Gastroschisis and endocrine disruptors

Abstract

Endocrine disruptors have been linked to a variety of human and animal disorders showing ongoing increases in frequency, but analysis is often complicated by retrospective issues of definition and ascertainment. Gastroschisis, a congenital abdominal wall defect with 10 to 20 fold or greater increases over the past half century, has origins linked to an estrogenic thrombophilia, suggesting a relationship to estrogen disruption. With this, a distinct morphology and other features make it particularly suited for retrospective studies. Early reports show ongoing increases starting in the U.S. and North-East Europe during the early 1960s. Exogenous and endogenous factors modify risks, which are primarily related to thrombophilia, although other mechanisms, especially vascular, may have a role. These factors can interact, often vary from study to study, and include geographic, genetic, ethnic and racial influences. Increases in other prenatal defects with similar origins indicate more than just an idiosyncratic effect limited to a single disorder. Findings suggest that a single agent is responsible for increases, although other factors can modify its effects. The same agent may affect other conditions through other mechanisms related to endocrine disruptors, which would be supported by a similar epidemiology. Bisphenol A is a possible candidate in terms of mechanisms, timing, increases, and distribution.

Introduction

We are experiencing an epidemic of epidemics: “Over the past 40 y we have seen a 57% increase in prostate cancer, 40% in breast cancer, 85% increase in hypospadias (penile defects), and a 50% reduction in sperm count. In addition, attention-deficit hyperactivity disorder (ADHD) has increased by 30%, autism spectrum disorders have doubled in the past 10 years, obesity has doubled in the past 30 years, and the number of US adults with diabetes has more than tripled since 1980.” And, for a cause, studies suggest environmental contaminants, with endocrine disruptors as particular concerns.¹

Unfortunately, retrospective studies have methodological issues with ascertainment and definition, while unknown latencies between exposures and illnesses further complicate analysis.

However, gastroschisis, a prenatally arising abdominal wall defect, may provide a useful model, with a distinct, well ascertained phenotype and other features that help minimize these problems. The greatest risk factors are consistent with first trimester maternal estrogen levels,^2,3 making endocrine disruption plausible, and multifold increases over half a century⁴ certainly give it epidemic status.

Here, information related to the epidemiology of gastroschisis is reviewed to investigate possible links to endocrine disruptors.

Justifications

Most disorders with suggested links to endocrine disruptors either involve sexual systems or functional disorders.¹ However, gastroschisis, a structural defect secondary to a prenatal thrombotic event³ with estimated 10 to twenty fold increases since 1960,⁴ may have similar origins. Maternal estrogen, a known thrombophilic agent, is associated with major risk factors such as decreased maternal age, parity (independent of maternal age), and body mass index, and increased alcohol intake. This is particularly noteworthy for body mass index, where gastroschisis is unusual among birth defects in decreasing with higher values.³ Increased incidence with stress,⁵ where estriol synergizes in vitro platelet aggregation by adrenaline,⁶ may explain certain life style related risks, while lower rates in Blacks than whites are consistent with different thrombophilic polymorphisms and thrombotic responses to estrogen.²

Several features make gastroschisis particularly useful for epidemiological studies:

1. Origins during the first trimester of pregnancy place causation temporally close to pathological events.

2. Obvious findings at birth, with immediate medical and surgical implications that make it impossible to miss.

3. With a few rare exceptions, a specific location to the right of, and adjacent to, the navel, and an absent visceral covering, readily distinguish it from other disorders.

4. Uncommon associations with genetic and chromosomal disorders⁷ help minimize heterogeneity.

5. With this, it should be possible to obtain parameters for gastroschisis that may help define other endocrine disruptor related epidemics.

Epidemic Timing

Gastroschisis was first delineated in 1953 by Moore and Stokes using 5 literature cases, plus 2 of their own.⁸ A decade later, Moore could only find 32 examples.⁹ Using later birth defect registry data, Castilla et al. concluded that “gastroschisis, once very rare, has increased worldwide for about 50 y, from the 1960s until the present time, and it is still increasing in most regions.”⁴ Their estimated 10–20 fold increases may be low: Moore's 32 global cases by 1963⁹ contrasts, for example, with 4,713 from 15 American states in 11 years reported half a century later.¹⁰

