|Home | About | Journals | Submit | Contact Us | Français|
Abbreviations: CI, confidence intervals; CMV, cytomegalovirus; EBNA, Epstein-Barr virus nuclear antigen; EBV, Epstein-Barr virus; HSV, herpes simplex virus; HSV-1, herpes simplex virus type 1; HSV-2, herpes simplex virus type 2; ICAM, intercellular adhesion molecule; IgG, immunoglobulin G; IgM, immunoglobulin M; VCA, viral capsid antigen.
Gastroschisis risk is highest in offspring of young women and is increasing in prevalence, suggesting that exposures that are increasingly common among younger females may be causal. Some infections by viruses in the herpes family are more common in the earlier childbearing years and have been increasing in prevalence over time. Data from the Finnish Maternity Cohort were linked to Finnish malformation and birth registers (1987–2012) for this study, a nested case-control study of mothers of offspring with gastroschisis and age-matched controls. Maternal antibody responses in early pregnancy (mean gestational age = 11.1 weeks) to Epstein Barr virus (EBV), herpes simplex virus types 1 and 2 (HSV-1 and HSV-2), and cytomegalovirus were measured. Conditional logistic regression models were used to estimate odds ratios (and 95% confidence intervals) for high immunoglobulin reactivity. Odds ratios for high immunoglobulin M (IgM) reactivity to EBV-viral capsid antigen and HSV-1 or HSV-2 (as indicators of recent infection) were 2.16 (95% confidence interval (CI): 0.97, 4.79) and 1.94 (95% CI: 0.74, 5.12), respectively. For higher immunoglobulin G (IgG) reactivity to EBV-viral capsid antigen and HSV-2 IgG, odds ratios were 2.16 (95% CI: 0.82, 5.70) and 2.48 (95% CI: 1.50, 4.10), respectively. Reactivities to HSV-1 IgG, cytomegalovirus IgM, or cytomegalovirus IgG did not appear to increase gastroschisis risk. Primary EBV infection was not associated with gastroschisis, but observed associations with both IgM and IgG reactivities to EBV and HSV suggest that reactivations may be risk factors for it.
Gastroschisis is an abdominal wall defect in which the contents of the abdomen are outside of the body at birth. The descriptive epidemiology of gastroschisis is intriguing because it most often occurs in the offspring of young women (1, 2), and its birth prevalence has been increasing over the past few decades (3, 4). These 2 epidemiologic features have led to hypotheses that behaviors common to younger childbearing females are the cause. Indeed, cigarette smoking, illicit drug use, and genitourinary infections such as Chlamydia trachomatis and other sexually transmitted infections are more common in young women and have been suspected as explanations for the inverse association with maternal age (5). However, these associations do not explain the strong inverse relationship with maternal age, and the question remains as to why teenaged women have 10 times the risk of those over age 30 years of having an infant born with gastroschisis.
The epidemiologic and immunological aspects of viral infections in the herpes family with regard to age and temporal trends suggest that these infections in women could be potential risk factors for gastroschisis in their offspring. Epstein-Barr virus (EBV), herpes simplex virus type 1 (HSV-1), and cytomegalovirus (CMV) are common in the general population, with a majority of individuals having been exposed by early adulthood. While primary infection in childhood is typically asymptomatic, adolescent and adult first infections can cause infectious mononucleosis (EBV and CMV) or cold sores (HSV-1), and symptoms worsen as age at first infection increases. Thus, a decrease in the proportion of children with evidence of past infection would increase the pool of adolescents and young adults who are at risk for primary infection (6–8). For example, EBV prevalence decreased from 65% in 2003–2004 to 59% in 2009–2010 among 12- to 14-year-olds in the United States. Likewise, HSV-1 prevalence decreased among US 12- to 13-year-olds from 40% in 1988–1994 to 36% in 1999–2004 (7). Over this same time period, CMV prevalence among 6- to 11-year-old girls decreased in non-Hispanic whites and Mexican Americans but not in non-Hispanic blacks (9). In contrast, herpes simplex virus type 2 (HSV-2) prevalence has been declining among US 14- to 19-year-olds (10). However, HSV-2 differs from EBV, HSV-1, and CMV in that infections are rare in children, because it is usually acquired from sexual activity and its seroprevalence is lower overall in the general population, which means the pool of adolescents and young adults at risk of primary infection is less affected by this temporal trend. More importantly, risk of first infection is highest for women in their younger childbearing years.
