism (unilateral or bilateral) at birth. Four Danish and 25 Finnish boys were still cryptorchid at 3 months of age, whereas the others (25/8) had spontaneous descent. Danish/Finnish controls (36/32), defined as boys without cryptorchidism at birth or 3 months, were included. In Denmark, the controls were selected randomly from the entire birth cohort of healthy boys. In Finland, the boys were selected prospectively by a case–control design in which the boys with cryptorchidism were matched to controls at birth for maternal parity, smoking (yes/no), diabetes (yes/no), gestational age (± 7 days), and date of birth (± 14 days). This design was chosen in Finland because of lack of sufficient funding to collect and store biologic samples from all. To ensure that all prospectively planned chemical analyses could be performed, only breast milk samples with a volume > 125 mL were included.

7The study was conducted according to the Helsinki II Declaration and was approved by the local ethical committees in both countries (Finland: 7/1996, Denmark: KF01-030/97) and the Danish Data Protection Agency (registration no. 1997-1200-074). The families were included after oral and written informed consent had been obtained from the parents.

8Human breast milk samples were collected from 1 to 3 months postpartum in Denmark and from 1 to 2 months postpartum in Finland as successive aliquots. All mothers were given oral and written instructions to feed the baby first and then sample milk aliquots (hind milk) by manual expression into a glass or porcelain container, avoiding the use of mechanical breast pumps. The aliquots were frozen consecutively in a glass bottle (250-mL Pyrex glass bottle with a Teflon cap (1515/06D, Bibby Sterilin, Staffordshire, England) and stored in household freezers. The samples were delivered frozen to the hospital at the 3 months’ examination and stored at –20°C.

9Exposure measurements in biologic samples from boys with congenital cryptorchidism at birth and controls were prospectively planned to include persistent and nonpersistent chemicals. Twenty-seven organochlorine compounds were selected by the following criteria: previous worldwide use, suspicion of endocrine-disrupting activity from animal and/or in vitro studies, and highly sensitive analytic methods available. As part of other substudies, the same breast milk samples were planned to be analyzed for other compounds with suspected endocrine-disrupting activity.

10The selected breast milk samples were thawed at room temperature for 12 hr, heated, and shaken for 30 min at 37°C to homogenize the samples, and then divided into smaller aliquots and refrozen at –20°C until further chemical analysis. Extraction, cleanup, and analysis of organochlorine pesticides in the milk samples were based on a method that has been described in detail elsewhere. Milk samples (10 mL) were extracted with 250 mL of a mixture consisting of acetone and n-hexane (2:1 v/v). The milk extracts were collected in flasks weighed in advance and evaporated using a rotary vacuum evaporator (water bath at 45°C). After evaporation, the flasks were placed into an exicator until stable weight was achieved. The lipid content was calculated on wet weight basis. The residual was dissolved in toluene, and gel permeation-chromatography followed by sandwich cartridge cleanup was used to remove lipids and other interferences from the extract. Finally, the organochlorine pesticides were measured by high-resolution gas chromatography/high-resolution mass spectrometry quantified by an isotope dilution method.

11Descriptive data of the mothers and the boys (anthropometric measurements) are reported as mean ± SD or number (percentage) (Table 1). We tested differences between boys with and without cryptorchidism by unpaired t-tests or chi-square tests. Descriptive statistics of pesticide concentrations are given as medians and ranges (minimum and maximum) because of skewed distributions. The sum of DDT metabolites was calculated as the sum of all six metabolites. We calculated the DDE(dichloro-diphenyldichloroethylene)/DDT ratio and the enantiomeric median ratio (ER) by simple division: p,p′-DDE/1,1,1-trichloro-2,2-bis(4-chlorophenyl)ethane (p,p′-DDT) and (+)-isomer concentration/-isomer concentration for the individual compounds, respectively. We tested the differences between cryptorchid and healthy boys using the Mann-Whitney U-test. We tested the differences in pesticide levels between cryptorchid and healthy boys using a logistic regression model in which the pesticide level (log-transformed) and, in some analyses, country and other potential confounders were entered as covariates. This approach had the advantage of allowing the inclusion of cases and controls based on their pesticide levels in the analysis. This selection does not introduce bias in the estimation. p-Values were not corrected for multiple testing.

