Early risk factors for miscarriage: a prospective cohort study in pregnant women

Early risk factors for miscarriage: a prospective cohort study in pregnant women

RBMOnline - Vol 17 No 1. 2008 101-113 Reproductive BioMedicine Online; www.rbmonline.com/Article/3031 on web 27 May 2008 Article Early risk factors f...

524KB Sizes 4 Downloads 22 Views

RBMOnline - Vol 17 No 1. 2008 101-113 Reproductive BioMedicine Online; www.rbmonline.com/Article/3031 on web 27 May 2008

Article Early risk factors for miscarriage: a prospective cohort study in pregnant women Petra Arck is Professor of Psychoneuroimmunology at the Charité, University Medicine Berlin, Germany, and has been awarded with a Canada Research Chair in Neuroimmunology. She currently pursues her research interests at the Brain Body Institute, McMaster University Hamilton, Canada as well as at the Charité, in order to advance transcontinental research. She received a MD degree from the University of Tuebingen, Germany, and completed her doctoral thesis in 1994. Dr Arck has received a number of research awards acknowledging her interdisciplinary research focus. Her scientific interests address the effect of stress on the neuro-immunological hemostasis in the context of reproduction. The authors report no financial or commercial conflicts of interest Dr Petra Arck Petra C Arck1,11, Mirjam Rücke1, Matthias Rose2,10, Julia Szekeres-Bartho3, Alison J Douglas4, Maria Pritsch5, Sandra M Blois1, Maike K Pincus6, Nina Bärenstrauch1, Joachim W Dudenhausen7, Katrina Nakamura8, Sam Sheps9, Burghard F Klapp1 1 Centre of Internal Medicine and Dermatology, Division of Psycho-Neuro-Immunology, Charité, University Medicine Berlin, Germany; 2Health-Assessment Laboratory and Quality Metric, Waltham, MA, USA; 3Department of Medical Microbiology and Immunology, and Reproductive and Tumour Immunology Research Group of the Hungarian Academy of Sciences, Pecs University, Medical School, Pecs, Hungary; 4School of Biomedical Sciences, University of Edinburgh, Edinburgh, UK; 5Department of Medical Biometry, University of Heidelberg, Germany; 6Department of Pediatrics, Division of Pneumology and Immunology, Charité, University Medicine Berlin, Germany; 7Department of Obstetrics, Charité, University Medicine Berlin, Germany; 8Department of Family Practice, Interdisciplinary Studies, University of British Columbia, Canada; 9Department of Health Care and Epidemiology, Faculty of Medicine, University of British Columbia, Canada; 10Department of Psychosomatic Medicine and Psychotherapy, University Medical Center Hamburg-Eppendorf, Germany 11 Correspondence: Tel: +49 30 450 553873; Fax: +49 30 450 553962; e-mail: [email protected]

Abstract Many pregnancies are lost during early gestation, but clinicians still lack tools to recognize risk factors for miscarriage. Thus, the identification of risk factors for miscarriage during the first trimester in women with no obvious risk for a pregnancy loss was the aim of this prospective cohort trial. A total of 1098 women between gestation weeks 4 and 12 in whom no apparent signs of a threatened pregnancy could be diagnosed were recruited. Demographic, anamnestic, psychometric and biological data were documented at recruitment and pregnancy outcomes were registered subsequently. Among the cases with sufficiently available data, 809 successfully progressing pregnancies and 55 subsequent miscarriages were reported. In this cohort, risk of miscarriage was significantly increased in women at higher age (>33 years), lower body mass index (≤20 kg/ m2) and lower serum progesterone concentrations (≤12 ng/ml) prior to the onset of the miscarriage. Women with subsequent miscarriage also perceived higher levels of stress/demands (supported by higher concentrations of corticotrophin-releasing hormone) and revealed reduced concentrations of progesterone-induced blocking factor. These risk factors were even more pronounced in the subcohort of women (n = 335) recruited between gestation weeks 4 and 7. The identification of these risk factors and development of an interaction model of these factors, as introduced in this article, will help clinicians to recognize pregnant women who require extra monitoring and who might benefit from therapeutic interventions such as progestogen supplementation, especially during the first weeks of pregnancy, to prevent a miscarriage. Keywords: body mass index, corticotrophin-releasing hormone, miscarriage, progesterone, risk analysis, stress perception