In general, these herpes viruses remain in host tissues in perpetuity (11). Following primary infection, the virus is held in a latent state but can reactivate when immune response is challenged as a result of comorbid conditions or other stressors. Specific antibody immunoglobulin responses can differentiate recent infection from past infection, but reactivations are more difficult to identify with corresponding antibodies. Typically, first exposure to a virus in the herpes family is followed within days by increasing levels of antigen-specific immunoglobulin M (IgM), which then decline and disappear approximately several weeks or months later. Antigen-specific immunoglobulin G (IgG) levels begin to rise shortly after IgM antibodies are produced, but their increase is slower and levels remain elevated for years (12). Thus, high levels of IgM coupled with low levels of IgG suggest a recent primary infection, whereas low levels of IgM together with high levels of IgG point to a past infection, as shown in Figure Figure11 for EBV viral capsid antigen (VCA). Reactivations are more difficult to identify with immunoglobulin patterns, because IgM antibodies may, although not always, appear (13).
The parallel age and temporal distributions of EBV, herpes simplex virus (HSV), and CMV in women with the presence of gastroschisis among their offspring and the propensity for stress-induced reactivation of maternal infections suggest that these infections might be risk factors for this congenital anomaly. Intrauterine infections with HSV or CMV are linked to congenital neurological anomalies (14, 15) but are not otherwise thought to be teratogenic (16). Thus, any association between maternal infections with these viruses and gastroschisis risk likely involves other pathways. In the present study, we used stored serum samples from pregnant women in Finland to examine maternal early-pregnancy antibody reactivities to EBV, HSV, and CMV as potential independent risk factors for gastroschisis in offspring.
This study was a case-control study nested in the Finnish Maternity Cohort, using linked data from Finland's congenital malformations and medical births registers. Since 1983, serum samples have been collected and banked from pregnant women in early gestation, covering more than 98% of all Finnish pregnant women. The mean gestational age at sampling was 11.1 weeks after the last menstrual period; 90% of samples were collected by the 15th week. The Finland National Institute for Health and Welfare gave permission to use these data. For this study, all live births, stillbirths, and fetuses from selective terminations of pregnancy and spontaneous abortions (1987–2012) with a diagnosis code for gastroschisis (International Classification of Diseases, Ninth Revision, code 756.73) were identified (17–19). Records were reviewed by a clinical geneticist, and diagnoses were transcribed to English. Infants born in the same year and geographic region in Finland to mothers of the same age (within 1 year) were randomly selected for each case, resulting in a ratio of 2–3 controls per case. Information on maternal smoking status, number of fetuses (singleton vs. multiple birth), and number of previous pregnancies was also obtained from the birth register. Cases were categorized as “isolated” or “multiple” based on the absence or presence of other, nonintestinal major congenital anomalies.
Samples were drawn from pregnant women at municipal maternity care centers and banked at −25°C until retrieval for this study. Because viral detection from a one-time sample is less useful for classifying infection history and for identifying reactivation, we examined virus-specific antibodies as a measure of infection. EBV-VCA IgG, EBV-VCA IgM, EBV-nuclear antigen (EBNA) IgG, CMV IgG, and CMV IgM antibodies were measured by chemiluminescence microparticle immunoassay (Abbott Architect; Abbott Laboratories, Abbott Park, Illinois). Antibodies were measured for HSV-1 or HSV-2 IgG with enzyme-linked immunosorbent assay (HerpeSelect 1 and 2; FOCUS Diagnostics, Cypress, California) and for HSV-1 or HSV-2 IgM with immunoenzymatic capture (Herpes Simplex 1+2 IgM; DIESSE Diagnostica Senese, Monteriggioni, Italy). Avidity of EBV-specific IgG antibodies was tested in samples reactive for anti-EBV VCA IgM by a protein-denaturing enzyme-linked immunosorbent assay (20–22) using the Enzygnost Anti-EBV/IgG kit (Siemens Healthcare Diagnostics, Tarrytown, New York) with a mixture of viral VCA antigen and in-house curve-fitting software for calculation of the IgG-avidity values (23). Similarly, samples with CMV IgM reactive antibodies were tested for avidity of CMV IgG antibodies by chemiluminescence microparticle immunoassay according to the manufacturer's instructions (Abbott Architect; Abbott Laboratories). IgG avidity is known to increase with duration of exposure to the antigen. Thus, a low avidity index (avidity IgG absorbance divided by standard IgG absorbance) of less than 25% in the presence of IgM reactive antibodies indicates a primary infection. Laboratory procedures were blind to the case-control status of samples.
