EVALUATING CHEMICAL AGENTS FOR POTENTIAL HAZARDS TO REPRODUCTION
Jack B. Bishop
The mission of the National Institute of Environmental Health Sciences (NIEHS) is to reduce the burden of human illness and dysfunction due to environmental causes through a better understanding of interrelationships between the interactive elements of environment, genetics (susceptibility) and time (age). The National Toxicology Program (NTP) at NIEHS strives to safeguard public health by identifying and characterizing the toxic effects of environmental chemicals and by providing quality data that regulatory agencies can use for risk assessments.
NTP tests that assess reproductive toxicity include (1) Total Reproductive Capacity Test (TRCT; Bishop et al., 1997); (2) Dominant Lethal Test (DLT; Generoso and Piegorsch, 1993; Lockhart et al., 1992); (3) Reproductive Assessment by Continuous Breeding (RACB; Chapin et al., 1997;1998); (4) Reproductive, Development, and General Toxicity (RDGT); and (5) Teratology Test (TER). These are primarily rodent tests conducted with rats and/or mice, except for the TER, which may also include rabbits, hamsters, or guinea pigs. Taken together, these tests cover periods of exposure from the gametes through, fertilization, implantation, in utero development, birth, lactation, postnatal development, and the adult, even out to a total of three generations.
Most of these tests are conducted at NTP contract laboratories under carefully specified protocols. However, these test protocols permit variations in the sex, species, and strain of test animal, route and dosing regimens for chemical administrations, and even the number and types of endpoints measured. A primary endpoint in all of these tests is pup survival, either number in utero or
number of live born. Other endpoints evaluated may include body weight, feed consumption, clinical signs, pup or fetal weights, anogenital distance, neurological functions, immunology, hormone analysis, sperm parameters, vaginal cytology, organ weights, gross and microscopic pathology, and visceral and skeletal anomalies. Gestation length can be determined with most of these assays, but often the matings are not specifically timed.
The NTP has conducted female mouse TRCTs on 55 chemicals, male mouse DLTs on 30 chemicals, female mouse DLTs on 30 chemicals, mouse RACBs on 80 chemicals, rat RACBs on 36 chemicals, rat RDGTs on 12 chemicals, mouse TERs on 46 chemicals, rat TERs on 51 chemicals, rabbit TERs on 23 chemicals, and hamster TERs on 2 chemicals. None of the TRCT, DLT, or TER studies identified any significant change in gestation length, although there were numerous studies with significant reductions in pup survival and/or reduced size and weight. Of 20 RACB and RDGT studies reviewed to date for which either “average gestation length” or “cumulative time to litter” data were recorded, only 3 exhibited a significant exposure-related change: one reduced the average time to litter (nitrofurazone), one increased time to litter (ethoxyquin), and one increased gestation length (sodium selenate).
Currently, plug checking and/or vaginal smears are conducted for DLTs, RDGTs, TERs and the RCAB F1 generation matings such that gestation length can actually be determined, but assessment of “prematurity” is difficult. It may be necessary to modify some of the current NTP reproductive test protocols to better collect this information. However, considering the infrequent observation of a significant effect, especially a shortening of gestation that might be associated with “preterm delivery, one has to question whether it would be worth the associated increased test cost.
Further, the role of infection in the etiology of premature birth has been estimated to be 40–60 percent; it is doubtful that rodent models will be of much use in predicting such environmental effects. In addition, initiation of parturition, a critical element in preterm deliveries, appears to be vastly different for rodents than for primates (Norwitz et al., 1999). Thus, I think we must constantly reassess our use of rodent animal models and the endpoints we measure for predictors of human ill-health effects, particularly for outcomes such as preterm birth.
MODELS THAT STUDY THE ACTIONS OF UTERINE LYMPHOCYTES DURING PREGNANCY
The mammalian uterus is endowed with all of the known lineages of immune-competent cells. In adult females, predictable, cyclic patterns of change occur in immune-competent uterine cells that can be correlated with ovulation or
pregnancy. In women, many nonhuman primates, and rodents, early pregnancy is accompanied by the transient appearance in the uterus of a population of granulated lymphocytes. These specialized lymphocytes become the dominant immune cells in the first half of gestation and have been estimated to represent >70 percent of the maternal bone marrow-derived cells at implantation sites. This population, which has been defined in women and mice as a subset within the natural killer (NK) lymphocyte lineage, has a strong association with uterine stroma. The granulated cells first appear and proliferate as uterine stromal cells transform from fibroblasts into decidual cells. Uterine Natural Killer (uNK) cells decline in the later part of pregnancy and are absent from the postpartum uterus.
Studies of pregnancies in mice genetically depleted in some or all lymphocyte lineages were important in identifying the lineage of the pregnancy-associated lymphocytes and providing insights to their functions. Mice lacking lymphocytes are immune deficient and must be housed under strict microbiological barriers to maintain their health. If this is done, most immune-deficient strains breed well and are excellent models for reproductive studies. Establishing and quantifying consistent phenotypes in the implantation sites of specific immune deficient mice provides powerful animal research tools in which reconstitution of implantation site structure can be assessed. These reconstitution approaches have many aspects that could be exploited in investigations of environmental toxicants. Four modeling approaches are briefly reviewed.
Model 1: Mice Genetically Deficient in Lymphocyte Lineages Have Quantifiable Implantation Site Defects. Mice deficient in NK cells were first reported in 1994–1995 (Wang et al., 1994; Di Santo et al., 1995). Our histological studies of implantation sites from NK cell-deficient strains, modified in different genes, showed that no uNK cells were present and that the architecture of implantation sites was disturbed. In particular, the decidual spiral arteries did not undergo the physiological changes of pregnancy and the decidua was hypocellular. To quantify these anomalies, vesse-to-lumen ratios and cell nuclei per square millimeter of decidua basalis are measured. Pregnancies proceed to term without evidence of dystocia. Litter sizes and weaning rates are normal (Greenwood et al., 2000).
To confirm that the absence of uNK cells was the key to the phenotype, reconstitution of the mice was undertaken, using marrow grafts from SCID (severe combined immuneodeficient) mice, a strain that produces no lymphocytes other than NK cells. When the grafted mice were mated, their implantation sites had high numbers of uNK cells and the spiral arteries and decidua matched those in normal pregnant mice (Guimond et al., 1998). This suggested that uNK cells produce molecules regulating gene expression in endothelium and smooth muscle cells of the decidual spiral arteries and in decidua. Interferon gamma (IFN-γ) is a key product of NK cells. Engraftment of alymphoid mice (RAGo/o/ γco/o) with marrow from mice deleted for IFN-γ production reconstituted uNK
cell numbers but failed to correct the vascular and decidual anomalies (Ashkar et al., 2000). In contrast, treatment of pregnant alymphoid mice with IFN-γ gave complete normalization of the vessels and decidua in the absence of any uNK cells. This indicates a central importance for IFN-γ, a regulator of expression of hundreds of genes, in implantation site health. It also identifies a potential target molecule for monitoring of environmental contaminant effects on the uterine immune system.
Model 2: Monitoring Sites of Cell Precursors and Their Mobilization to the Pregnant Uterus in Vivo. Availability of animals having no ability to generate NK cells permits study of the origin of uNK cells by transplantation of tissues from mice with normal lymphocyte precursors. Uterine segment transplantation was used to show that the precursors do not reside in the uterus, while lymphoid tissue transplants to mated RAGo/o/γco/o showed that all lymphoid tissue had at least some minor precursor cell content. Spleen cells, but only those harvested from donors who were in the early (gestation day [gd] 3 or 5) stages of pregnancy, were the richest source of uNK cell precursors (Croy et al., 2001). This has provided a model in which molecules responsible for homing patterns of cells can be assessed. Spleen cells are collected from gd 3 pregnant donors, genetically deleted for the genes of interest, and transplanted into mated alymphoid mice. Histological specimens from implantation sites are evaluated for numbers of uNK cells, their stages of maturity (granularity and cell diameter), and their distribution relative to the decidual spiral arteries in comparison to controls. Evaluation of environmental agents for interference with cell homing patterns to the uterus could be evaluated using this assay.
