Gastroschisis and maternal intake of phytoestrogens
Abstract
The prevalence of gastroschisis has increased significantly in the past few decades. The strongest risks have been observed for women <25 years old or of low body mass index, and maternal diet also been proposed to be associated with risk. The objective of this study was to evaluate whether the risk of gastroschisis is associated with maternal dietary intake of phytoestrogens. The analysis includes data on mothers of 409 gastroschisis cases and 3,007 controls who delivered their infants from 2005 to 2010 and participated in the National Birth Defects Prevention Study, a multistate, population-based, case-control study. Detailed information was obtained from maternal telephone interviews that included a validated food frequency questionnaire. We conducted logistic regression analyses that included each phytoestrogen in its continuous form (to test for linearity) and quadratic form (to test for non-linearity), adjusted for maternal energy intake, age, BMI, race-ethnicity, and smoking in 1st trimester. Logistic regression analysis indicated that biochanin A, formonoetin, and coumestrol had a significant non-linear association with gastroschisis (P-value <0.05 for quadratic term). Lower intakes were associated with increased risk, with somewhat stronger but relatively modest associations at the lower end of the distribution; for example, the ORs for the 10th versus 50th percentiles ranged from 1.1 to 1.2. Associations were not significant for the other phytoestrogens. This study provides some evidence for association with certain phytoestrogens, after adjusting for covariates. The implications of our findings for clinical practice are uncertain pending other studies examining this association. © 2016 Wiley Periodicals, Inc.
INTRODUCTION
Gastroschisis is a congenital defect of the abdominal wall through which the intestines and other abdominal organs can protrude. The birth prevalence ranges from 2 to 3 per 10,000 live births and has shown a significant increase worldwide over the past three decades [Roeper et al., 1987; Laughon et al., 2003; Kilby, 2006; Loane et al., 2007; Vu et al., 2008; Jones et al., 2016]. Several population-based studies have found gastroschisis risk to be strongly increased with young maternal age, nulliparity, low body mass index (BMI), and lower socioeconomic status [Werler et al., 1992; Torfs et al., 1998, 1994; Lam et al., 1999; Stoll et al., 2001; Siega-Riz et al., 2006; Lubinsky, 2015]. While gastroschisis has been reported to occur more frequently in non-Hispanic whites than blacks or Asians, it is often difficult to compare prevalence data from different registries [Castilla et al., 2008].
Many different theories have been introduced over the years to attempt to explain the pathogenesis of gastroschisis. These hypotheses include anomalous mesenchymal differentiation from the fold of the abdominal wall; embryologic sequential change; defect of the umbilical ring; intrauterine interruption of the omphalomesenteric artery; failure of yolk sac incorporation into the umbilical stalk; and abnormal folding of the body wall causing ventral body wall defect through which the gut herniates [Duhamel, 1963; deVries, 1980; Hoyme et al., 1981; Feldkamp et al., 2007; Stevenson et al., 2009; Rittler et al., 2013]. Despite the variety of these theories, the pathophysiology of gastroschisis is poorly understood. Lubinsky [2012] proposed a dual pathogenesis based on a structural vascular predisposition with thrombosis. He suggests that normal involution of the right umbilical artery creates a localized predilection to a thrombotic event, weakening the umbilical ring and generating an area of possible herniation. In particular, he proposes that estrogen-related thrombophilia may contribute to risk of gastroschisis given that estrogen levels in early pregnancy are typically higher in women who are younger, nulliparous, or have lower BMI [Troisi et al., 2008; Toriola et al., 2011].
Various nutritional factors have been proposed to contribute to risk of gastroschisis, in part because worse dietary intake is expected among the youngest mothers [Bodnar and Siega-Riz, 2002]. However, results from studies examining a variety of nutrients have been inconsistent [Torfs et al., 1998; Feldkamp et al., 2011; Paranjothy et al., 2012; Canovas-Conesa et al., 2013; Feldkamp et al., 2014].
Another class of nutrients potentially associated with gastroschisis is phytoestrogens, which are structurally similar to estrogen. Diets high in soy products and some types of seeds and nuts are particularly high in phytoestrogens. The main types are isoflavones, lignans, and coumestans, and each form has varied effects on the levels and activities of estrogen. Phytoestrogens produce estrogenic and anti-estrogenic effects and as such have been compared to natural selective estrogen receptor modulators [Setchell, 2001]. In addition, some phytoestrogens have been shown to be protective against thrombosis [Hertrampf et al., 2005].