Distributions of each immunoglobulin were log-transformed and compared between mothers in the case group and mothers in the control group with restricted cubic spline regression to identify cutpoint values for categorization (24). High, medium, low, and nonreactive categories were established for EBV-VCA IgM and IgG, HSV-1 or HSV-2 IgG, HSV-1 or HSV-2 IgM, and CMV IgM; reactive and nonreactive categories were created for CMV IgG. Combinations of antibody reactivities were also compared between cases and controls. Specifically, we examined patterns of 1) EBV-VCA IgM, EBV-VCA IgG, and EBNA IgG; 2) HSV-1 IgG, HSV-2 IgG, and HSV-1 or HSV-2 IgM; and 3) CMV IgG and IgM. As EBV-VCA IgM increases sharply and IgG increases more slowly after first exposure, EBNA IgG lags even further behind and as a rule remains detectable in perpetuity (13, 25).
For relative risk estimation, conditional odds ratios and 95% confidence intervals, stratified by case-control set, were calculated for each viral antibody category, using the lowest category as the referent. Maternal age (years, continuous), maternal smoking status (none, quit during first trimester (1991–2002), smoked during pregnancy, unknown), and gravidity (0, 1, 2, ≥3, unknown) were included as potential confounders. We also examined 2 subgroups: 1) isolated cases, because gastroschisis most often occurs without other major anomalies, and 2) live-born cases, because all infants in the control group were live-born.
Initially, 427 potential cases were identified, and 118 of them were excluded for diagnostic reasons (e.g., early embryonic defect, insufficient information, chromosomal defect). Of 309 cases, 16 cases (5%) were excluded because their serum sampling dates were unclear, either occurring before or after the first trimester. Information on maternal smoking status or gravidity was missing for 22 (7%) cases and 14 (1.7%) controls, resulting in 271 cases and their 753 matched controls for the multivariable-adjusted risk estimation. Isolated cases made up the majority of cases (n = 229; 85%). Selected demographic factors are shown for cases and controls in Table Table11.
Selected antibody categories for EBV, HSV, and CMV were compared between case mothers and control mothers; results are shown in in Table Table2.2. The majority of case and control mothers had nonreactive or low EBV-VCA IgM values and high EBV-VCA IgG values. Approximately twice as many cases were in the highest category for EBV-VCA IgM; the odds ratio was 2.16 (95% confidence interval (CI): 0.97, 4.79). Medium and high levels of EBV-VCA IgG were observed in 98.2% of case mothers and 96.0% of control mothers; the odds ratio was 2.16 (95% CI: 0.82, 5.70). Reactive EBV-VCA IgM suggests a possible primary infection. The VCA IgG avidity index was over 40% avidity for all women with reactive VCA IgM antibodies, ruling out recent primary infection.
Most women had low or nonreactive levels of IgM antibodies to HSV-1/HSV-2. Based on 7 (2.6%) cases and 11 (1.5%) controls with high levels, the odds ratio was 1.94 (95% CI: 0.74, 5.12). High IgG antibodies to HSV-2 were observed among 13.3% of case mothers and 6.1% of control mothers, and the odds ratio was 2.48 (95% CI: 1.50, 4.10). HSV-1 IgG was not associated with gastroschisis risk for the high-levels category.