Model 3: Evaluation of Uterine Stromal Cell-Mediated Effects. A cDNA microarray analysis to identify IFN-γ-regulated genes in decidua basalis from normal mice at gd 6 and 10 (the genes potentially regulated by uNK cells) revealed large gains in expression of two members of the α2-macroglobulin α2M) gene family. Products of this molecular family are found in high concentrations in blood plasma where they serve as protease inhibitors and cytokine transporters. Implantation sites from mice simultaneously lacking the two most highly expressed known members of the family, MAM and MUG-1 (Umans et al., 1999) were studied and found to have a phenotype related to that in NK cell-deficient mice. In MAMo/o/MUG-1o/o mice, the spiral arteries dilated but were thick walled due to a cuff of trophoblast-like cells while decidua was hypocellular. In preliminary studies, infusions of human α2M into NK cell-deficient mice promoted spiral artery dilation, but was less effective than infusions of IFN-γ (V:L ratios at gd 12 for PBS, αtwoM; and IFN-γ infusions were 2.8, 1.8, and 1.2, respectively; in normal mice, the V:L ratio is 1.2). A novel member of the αtwoM family, AM-X, was also identified and is being cloned. AM-X is expressed by stromal cells as they decidualize and by cells surrounding the spiral arteries. These studies highlight that uNK cells have regulatory actions on decidual cells that can be dissected and monitored.
It is also true that stromal cells act on uNK cells. Uterine stromal cells produce interleukin-15 (IL-15), the major survival signal for NK cells. Regulation of IL-15 in the uterus is unusual because it appears to be independent of the transcription factor IRF-1 used in lymphoid tissue (Croy et al., 2001). Mice, in which overexpression of genes such as IL-15 has been induced, are also valuable because overactivity of uNK cell behavior can be quantified.
Model 4: Use of Mouse Uterine and Lymphoid Tissues to Evaluate Human Cells. Tissue sections from mice can be used to assess functional interactions between endothelium and human lymphocytes in vitro under conditions that mimic shear flow within blood vessels (Frey et al., 1998). We found that pregnancy of the mouse donating the tissue enhanced this interaction. In peripheral lymphoid tissues, elevated adhesion was observed by gd 3, peaked at gd 6–8, and was sustained until birth. Postpartum adhesion levels were the same as in tissues from virgin mice. When adhesion of human blood lymphocytes, prelabeled with the NK cell marker CD56, was examined on uterine tissue sections from virgin mice, the CD56bright cell subset was enriched from 1.5 to 25 percent of the total cells. Further threefold enrichment occurred if the uterine sections came from pregnant mice. Antibodies to L-selectin or α4-integrin blocked adhesion. The dynamic changes in endothelium, demonstrated in these experiments, were induced equivalently in ovariectomized females treated by either estrogen or progesterone or both. The finding that steroid hormones change the functional properties of endothelium in a way that is recognized by lymphocytes has been consistent across 20 human lymphocyte donors using cell lines expressing single adhesion receptor molecules. This assay, in a modified form, could have applications in toxicology to monitor environmentally induced changes in endothelial cells within the uterus that would be recognized by human cells.
Acknowledgments. The studies reported include work done in collaborations with Dr. S.S. Evans, Roswell Park Cancer Institute, Buffalo, New York; Dr. F. van Leuven, K.U. Leuven, Leuven, Belgium; Dr. A.M. Hayes, University of Guelph, Guelph, Canada; and Dr. Michael Caligiuri, Ohio State University, Colombus, Ohio. The studies were conducted by research trainees in my laboratory and supported by grants from Natural Sciences and Engineering Research Council of Canada (NSERC), Ontario Ministry of Agriculture Food and Rural Affiars (OMAFRA), and Canadian Institutes of Health Research (CIHR).
REGULATION AND ASSESSMENT OF UTERINE CONTRACTILITY AND CERVICAL RIPENING DURING PREGNANCY
The problems associated with labor during pregnancy are among the most important health issues facing health care providers of women. Understanding the regulation of the uterus and cervix in pregnancy and developing methods to
control their function are essential to solving problems relating to labor. We have evaluated uterine contractility and cervical ripening regulatory mechanisms during pregnancy in animal models and in humans under a variety of conditions. These studies suggest that environmental toxicants could alter uterine and/or cervical function and thereby accelerate or prolong the course of pregnancy. Our studies, in accordance with many others, suggest that both the uterus and the cervix pass through irreversible changes in preparation for normal-term labor as well as during preterm labor. The preparatory step in the uterus occurs just prior to labor and consists of an upregulation of systems important for augmented myometrial contractility, including increased cell-to-cell coupling, changes in ion channels, receptors for stimulants, and a downregulation of inhibitory systems such as nitric oxide. In the cervix and also in fetal membranes, the preparatory step during term delivery is temporally different from that in the uterus, indicating separate control systems. In the cervix the preparation step includes a slow progressive softening during the last half of pregnancy due to a decrease in collagen concentration prior to dilation, which occurs immediately before or accompanies labor.
Unfortunately, at the moment, we have only crude, inaccurate, and subjective methods to clinically measure the functional changes in the uterus and cervix that occur in preparation for labor. Without better diagnostic tools we may never advance our understanding of the uterus or cervix and may never find better treatments to modify their activity. In the past several years, we have developed noninvasive methods to quantitatively evaluate the uterus and cervix based, respectively, on recording of uterine electrical signals from the abdominal surface (uterine electromyography [EMG]) and measurement of light-induced cervical collagen fluorescence (LIF) with an optical device (collascope). The EMG method uses electrodes placed on the abdominal surface and systems for digitizing, recording, and analyzing the electrical bursts of the underlying uterine EMG activity which produce the contractile events. The system and methods are similar to recording cardiac electrical activity with electrocardiography. The collascope is a specially constructed device that illuminates a small area of the ectocervix and excites the collagen in the tissue to fluoresce. The fluorescent signal is analyzed by the device and is proportional to the amount of collagen in the cervix. Using an EMG labor monitor and collascope, we have examined more than 400 pregnant patients in the clinic and hundreds of animals in the laboratory. Both methods are rapid and allow accurate assessment of uterine contractility and cervical ripening. Our studies indicate that uterine and cervical functions can be monitored successfully during pregnancy using these approaches and that these techniques might be used in a variety of conditions associated with labor to better define management. The potential benefits of the proposed instrumentation and methods include quantitative procedures to predict the onset and progress of the preparatory steps leading to normal or preterm labor, reduction in the rate of preterm delivery, improvement in maternal and perinatal out-
come, monitoring treatments, decreasing cesarean section rate, and improved research methods to better understand uterine and cervical function.
FUNCTION OF STEROIDS IN PARTURITION AND PRETERM LABOR
Increased uterine contractility at term and preterm labor results from activation and then stimulation of the myometrium. Activation can be provoked by mechanical stretch of the uterus and by an endocrine pathway resulting from increased activity of the fetal hypothalamic–pituitary–adrenal (HPA) axis. Stimulation of the myometrium can result from a number of factors one of which is likely prostaglandins (PGs). The focus here is on the role of steroids in parturition, with particular emphasis on the regulation of prostaglandin production. In experimental animals, there is a decrease in the concentration of progesterone in the maternal plasma prior to the onset of labor, but in the human such a drop in progesterone concentrations does not occur. However, at the cellular level there are a number of similarities in the action of steroids in animal and human parturition in regulating PG formation and metabolism. In parturition the type 2 isomer of prostaglandin H synthase (PGHS-2) is particularly important and its expression has been found to increase in term and preterm labor. In fetal sheep, increased cortisol regulates PGHS-2 expression in the placenta in an estrogen-independent manner, resulting in increased levels of PGE2 in the fetal circulation. Later increases in maternal uterine expression of PGHS-2 require estrogen and lead to increased concentrations of PGF2α in the maternal circulation. In women, the fetal membranes surrounding the amniotic cavity are thought be an important source of prostaglandins involved in parturition. Formation of prostaglandins in these tissues is regulated by paracrine and autocrine mechanisms. In women, cortisol can contribute to increased PG production in fetal tissues through upregulation of PGHS-2 and downregulation of 15-hydroxyprostaglandin-D dehydrogenase (PGDH), the principal enzyme involved in the metabolism of prostaglandins. The effect of cortisol on chorion expression of PGDH reverses a tonic stimulatory effect of progesterone and likely occurs through the glucocorticoid receptor. By competing with progesterone inhibition, cortisol also increases expression of placental corticotropin-releasing hormone (CRH), which can in turn increase prostaglandin production. The 11βhydroxysteroid dehydrogenase, an enzyme which interconverts the active glucocorticoid, cortisol, and the inactive glucocorticoid, cortisone, appears to be important in regulating the effect of cortisol on prostaglandin metabolism in the placenta and chorion. In turn, this enzyme is upregulated in the fetal membranes by prostaglandins, forming a feed-forward (positive feedback) loop leading to increased prostaglandin production. Other agents such as pro-inflammatory cytokines similarly upregu-
late PGHS-2 and decrease expression of PGDH, indicating the presence of several mechanisms by which labor at term or preterm may be initiated.