Our objective was to examine the novel hypothesis that phytoestrogen intake is associated with risk of gastroschisis, using data from the National Birth Defects Prevention Study.
MATERIALS AND METHODS
Study Design
The NBDPS is a multicenter, population-based, case-control study of structural congenital malformations involving centers from 10 states (Arkansas, California, Georgia, Iowa, Massachusetts, New Jersey, New York, North Carolina, Texas, and Utah). Recruitment, data collection, and current analysis includes deliveries starting with estimated due dates in 2005 and continuing through 2010. Details of the study methods and descriptions of the surveillance systems in the NBDPS study centers have been published previously [Yoon et al., 2001].
Case Review and Classification
Medical record information was reviewed by a clinical geneticist at each study center. Cases with recognized single-gene disorders or chromosomal abnormalities were excluded. All cases with phenotype suggestive of amniotic band sequence or limb–body wall complex were excluded. No exclusions were made based on the presence of other birth defects. Each case received a final review by one clinical geneticist to ensure that cases from each study center met eligibility criteria [Rasmussen et al., 2003].
Control Selection
Each participating center randomly selected healthy liveborn infants without birth defects for each study year from birth certificates (Arkansas 2000–current, Georgia 2001–current, Iowa, Massachusetts, North Carolina, Utah) or birth hospitals (Arkansas 1997–1999, California, Georgia 1997–2000, New York, Texas) as controls to represent the populations from which cases were derived.
Maternal Interviews
Maternal interviews of case and control mothers were conducted by trained interviewers by computer-assisted telephone interview, in English or Spanish, from 6 weeks to 24 months after the infant's estimated due date. Exposures to many factors were assessed, relative to the woman's estimated date of conception. This date was derived by subtracting 266 days from the woman's expected due date and was primarily self-reported. Median time to interview was 7.7 months for cases and 9.5 months for controls, with 65% participation.
Food Frequency Questionnaire
Mothers reported their average intakes of foods using a shortened version of Willet food frequency questionnaire that was validated by the Nurses' Health Study [Willett et al., 1985]. The questionnaire includes 58 food items, and participants reported how often, on average, they consumed food items in the year prior to becoming pregnant. Fifteen response categories were possible ranging from once per month to six or more per day. Foods eaten less than once a month were recorded as “never or none.” Additional, more detailed questions assessed intake of sodas, breakfast cereals, food supplements, and caffeinated tea and coffee. Nutrient values were derived from version SR 25 of the US Department of Agriculture National Nutrient Database [U.S.-Department-of-Agriculture-Agricultural-Research-Service, 2007]. Starting with births in 2005, soy and soy-containing food items were included in the questionnaire.
Phytoestrogen Intake
Using recent, laboratory-derived phytoestrogen values published by Kuhnle et al. [2008,2008, 2009,2009] we estimated maternal phytoestrogen intake. These values were available for a relatively consistent set of specific phytoestrogens across all food items, including most of the items in this study's food frequency questionnaire (if not, proxies were selected). Phytoestrogen values were assigned to each food item in the food frequency questionnaire, with details published previously [Carmichael et al., 2011]. We assessed intake of coumestrol; isoflavones biochanin A, daidzein, formonoetin, genistein, and glycetin; lignans mataresinol and secoisolariciresinol; as well as total isoflavones, lignans, and phytoestrogens.
Analyses
Analyses were restricted to 454 cases and 3,352 controls born from 2005 to 2010, as these were the years in which soy and soy-containing food items were included in the food frequency questionnaire. Mothers with less than 500 or greater than 5,000 kcal of daily energy intake and mothers with more than one food item missing were excluded, leaving 430 cases and 3,168 controls available for analyses. A total of 409 cases and 3,007 controls had complete data on all covariates. Odds ratios (OR) and 95% confidence intervals (CI) were estimated for an association of gastroschisis with phytoestrogen intake from logistic regression analyses. We examined models that specified phytoestrogen intake as continuous, as a test for linearity, with and without the inclusion of a squared (quadratic) term, as a test for non-linearity of the association. We considered P < 0.05 as significant for the linear term and P < 0.10 for the quadratic term. We estimated ORs for the 10th, 25th, 75th and 90th percentiles, all relative to the 50th percentile, for each phytoestrogen, given that the direction of the hypothesized associations was uncertain. Data were also analyzed using spline regression, as a confirmatory approach.