For CMV IgM, medium and high categories were combined due to small numbers; the odds ratio was 1.11 (95% CI: 0.59, 2.12). In the CMV IgG reactive category, the odds ratio was 0.87 (95% CI: 0.63, 1.19). Of the 14 cases and 35 controls with medium-high levels of CMV IgM, IgG avidity testing identified merely 3 cases and 2 controls with low avidity suggestive of a recent, primary infection. Although these 3 case mothers represented a larger proportion (1.1%) than that for controls (0.3%), the numbers were small.
Patterns of antibody reactivities to EBV, HSV, and CMV are presented in Table Table3.3. For women whose pattern indicated a likely EBV reactivation (because they were positive for all 3 immunoglobulin markers), the odds ratio was 4.24 (95% CI: 1.24, 14.48). The odds ratio for a past HSV-2 infection (IgM-negative (IGM–), IgG-positive (IgG+)) was associated with an approximately 2.4-fold increase in risk (95% CI: 1.52, 3.95). Although slightly higher odds ratios were observed for patterns suggestive of a past EBV or HSV-2 infection (IgM−, IgG+), recent primary HSV-2 or HSV-1 infection (IgM+, IgG−), and HSV-2 or HSV-1 reactivations (IgM+, IgG), the numbers were small, and their wide confidence intervals included the possibilities of an inverse association.
When restricting the analyses to isolated cases (Table (Table4),4), odds ratios for IgM and IgG categories were generally similar to those for cases overall. Positive associations were apparent for the highest categories of EBV IgM and HSV-2 IgG. Odds ratios for EBV IgG and HSV-1 or HSV-2 IgM remained elevated but were less stable. Live-born cases and their matched controls were similarly distributed across all of the serotypes. Odds ratios were more than doubled for the highest category of IgM antibodies to EBV-VCA and HSV-1 or HSV-2 and that of IgG antibodies to EBV-VCA and HSV-2, although the confidence intervals around EBV estimates crossed the null. Of note, all 7 case mothers with HSV-2 IgM values of at least 1.2 had infants that were live-born, and the corresponding odds ratio of gastroschisis was 4.45 (95% CI: 1.44, 13.80).
Our exploration of some common herpes viruses in relation to the risk of gastroschisis was prompted by the elevated risk of both among adolescents. Specifically, gastroschisis occurrence and EBV, HSV, and CMV incident, primary infections have been increasing in young women of childbearing ages (26). Results from the present study, however, do not support associations between maternal primary infection with EBV or CMV and gastroschisis in offspring. The lack of such associations is consistent with the clinical picture of gastroschisis, which does not include evidence of congenital infection (27). However, maternal-fetal transmission of infection is not the only mechanism by which adverse birth outcomes can be influenced by maternal infection. Innate and adaptive immune responses to infection can produce fever, inflammation, circulating cytokines, and alteration in epigenetic regulation, which have been linked to birth defects and other adverse reproductive outcomes (28, 29). Our exploration of the full distributions of IgM and IgG antibody reactivities to these viruses identified possible novel risk factors for gastroschisis.
Specifically, high IgM and IgG levels were each observed to be associated with higher gastroschisis risk; elevations in the levels of both immunoglobulins suggest possible maternal reactivation, particularly in the presence of high-avidity EBV-VCA IgG. Further evidence to support EBV reactivation as a risk factor for gastroschisis is the greater than 4-fold higher odds ratio for women with simultaneously high levels of VCA IgM, VCA IgG, and EBNA. Similar to EBV, odds ratios for high HSV-1/HSV-2 IgM and HSV-2 IgG were elevated. High HSV-1 IgG values, on the other hand, did not appear to be associated with higher gastroschisis risk. Patterns of HSV reactivities revealed elevated odds ratios, albeit with 95% confidence limits below 1.0, for those consistent with new or recurrent HSV-1 or HSV-2 infections and old HSV-2 infections, but there was no association with past HSV-1 infection. Maternal exposure to CMV did not appear to be associated with gastroschisis. It is interesting that, of the herpes viruses we examined, CMV is the only one known to cause adverse pregnancy outcomes (14) and yet the only one showing odds ratios below 1.0 in relation to gastroschisis. If early gestational CMV infection (in the fetus or the mother) causes fetal loss, as has been reported (30), and the magnitude of such an association is greater than that for gastroschisis, the observed “protective” odds ratios could reflect a downward bias. Our observed associations with EBV and HSV remained when we restricted our analyses to the presumably more homogeneous group of isolated cases and, in order to address the possibility that fetal deaths differ in ways that might bias the overall findings, to live-born cases.