Introduction Human reproduction is relatively inefficient, with fecundity, i.e. the highest probability of conception during one menstrual cycle, as low as 30%, and with a high incidence of pregnancy loss after initial blastocyst implantation. Some studies report

percentages of fetal loss in more than 50% of pregnancies (Wilcox et al., 1999; Red-Horse et al., 2004; Nepomnaschy, 2006). Genetic abnormalities of the fetus or pre-existing medical diseases, higher age, smoking or uterine malformations

© 2008 Published by Reproductive Healthcare Ltd, Duck End Farm, Dry Drayton, Cambridge CB3 8DB, UK

101

Article - Early risk factors for miscarriage - PC Arck et al. of the mother may all be responsible for such pregnancy losses, although the majority remain unexplained (Sthoeger et al., 1993; Stern et al., 1996; Cnattingius et al., 2000; Heffner, 2004; Chrsitiansen et al., 2006; Iyer et al., 2007; Johns et al., 2007). Currently, individual risk factors for miscarriage have been studied extensively but miscarriage prevention is still largely under-theorized; clinicians mainly lack access to a testable set of risk factors, especially in pregnant women who have a low risk of suffering a miscarriage during early pregnancy. This clinical problem is further impaired by the fact that women are increasingly delaying pregnancy until their 30s. Hence, even one miscarriage may jeopardize subsequent successful reproduction in individual couples. By historical consensus, clinical investigation and therapy are not offered before three or more miscarriages are sequential (Hannes et al., 1992; Daya, 1996), and with advancing age this may be too late (Buletti et al., 1996). The price of access to preventive services seems exceptionally high for women, considering that risk of miscarriage is estimated to increase with number of previous losses, possibly adding 40% risk to a subsequent pregnancy (Daya, 1996). Thus, clinicians and pregnant women alike would benefit from the development of standard risk measurements for widespread clinical use. It is well accepted that the maintenance of early pregnancy is mediated by hormones and by endocrine–immune interactions (Norwitz et al., 2001; Schindler, 2005; Piccinni, 2006). Secretion of human chorionic gonadotrophin (HCG) and corpus-luteumderived progesterone and oestradiol sustain early gestation, and a major shift occurs from luteal to placental progesterone production in the later stages of the first trimester. Progesterone plays many crucial roles during implantation: it sustains decidualization, controls uterine contractility and cervical competence, and functions as an ‘immunosteroid’ by promoting maternal immune tolerance to the fetal semi-allograft and controlling the bias towards a pregnancy protective immune milieu via a protein known as progesterone-induced blocking factor (PIBF) (Szekeres-Bartho et al., 2001; Blois et al., 2004, 2007; Piccinni, 2006; Arck et al., 2007). Hence, it is not surprising that elective abortions can be rapidly induced in humans if progesterone-receptor antagonists are administered before 7 weeks of gestation (Baulieu, 1997). Likewise, it was proved decades ago that pregnancy failure in mice can result from impaired progesterone synthesis by the corpus luteum of the ovary (Csapo and Pinto-Dantas, 1965).

102

A well-balanced equilibrium between the endocrine and immune systems is essential to pregnancy maintenance, and both are susceptible to stress-triggered dysregulation (Padgett and Glaser, 2003; Nepomnaschy et al., 2007). Epidemiological studies in humans indicate that the onset of a miscarriage may be attributable to high levels of perceived stress (Stray Pedersen and Stray Pedersen, 1984; Neugebauer et al., 1996; Maconochie et al., 2007). Evidence from basic science research employing rodents strongly supports the notion that stress exposure inhibits progesterone production and challenges the equilibrium of endocrine–immune cross talk, defying maternal immune tolerance and subsequently resulting in fetal rejection (Blois et al., 2004). Clearly, the concept of a stress trigger for miscarriage in humans is controversial and a matter of intense discussion and research effort (Nelson et al., 2003; Nepomnaschy et al., 2007). A lack of appropriate tools