Early clinical data provides additional information. While imperfect, there are some compensations. A need for complex neonatal surgery and care meant referrals in specific areas to a few regional centers. And here, different mortality and complications from omphalocele, a somewhat similar appearing lesion with similar surgical issues (see below), meant that once surgeons became aware of gastroschisis, they were strongly motivated to differentiate the 2 conditions. With this, there is internally consistent evidence that increases began at a fairly specific point in time:

1. Before 1963, individual reports involved one or 2 cases save for Berman's 4,¹¹ but US groups began noting multiple examples in the 1960s: Los Angeles had 23 patients from 1960–70,¹² Chicago 8 in 2.5 years,¹³ Miami 17 in 8 years,¹⁴ and Kentucky 8 for 1968–69.¹⁵

2. Specific areas showed ongoing increases: St. Louis had 4 cases from 1962–65, 27 more to 1972,¹⁶ and then 7 in 4 months,¹⁷ while Columbus, Ohio saw 20 from 1948–68, and 27 more by 1973.¹⁸

3. Omphalocele, another congenital abdominal wall defect with similar surgical needs and relatively stable frequencies, provided a benchmark. Gastroschisis was rarer at first, reflecting pre- or early pandemic rates, with ratios of 3:15 (excluding one atypical case) from 1954–64 in New Haven, Conn.¹⁹ and 4:31 for 1953–65 in Cleveland.²⁰ But it was 23:27 for Los Angeles from 1960–70,¹² and the next decade, 47:28 in Utah for 1971-79,²¹ and San Diego 54:19 over 16 years.²²

4. In Moore's review, 20% of cases,⁹ including two of Berman's four,¹¹ had atypical locations, but Shi et al. found a subsequent dearth of such findings, and only 22 examples from 1993 on,²³ and Fillingham and Rankin saw only one out of 143 in 2008,²⁴ indicating that typical right-sided defects had increased disproportionately.

A survey of surgical centers covering 1954-77 found considerable international variability, but the highest gastroschisis to omphalocele ratios were in the US, followed by Northern Europe and Scandinavia,²⁵ consistent with earlier onsets in these areas. Systemic birth defect surveillance began in the mid-1960s, and upward trends in British Columbia, Norway, and Sweden were noted starting around 1970. Frequencies became high in the Western hemisphere, North-Central Europe, Japan, Australia,⁴ New Zealand,²⁶ and South Africa,²⁷ but not China.²⁸

Local differences can be considerable, e.g., despite rises elsewhere in Italy, 4 regions had low, stable rates for over 25 years.²⁹ Before1978, rural-to-urban ratios varied from 13% to 61%;²⁵ more recently, higher rural rates occurred in some, but not all, areas,^30,31 as did seasonality.³²

Overall, the data supports increases in parts of the US and North-East Europe by the early 1960s, with rapid, but variable, extension across the globe. While changes from a very low baseline would be difficult to detect initially, significant increases are unlikely to have begun much earlier.

Possible Lessons From Gastroschisis

The same rises and maternal age and parity effects in septo-optic dysplasia, another likely prenatal disruptive disorder,³³ suggest that gastroschisis is not unique. Although these disorders involve specific thrombotic pathways, estrogen disruptors can have multiple effects,³⁴ so the same agent could affect other disorders through additional mechanisms with similar timing and distribution, while risk factors, including the maternal age effect, would be expected to differ.

With this, there should be associations with other disorders with suspected relationships to endocrine disruptors. However, many, such as prostate and breast cancer,¹ are adult diseases, making correlations difficult. For others, such as childhood diabetes and obesity,¹ already significant incidences in the general population would make increases with gastroschisis less obvious, and there are confounders such as a greater incidence of gastroschisis with younger mothers with lower socioeconomic status, and a negative relationship with maternal obesity.

Overall, the epidemiology of gastroschisis seems intrinsically complex even when methodological issues are minimized. Risks can be affected by multiple factors, ranging from specific maternal exposures to demographic factors, and specific findings vary from study to study.⁷

Even decreased maternal age, the strongest and most consistent factor, varies. For example, compared to those 25 to 29 years old, mothers 12 to 15 had 4.2-times greater rates in California from 1987 to 2003,³⁵ while a European study found a relative risk of 7.0 for mothers under 20 from 1995 to 2002.³⁶ This was reduced in US. Blacks,¹⁰ and absent in the Arkhangelskaja Oblast, Russia.³⁷