Primary EBV infection in pregnancy is generally considered to confer a very small risk of maternal-fetal transmission, with very few case reports of fetal anomalies (16, 31). Reactivations—based on the simultaneous presence of EBV early antigen and EBNA antigen antibodies (32, 33), circulating EBV DNA (34), or the simultaneous presence of antibodies to EBV-VCA IgM, EBV-VCA IgG, and EBNA (35)—are common in pregnant women (32–34) and in general populations (35). To our knowledge, this is the first study to have examined exposure to common herpes viruses in early pregnancy in relation to gastroschisis (26).
Congenital herpes infections have been documented, but maternal-fetal transmission is associated with either recurrence or primary infection in late pregnancy and parturition. Epidemiologic evidence suggests that HSV-1 or HSV-2 primary infections in early pregnancy have very little if any association with structural defects (16). In a Hungarian study, Acs et al. (36) reported that oral herpes was associated with cleft lip and/or palate but found no associations for genital herpes. Antiherpetic medication use, based on prescription data, was reported to have no association with birth defects overall in a Danish study (37), but abdominal wall defects were more common in the offspring of exposed women than in those of unexposed women. A case-control study of gastroschisis identified increased risks for genital herpes and for antiherpetic medication use (38). Although the case-control study relied on maternal report and the data-collection instrument did not include specific prompts for genital herpes or antiherpetic medications, its results are compatible with the present biomarker-based findings.
Both primary infection and reactivation (or reinfection) with CMV in pregnancy are associated with a congenital infection syndrome that includes neurosensory deficits (14). Other adverse outcomes in the absence of congenital CMV infection are not well studied, despite many case reports (39). Three case reports of CMV-infected newborns with gastroschisis (40–42) are worth noting, but they do not support or detract from our observed null findings.
If a true association between reactivation of herpes infection and gastroschisis exists, the underlying mechanism is unclear. The pathogenesis of gastroschisis is not known, despite several hypotheses, including vascular disruption in early embryogenesis (43–46). Reactivation of herpes viruses induces a cell-mediated immune response that includes monocyte production of cytokines and intercellular adhesion molecules (ICAMs) (47). ICAM 1 is expressed by some types of white blood cells and endothelial cells, and high levels are associated with vascular pathology (48). EBV and CMV reactivations in particular have been associated with higher levels of ICAM 1 (47, 49). Reactivation of other viruses in the herpes family, including HSV-2, has stimulated in vitro secretion of proinflammatory cytokines (50). Gastroschisis was linked to ICAM in a study that explored 32 blood vessel–related genes in affected babies. Having at least 1 copy of a single nucleotide variant of the ICAM-1 gene approximately doubled gastroschisis risk (51). Plasminogen activator inhibitor 1 and endothelial leukocyte cell adhesion molecule 1 were also observed to increase risk, with atherosclerotic and inflammatory roles, respectively (51).
In addition to the increasing occurrence of gastroschisis and its strong inverse association with maternal age, a wide range of maternal exposures have been associated with gastroschisis risk in epidemiologic studies. After adjusting for maternal age, several factors have consistently been shown to increase gastroschisis risk, including smoking (52), salicylate use (5), genitourinary or sexually transmitted infection (53–55, 57), chlamydia (56, 57), and low body mass index (58, 59). In addition to these better-established risk factors, age-adjusted positive associations have also been reported for alcohol drinking (60–62), nongenitourinary infections (54), use of cold medications (61, 63, 64), illicit drug use (54, 61, 65, 66), pesticide exposure (67–69), antidepressant use or mental health conditions (54, 70, 71), diets low in fruits and vegetables (72, 73), and no or low supplementation with folic acid (73, 74). None of these associations appears to explain the inverse association with young maternal age. This wide range of suggested risk factors, together with the lack of a clear pathogenetic mechanism, suggests that gastroschisis may have multiple causes. It is worth considering, however, whether the range of exposures shares any commonalities. One possibility is that they may all be related to immune system stress. Cigarette smoking (75), high alcohol intake (76, 77), cocaine use (78), and pesticide exposure (79) suppress immune response. The reports of genitourinary infections as a risk factor likely represent many different infectious agents, but all of those infections would elicit an immune response. Increased risks associated with the use of salicylates and cold medications may be confounded by underlying injury or infection. Also, increased levels of inflammation and proinflammatory cytokines co-occur with depression (80). The strong inverse association between body mass index and gastroschisis could also involve inflammation, as suggested by high levels of cytokines and cell adhesion molecules in women with anorexia nervosa (81). Finally, multivitamin supplementation and a diet high in fruits and vegetables may reduce inflammation, as a result of their antioxidant properties (82).