to evaluate stress perception is a limitation in cohort studies, and contradictory results from different studies may arise from diverse experimental designs and from the difficulty of establishing temporal coincidence of stressful life events or self-reported stress perception to the onset of a miscarriage. Mediators of the hypothalamus–pituitary–adrenal axis, such as corticotrophin-releasing hormone (CRH) and cortisol, may serve as better stress indicators (Nepomnaschy et al., 2006). Clearly, stress perception can also be indicated by other social and societal mechanisms such as smoking, social status, work satisfaction, poor eating habits and low or high body mass index (BMI). Hence, the effects of stress may be interactive or may modify other effects, possibly playing a number of roles, as is also postulated for the role of age. The present study followed a cohort of 864 pregnant women at low risk for a miscarriage through their pregnancies, in order to identify risk associations and the interactive effects of possible risk factors. To provide a data source for the identification of possible risk factors, the individual medical history, age, BMI, social and work status, family, education, smoking habits, and results of psychosocial evaluation were documented at recruitment during the first trimester. Serum was cryopreserved and subsequently analysed for progesterone, PIBF and the stress hormone CRH. Significant associations are presented in the discussion, in the context of the suspected social and biological mechanisms and interactive effects.

Materials and methods Subjects and study conduct A prospective cohort study was conducted by the Departments of Internal Medicine, Obstetrics and Psychosomatics at the Charité Hospital, University Medicine, Berlin, Germany. Recruitment of pregnant women was carried out by 99 obstetricians in private practice in Berlin. The obstetricians were asked to recruit women who had scheduled an appointment to obtain confirmation of pregnancy. Women who were already beyond the 12th week of gestation or who were enrolled in an IVF programme were excluded from the study. Written informed consent was obtained from all the women, and the study was approved by the ethics committee of the Charité Hospital. Based on the focus of this study to include stress perception in the risk analyses, power analysis prior to the onset of the study revealed that a sample size of 892 cases will have a power of at least 80% to detect significant differences with regard to the stress score obtained from the Perceived Stress Questionnaire (PSQ) among the women with normally progressing pregnancies compared with women subsequently suffering from miscarriage. This assumes a common standard deviation and the application of a two-sided t-test with a level of significance at 0.05. Hence, initially 1050 or more study participants needed to be recruited to compensate for an expected drop-out of 15%. Medical, obstetric and gynaecological history and demographic details were thoroughly documented at recruitment using two forms: one completed by the clinician and one by the pregnant woman. Pregnancy was confirmed via a positive blood or RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al. urine test. A pre-stamped card was also placed in a maternal pass handed out to every pregnant woman in Germany, as is common also in other countries. The women as well as the obstetrician were asked to register their pregnancy outcome by returning the completed card. On this card, the gestational week of delivery, the method of delivery, and the gender, weight and length of the baby could further be recorded. The card asked if the pregnancy had been maintained without complications, or if the following complications had occurred: miscarriage, pregnancy-induced hypertension, preeclampsia/HELLP (haemolysis, elevated liver enzyme levels and a low platelet count) syndrome, fetal growth retardation or gestational diabetes. In order to confirm the pregnancy outcome, obstetricians also registered and returned a card when the outcome was known. The cases of miscarriage were confirmed by the physician and were defined as when women reported bleeding and/or expelling tissue and a viable intrauterine pregnancy was undetectable in subsequent transvaginal ultrasound, accompanied by decreasing concentrations of HCG prior to 20 weeks of gestation. In some women, an incomplete miscarriage was diagnosed by an empty sac, which was then treated by pharmacological or surgical intervention.

Psychosocial evaluation

Progesterone analysis Serum progesterone concentration was assayed using an ELISA (enzyme-linked immunosorbent assay) kit (DRG Instruments GmbH, Germany). Briefly, serum samples were thawed and steroids were extracted and assayed according to the manufacturer’s protocol.