CLINICAL AND PUBLIC HEALTH INTERVENTIONS—WHY NOTHING HAS WORKED
Robert L. Goldenberg
Conceptually, the origin of preterm birth can be divided into deliveries initiated by the clinician for the benefit of either the infant or the mother (indicated preterm birth) or those that follow either spontaneous preterm labor or spontaneous rupture of the membranes. The latter two categories taken together are often called spontaneous preterm birth. Approximately 20 percent of preterm births are indicated, approximately 30 percent follow spontaneous rupture of the membranes, and approximately 50 percent follow spontaneous preterm labor. The ultimate method of delivery, whether vaginal or by cesarean section, is not part of this definition.
Indicated preterm births usually occur because the mother is severely ill with a life-threatening condition, usually preeclampsia, or the fetus shows signs of deterioration and risk of fetal death, often in conjunction with maternal preeclampsia or fetal growth restriction. Attempts to reduce the prevalence of the conditions leading to indicated preterm birth, especially preeclampsia and fetal growth restriction, have generally failed. Examples of failures to reduce the severity or prevalence of these conditions include nutritional interventions, maternal bed rest, low-dose aspirin, and calcium supplementation. Because the prevalences of preeclampsia and fetal growth restriction appear to have remained unchanged, while there is increasing evidence that early delivery for fetal distress associated with these conditions leads to a reduction in stillbirths, the percentage of infants born following an indicated preterm birth seems to be increasing. It is emphasized that since the ultimate goal of obstetric care is to increase the number of living infants born without handicap, not to reduce preterm birth, this increase in preterm birth may, in fact, be beneficial. In any case, there is nothing on the horizon to suggest that a reduction in the prematurity associated with these conditions will occur in the near future.
Spontaneous preterm births are often divided into those that occur early and those that occur later. Early preterm births, those at less than 28 or 30 weeks’ gestation, generally occur in association with an intrauterine infection or placental hemorrhage. In a study at our institution, 83 percent of the spontaneous preterm births weighing <1,000 grams were associated with bacteria in the fetal membranes prior to membrane rupture. Therefore, a chronic intrauterine infection with relatively low-virulence organisms such as ureaplasma, mycoplasma, and bacteroides is associated with and probably causal for most early preterm births. Later preterm births, especially those that occur at 35 and 36 weeks’
gestation, are, for the most part, not associated with infection, placental hemorrhage, or a specific etiologic factor. Instead, these preterm births appear to occur through the normal mechanisms responsible for term labor, which, however, occur earlier than usual. Women having a preterm birth at these gestational ages often have an increased number of risk factors, such as maternal thinness, smoking, and various psychosocial characteristics, but often have no specific precipitating cause for the spontaneous labor or rupture of membranes.
Interventions to reduce spontaneous preterm birth are categorized as targeted or general. General interventions might include providing prenatal care to a population that previously had none, improving the general quality of prenatal care, providing across-the-board nutritional supplementation, or providing some sort of social support or home visiting to a population of pregnant women. These types of interventions, likely because they are so nonspecific, have rarely been shown to be beneficial in reducing preterm birth.
Nutritional interventions deserve particular attention. Although preterm birth is clearly associated with maternal thinness, in developed countries neither nutritional counseling nor caloric, vitamin, or mineral supplementation has had much, if any, impact on the preterm birth rate. High protein nutritional supplementation has been associated with increased risk of preterm birth. These findings likely do not apply in developing countries where randomized trials of both vitamin or mineral and caloric supplementation have been associated with improved outcomes.
In recent years, the general trend for attacking preterm birth has been to identify specific risk factors and define an intervention to either eliminate or treat the risk factor in an attempt to reduce the associated preterm birth. Risk screening encompasses various demographic and medical history questionnaires such as that developed by Creasy et al. in the early 1980s, home uterine activity monitoring to determine an increase in uterine contractions, cervical ultrasound to demonstrate shortening of the cervical length, fetal fibronectin screening looking for a fetal protein in the vagina (where it should not normally occur), and a wide variety of other types of attempts to define populations at increased risk. Cervical or vaginal fetal fibronectin is probably the most potent risk factor, followed by a short cervical length. To date, however, despite a number of attempts, no intervention targeted to these risk factors has consistently been shown to reduce preterm birth.
As stated earlier, intrauterine infection is associated with a very large proportion of the earliest preterm births. Often, this intrauterine infection is linked to the presence of bacterial vaginosis, which, in nearly 20 studies, is associated with an approximately twofold increased risk of a preterm birth. Various attempts using antibiotics to treat either the bacterial vaginosis (BV) or the intrauterine infection thought to be associated with bacterial vaginosis have produced mixed results. In developed countries, if antibiotic treatment of BV actually works to prevent prematurity, it appears to work only in those women who have
had a prior preterm birth. Results from developing countries, such as in the Rakai region of Uganda, suggest that mass treatment with antibiotics of a population of pregnant women may reduce the preterm birth rate.
The use of antibiotics in women in early preterm labor in an attempt to reduce preterm delivery has more often than not been unsuccessful. There are, however, two randomized studies, both using metronidazole and ampicillin, that suggest benefit. Nevertheless, it appears that in women in preterm labor, the use of antibiotics alone is not likely to have a major impact on the early preterm birth rate. If infection exists and antibiotics do not seem to cure the infection and prevent the preterm birth, the question is, of course, Why not? Many potential explanations can be given, but it may be that once the intrauterine infection is established, antibiotics may actually increase the inflammatory response and hasten the premature delivery rather than delay it. An alternate explanation may be that the intrauterine bacteria are in an anatomic space, or in an environment such as a biofilm, that protects them from antibiotic treatment. There are many examples of a “structural” protection of bacteria from antibiotics in other parts of the body.
In summary, it is clear that at the present time, the interventions that have been applied generally to populations and to specific targeted high-risk populations have not achieved any real reduction in the preterm birth rate. The likely explanation is that the events leading to preterm birth are far more complicated than many of us realize and that, in general, we know far too little about the sequence of events leading to spontaneous preterm birth. Only when we better understand the biology of both early and late spontaneous preterm birth will we be likely to develop effective interventions.
ADVERSE CHILD OUTCOMES ASSOCIATED WITH PRETERM BIRTH
Technologic and therapeutic advances in neonatal intensive care have led to the improved survival of preterm infants. Whereas prior to the 1990s very few infants survived with birth weights less than 2 pounds, the vast majority of such infants now survive. The improved survival has however been associated with an increase in neonatal complications including brain injury and chronic lung disease, which are most prevalent among the smallest and least mature infants.
Neonatal complications of prematurity, including respiratory distress syndrome, chronic lung disease, brain hemorrhage, infections, and poor growth, result in health and developmental problems during childhood. These include an increase in respiratory infections and asthma, poor growth attainment, and neurological and developmental handicaps. The rates of cerebral palsy, poor vision, deafness, and mental retardation are higher among preterm than term-born chil-
dren. These impairments increase with decreasing gestational age and birth weight. Predictors of poor neurodevelopmental outcomes include the neonatal complications of prematurity as well as socioenvironmental risk factors. The same socioenvironmental factors that predispose to preterm birth usually continue after birth and may have deleterious effects on childhood health and development. The major neurodevelopmental impairments such as cerebral palsy and mental retardation can be diagnosed in early childhood. However, more subtle problems in behavior and functioning may present later at school age even among children who have no overt neurodevelopmental sequelae. These school age problems include poor visual motor and gross motor functioning, poor math abilities, behavioral problems mainly related to poor attention and hyperactivity, and social–emotional immaturity. Although respiratory and other health problems tend to diminish during childhood and catch-up growth occurs among many of the children, the school functioning problems do not resolve and persist into adolescence. More preterm than term-born children fail grades and fewer complete a high school education. Children born preterm have increased special health care needs compared to term-born children. During infancy and early childhood these needs pertain to both medical and educational services. However, during the school age and adolescent years they pertain mostly to special education, physical and occupational therapy, and counseling. Differences in intelligence between term and preterm children are evident even among preterm children who weigh between 3 and 5 pounds at birth and who do not require neonatal intensive care.
A FRAMEWORK FOR SOCIAL AND CULTURAL DETERMINANTS OF PREMATURITY
Carol J. Rowland Hogue
Although stressful events and lack of social support during pregnancy have been associated in some studies with increased risk of preterm delivery, interventions to increase social support have, in general, not lengthened gestation. Lack of effect may suggest that the causal hypothesis is faulty. Alternatively, the trials may have been diluted by inclusion of women in the intervention group who did not need social support. Some have argued that currently available interventions need to be offered only to women who both lack intimate support and are at high risk of preterm delivery. Other reasons for inconsistent results in observational studies and clinical trials may be incomplete stress exposure assessment or inaccurate delineation of the stress–health causal model, leading to measurement error and failure to account adequately for confounding and effect modification.