Rate increases did not differ by age in a European study (1980–2002,³⁶ but were greater for younger mothers in smaller cohorts in Western Australia (1980–93),³⁸ Eastern Ireland (1981–2000),³⁹ and Spain (1980–2004),⁴⁰ and in a large US study (1995 through 2005).¹⁰

Interactions between risk factors can be complex. Obesity protects against gastroschisis, and may be synergistic with rising age,⁴³ and some risks are only apparent in older mothers, e.g., non-aspirin NSAIDs, bronchodilators, atrazine, and smoking,^41,42 the last interacting with vasoconstrictor and vascular related genetic polymorphisms.^44,45

Population differences can be profound, with interdependent geographic, genetic, ethnic and racial influences. Again, despite increases in surrounding areas, rates in parts of Italy remain low.²⁹ Besides a relative decreased risk for US. Blacks compared with whites,¹⁰ Indigenous mothers had a 2.5 times higher risk than other Australians of the same average age.⁴⁶ US whites and native Hispanics had lower rates than foreign born Hispanics in the US less than 5 years, or over 19 years old at entry,⁴⁷ suggesting environmental factors rather than ethnicity per se. Similarly, Asian origins lower risks,³⁵ yet rates increased in Japan,⁴ but not China.²⁸ Genetic polymorphisms also have a role.⁴⁴

Possible Primary Agents

There are some clues about primary causes. Sudden changes from a negligible baseline support a new exogenous factor, and widespread major increases indicate a significant environmental presence. Different mechanisms and pathogenetic factors may have variable contributions,³ but the data suggests a primary agent with: (1) Appearances in the early 1960s or so. (2) Early presence in the US and North-east Europe. (3) Increases over time and initial geographic extensions to other industrialized countries.

Life style related causes associated with decreased maternal age are unlikely. Known risks here are too small and too limited, generally had significant presences when gastroschisis was rare, and have not increased commensurately. Less of a maternal age effect in US. Blacks than in whites¹⁰ also goes against a primary role for what are typically also socioeconomic factors.

Plausible mechanisms and biological availability suggest a primary role for estrogen disruptors, and epidemic and epizootic effects are already a concern with synthetic forms that include DDT, dioxin, bisphenol A, PCBs (polychlorinated biphenyls), PBBs (polybrominated biphenyls), phthalate esters, endosulfan, atrazine and zeranol,⁴⁸ plus others with poorly (if at all) understood effects, distributions, bio-availability, by-products, and interactions.⁴⁹

Temporal considerations eliminate most of these, with 2 major exceptions.

One is atrazine, 2-chloro-4-(ethylamino)-6-(isopropylamino)-s-triazine, a herbicide introduced in 1959 and rapidly accepted, e.g., going from 1% of all Illinois corn in 1961 to 55% in 1973.⁵⁰ However, despite epidemiological links to gastroschisis, exposures are limited, uses have leveled off, and relatively low risks seem limited to older mothers. Also, after a European Union ban in 2004,⁴¹ gastroschisis rates continued to rise.³⁶

Another substance, 2,2-bis (4-hydroxyphenyl) propane, or bisphenol A (BPA), is a more likely candidate for several reasons:

1. Initial timing and distribution: BPA was first made commercially in 1957 in the U.S. and 1958 in Europe to produce epoxy resin and polycarbonate plastic,⁵¹ both of which can directly or indirectly lead to human uptake of BPA.⁴⁹ In Europe, polycarbonate was initially used in electrical insulators, but by 1963 it was “widely used for kitchenware and camping utensils,” with a “wide variety of products,” including food containers such as milk jugs, added the next year.⁵² I have not found documentation for the U.S., but a similar course is likely.

2. Increasing volume: BPA exposures are ubiquitous in many areas.⁴⁹ Production rose steadily, making it probably the highest volume synthetic estogen disruptor. World-wide, it reached 2.8 million metric tons in 2002, with an estimated 5.5 million metric tons (about 12 billion pounds) in 2011,⁵³ a doubling in 9 years. (A metric ton is 1000 kq, a bit over 2, 204 pounds, or 1.10 tons, US).

3. Biological availability: BPA is widespread in the environment, and can be found in maternal and fetal serum. In some areas, exposures are close to universal.⁴⁹

4. Plausible mechanisms: Besides complex estrogenic effects,⁵⁴ BPA affects aryl hydrocarbon and other receptors, as well as immune functions, providing a basis for additional influences.⁵⁵

5. Other links: Indications of ties to a host of reproductive, endocrine, metabolic and other conditions⁵³ suggest that BPA is, in fact, a feasible agent for human disorders. And, again, since BPA has multiple modes of action, other disorders need not involve thrombophilic or vascular mechanisms.

