This study reports the novel findings that paternal pesticide exposure is a risk factor for cryptorchidism and that paternal smoking is associated with hypospadias in the offspring. A strength of the study is that the results are based on a case–control study nested within a large birth cohort in the general population of Rotterdam. Because 95% of all consecutive newborn boys in Rotterdam were prospectively subjected to a standardized examination of the external genitalia, bias in case identification by exposure is unlikely. The prevalence of 1.1% cryptorchidism and 0.8% hypospadias in our population has been described elsewhere and is within the range reported by comparable studies (Pierik FH et al., unpublished data; Pierik et al. 2002
). A good accuracy of the diagnosis of both abnormalities by CHC physicians is expected because of the standardized and systematic examination of a birth cohort. A high accuracy (88% verification) of the hypospadias diagnosis by CHC physicians has been demonstrated previously (Pierik et al. 2002
), whereas the accuracy of cryptorchidism diagnosis was not assessed. Because the case status was assessed prospectively before data on determinants were collected, the misclassification by CHC physicians is probably nondifferential, which would bias the results toward unity in our analyses. Resources were insufficient to have CHC physicians report the exact location of the urethral opening and the left and right testis for the nearly 9,000 subjects. Another strength of the present study is that both maternal and paternal determinants were included. A weakness of the study is that the paternal determinants were missing for 26% (n
= 116) of the subjects, and in the subjects with paternal information, the paternal determinants were presented by the fathers themselves in only 28% (n
= 91). Differential misclassification between mothers and fathers on self-reported paternal exposure to solvents cannot be ruled out because fathers and mothers reported a paternal exposure prevalence of 31 and 20%, respectively. However, the hypospadias risk for paternal solvent exposure reported by the father (OR = 1.9; 95% CI, 0.6–6.2) or mother (OR = 2.5; 95% CI, 1.0–6.2) was comparable in size, although the 95% CI was wider in these smaller subsets. For other paternal occupational exposures and lifestyle factors, such as smoking and alcohol use, no differences were observed between reporting mothers and fathers.
The multivariate analyses suggest an important role of paternal smoking and occupational exposures. Paternal smoking was significantly associated with hypospadias (OR = 3.8; ). Paternal smoking has previously been associated with the occurrence of single and multiple birth defects (Zhang et al. 1992
), but not specifically with hypospadias. Paternal smoking could have an effect through passive exposure of the mother, but this is unlikely because active smoking by the mother was not a risk factor. We cannot exclude that mothers have underreported their smoking. When mothers of cases under-report their own smoking more than that of their partner, paternal smoking may partly be a spurious risk factor.
After correction for other significant risk factors, paternal pesticide exposure based on the JEM was significantly associated with cryptorchidism (OR = 3.8; ), and self-reported paternal solvent exposure was borderline associated with hypospadias (OR = 2.0; ). The exposure classifications of solvents and pesticides were too broad to allow identification of specific (groups of) chemical agents to be held responsible for the increased risks of either anomaly. Because parents of cryptorchidism and hypospadias cases may have been more concerned with and knowledgeable about environmental risk factors than were parents of controls, differential reporting between cases and controls may have occurred. However, several reasons argue against information bias explaining the observed associations. First, the increased cryptorchidism risk for self-reported pesticide exposure (OR = 2.8) was confirmed by the independent JEM-based pesticide exposure (OR = 4.5). Unfortunately, no JEM judgment was available to validate self-reported solvent exposure. Second, parents were not informed about potential risk factors or the JEM classification. Third, the agreement between self-reported and JEM exposures was not different between cases and controls.
The JEM was developed for a study on occupational risk factors for hypospadias, with a focus on EDs (Van Tongeren et al. 2002
; Vrijheid et al. 2003
). The interexpert agreement among the industrial hygienists developing the JEM was good for pesticides (κ= 0.77) (Van Tongeren et al. 2002
). Although the JEM may misclassify occupational exposures, nondifferential misclassification leads to attenuation of the ORs when both the outcome and the determinant are dichotomous variables (Chen 1989
; Greenland 1980
), and cannot explain the observed association between cryptorchidism and JEM-based pesticide exposure.