12To determine whether a given measurement of a sample indicated the presence of a pesticide, both the limit of detection (LOD) and the limit of quantification (LOQ) were determined for every sample. We defined the LOD as three times the background noise of the analytic instrument. Samples with values below the LOD were nondetectable. Samples above LOD, in which the pesticide concentration could not be reliably quantified, were assigned a value below the LOQ; the LOQ represents the level at which a compound concentration was determined with sufficient precision. We defined the LOQ as three times the value of a blank sample, which was the blank matrix used in sample preparation. Detection rate was defined as percentage of samples with a detectable and quantifiable value. Because statistical handling of measurements below the LOD or below the LOQ may influence results, we tried different selection schemes to test whether conclusions were sensitive to the actual values of the unquantifiable measurements. We performed three analyses to compare the exposures of cryptorchid and normal boys. The first included all data using the LOD for samples with nondetectable values and the detected value for unquantified samples. The second included all data except for samples with values below the LOQ. The third analysis excluded all nondetectable samples. Exposure levels reported in Table 2 are based on the third analysis excluding samples with values below the LOD or below the LOQ. Exposure patterns between cases and controls using the other two statistical approaches did not substantially differ from those in Table 2 (data not shown).

13To investigate a combined effect of persistent pesticides, we used eight pesticides for which all individuals had measurable and quantifiable levels [p,p′-DDE, p,p′-DDT, β-hexachlorocyclohexane (HCH), hexa-chlorobenzene (HCB), α -endosulfan, cis-heptachloroepoxide (cis-HE), oxychlordane, and dieldrin]. A statistical test of the null hypothesis that there were no differences in the median exposure levels of cases and controls was carried out as a Monte Carlo permutation test. In the permutations, cases and controls were randomly assigned exposure profiles from the observed profiles, and median levels between cases and controls were compared within countries. We then compared the observed test statistic (the number of median exposures, which were higher among cases than controls) with the permutation distribution. The permutation scheme of the test takes into account the within-individual association structure between different persistent organochlorine pesticides—that some individuals tend to have high exposure to many pesticides, whereas others have low exposures.

14Study population characteristics are given in Table 1. We found no significant differences between the mothers giving birth to a cryptorchid boy versus a healthy boy. Gestational age of cryptorchid boys was slightly lower than in normal boys (Denmark: p = 0.029; Finland: p = 0.688). No systematic difference between cases and controls with regard to year of birth was observed (p = 0.543). In both countries, the selected controls did not differ from the healthy mothers and boys in the entire cohorts with respect to maternal age, parity, smoking, gestational age, and birth weight (data not shown).

15Table 2 shows the results of pesticide measurements in breast milk samples of mothers with cryptorchid and normal boys. The lipid content (percent weight per weight) did not differ significantly (p = 0.707) between cases [3.7 (range, 1.1–7.9)] and controls [3.8 (range, 0.4–10.1)]. Concentrations of persistent pesticides (nanograms per gram lipid) in breast milk showed large interindividual variations and skewed distributions as well as differences in absolute levels.

16Eight compounds—p,p′-DDE, β-HCH, HCB, α -endosulfan, oxychlordane, p,p′-DDT, dieldrin, and cis-HE (listed with decreasing concentrations)—were quantifiable in all samples. The concentrations of these eight compounds were higher than those of any of the remaining 19 compounds, which were all measured in very low concentrations. The sum of all DDT metabolites was slightly higher for cases than controls, but the difference did not reach statistical significance. The ratio between p,p′-DDE and p,p′-DDT was higher among controls than cases, but not significantly. Seventeen of 21 organochlorine pesticides [p,p′-DDT; p,p′-DDE; 1,1-dichloro-2,2-bis(4-chlorophenyl)ethane (p,p′-DDD); 1,1,1-trichloro-2-(2-chlorophenyl)-2-(4-chloropheny l)ethane (o,p′-DDT); HCH (α, β, γ ); HCB; pentachloroanisole (PCA); α -endosulfan; cis-HE; chlordane (cis-, trans-); oxychlordane; methoxychlor; octachlorostyrene (OCS); and dieldrin] were measured in slightly higher median concentrations in milk from mothers giving birth to cryptorchid boys than in mothers giving birth to healthy boys with only trans-chlordane reaching statistical significance (p = 0.012) (Table 2).