In other words, exposures and experiences that stress the immune system in early pregnancy might trigger a cascade of effects that ultimately impair abdominal wall development. Adolescent pregnancy, independent of harmful behaviors, is often a stressful event (83), which could be the basis for the higher occurrence of gastroschisis in the youngest women. Stress-induced EBV and HSV reactivations are well-documented phenomena (84) and could be part of such a cascade. It is also possible that gastroschisis risk associated with high levels of IgG and IgM antibodies to EBV and HSV reflects direct effects of correlated stressors. If EBV and HSV reactivations are intermediates between stressors and the development of gastroschisis, interventions that boost cell-mediated immune response may be protective. If the observed positive associations are true, our results bring us closer to understanding risk factors for gastroschisis but continue to leave the majority of cases unexplained. Conversely, if herpes virus reactivations are intermediates between the many maternal exposures previously shown to increase gastroschisis risk, we would expect them to be more common than we observed.
The vast majority of epidemiologic studies of maternal risk factors for gastroschisis in offspring rely on maternal report. To our knowledge, this was the first study to examine biomarkers of herpes virus infections in a large, population-based sample of women in early pregnancy. The timing of blood draws in early pregnancy was ideal for a single measurement for the study of herpes virus infection in relation to gastroschisis (27). Nevertheless, the duration of elevated IgM antibody levels may be as short as 7 days (85), and the shorter the duration, the more likely we are to miss some reactivations. Any resulting misclassification of antibody patterns would bias the observed odds ratios downward. EBV early antigen, which can also be helpful in the identification of reactivations, was not assayed. At present, there is no reliable, specific method for measurement of reactivations. Another limitation of our measure of infection status is the lower sensitivity and predictive value of HSV IgM assays and the inability to differentiate between types (86). Viral detection was also not part of this study, but serological evidence of virus-specific antibodies is considered superior in sensitivity for measuring infection, perhaps due to recent evidence of incomplete viral replication during reactivations (50).
In conclusion, this population-based study of early pregnancy antibody reactivities to common viruses in the herpes family provides no evidence that gastroschisis results from primary infection with EBV, HSV-1, HSV-2, or CMV in early pregnancy. Our findings do suggest that past or reactivated EBV and HSV-2 infections, but not CMV infections, may play a role in the development of gastroschisis. It is not clear whether the observed associations, if true, reflect a specific role for reactivations or reinfections or that reactivations or reinfections are a general marker of stress resulting from other known risk factors. Future studies should include more detailed and precise measurements of reactivations and consider interactions with other maternal exposures.
Author affiliations: Department of Epidemiology, School of Public Health, Boston University, Boston, Massachusetts (Martha M. Werler, Samantha E. Parker); Department of Virology, Haartman Institute, University of Helsinki, and Helsinki University Hospital, Helsinki, Finland (Klaus Hedman); Information Services Department, National Institute for Health and Welfare, Helsinki, Finland (Mika Gissler, Annukka Ritvanen, Heljä-Marja Surcel); and Prenatal Serology Laboratory, National Institute for Health and Welfare, Oulu, Finland (Heljä-Marja Surcel).
This project was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant R21-HD072460 and T32-HD052458).
We thank Lea Hedman, Sara Kuusiniemi, and Mari Sujala for their laboratory assistance.
These results were presented in part at the 2015 Annual Meeting of the National Birth Defects Prevention Network, October 19–21, 2015, Arlington, Virginia.
Conflict of interest: none declared.