The classic “host, environment, agent” triangle of epidemiologic causality can illustrate what might be included in a more comprehensive test of the stress–
preterm delivery hypothesis. The host is the individual woman, the environment is her social and cultural context, and the agent is the immediate stressful event(s) requiring her response. This framework provides a context to determine the extent to which a given theory may include important and potentially interrelated factors.
Host Susceptibility. Individuals may differ in psychological and physiological responses to the same stressor, reflecting differences in the context within which the stressor occurs and differences in individual likelihood to respond to that type of stressor. Differential host susceptibility may be specific to the type of stressor. Potential risk factors for host susceptibility may include factors present at birth or related to personality, early life experiences, coping strategies, and circumstances of the pregnancy. In this framework, external resources available for coping with stress, such as levels of social support, are part of the environmental context.
Environments or Contexts. The social and cultural context of a pregnant woman may add to or ameliorate her level of distress. Ameliorating environmental factors include intimate social support, which has been associated with improved pregnancy outcome in a number of studies, although attempts to replicate social support in clinical trials to prevent poor pregnancy outcomes have not been effective. The chronic stress associated with environmental or social stressors may be measured at the individual level, for example, through inquiry into daily hassles as well as into perceptions of environmental issues. However, the negative health impact of environmental stressors is dependent not only on individual perception, but also on the actual environmental risks. For this reason, measurement of environmental risks must include contextual variables. Stressful enviromental factors include those associated with gender, with socioeconomic status, and with race or ethnicity.
Agents. Agents of stress are those events that provoke immediate individual response to a physical or emotional challenge. Stress is necessary for growth, development, and maintenance of brain and muscle activity. However, prolonged stress or an acute stressor that overwhelms the individual’s ability to return to homeostasis can result in ill health. Stress results from host and environmental risks as well as acute stressful events. Whether the individual is overwhelmed depends not only on the strength of the agent but also on host susceptibility to stress, as well as the background level of environmental and contextual stress, as mediated by the background level of host, environmental, and contextual resources for handling stress.
Two kinds of stress agents have been studied extensively with respect to their impact on pregnancy outcomes. These are acute, stressful life events and hard physical labor—a type of individual, chronic exposure. A third kind of agent, individual experiences of racism, holds potential for explaining at least part of the excess preterm delivery rate among African Americans and the in-
creased preterm delivery rate among Mexican Americans who have lived in the United States for most or all of their lives.
Connections Among Host, Environments, and Agents. Host factors, environmental stressors, and acute stress triggers that have been hypothesized to increase women’s risk of delivering prematurely do not operate in isolation, but are often interconnected. For example, the environmental risks of poverty and racism may operate through increasing host susceptibility by lowering self-esteem, lessening a sense of personal control, and increasing fatalistic views. To put together the many ways in which stressors and stress mediators may interact will require a unified theory of stress that includes not only various types of stress contagion but also the causal pathways for contextual stressors and other stressors. There is great need for better theoretical work in this area, coupled with a need to link epidemiologic theory development with biophysical markers of stress in studies of health outcomes.
CURRENT APPROACHES TO REPRODUCTIVE AND DEVELOPMENTAL TOXICITY TESTING AND RISK ASSESSMENT
Carole A. Kimmel
Approaches for reproductive and developmental toxicity testing of environmental chemicals as well as pharmaceutical agents include evaluations of fertility, pregnancy maintenance, parturition, lactation and survival, growth, and development of offspring. Rats and rabbits are the most frequently used animal models. Gestation length can easily be determined because the time of mating and the day of parturition are specifically recorded.
Testing protocols for reproductive and developmental toxicity evaluation of pesticides and industrial chemicals include evaluations of gestation length and of survival, growth, and development of offspring. Recent changes in the prenatal developmental toxicity testing protocol further increase the possibility of detecting alterations in gestation length. Treatment of pregnant animals was extended from the end of organogenesis to the day before termination in both rats (from gd 6–15 to gd 6–20) and rabbits (gd 6–19 to gd 6–28 or 29). Thus, effects of chemicals in late gestation on litter size, survival, growth, and development of the fetus, as well as on the occurrence of resorptions, abortions, or premature delivery in the females, would more likely be detected. When a developmental neurotoxicity study is conducted, typically in rats, the testing protocol includes exposure from early pregnancy throughout gestation and for part or all of the lactation period. Thus, animals are exposed until the end of gestation and gestation length is recorded, along with litter size, survival, and growth of the offspring. For the reproduction and fertility effects testing protocol, rats are typically used and are exposed from before mating and throughout gestation and
lactation. F1 pups are continued on the same exposure level into adulthood, and then are mated, with the same level of exposure continuing throughout their pregnancy and lactation periods. Length of gestation is recorded for both parental and F1 females, as are litter size, survival, growth, and development of the offspring.
Similar protocols are used by the Food and Drug Administration (FDA) for testing food additives and contaminants for reproductive and developmental toxicity. The standard testing protocols for pharmaceutical agents are designed to cover all stages of development, from the premating period through gestation and lactation, although long-term multigeneration studies are not typically conducted. Recently, pediatric evaluation of pharmaceuticals has gained a greater emphasis within the FDA. Testing guidelines have not been established as yet, but guidance on areas of concern is available, and includes the potential toxicity of drugs in premature infants.
The Environmental Protection Agency (EPA) risk assessment guidelines for developmental toxicity (U.S. EPA, 1991) and reproductive toxicity (U.S. EPA, 1996) discuss gestation length, premature delivery, and growth retardation of offspring and their consideration in hazard characterization as important endpoints of reproductive and developmental toxicity. Significant shortening of gestation can lead to adverse outcomes of pregnancy such as decreased birth weight and offspring survival. Several examples exist in the literature of agents that cause premature delivery in commonly used laboratory animal species, including mice, rats, and rabbits. Alterations in gestation length appear to be effects that occur at higher dose levels, with effects on offspring survival, birth weight, or other measures of growth and development as more sensitive indicators of reproductive and developmental toxicity.
ASSESSMENT AND RELEVANCE OF ENVIRONMENTAL CHEMICAL EFFECTS ON UTERINE MUSCLE
Although regulation of uterine contractility is fundamental for successful pregnancy, relatively little attention has been given to environmental chemical effects on the uterine muscle. The bulk of the uterine wall is comprised of the myometrial smooth muscle layer. During pregnancy, the myometrium increases substantially in size to accommodate the growing fetus and to prepare for the task of childbirth. With completion of the gestational term, the myometrium generates the forceful, oscillatory, repetitive, and coordinated contractions over a sustained duration that are necessary for successful parturition. Preterm development of such contractions by environmental chemicals could promote premature birth in the absence of successful intervention.
As in all muscle, sufficient elevations of myometrial intracellular calcium concentration initiate cell contraction. Gap junctions in the myometrium increase at parturition to form a communication network of intercellular channels that promote the coordination of contractions of individual myometrial cells. Although much diversity exists among species regarding the fetal and maternal signals that initiate parturition, mechanisms of parturition appear to converge at the level of the myometrium. Consequently, my laboratory has used in vitro experimental systems of myometrial cells and tissues of laboratory animal species to study direct toxicant actions in uterine muscle.
Uterine contractility responses are assessed by suspending in standard muscle baths uterine strips cut along the longitudinal axis of uteri from pregnant rats, and monitoring changes in the force and frequency of spontaneous oscillatory contractions in response to the test chemical. Using this approach, polychlorinated biphenyl (PCB) mixtures, lindane (γ-hexachlorocyclohexane), β-hexachlorocyclohexane, and various 1,1,1-trichloro-2,2-bis (p-chlorophenyl) ethane DDT and PCB isomers were shown to exert rapid and direct actions on uterine contractions. By using biochemical, fluorometric, and pharmacological approaches in myometrial cell culture and tissue contractility experiments, elevation of intracellular calcium concentration via activation of voltage-operating calcium channels were linked to the stimulatory activities of Aroclor 1242. These results suggest that some environmental chemicals may directly stimulate uterine contraction by activating calcium-dependent mechanisms. In addition, prolonged exposure to the estrogenic PCB 4-hydroxy-2′,4′,6′-trichlorobiphenyl increased oxytocin-induced oscillatory uterine contraction frequency in an estrogen receptor-dependent manner, showing the potential for estrogenic environmental chemicals to stimulate uterine contractions via indirect mechanisms involving oxytocin.
Conversely, the ability of acute exposures of lindane and 4-hydroxy-2′,4′,6′-trichlorobiphenyl to abolish spontaneous oscillatory uterine contractions was associated with inhibition of myometrial gap junction-mediated intercellular communication. In the case of lindane, the inhibition of myometrial gap junction communication was via an oxidative stress-mediated mechanism. Whether activation of gap junction communication and alteration of cell redox are mechanisms of toxicant-induced stimulation remains to be investigated.