17Table 3 presents the overall exposure pattern of cryptorchid and healthy boys depending on inclusion of samples with values below the LOD and/or LOQ. The overall exposure pattern in the three analyses was comparable, showing that most compounds were measured in higher concentrations in milk from case mothers than in milk from control mothers.

18A combined analysis (a Monte Carlo permutation test) of the eight most prevalent organochlorine pesticides revealed that the difference between milk from case mothers and control mothers was not likely due to chance (p = 0.032).

19OCS, pentachlorobenzene (PeCB), p,p′-DDD, o,p′-DDT, and mirex were detected in almost all samples (detection rates 90.8–99.2%), but at low concentrations. Aldrin, β-endosulfan, trans-heptachloroepoxide, and ɛ -HCH were not detected in any sample. Heptachlor and δ-HCH were measured at very low concentrations in three and five samples, respectively. None of these six pesticides are included in Table 2. For the remaining eight compounds [methoxychlor, cis-chlordane, PCA, γ -HCH, 1,1-dichloro-2-(2-chlorophenyl)-2,2(4-chlorophenyl)ethane (o,p′-DDD), trans-chlordane, α -HCH, 1,1-dichloro-2-(2-chlorophenyl)-2-(4-chlorophenyl)ethylene (o,p′-DDE)], the total detection rate varied between 26.9 and 71.5%. Detection rates did not differ significantly between cases and controls for single pesticides (data not shown).

20For oxychlordane and cis-HE, the enantiomeric concentrations were available. The absolute concentrations of the enantiomeric isomers were higher, but not significantly, for cryptorchid boys than for controls (data not shown). The ER for oxychlordane (cases/controls) was 1.36/1.28. For cis-HE, the corresponding figures were 2.48/2.19 (p = 0.103 and p = 0.467, respectively).

21Most persistent organochlorine pesticides were found in higher concentrations in boys with cryptorchidism than in controls, although no individual compound was significantly correlated with cryptorchidism. For eight chemicals (p,p′-DDE, p,p′-DDT, β-HCH, HCB, α -endosulfan, cis-HE, oxychlordane, and dieldrin), which were measurable in all samples, the differences between cases and controls were unlikely to be due to chance. Although we cannot exclude the possibility that individual chemicals alone may cause cryptorchidism, our study suggests that exposure to more than one chemical at low concentrations represents a risk factor for congenital cryptorchidism. There may also be simultaneous coexposure to other environmental chemicals that contribute to the effect on testicular descent; the pesticide measurements may represent a proxy marker (sentinel) for other exposures, such as brominated flame retardants.

22We were able to quantify most organochlorine pesticides at low levels in breast-milk samples. This indicates that these chemicals are still relevant despite the fact that most have been banned or restricted in the study areas for many years. High DDE/DDT levels support the assumption the current exposure level primarily originates from previous contamination, environmental persistence, and long-range atmospheric dissipation, and less from imported food products and increased traveling activity to areas with ongoing use of pesticides such as DDT for malaria control.

23Levels of the investigated pesticides have generally declined in the study area, whereas the birth prevalence of cryptorchidism appears to have increased. This may explain why no single compound was strongly correlated with cryptorchidism. However, low concentrations of a mixture of chemicals over time may still be harmful to the fetus. In addition, the overall exposure to other chemicals with endocrine-disrupting activity may have increased in the same period. Thus, the persistent pesticides investigated here may reflect current overall exposure: Women with the highest levels of persistent pesticides may also be the ones with the highest concentrations of other endocrine-disrupting chemicals. This hypothesis is supported by a previous study that found the concentrations of HCB, p,p′-DDE, p,p′-DDT, and β-HCH in individual samples increased with increasing poly-chlorinated biphenyl concentrations.