Some studies have reported on the association between occupational exposure and birth defects. Paternal solvent exposure has been associated with cleft palate, neural tube defects, and preterm birth (Kristensen et al. 1993
; Olshan et al. 1991
). A study among gardener and farmer families applying pesticides reported an increased risk of cryptorchidism and hypospadias in their offspring (Kristensen et al. 1997
) but could not distinguish paternal from maternal exposure. Another study observed an increased risk of cryptorchidism in sons of female gardeners and farmers but not in sons of men working in farming or gardening (Weidner et al. 1998
). Neither paternal nor maternal occupation was associated with hypospadias. Because exposure assessment was limited to job title, limited information was available on the role of specific occupational exposures, such as pesticide use (Weidner et al. 1998
It remains to be established whether the associations between external agents and cryptorchidism and hypospadias are causal or based on confounding (e.g., by unknown but related occupational risk factors). Several plausible biologic mechanisms that could mediate the observed effects of paternal smoking and occupational exposure on the offspring have, however, been described. There is growing human evidence that paternal environmental factors around the time of fertilization play a role after fertilization. More than 100 chemicals, including pesticides and solvents, have been related to male-mediated adverse reproductive outcomes (Davis et al. 1992
). Animal studies provide extensive evidence for male-mediated developmental effects (i.e., spontaneous abortions, growth retardation, malformations, and behavioral abnormalities) of environmental agents (Robaire and Hales 2003
). Several modes of action of chemicals have been shown, the most likely being genetic (e.g., germline DNA modification) or epigenetic (e.g., DNA repair, chromatin structure, apoptosis) effects on germ cells, whereas exposure of the oocyte or embryo to contaminated seminal fluid could also play a role (Davis et al. 1992
; Robaire and Hales 2003
). A study in mice demonstrated that environmental pollution resulted in DNA mutations that were inherited by the offspring, primarily through the paternal germline (Somers et al. 2002
On the basis of the xenoestrogen hypothesis (Sharpe 2003
), we anticipated that maternal exposure to EDs during fetal life could be a causal pathway leading to cryptorchidism and hypospadias. As of yet, few human data are available to confirm or refute this hypothesis. We did not find an association between maternal occupational exposure and either abnormality, perhaps due to the small proportion of exposed mothers. A previous study reported a maternal vegetarian diet as a risk factor for hypospadias and suggested a higher phytoestrogen intake as explanation (North and Golding 2000
). We specifically assessed dietary phytoestrogen intake, which was not a significant risk factor for hypospadias or cryptorchidism. However, the nutrition data may suffer from inaccuracies because nutrition was assessed only once, whereas considerable intraindividual variation has been described with food-frequency questionnaires (Goldbohm et al. 1995
). The findings in our case–control study suggest an association between cryptorchidism and hypospadias and lower socioeconomic status, as reflected in low education level and suboptimal general health status of both parents. The effect of socioeconomic status may be confounded by selection bias, especially because of differential response between cases and controls. For the impact of education to be spurious, this would require approximately a 2-fold higher response among parents of cases than of controls in subjects with a low education.
A similar differential response bias may have contributed to the observed effect of Turkish origin on cryptorchidism and hypospadias. Based on the nationalities of all 8,695 examined boys, Moroccan, Turkish, and other minorities were underrepresented by about 40–50% among controls. To exclude confounding by country of origin, we repeated the regression analysis in Dutch subjects only, which did not yield significantly different results, although standard errors increased because of a smaller sample. Among Dutch subjects paternal exposure to pesticides has a similar effect (OR = 3.4; 95% CI, 0.3–43.0) on cryptorchidism but failed to reach the level of conventional significance. Paternal smoking (OR = 6.5; 95% CI, 2.0–21.7) and self-reported paternal exposure to solvents (OR = 3.3; 95% CI, 1.2–9.5) remained significant risk factors for hypospadias among Dutch subjects.
Previous studies have reported ethnic variations in the occurrence of cryptorchidism and hypospadias (Chia et al. 2003
; Fredell et al. 2002
). Familial aggregation has been described for both abnormalities, supporting the importance of genetic factors (Fredell et al. 2002
; Weidner et al. 1999
). The association between Turkish origin and cryptorchidism and hypospadias may be the result of a genetic or environmental factor among Turkish people that predisposes toward these abnormalities. A higher maternal age was a significant risk factor within the Turkish minority, but not in the overall group of non-Turkish origin. We cannot exclude the possibility that the response may have been different with age among Turks.
In the multifactorial models without adding paternal risk factors, preterm delivery was associated with cryptorchidism (OR = 3.1; ), and being SGA was associated with hypospadias (OR = 7.3; ). These associations are well known from previous studies (Weidner et al. 1999
). Some authors point to reduced placental function as underlying etiology for low birth weight, cryptorchidism, and hypospadias (Fredell et al. 1998
Some earlier studies looking at large groups of cases have reported ORs ranging from 1.1 to 1.9 for low birth order and a higher maternal age as risk factors for cryptorchidism or hypospadias cases (Akre et al. 1999
; Biggs et al. 2002
; Kallen 2002
; Møller and Skakkebaek 1997
), although others did not observe these excess risks (Berkowitz et al. 1995
; Jones et al. 1998
). Birth order and parental age were not significantly related to cryptorchidism or hypospadias in our study, which may be because of the relatively small effect and limited population size.
Our observation that a longer time to pregnancy was associated with hypospadias () may be explained by familial aggregation of hypospadias (Fredell et al. 2002
) and its association with subfertility (Skakkebaek et al. 2001
). Previous studies have reported a higher incidence of hypospadias in boys born after intracytoplasmic sperm injection (Ericson and Kallen 2001
; Wennerholm et al. 2000
), which may be explained by a lower birth weight that occurs more frequently after ART. In our study, the frequency of ART was too low to evaluate its association with hypospadias or cryptorchidism.
This study suggests that paternal environmental exposures may increase the risk of cryptorchidism and hypospadias in newborn boys, which may indicate an effect on the paternal germline. Cryptorchidism was associated with paternal exposure to pesticides, and hypospadias was more frequent in fathers that were active smokers. The pregnancy-related risk factors of low birth weight and SGA birth for hypospadias and preterm delivery for cryptorchidism have consistently been found in previous studies (Weidner et al. 1999
). Future studies on environmental risk factors for cryptorchidism and hypospadias should not only focus on maternal exposure during fetal life but also include the paternal pathway to substantiate whether the observed associations are causal.