Progesterone-induced blocking factor analysis Serum PIBF was determined using a competitive enzymelinked immunosorbent assay. In brief, 96-well microtitre plates were coated with recombinant human PIBF. Sera and standard recombinant PIBF were incubated with an equal volume of biotin-labelled anti-recombinant PIBF IgG for 1 h at 37°C and then added to the plates. After 60 min the plates were washed and horseradish-peroxidase-conjugated streptavidin (AP Hungary Ltd, Budapest, Hungary) was added. After 30 min incubation at 37°C, the reaction was developed by optical densitometry and read at 495 nm. Absorbance was determined at 492 nm, and the concentrations of PIBF were calculated from a corresponding standard curve with a range from 0.005 to 500 ng/ml.

At recruitment, participating women were asked to complete established questionnaires that evaluate stress perception, including the previously validated PSQ (Levenstein et al., 1993; PSQ short version). Besides the sum score, there was a focus on the PSQ subscale for ‘external demands’, as this reflects the effect of external stress factors (Fliege et al., 2005). Additional questionnaires evaluating the quality of life (QoL), the Short-Form 12 QoL-SF12 (Ware, 2003), depressive symptoms (‘Allgemeine Depressionsskala’; Hautzinger and Bailer, 1993) and social support (SOZU) (Fydrich et al. 1999; Elsenbruch et al., 2007) were included.

Analysis of corticotrophin-releasing hormone concentration

Eligibility

Statistical analyses

Data from pregnant women included in the statistical analyses for cases that met these inclusion criteria were as follows (i) medical: week of gestation between 4 and 12, no fertility treatment, no hepatitis B/C or HIV infection, no signs of an imminent miscarriage such as vaginal bleeding, low β-HCG, missing embryonic/fetal heart rate confirmed during ultrasound screening; (ii) formal: documentation of medical, obstetric and gynaecological history and demographic data available from two forms (clinician and patient); and (iii) psychosocial: at least two completed questionnaires and at least one form documenting the outcome of the pregnancy (pre-stamped card, mailed by either the physician or the pregnant woman) available for each recruited case.

Continuous variables were calculated as mean, standard deviation and median; the median was used for data presentation. Comparisons between groups (eligible versus non-eligible study participants; or subgroups of women with normally progressing pregnancies versus subsequent miscarriage) were carried out by the Mann–Whitney U-test for ordinal or continuous data and by Fisher’s exact test for categorical data. This analysis is one form of a univariate analysis of possible risk factors for a miscarriage. For further univariate and multivariate analyses of risk factors, logistic regression was used. Dividing the distribution of the continuous factors into categories defined by distribution quantiles, and further analyses of these categorized factors revealed that a dichotomization of all considered continuous parameters was justified, with the determination of cut-offs supported by diagnostic criteria such as the Youden index. Dichotomized factors were then coded in such a way that a higher risk resulted in an odds ratio of >1. A model that included interaction effects was created according to Hosmer and Lemeshow (1989). Although none of the P-values can be interpreted as error probabilities in a confirmative sense, a Pvalue of <0.05 is considered as statistically significant in all analyses.

Blood withdrawal At recruitment, blood was taken by venous puncture from all women and delivered to the laboratory within 1–3 h by courier. Serum was harvested from all blood samples after centrifugation (1100 g/20 min) and stored at –80°C until further use.

RBMOnline®

Analysis of serum CRH concentration was performed in a subgroup of women with subsequent miscarriage: this selection was due to the limited availability of serum. This subgroup was matched with women with normally progressing pregnancies based on gestational age and age. Unextracted serum was analysed for CRH concentration using a commercially available competitive enzyme immunoassay kit (Phoenix Pharmaceuticals Inc., Belmont, USA).

103

Article - Early risk factors for miscarriage - PC Arck et al.