24To our knowledge, no previous studies have compared levels of persistent organochlo-rine pesticides in breast milk with the birth prevalence of cryptorchidism. Breast milk was chosen as a surrogate biomarker of previous maternal exposure to persistent pesticides because these compounds accumulate in lipid-rich tissue and thereby in breast milk. There is a dynamic equilibrium between levels of persistent compounds in maternal adipose tissue and breast milk; therefore, daily intake of pesticides during lactation has little influence on the levels measured in milk. Both human and animal studies have demonstrated that pesticides during pregnancy can be transferred to the fetus by crossing the placenta. Levels measured in breast milk were positively correlated to levels measured in umbilical cord samples. Concentrations of compounds in breast milk are a suitable proxy for fetal exposure during pregnancy. As persistent pesticides are accumulated in the lipid fraction of the breast milk, any variations in the lipid content may affect the levels measured. In our study, the women were carefully instructed to collect only hind milk, and they collected many small aliquots that were pooled over time. Thus, the breast milk samples in this study represent an average content over a long period. Because we found no differences in mean lipid content between cases and controls, the potential bias induced by lipid variation due to collection is negligible. In both countries, the selected controls did not differ from the healthy mothers and boys in the entire cohorts with respect to maternal age, parity, smoking, gestational age, and birth weight, and therefore we believe that the samples are representative.

25Because of matching for the most common confounders, such as parity, in the Finnish population, there were no significant differences between cases and controls. In the Danish group, only gestational age differed slightly (p = 0.029). Primiparae, slim women, and smokers tend to have higher pesticide levels. As more mothers of healthy boys were primiparae and nonsmokers and had lower mean body mass index (BMI) than mothers of cryptorchid boys, our study may underestimate the effect of organochlorine pesticide exposure on cryptorchidism. No definitive relation between maternal age and the level of organochlorine pesticides in breast milk has been reported; some studies have found that older mothers have higher levels, whereas others have not. In our study, case mothers were slightly older than control mothers, but fewer of them were primiparae. Older mothers had higher concentrations [significant difference in three compounds (p,p′-DDE, oxychlordane, and α -endosulfan)]. However, older primiparae mothers [n = 7:3 cases (4.8%); 4 controls (5.9%)] especially contributed to this difference. Because they were equally distributed in the two groups, this cannot explain the difference between cases and controls. Milk from mothers giving birth to premature babies may be different in composition. One Danish study described higher levels of HCB in milk from mothers giving birth to premature infants. In our study, levels were higher in milk from mothers giving birth to premature infants. The differences were not significant, and the number of premature infants was low (five cases and two controls) and cannot therefore explain the difference we found.

26The sensitivity of the analytic method allowed the detection of traces of organochlorine pesticides below the LOQ. This phenomenon was, as expected, most frequent for pesticides detected in generally low concentrations. Because the statistical handling of these measurements can profoundly change the results, we evaluated our data carefully with different approaches and found that the overall findings remained unchanged.

27Few other studies have investigated the possible relationship between persistent pesticide levels in biologic samples and the prevalence of cryptorchidism. compared levels of pesticides [DDT and metabolites, toxaphene, HCH (α, β, γ ), HCB, PCA, PeCB, and several chlorinated cyclodienes such as heptachlor] in fat biopsies from 18 cryptorchid boys and 30 controls. Their findings were comparable with ours: Pesticide concentrations were higher among cases than controls, but reaching significance for only a few (HCB and heptachloroepoxide). However, the relatively limited total number of samples and especially the broad age range of boys (0.1–15 years) limits the interpretation of the data. For some participants, biomonitoring was conducted a long time after the relevant prenatal exposure window for testicular maldescent.

28Two studies have been published comparing levels of DDE and DDT in maternal blood and cryptorchidism and hypospadias in the offspring based on two large birth cohorts conducted in the United States in the 1960s. Although both studies were based on biologic samples collected in a period during which DDT was still being used, neither study found firm associations. did not find any association. found adjusted odd ratios to be elevated, although not significantly, and concluded that the results were consistent with a modest to moderate association between DDE/DDT and cryptorchidism. Both studies were performed as nested case–control studies within large prospective birth cohort studies; the women were recruited during pregnancy and the diagnosis of cryptorchidism was well ascertained. Although these studies applied a different biologic matrix (blood) than ours (breast milk), the concentrations in ours were lower. This would also be expected, as the samples in both studies were collected in the period from 1959 to 1966, when DDT was still permitted.