These laboratory studies provide insight into mechanisms by which environmental chemicals could stimulate preterm birth. In general, experimental toxicology best contributes to our understanding of risks to parturition from environmental chemicals when interpreted in conjunction with epidemiology findings, providing biological plausibility for epidemiological associations of exposures and outcomes as well as information on chemicals of concern for future epidemiology investigations.
EXPOSURES TO ENVIRONMENTAL AGENTS AND PRETERM DELIVERY
Matthew P. Longnecker
Data regarding occupational and environmental agents in relation to the risk of preterm delivery in humans were identified in the MEDLINE database using the PubMed search engine. The reference lists of articles identified were searched for additional reports. For several exposures the data suggested that environmental factors contribute to the risk of preterm delivery.
In three studies of air pollution effects, including one from the United States, particulate matter (either total suspended particles or particulate matter less than 10 microns in diameter) was associated with a modest increase in risk of preterm delivery; suggestive data also implicated sulfur dioxide. All three studies had ecologic–time series type designs, employed a limited number of air monitoring stations, and had little personal-level data.
In a recent report about a United States population first studied in the 1960s, maternal pregnancy serum level of the 1,1,1-trichloro-2,2-bis (p-chlorophenyl) ethane DDT metabolite DDE was associated with increased risk of preterm birth. The association was also seen in some earlier, smaller studies, but the exposure levels were lower. Whether the association is causal has not been resolved, although if it were causal, the effects are probably evident at levels of exposure seen only in countries where DDT has been used recently.
Potential risk factors for preterm birth identified in more than one occupational study were some maternal occupation (such as metal, electrical, janitorial, food service, textiles); maternal occupational exposures (e.g., solvents); some paternal occupation (such as food service, textiles); and paternal occupational exposures (pesticides, e.g., atrazine in one study). As in other occupational studies of preterm birth, only very general information about exposure was available.
Data from a recent ecologic study suggest that high levels of arsenic in drinking water increase preterm birth. The association with arsenic was in a population with water arsenic levels that were about two times higher than those in high-exposed areas of the United States (e.g., selected counties in Utah).
Risk factors that have been studied and appear not to be associated with increased risk of preterm birth were chlorinated water disinfection by-products and use of video display terminals. Data regarding maternal lead exposure were inconsistent, though findings in the largest prospective studies that adjusted for multiple confounders do not support a relation.
While many of the potential factors studied so far merit further investigation, additional factors for which data were considered inadequate were cadmium, paternal occupational lead exposure, polychlorinated biphenyls, and various occupational chemical exposures among mothers and fathers.
None of the factors presented are established as related to risk of preterm
delivery. Strengths and weakness of the review strategy will be discussed as will priorities for future research directions.
FETAL SIZE AND PRETERM BIRTH
Stephen J. Lye
The onset of labor requires both the activation and the stimulation of the myometrium to generate the intense and coordinated contractions needed to bring about the delivery of the neonate. Activation involves a switch in the contractile state of the myometrium from a state of inactivity to one in which the muscle develops increased excitability, increased responsiveness to uterotonic agonists, and enhanced intercellular communication. Once activation is induced, stimulation is achieved through the increased production of agonists such as stimulatory prostaglandins and oxytocin. We have predicted and have gained evidence to suggest that the process of myometrial activation results from the synchronous increased expression of a cassette of genes encoding “contraction-associated proteins,” or CAPs, including ion channels (which regulate the resting membrane potential of myocytes and hence their excitability), agonist receptors (to enhance myometrial responsiveness to stimulatory hormones), and gap junctions (to provide for enhanced cell–cell communication and therefore synchronization of uterine contractions). Clearly, knowledge of the mechanisms that control the process of myometrial activation is central to the development of strategies to prevent preterm labor.
We present evidence that myometrial activation and the onset of labor is ultimately controlled by the fetal genome through two pathways—one involving fetal endocrine signals and the other mediated by mechanical signals induced by stretch of the uterine wall by the growing fetus.
The contribution of fetal endocrine signals in the onset of labor was first demonstrated by Liggins some 30 years ago. This pathway involves activation of the fetal hypothalamic–pituitary–adrenal axis leading to increased synthesis and release of cortisol from the fetal adrenal gland. Cortisol in turn induces the expression of PGHS-2 in the fetal placental, which leads to the production of PGE2. As PGE2 levels rise there is an induction of P450C17 activity, which in most species ultimately leads to the metabolism of progesterone through to estrogen. We have shown that estrogen positively regulates the expression of CAP genes, while progesterone suppresses their expression. The mechanisms by which these steroids act remains to be determined although there is evidence to suggest that transcription factors of the AP-1 family contribute to the activation of CAP genes. In contrast to virtually all other species, progesterone levels do not fall at term in humans; nevertheless there is evidence to suggest that progesterone is required for the maintenance of pregnancy in women. It is likely that some mechanism exists in humans to block progesterone signaling at term and there-
fore induce a functional withdrawal of progesterone, permitting the activation of CAP genes, myometrial activation, and the onset of labor.
Although this endocrine pathway appears to be necessary, data we obtained several years ago suggested that this pathway is not, in itself, sufficient for the initiation of labor. In studies in unilaterally pregnant rats we found that even though the endocrine changes at term occurred normally, CAP gene expression increased only in the gravid horn and not the empty horn. However, if a plastic tube was placed in the lumen of the empty horn to induce stretch, the increase in CAP gene expression occurred to the same extent as in the gravid horn. These data suggested that both endocrine and mechanical signals (as a result of fetal growth) are required for the initiation of myometrial activation and the onset of labor. This led to the question of how pregnancy could be maintained to term if fetal growth led to increased CAP gene expression. As a result of a series of studies we now believe that during pregnancy, tension on the uterine wall leads to growth of the uterine wall, which in turn reduces the tension and thus prevents premature activation of CAP gene expression. This stretch-induced growth of the myometrium results from a hypertrophy of the myocytes and requires the presence of progesterone. The mechanisms by which myometrial cells sense that they are under tension and are able to transduce this into a signal that leads either to genes controlling myocyte hypertrophy or to myometrial activation remain to be determined, though our evidence suggests an involvement of the focal adhesion signaling through mitogen-activated protein kinsase (MAPK) and increased AP-1 protein expression.
Although there are many pathways by which preterm labor might be induced, increased myometrial contractility is likely common to all of them. We suggest that, at least for idiopathic preterm labor, premature activation of the endocrine and/or mechanical pathways contributes to preterm birth. It is well established that fetal hypoxia (such as might be induce by placental insufficiency) can lead to increased activity of the fetal HPA axis leading to increased cortisol synthesis. Initially negative feedback pathways would block further elevations in this steroid, although chronically elevated cortisol might negatively impact fetal growth. Continued hypoxic stimulation would eventually lead to cortisol induction of placental endocrine changes, myometrial activation, and the initiation of preterm labor with the birth of a baby that is small for its gestational age.
Multifetal pregnancies are known to be at increased risk of preterm birth. We believe that our data on stretch-induced increase in myometrial CAP gene expression offers a possible explanation for this observation. While the increased tension in the uterine wall would lead to increased myometrial growth there may be some limits to this capacity. At the very least, the stress placed on the mechanical pathway would likely make such pregnancies more susceptible to preterm birth from other causes. This model also suggests that large babies or pregnancies complicated by increased intrauterine volume (e.g., polyhydramnios)
might be at risk of preterm birth through premature activation of mechanical signals.
Recent clinical data from the group of Dr. Robert Gagnon (University of Western Ontario, Canada) supports a role for both the endocrine and the mechanical pathways in the onset of preterm birth. These data showed a twofold increase in the incidence of preterm birth in babies who were larger (>97 percentile) or smaller (<3rd percentile) than those who were appropriately grown.
SPECIES DIFFERENTIATION AND ANIMAL MODELS OF PARTURITION
Peter W. Nathanielsz
Here is a working definition of parturition: Parturition is a multifactorial process that involves fetal, placental, and maternal mechanisms. Parturition involves the recruitment of interactive positive feed-forward loops and the removal of pregnancy maintenance mechanisms. Changes in these stimulatory and inhibitory mechanisms exhibit critical tissue-specific time relationships to each other.
There are four central features in this definition.
is a multifactorial process,
recruits interactive positive feedforward loops,
involves the removal of pregnancy maintenance mechanisms, and
mechanisms have critical time-dependent tissue-specific interrelationships.
The three indispensable processes involved in normal labor and delivery are
A switch in myometrial contractility pattern from contractures to contractions,
Rupture of the fetal membranes, and
Dilation of the cervix.