All models were adjusted for the following covariates, which were selected a priori on the basis of potential associations with gastroschisis and/or nutritional status: maternal age (years); energy intake (kilocalories), prepregnancy body mass index (weight [kg]/height [m2]), race/ethnicity (non-Hispanic white, Hispanic), and first trimester cigarette smoking. Models were also examined that were stratified by maternal age, categorized as less than 20 years, 20–25 years, and >25 years.
RESULTS
We included 409 cases and 3,007 controls who had complete data on all covariates. Mothers of cases were more likely to be Hispanic, young (<25 years), have lower BMI, and report first trimester cigarette smoking, relative to control mothers (Table I).
| Cases | Controls | |||
|---|---|---|---|---|
| No. | %a | No. | %a | |
| Maternal race/ethnicity | ||||
| Non-Hispanic white | 194 | 47 | 1,762 | 59 |
| Hispanic | 126 | 31 | 658 | 22 |
| Other | 89 | 22 | 587 | 20 |
| Maternal age, years | ||||
| <20 | 139 | 34 | 252 | 8 |
| 20–25 | 194 | 47 | 850 | 28 |
| >25 | 76 | 19 | 1,905 | 63 |
| Maternal prepregnancy BMIb | ||||
| Underweight (<18.5) | 38 | 9 | 154 | 5 |
| Normal weight (18.5–<25.0) | 264 | 65 | 1,508 | 50 |
| Overweight (25–<30.0) | 81 | 20 | 726 | 24 |
| Obese (≥30.0) | 26 | 6 | 619 | 21 |
| First trimester cigarette smoking | ||||
| Yes | 136 | 33 | 433 | 14 |
| No | 273 | 67 | 2,574 | 86 |
- BMI, body mass index.
- a Numbers may not add to 100% because of rounding.
- b BMI: weight (kg)/height (m)2.
The logistic regression analyses indicated that biochanin A, formonoetin, and coumestrol were significantly associated with gastroschisis. For all three of these phytoestrogens, the association was non-linear (the quadratic term was significant, P-values 0.029–0.045). Table II presents adjusted ORs for intake of phytoestrogens and gastroschisis. Results suggest that lower intake was associated with increased risk, with somewhat stronger but still relatively modest associations at the lower end of the distribution; for example, the ORs for the 10th versus 50th percentiles ranged from 1.1 to 1.2. The results for spline regression analysis were in agreement with logistic regression; P-values, which reflect evidence for non-linearity of association, were 0.012–0.017 for the three noted phytoestrogens (data not shown), and associations were not significant for the other phytoestrogens (P > 0.05). Stratification by maternal age (<20, 20–25, and >25 years) did not substantially alter the observed pattern of results (data not shown).
| AOR (95%CI) Percentile value (μg/day) | ||||||
|---|---|---|---|---|---|---|
| Phytoestrogen | 10th percentile | 25th percentile | 50th percentile | 75th percentile | 90th percentile | P-value for quadratic term |
| Biochanin A | 1.1 (0.9, 1.3) 7.3 | 1.1 (1.0, 1.2) 10.1 | Reference 14.3 | 0.9 (0.8, 1.1) 20.1 | 0.9 (0.7, 1.2) 26.5 | 0.032 |
| Formonoetin | 1.2 (1.0, 1.4) 2.7 | 1.1 (1.0, 1.2) 3.9 | Reference 5.5 | 0.9 (0.8, 1.0) 7.6 | 0.9 (0.7, 1.1) 10.1 | 0.045 |
| Coumestrol | 1.2 (1.0, 1.4) 3.5 | 1.1 (1.0, 1.2) 5.2 | Reference 7.9 | 0.9 (0.8, 1.0) 12.1 | 0.9 (0.7, 1.1) 16.3 | 0.029 |
- AOR, adjusted odds ratio; CI, confidence interval.
- a Odds ratios were adjusted for the following covariates: maternal energy intake, age, BMI, race-ethnicity, and smoking in 1st trimester.