With this, a primary role for BPA seems plausible, and further studies justified. Reconstruction of historical product exposure would help- while absences of increasing frequencies might reflect confounding factors, or an early lack of recognition, increases without BPA exposure would go against the hypothesis. Currently, maternal BPA and estrogen levels with prenatally detected gastroschisis would be temporally close to a causative event, and should be most significant with younger mothers, although it would be necessary to show that such levels accurately reflect earlier pregnancy levels.

A report on the reproductive and developmental toxicity of bisphenol A⁴⁹ concluded that “For birth defects and malformations, the Expert Panel has negligible concern” (p. 383). This was largely based on animal data, and there are no large scale human epidemiological studies on the relationship of BPA to birth defects to justify their statement, let alone investigations of specific malformations that could increase without noticeably affecting the overall rate.

Also, alcohol shows that human teratogens can have extensive, albeit often subtle, developmental effects that are clues to relatively small increases in obvious structural malformations.⁵⁶ And here,⁴⁹ “No studies were located on possible human developmental effects of bisphenol A” (p. 236).

Finally, the complex epidemiology of gastroschisis suggests a single primary agent with significant effects from other factors, a situation that straight-forward BPA animal studies fail to reproduce.

In considering future studies, it should be noted that BPA can be replaced as a plasticizer by bisphenol S, where similar mechanisms³⁴ need to be taken into account.

Conclusions

1. An estrogen disruptor facilitating thrombophila is a plausible cause for recent epidemic increases in gastroschisis.

2. It is likely that a single agent is primarily responsible.

3. Exogenous and endogenous factors including geographic, genetic, ethnic and racial influences can modify risks. Factors can interact, often vary from study to study, and are probably not causal in the absence of the primary agent.

4. Changes were first noted in the U.S. and North-East Europe during the early 1960s, and it is unlikely that they began very much earlier.

5. Other epidemics may be caused by the same agent. Different mechanisms and secondary factors may be involved, but timing and geography should be similar.

6. BPA is a candidate for causation in terms of properties, timing, increases, and geographical distribution.

Addendum: ICD codes and accuracy

Reviewers were concerned that “data reporting on the prevalence of gastroschisis 50 y ago is very scant and poor, one of the possible reasons being that the defect had the same ICD code (756.7) as omphalocele; consequently gastroschisis was often misdiagnosed in registries and chart summaries as an omphalocele, and even, but rarely, as limb-body-wall complex.”

The issue here is that cases that once would have been diagnosed as omphalocele were later recognized as gastroschisis, which would then appear to be increasing. However, the early findings summarized here were not based on hospital registries and ICD codes, but typically on reviews by individual surgeons or surgical groups at a single institution on their own experiences.

And here, surgeons are excellent at ascertaining what they feel is relevant to their craft, and differences between omphalocele and gastroschisis were important. They required different techniques, and choices influenced survival rates, which surgeons certainly cared about. Once it became apparent that these children could survive, there were concerted efforts to get better results, and gastroschisis treatment evolved rapidly, making it a fertile field for innovation. There were also different needs for pre- and post- operative care.

At the start, as gastroschisis increased, omphalocele rates were basically unchanged. Increases in the 2 combined showed that something was changing, so more than a simple switch in categories was involved. Similarly, gastroschisis numbers later often surpassed those of omphalocele; if this was a category change, omphalocele rates should have decreased, which did not occur.

Castilla et al., reviewing the world literature, concluded that “Even if some cases were missed or misdiagnosed, the birth prevalence of gastroschisis in the 1960s was most probably similar to that of very rare anomalies such as sirenomelia, cyclopia, acardio-acephalus, and conjoined twins.”⁴ With this, classification issues may have reduced apparent incidences, but greater accuracy would still have only raised rates to this low level.

Finally, while retrospective data is imperfect, there does not appear to be anything better here. Uncertainties are important to recognize, and may make conclusions tentative, but there still may be important clues to significant issues, and if we want perfection, we will never advance. Despite ICD code changes and imperfections in the data, there seems to be enough information to at least consider possible links between gastroschisis and BPA.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.