The Fetal Role in the Determination of the Duration of Pregnancy. In sheep, interruption of the fetal hypothalamic–pituitary–adrenal axis at any level will prolong pregnancy (Challis et al., 1994; Nathanielsz, 1996). Infusion of adrenocorticotropic hormone (ACTH) or cortisol to the ovine fetus will shorten the length of pregnancy. Taken together, these studies have been interpreted as indicating that the fetus determines the length of ovine pregnancy. However, while demonstrating that the fetus is involved in determining the duration of
pregnancy, each of these studies shows only that these fetal endocrine tissues are involved in the promotion of parturition.
Since parturition is a multifactorial system, it is difficult to determine precisely what is the initiator of the cascade that leads to delivery. In one very real sense, the signal for parturition is fertilization. Following fertilization, a sequence of genetically programmed events is set in train. This sequence unfolds in a manner and at a speed determined by both the genome and the environment, nature and nurture. As with most biological processes, though not all, the duration of pregnancy is gene influenced, not gene determined. In closely inbred groups of animals maintained under carefully regulated nutritional and environmental conditions, the duration of pregnancy is very homogeneous. This suggests that in such precisely defined environments, the fetal developmental steps that regulate the system are very precisely regulated by the genome. However, external stresses, nutritional challenges to the fetus, and other conditions can alter the duration of pregnancy by speeding up or interfering with one or more of the normal mechanisms.
A very pronounced example of this is the prolonged pregnancy induced when pregnant sheep ingest the corn lily on the fourtheenth day of pregnancy and expose the developing fetus to the alkaloid toxin 2-deoxyjervine, resulting in a cyclopian deformity and distorted brain development. Another good example is human anencephaly. There is a high incidence of polyhydramnios in anencephaly. When polyhydramnios occurs, the uterine muscle is stretched and premature labor is common. However, when anencephaly is not accompanied by polyhydramnios, human pregnancy is prolonged. These observations suggest some similarity between the neuroendocrine involvement in parturition in the sheep described below and the mechanisms that are critical to labor and delivery in human pregnancy (Nathanielsz, 1996).
Because of the inability to perform carefully controlled, invasive studies in human pregnancy, it is necessary to conduct studies in animal models, especially nonhuman primates. This presentation compares the evidence of fetal involvement in rhesus monkey pregnancy and sheep pregnancy.
Evidence for increased fetal adrenal function in late gestation. There is a remarkable similarity in the rise of fetal cortisol in the sheep and fetal androgen in the fetal monkey over the final 30 percent of gestation (Challis et al., 1994; Nathanielsz, 1996). This and other observations have led us to hypothesize that the similarity between nonhuman and sheep pregnancy lies in the rise in estrogen production prior to labor.
In sheep, fetal cortisol stimulates placental conversion of progesterone to estrogen (Anderson et al., 1975). In primates the placenta is an incomplete steroidogenic organ and estrogen synthesis by the placenta has an obligate need for androgen precursor (Novy et al., 1981).
Evidence for increased estrogen production in late gestation. Longitudi-
nal measurement of maternal estrogens immediately before spontaneous term labor shows an increase in all primate species studied in a detailed fashion.
Patterns of myometrial activity that occur throughout gestation. In all mammalian species studied to date, myometrial activity throughout pregnancy is of the contractures type, exemplified by long-lasting, low-frequency epochs of activity that have a very different temporal and amplitude pattern from contractions (Nathanielsz, 1996). At labor and delivery, contractures must switch to contractions to produce efficient delivery of the fetus. In the sheep this switch occurs once, generally at nighttime, and the ewe proceeds to delivery. In the monkey this switch occurs and augments for a few hours each night until delivery occurs after several switches (Nathanielsz, 1996).
Initiation of premature labor in rhesus monkeys by stimulation of estrogen production. 1. Estrogen as the central mediator of parturition mechanisms: In a series of studies in the pregnant rhesus monkey we observed that situations which resulted in elevated maternal estrogen concentrations in late pregnancy were generally accompanied by a switch in myometrial activity from the contractures to the contractions mode. These include following laparotomy and fetal catheterization surgery (Nathanielsz et al., 1984), during hypoglycemia induced by food withdrawal (Binienda et al., 1988), predelivery (Taylor et al., 1983), and following androstenedione administration to the pregnant monkey (Figueroa et al., 1989; Mecenas et al., 1996; Nathanielsz et al., 1998). Novy and colleagues were unable to induce premature delivery in the pregnant rhesus monkey by the administration of estrogen (Novy et al., 1983). We therefore hypothesized that under normal circumstances, estrogen exerts both paracrine effects at its site of generation from androgens and classical endocrine effects. Thus, we carried out androgen infusion to ensure that estrogen was produced at its normal locus of production.
2. Infusion of androgen into the pregnant rhesus monkey to elevate estrogen biosynthesis. Continuous infusion of androstenedione into the pregnant rhesus monkey at 0.8 of gestation produces labor associated with the three indispensable processes described above: a switch of myometrial contractures to contractions, rupture of the fetal membranes, dilation of the cervix, and delivery of live young (Nathanielsz et al., 1998).
It is interesting to note that although the infusion of androgen was continuous and the maternal plasma estrogen concentration was elevated throughout the 24-hour day, the switch from contractures to contractions occurred only around the hours of darkness. This rhythm in the switch is likely to be driven by oxytocin since the switch can be abolished by oxytocin antagonists such as Atosiban.
3. Inhibition of the effects of androstenedione by aromatase inhibitors. The proof that local estrogen production is key to labor and delivery is furnished by studies in which the promotion of labor by infusion of androstenedione is inhibited by aromatase inhibitors (Nathanielsz et al., 1998).
4. Failure of estradiol infusions to produce premature delivery in the rhesus monkey: Systemic infusion of estrogen does not precipitate labor and delivery in rhesus monkeys. This finding further supports a role for local paracrine actions of estrogen (Novy et al., 1983).
The Mother Determines the Precise Time That Labor Begins. We have demonstrated that the switch from contractures to contractions is regulated by maternal oxytocin acting on a myometrium that has been prepared by the rise in maternal estrogens described above. Firstly, the switch can be initiated prematurely by infusing androgen into the pregnant rhesus monkey to raise maternal estrogen. In this experimental paradigm, the switch is prevented if aromatase inhibitors are administered with the androgen to inhibit estrogen production (Nathanielsz et al., 1998).
Oxytocin antagonists will inhibit both the normal term switch from contractures to contractions and the switch that is produced prematurely by androgen infusion. Maternal plasma oxytocin concentrations in late gestation show a pronounced 24-hour rhythm with a peak in the early hours of the evening, coincident with the switch from contractures to contractions. A similar rhythm has been shown in pregnant women in late gestation. It is now clear that the failure to demonstrate a rise in oxytocin in late human pregnancy in previous studies can be attributed to a failure to obtain blood samples at the correct time of day.
The Role of Prostaglandin Synthesis, Degradation, and Receptor Activation. Prostaglandins have been shown by many investigators to be essential for the normal completion of parturition. Various animal models have been used to highlight the changing role of prostaglandins. Since these regulatory compounds work at the paracrine and autocrine level, it is essential to compare effects both in the whole animal and in vitro. Our recent studies have dealt with PG synthesis in the pregnant sheep and baboon and with changes in the various PG receptors in different intrauterine tissues throughout the last third of gestation and in labor in the sheep and baboon (Smith et al., 1998, 2001a, 2001b, 2001c).
TOXIC SOCIAL ENVIRONMENT: A FACTOR IN PRETERM BIRTH?
Two of the largest and most intractable risk factors for preterm birth are maternal social class and race or ethnicity. Attempts to “unpack” class and race into specific risk components have failed to explain more than a fraction of their predictive power. New approaches to the conceptualization and measurement of psychologic and social stressors should yield more insight into the elevated risk of preterm birth that accompanies poverty and minority status in the United States. Approaches to measuring maternal stressors and stress responses during
pregnancy are being refined. Previously neglected stressors, such as maternal experiences of violence and racism, are being addressed in new studies. The measurement of women’s social position is being refined. New models of the interaction between psychosocial stress and physiologic pathways to prematurity are being tested, including potential associations of maternal stress with CRH levels and risk of infection. Finally, there is growing acknowledgment that chronic stress before pregnancy may be at least as important to pregnancy outcome as stressors occurring during pregnancy. As the science of measuring psychosocial predictors of health grows more sophisticated, we may find new explanations for the long-standing social and ethnic gaps in risk of preterm birth in the United States.
CAUSES AND MECHANISMS OF PREMATURE LABOR
James M. Roberts
This presentation reviews the currently popular theories of preterm labor to provide insight as to where toxicants could act to lead to preterm birth. In addition, it considers the fact that almost one-fifth of preterm births are iatrogenic, usually secondary to indicated delivery for preeclampsia, making preeclampsia another possible target for toxicants to increase preterm birth.