DISCUSSION
This study provides some evidence for association with phytoestrogens and gastroschisis. Results suggested that lower intakes of biochanin A, formonoetin, and coumestrol were associated with modestly increased risk of gastroschisis.
While this study examined the hypothesis that phytoestrogens are associated with gastroschisis, previous studies focused instead on a variety of other nutritional factors, with mixed results. Small case-control studies have found risk reduction to be associated with higher intake of fruits, vegetables, oleic acid, alpha-carotene, and total glutathione and lower intake of nitrosamines [Torfs et al., 1998; Paranjothy et al., 2012]. Previous analyses based on NBDPS data did not find an association with intake of fats or various macro- and micronutrients [Siega-Riz et al., 2006; Feldkamp et al., 2011], although one did report an association with diet quality, but only among Hispanic women [Feldkamp et al., 2014].
Lubinsky's [2014] hypothesis suggests that estrogen-related thrombophilia may contribute to risk of gastroschisis. Estrogen levels in early pregnancy are typically higher in women who are young, nulliparous, and have lower BMI, all of whom are at elevated risk for gastroschisis. While Lubinsky's proposal suggests that increased estrogen may directly contribute to gastroschisis, the mechanism by which phytoestrogens may protect against gastroschisis is not clear. Some studies have indicated possible protection by phytoestrogens against various health outcomes, including cancer and heart disease; antioxidant and antiproliferative properties may contribute to these associations [Harper et al., 1999; Adlercreutz, 2007; Helferich et al., 2008; Park et al., 2008; Cano et al., 2010; Ward et al., 2010]. Furthermore, recent studies in mice have either protective or teratogenic effects of phytoestrogens on fetal development [Xing et al., 2010, Zhao et al., 2010; El-Nahla et al., 2014]. The nonlinear association that we found was modest, though statistically significant. It suggests a protective effect of phytoestrogens. Phytoestrogens may have estrogenic or anti-estrogenic effects [Setchell, 2001; Bradlow et al., 2006; Adlercreutz, 2007]; the latter type of effect could be protective against estrogen-related thrombophilia in the pathogenesis of gastroschisis suggested by Lubinsky, and thus serve as a potential explanation for the observed inverse association with phytoestrogen intake.
Critical strengths of this study are population-based study design, and careful case review. Our study is also strengthened by its ability to estimate phytoestrogen intake using the phytoestrogen databases published by Kuhnle et al. [2008,2008, 2009,2009]. These databases provided values for multiple specific phytoestrogens, for a wide variety of foods, based on consistent methodology. This approach is preferable to the more common approach of estimating intake based on a variety of source data. Another strength was that the dietary assessment included soy and soy-containing products. NBDPS began including an assessment of soy products starting with births in 2005 and the fact that these products contain phytoestrogens of significant quantity prompted the exclusion of cases prior to 2005. Also a strength is the adjustment for several covariates in our analysis. There were several limitations to the study. It cannot validate phytoestrogen intake against a gold standard, such as serum or urine values. The study relied on self-reported retrospective dietary recall for women in the year prior to pregnancy, the quality of which may be subject to error; however, we expect the error in dietary recall to be non-differential with respect to case status. While the estimates of phytoestrogen intake were based on a relatively comprehensive dietary assessment, they did not include phytoestrogens from certain potentially relevant foods, such as coffee and tea. Another potential limitation is that there might be other confounding factors that were not recognized and therefore not accounted for in the analysis. The study is also limited by small sample size for some cells, which in particular limited our ability to examine potential differences of association within strata of maternal age.
Our study found a modest association of maternal phytoestrogen intake with risk of gastroschisis. It lends some support to the hypothesis that maternal estrogenic exposures may increase gastroschisis risk [Lubinsky, 2015]. As the first analysis of maternal phytoestrogen intake and gastroschisis, results should be interpreted with caution and clinical implications are uncertain.
ACKNOWLEDGMENTS
This work was supported by the Centers for Disease Control and Prevention Centers of Excellence No. U50/CCU925286 and U01DD001033. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention or the California Department of Public Health. We thank the California Department of Public Health, Maternal Child and Adolescent Health Division for providing surveillance data from California for this study. We thank Dr. Mary Cogswell, National Center for Chronic Disease and Health Promotion, Centers for Disease Control and Prevention, for her valuable contributions to developing the phytoestrogens database.