29Other studies without access to biologic samples have also reported associations between cryptorchidism and parental pesticide exposure. Most of these studies were register-based and retrospective. One was performed as a nested case–control study in a large birth cohort study comparing parental exposure (registered by interview of the parents after birth) among boys with cryptorchidism and hypospadias with the exposure of parents of normal boys. conducted an ecologic study in which orchidopexy rates in different municipalities in Granada, Spain, were compared. Pesticide use was categorized into four groups; areas with high pesticide exposure had higher orchidopexy rates. Generally, the exposure assessment was indirect and based on job titles, pesticide purchase, or self-reported exposure with the possibility of recall bias in case–control studies. Distinction between maternal and paternal exposure is made only in some studies: found that cryptorchidism was significantly associated with paternal pesticide exposure but not maternal. described a significantly increased risk of cryptorchidism in sons of female gardeners. Finally, found increased risk with maternal pesticide exposure during pregnancy (relative risk = 4.6); however, this did not reach significance. Because most of these studies have been conducted recently, the pesticide exposure may reflect newer, nonpersistent pesticides. In addition, register data of cryptorchidism are less reliable because of variation in ascertainment and reporting.

30We found a large interindividual difference in concentrations of pesticides in breast milk. This was most likely related to individual differences in exposure and metabolism. Although no significant difference in factors such as lipid content in breast milk, maternal age, maternal BMI, parity, and smoking habits were found between the mothers of cryptorchid and healthy boys, these factors may have contributed to some of the interindividual differences.

31Compared with previous reports on Danish and Finnish breast milk, the average levels of persistent pesticides found in our study were low. This finding agrees with other reports on declining concentration of some persistent chemicals. The ERs indicated to what extent the chiral persistent pollutants were metabolized in the mother’s body after the exposure, for example, because-cis-HE is more readily degraded than (+)-cis-HE in the human body; and if the two samples have the same levels of cis-HE with different ERs, the higher ER sample could have been exposed more heavily in the past. It is an interesting result that the case samples have higher levels of cis-HE with higher ERs (medians). This means, considering the already metabolized part, that case mothers should have been exposed to cis-HE more heavily than control mothers in their exposure histories.

32The most recent data (1993–1994) from the Danish National Board of Health (DNBH) of breast milk samples from 86 primiparae 25–29 years of age showed median concentrations of 178, 38, 33, and 8 ng/g lipid for p,p′-DDE, β-HCH, HCB, and dieldrin, respectively. In our study, the corresponding figures were 135, 16, 12, and 5 ng/g lipid for primiparae of the same age. DNBH estimated that the average daily intake of p,p′-DDE by the child was 1.1 μg/kg/day; in our study, the corresponding value was 0.7 μg/kg/day. This value is still above the acceptable daily intake/tolerable daily intake (ADI/TDI) suggested in the same report (0.5 μg/kg/day). For μ-HCH, HCB, and dieldrin, estimated average daily intake in our study (0.09, 0.07, 0.03 μg/kg/day) was below the recommended ADI/TDI (0.6, 0.17, and 0.05 μg/kg/day).

33In contrast to a Finnish study from 1988 in which mirex was not detected in any of the milk samples, we were able to measure mirex in all samples, although in low concentrations. This difference could be due to our more sensitive analytic methods. To our knowledge, data on endosulfan, PCA, and PeCB in Danish and Finnish milk samples have not been published before. The levels were low compared with studies from other industrialized countries.

34In this study we focused on pesticides with suspected endocrine-disrupting activity or pesticides that have been used worldwide. The relative potency of each pesticide is often much lower than that of natural hormones. However, several of the included compounds act as both estrogens and antiandrogens, which might increase their possible adverse effect on testicular descent. Furthermore, different compounds may interact, thereby enhancing their effects. Existing data on possible mixture effects of the specific organochlorine pesticides in vitro are limited. A combination of 10 compounds including endosulfan, dieldrin, methoxychlor, and some DDT metabolites demonstrated a cumulative effect. Similarly, a number of studies have found indications of additivity for some of the pesticides included in our study. Others have not been able to demonstrate additivity between dieldrin and endosulfan or methoxychlor. Investigating the possible effects of mixtures is complicated because mechanisms of actions for individual compounds are often poorly known, and some chemicals may act through different routes depending on dosage.

35In conclusion, our study suggests an association between congenital cryptorchidism and some persistent organochlorine pesticides present in mothers’ breast milk. Although our study cannot provide proof for a causal relationship, our data are in line with results from animal studies. Thus, prenatal exposure to persistent organochlorine pesticides may adversely affect testicular descent in boys.