Currently, two theories hold most interest for the genesis of preterm birth. The inflammatory theory of preterm birth has received most attention. Another hypothesis that has received less attention is the relationship of preterm birth to abnormal implantation. One other hypothesis considered here, which had long guided thinking about preterm birth, is that preterm labor is merely term labor occurring early, essentially an abnormality of the biological clock.
The hypothesis that inflammation might contribute to preterm birth was originally stimulated by the concept that infection, more specifically subclinical infection, might lead to preterm birth. The association of bacterial colonization of fetal membranes with early spontaneous preterm birth, but not induced preterm birth, provided strong evidence for this hypothesis. The association of intraamniotic bacteria and inevitable delivery provided further support. The role of bacteria was biologically plausible since these organisms have the capacity to activate phospholipase and generate prostaglandins implicated in uterine contractility. It soon became evident that increased cytokines in amniotic fluid, even without microorganisms, were an excellent predictor of preterm birth. This raised the possibility that these cytokines, in addition to (or rather than) microorganisms, might be responsible for the activation of labor mechanisms. The concept that cytokines could be the important mediators raised the possibility that causes of inflammation, other than infection, could be involved in the genesis of preterm birth. The relationship of distal inflammatory stimuli such as periodontal infections was made plausible by this mechanism. In addition, the relationship of
preterm birth to maternal stress was suggested to be secondary to activation of the inflammatory response by CRH, for example. Cytokines could also be released by immune interactions between mother and infant. Furthermore, the recognition that even normal pregnancy is associated with a marked activation of the inflammatory response suggests that some women with preterm birth may simply be at the wrong end of the normal distribution of pregnancy-induced inflammatory responses. Environmental toxicants could also augment this inflammatory response.
Preterm birth is also associated with abnormal implantation. This is indicated by studies of placental bed biopsies, and placental pathology and by the increased frequency of growth restriction in infants with spontaneous preterm birth. This suggests similarities with other abnormal implantation disorders such as habitual abortion and preeclampsia. Although neither relationship has been extensively explored, we have found that as with preterm birth, early-onset preeclampsia (indicated delivery prior to 34 weeks) is associated with an increase of the acute-phase reactant feritin at 16 to 19 weeks’ gestation. In addition, women who have had early-onset preeclampsia in their first pregnancy deliver in their subsequent “normal” pregnancy on average one week earlier than women with a normal first pregnancy. Information on control of human implantation is increasing rapidly and suggests targets for toxicants that lead to preterm birth.
Preterm birth is recurrent, and increased risk is inherited and varies with race. Although part of this increased risk can be environmental it is also likely that this is genetically influenced. In years past the impact of genetics was used to support the possibility that in some individuals the biological clock determining the onset of labor was abnormal. Much effort was directed at understanding term birth with the hope that this information would extrapolate to preterm birth. Distal mechanisms are undoubtedly similar and the understanding of term labor is pertinent to these. However, current thinking has largely abandoned the concept of an abnormal biological clock mechanism in favor of distinct factors stimulating term and preterm birth. Our level of understanding the mechanisms of preterm labor, however, does not justify abandoning this (or any) concept. The genetic contribution to preterm birth could also be contributions to abnormal implantation or inflammation. Especially pertinent to this workshop, interaction of environmental toxicants and genetically altered metabolism could contribute to preterm birth through any of the suggested mechanisms.
Preeclampsia accounts for 15 percent of preterm births. As a pregnancy-specific disorder that puts maternal and infant well-being in jeopardy, delivery is the only known effective treatment. In about 10 precent of cases this will necessitate preterm delivery. Preventing preeclampsia would prevent these preterm births. Mechanistic studies of preeclampsia have made considerable progress in the last 10 years. Large clinical trials testing one of the postulated pathogenic mechanisms, oxidative stress, are about to begin. Preeclampsia is also associated with exposure to environmental toxins. Women working in occupations with
increased exposure to organic solvents have a fourfold increased risk of the disorder. Anecdotally, in the portion of Mexico City with the most severe air pollution, preeclampsia–eclampsia accounts for 80 percent of maternal deaths.
Preterm birth is the major pregnancy problem in developed countries. There has been little progress in reducing the frequency of the disorder. It is quite probable that all preterm labor is not from a single cause and that all of the currently suggested precursors of preterm birth (and likely others) are important in different women and may be interactive. Understanding how environmental exposures may act upon these mechanisms could provide useful insights for prevention.
MOLECULAR MECHANISMS AND CELLULAR SIGNALING PATHWAYS ASSOCIATED WITH PARTURITION
Barbara M. Sanborn
Prior to parturition, there is a transition period during which a series of coordinated events prepare the uterine smooth muscle (myometrium) to respond to contractant signals and to be less responsive to relaxant signals. These events include alterations in the concentration of the many proteins that constitute components of the signaling pathways involved. During parturition, maternal and fetal compartments express the signaling pathways and produce the molecules that stimulate the myometrium to contract. The regulation of myometrial contraction is of paramount importance for the maintenance of pregnancy and for parturition. Understanding this regulation involves delineating the pathways that control contraction and relaxation and defining their interaction.
In myometrium, an increase in intracellular calcium (Ca2+i) favors contraction. The major contractant pathways, including those stimulated by oxytocin, target activation of phospholipase C (PLC), resulting in release of intracellular calcium from intracellular stores by inositol 1,4,5-triphosphate(IP3) (Sanborn, 2001; Sanborn et al., 1998). Signaling pathways that activate PLC can also stimulate calcium entry through calcium release-activated channels, either directly or indirectly. The introductory calcium level (Monga et al., 1999) Ca2+i is lowered by the actions of plasma membrane and sarcoplasmic reticulum calcium pumps.
Agents that stimulate cyclic adenosine 5-monophosphate (cAMP) inhibit myometrial contractile activity, contractant-stimulated phosphatidylinositide turnover, and increases in Ca2+i (Sanborn, 2001; Sanborn et al., 1998). A major point of cAMP inhibition is at the level of PLC. Although PLCβ1 is not a substrate for cAMP dependent protein kinase (PKA), PLCβ3 is phosphorylated by this enzyme, as well as by cGMP-dependent protein kinase and protein kinase C (Yue et al., 1998, 2000; Xia et al., 2001). Myometrial plasma membranes possess A-kinase-associated-proteins (AKAPs), that serve to localize PKA there (Dodge et al., 1999a). The PKA that inhibits PLC in myometrial plasma mem-
brane appears to be associated with AKAPs, both in human cells and in pregnant rat tissue. The ability of activated PKA to inhibit the phosphatidylinositide turnover pathway is markedly diminished at term in the rat. This change is accompanied by a loss in PKA both from the plasma membrane and associated with AKAP150 (Dodge et al., 1999b). These changes are not accompanied by a change in AKAP expression in the membrane. Rather, there is a change in the ratio of PKA and protein phosphatase 2B (PP2B) associated with the plasma membrane and with AKAP150.
These data point to an important new mechanism, namely a change in the localization of AKAP-associated PKA in plasma membrane that may serve as the final regulatory checkpoint on the inhibitory action of cAMP–PKA on contractant-stimulated PLC activity. Recent evidence suggests that this change is correlated with the timing of parturition in the rat and that hormones play a role. Understanding the control of the scaffolding mechanism and its relevance in primates will be critical to understanding the actions of the agents utilizing this pathway to promote uterine relaxation. Such mechanisms may play a critical role in controlling the transition from myometrial relaxation to contraction.
Note this work was supported by HD09618 and HD38970.
BEHAVIOR, NUTRITION, INFECTION, AND STRESS: EPIDEMIOLOGIC CLUES TO THE STUDY OF THE ENVIRONMENT AND PRETERM BIRTH
David A. Savitz
To help focus the study of environmental agents that might influence the risk of preterm birth, the extensive literature on nonenvironmental factors warrants scrutiny. Epidemiologic research on a range of behaviors (tobacco use, cocaine use, physical activity), nutrition (iron, folate), infection (bacterial vaginosis), and psychological stress has been conducted for some time and offers insights into methodologic challenges, suggests possible etiologic pathways, and identifies potential confounding factors that must be considered. At present, strong predictors of preterm birth are limited to multiple gestation, prior preterm birth, and African-American ethnicity; weaker but modifiable influences include infection, tobacco use, low prepregnancy weight, lower socioeconomic status, and other prior adverse pregnancy outcomes. Despite extensive research, the roles of nutrition, cocaine use, physical exertion, and psychological stress remain unresolved. Nevertheless, these efforts raise questions and suggest approaches relevant to the study of environmental contributors.
1. The diversity of identified predictors of preterm birth—namely, multiple gestations, prior preterm birth, tobacco use, bacterial vaginosis, low prepregnancy weight, African-American ethnicity, and economic deprivation—suggests a
multiplicity of pathways, each making small, probabilistic contributions to the same endpoint. The relevant model seems to be one of multiple contributing factors operating through diverse mechanisms rather than a single cause or even a single pathway. The exploration of environmental contributors may expand the array of contributing factors, but there is no expectation that “the cause” of preterm birth will be found or isolated in the environment.
2. Influences on preterm birth have been more difficult to identify than causes of reduced birth weight. The predictors of reduced birth weight are often shared with predictors of preterm birth, but often have weaker associations. Such predictors of preterm birth as socioeconomic status and tobacco use seem to have strong influences on birth weight but only modest influences on preterm birth, indicating that the pathways probably overlap to some extent but are not identical. Direct examination of preterm birth, low birth weight, and small-for-gestational-age deliveries suggests modest overlap, such that synthesizing the literature requires considering research subsets defined by the pregnancy outcome under examination. One reason for the more limited success in the study of preterm birth may be the markedly greater uncertainty in measurement of duration of gestation as compared to birth weight, reflected in greater magnitude of error in vital records, lower quality of maternal reports of gestational age as compared to birth weight, and the more limited availability of gestational age from less developed countries. Duration of gestation is usually based on the last menstrual period, a fallible marker of the time of ovulation, although widespread use of ultrasound for dating has been helpful in increasing accuracy in many settings, a benefit to those studies that can incorporate this information over those based solely on last menstrual period.
3. Isolating specific causal agents from nonspecific influences of a healthful life-style and favorable socioeconomic conditions has been very difficult. As for many health endpoints, those who have a range of favorable attributes have more favorable outcomes. However, the lack of specificity in what aspects of “favorable circumstances” are responsible has been very challenging to researchers addressing preterm birth. For example, there are consistent findings of an association between cocaine use and preterm birth, reduced risk associated with leisure time physical activity during pregnancy, reduced risk with favorable nutritional status and use of prenatal vitamins, and increased risk associated with physically demanding occupations, yet none are necessarily causal, and isolating a true etiologic effect from a spurious association due to other unmeasured or unknown factors has been unsuccessful thus far. Isolating environmental agents from the circumstances that give rise to exposure will pose a serious challenge for the many environmental agents associated with socioeconomic deprivation and less favorable life-styles.
4. Preterm birth is clearly a heterogeneous entity, with distinctive contributing pathways, yet the most useful approach to subdividing the outcome for identification of etiologic factors is unclear. Many studies separate spontaneous from
medically indicated preterm delivery, a reasonable strategy, yet the common indications for medical intervention and early delivery (fetal growth restriction, hypertension) are also independent risk factors for spontaneous preterm birth. Some risk factors may be shared across spontaneous and indicated preterm births, whereas others are likely to differ. Other divisions of preterm birth are based on clinical presentation (idiopathic preterm labor, preterm premature rupture of membranes) or on severity of prematurity defined by duration of gestation. More novel approaches to categorization consider the underlying etiologic process (e.g., infection or inflammation, vascular compromise). There are considerable logistical and conceptual challenges to refining the endpoint for epidemiologic study.
5. Biological markers of exposure have much to offer since true prospective studies beginning early in pregnancy allow for assessment in the etiologically relevant time interval. Biological markers of nutritional status (serum folate, ferritin, transferrin receptor saturation), stress (corticotropin-releasing hormone, cortisol), tobacco use (urinary and serum cotinine), cocaine use (cocaine and benzoylecognine in hair and urine), and infection or inflammation (fetal fibronectin, cytokines) have all been incorporated into epidemiologic studies of preterm birth to great benefit. Participation in prenatal care offers unusually favorable opportunities to incorporate collection of biospecimens in large clinically based populations as opposed to community samples, with clear applicability to the study of environmental agents.
The study of environmental agents in relation to preterm birth is at a very early stage of development, with far less interest in the past than in potential environmental contributors to pregnancy loss or male infertility, for example. It is important as these research avenues move forward to ensure that sophisticated environmental approaches are combined with rigorous evaluation of the reproductive health endpoints and potential confounding factors, generally requiring a multidisciplinary research team.
GENE–ENVIRONMENT INTERACTIONS AND PRETERM DELIVERY
Although the causes of preterm delivery remain unclear, preterm delivery appears to be a highly complex entity determined by multiple environmental and genetic factors, as well as gene–environment interactions. Most previous studies have focused on socioenvironmental or clinical variables. The role of genetic susceptibility and gene–environment interactions in relation to preterm delivery is largely unexplored. This presentation provided a brief overview of gene–environment interaction. Then, it summarizes findings of gene–environment interac-
tions on preterm delivery from our ongoing molecular epidemiologic studies of preterm delivery in both Chinese and U.S. populations, specifically, interactions between maternal metabolic genes and benzene exposure and interactions between maternal metabolic genes and cigarette smoking. In addition, important methodological issues in this research field were discussed (see chapter 4 of this summary for more details). Our data have shown a consistent evidence of gene-environment interactions in diverse populations. However, more studies are needed and multidisciplinary collaborations are required in order to jointly and comprehensively assess the role of environmental factors, genetic factors, and gene–environment interactions in preterm delivery among populations with marked differences in social status and environmental exposures.
ROLE OF NITRIC OXIDE IN UTERINE ACTIVITY AND PREMATURE PARTURITION
Preterm labor and delivery remain an important problem in obstetrics, with prematurity contributing to 8–10 percent of neonatal deaths and responsible for 60–70 percent of neonatal morbidity in the United States. There is no effective treatment for preterm labor. Most commonly used tocolytics, β2-adrenergic agonists such as ritodrine and terbutaline, do not prolong pregnancy or improve neonatal outcome and have serious side effects. Nitric oxide has been studied in the last eight years as a potential tocolytic. This presentation examines the studies on (1) synthesis of NO and expression of NO synthase enzymes in the uterus of various species; (2) effects of NO on uterine relaxation; (3) regulation of NO synthesis and effects of NO on the uterine relaxation during pregnancy and labor; (4) effects of NO donors on preterm labor and possible prolongation of gestation; and (5) interaction of NO with other uterotonins in modulating uterine activity during pregnancy and preterm labor.
Nitric oxide is a simple, but highly reactive, endogenous chemical, hitherto known as an environmental toxicant. Since the first description of the therapeutic use of glyceryl trinitrate (GTN) appeared in 1879, organic nitrites have remained the cornerstone for the treatment of angina pectoris. GTN has been used in obstetrics as a uterine relaxant in cases of breech extraction. In 1993, we and others reported the presence of the NO system in the uterus of a variety of animals, suggesting that endogenous NO synthesized in the uterus could play a role in uterine quiescence during pregnancy. Morphological and biochemical studies of uterine samples demonstrated the presence of the NO system in the rat, rabbit, mouse, sheep, and human. Several studies demonstrated increases in NO synthesis in the uterus and in NO-induced uterine relaxation during pregnancy. Both NO synthesis and uterine sensitivity to NO are substantially reduced at term, indicating a role for NO in uterine quiescence during pregnancy and labor.
Similar changes have also been reported in the human, indicating that the NO system is a potentially important system for uterine relaxation during pregnancy.
Studies on the use of NO donors as tocolytics in the threatened preterm labor in women have been limited to very few. In 1994, transdermal patches of GTN were reported to prolong pregnancy for a mean of 59 days in a group of 13 women. In 1996, in another group of 10 women, pregnancy was prolonged for a mean of 46.2 days. The effects of GTN appeared to be due to reductions in uterine contractions in these preterm labor women. It is unknown from these studies whether NO donors are superior to other tocolytics; however, the side effects appear to be less severe.
It is unclear if the intact uterine NO system is critical for maintaining uterine quiescence during normal pregnancy. Studies in the rat and sheep indicate that inhibition of NO synthesis prior to term does not result in parturition; however, this does occur in mice. It is possible that NO may interact with other uterotonins and that the resultant effects may depend on these interactions. Moreover, uterine NO synthesis has been shown to be regulated by estradiol and progesterone, key hormones in pregnancy. Progesterone not only enhances NO synthase expression and NO synthesis in rat uterus, but also enhances NO-induced uterine relaxation. On the other hand, antiprogesterones decrease NO synthase enzymes, inhibit NO synthesis, and reduce uterine relaxation responsiveness to NO. Furthermore, inhibition of NO synthesis during pregnancy in the presence of a low dose of antiprogesterone leads to preterm labor in the rat. On the other hand, NO donors prevented prostaglandin F2α-induced preterm labor in the rat. These studies in the rat suggest that NO may interact with other uterotonins in the maintenance of uterine quiescence during pregnancy. In conclusion, uterine NO system may play a role in maintaining uterine quiescence during pregnancy, and perturbation of this system could facilitate preterm parturition.