Results First, it was determined whether the eligible study population displayed in Figure 1 was representative for the entire population recruited for the present study, in order to exclude a possible selection bias. All parameters of the questionnaires presented to the pregnant women at time of recruitment were compared between the study participants fulfilling the inclusion criteria (n = 988) and those women who fulfilled the medical inclusion criteria but who were excluded from the study because of missing forms (n = 159). As shown in Table 1, no significant selection effect could be observed after excluding 159 non-eligible women due to incomplete documentation. Non-significant, but noticeable characteristics in this group of excluded women are a higher number of non-planned pregnancies, a slightly shorter time of living together with the child’s father, a slightly higher mean score in the PSQ and a higher number of women with no profession. Of the total number of women eligible to participate in the present prospective study (n = 864), 809 continued to have a successful pregnancy, defined as the birth of a viable baby, and 55 suffered from a miscarriage after recruitment (Figure 1). Hence, the eligible women were assigned to two subgroups: (i) women with normally progressing pregnancies; and (ii) women with subsequent miscarriage; all parameters recorded at recruitment were compared between the two subgroups cross-sectionally.

104

Comparison of demographic, anamnestic, biological and psychosocial factors between the two groups revealed that women with subsequent miscarriage were of a higher age, had a significantly lower BMI and significantly lower concentrations of serum progesterone compared with women with normally progressing pregnancies (Table 2). The group of women with subsequent miscarriage also showed a slightly higher frequency of chronic diseases. Psychometric analyses showed that women with subsequent miscarriage reported higher levels of stress perception, especially higher scores in the ‘demands’ scale of the perceived stress questionnaire, but levels of significance were not reached (Table 2). Since it is well known that the risk for a subsequent miscarriage is considerably higher early during gestation, it was determined whether the risk factors for a subsequent miscarriage could be confirmed in the subgroup of women recruited during early gestation (week 4–7) in the present study (n = 335). Among these women, the risk factors of low BMI and low progesterone could be confirmed. In addition, increased levels of stress, as reflected by a high perception of demands, also reached levels of significance in this subset (Table 3). The putative role of high stress perception as a risk factor for miscarriage could be further substantiated by the observation that serum concentrations of the stress hormone CRH were significantly increased in women subsequently suffering from miscarriage, compared with women with normally progressing pregnancies, as well as in the subgroup recruited during gestation weeks 4–7 (Figure 2a, c). Additionally, concentrations of the pregnancy-protective protein PIBF were reduced in women with subsequent miscarriage irrespective of the time of recruitment (Figure 2b, d). Further, women

with a university degree had an increased relative risk of experiencing a miscarriage (Tables 2 and 3). Living with a partner or employment could not be identified as risk factors for miscarriage. Analyses of additional psychometric parameters employed in the study such as social support, depressive symptoms or low quality of life did not reveal a significant effect on the probability of subsequent miscarriage (data not shown). Univariate analysis confirmed higher maternal age, low BMI and lower progesterone concentrations as risk factors for subsequent miscarriage in the entire population of eligible participants and in the subgroup of women recruited during early gestation, between weeks 4 and 7 (Table 4). Multivariate analysis of all eligible study participants and women recruited between gestation weeks 4 and 7 ruled out the possibility that the univariate effects of progesterone and BMI were due to temporal differences in distribution of the week of gestation between the subgroups (Table 5). Moreover, it was observed that the correlation between progesterone and gestational age was weak but statistically significant when analysing the data of the entire eligible study population (Spearman rank correlation r = 0.14, P < 0.0001 for progesterone) (Figure 3). Interestingly, the correlation between progesterone and woman’s age was statistically significant, irrespective of the gestational age (Table 5). As depicted in Table 6, low BMI (≤20 kg/m2) proved to be a risk factor for miscarriage during the early weeks of gestation (4–7 weeks) irrespective of the age of the woman [≤20 kg/m2, odds ratio 0.36, 95% confidence interval (CI) 0.19 to 0.67, P = 0.001]. The interaction between age and progesterone also required further differentiation into subgroups consisting of younger (≤33 years) and older (>33 years) women. The influence of progesterone on the risk of miscarriage was dependent on the age of the woman, irrespective of the week of gestation. Thus, progesterone concentrations of ≤12 ng/ml were associated with an increased risk of miscarriage in older women (>33 years old) during the entire first trimester (≤12 ng/ml OR 0.51, 95% CI 0.28 to 0.92, P = 0.0257). The risk of miscarriage was clearly elevated when two or three of the risk factors were present (group with early weeks of gestation: odds ratio 7.1, 95% CI 2.8 to 17.9), compared with the absence of all factors. The population in this interaction model had a slightly higher ratio of miscarriage (52:833 total women) due to some exclusions for missing parameter values. Besides risk factors in the first trimester that may be described as pre-existing conditions, such as higher maternal age and BMI, the data revealed that low concentrations of progesterone increased the risk for a subsequent abortion. The latter parameter may clearly be modulated by pharmacological interventions. Therefore, it was crucial to determine the time window available for such approaches upon identification of women at high risk for a miscarriage. As shown in Table 7, the time gap between the recruitment visit to the physician and the onset of a miscarriage ranged from 2.7 to 10.8 weeks; future studies may show whether pharmacological interventions during this time window, such as progesterone supplementation, lead to a prevention of miscarriage.

RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al.

Figure 1. Study population flow chart.

Table 1. Comparison of demographic, anamnestic and psychosocial parameters at time of recruitment between eligible and non-eligible study participants. Parameter

Eligible study Non-eligible study participants participants

Age, median (years) BMI, median (kg/m2) Planned pregnancy (%) Previous miscarriage (%) No permanent partner (%) Years with partner (median) University degree (%) ADS score (median) Social support score (median) PSQ score (median) QoL-SF12, PCS (median) Chronic diseasea (%) Medicationb (%) Employed (%)

29.4 (862) 22.9 (835) 65.8 (842) 26.9 (446) 2.1 (850) 5.5 (837) 31.1 (853) 11.4 (839) 4.5 (856) 32.9 (844) 48.7 (820) 7.5 (852) 13.8 (855) 83.3 (852)

28.2 (120) 22.9 (132) 50.0 (116) 25.4 (59) 1.8 (112) 4.4 (112) 25.9 (112) 11.8 (111) 4.4 (111) 35.7 (115) 49.2 (106) 10.2 (118) 12.4 (140) 72.3 (112)

Values in parentheses are the number of study participants for whom data were available. ADS = Allgemeine Depressionsskala; BMI = body mass index; PSQ = Perceived Stress Questionnaire; QoL-SF12 = Quality of Life-short form 12; PCS = Physical Component Summary. a Includes asthma, diabetes mellitus, hypertension, inflammatory bowel diseases, lupus erythematosus, migraine, multiple sclerosis, renal disease, rheumatic diseases, sarcoidosis, thyroid diseases. b Includes bronchial spray, heparin, insulin, pain medication, thyroid hormones.

105 RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al.

Table 2. Comparison of demographic, anamnestic, biological and psychosocial parameters at time of recruitment between eligible study participants (recruited between gestation weeks 4 and 12) with normally progressing pregnancy or subsequent spontaneous abortion. Parameter

Normally Subsequent progressing spontaneous pregnancy abortion

Age, median (years) BMI, median (kg/m2) Progesterone, median (ng/ml) QoL-SF12, median (PCS) Previous miscarriage (%) Living with partner (%) University degree (%) ADS (score) Social support (score) PSQ (score) PSQ demands (score) Chronic diseasec (%) Medicationd (%) Employed(%)

29.7 (807) 22.2a (873) 15.3b (809) 50.2 (767) 28.3 (420) 97.9 (795) 30.5 (798) 10.0 (785) 4.6 (801) 30.0 (790) 33.3 (801) 9.6 (809) 8.5 (809) 83.6 (798)

30.7 (55) 21.2a (54) 13.5b (53) 52.2 (53) 25.0 (32) 98.2 (55) 40.0 (55) 9.5 (54) 4.6 (55) 31.7 (54) 40.0 (54) 16.4 (55) 3.6 (55) 80.0 (55)

Values in parentheses are the number of study participants for whom data were available. Values not in parentheses are median values, unless otherwise stated. Values with the same superscript letter are significantly different: aP = 0.036, b P = 0.016. c Includes asthma, diabetes mellitus, hypertension, inflammatory bowel diseases, lupus erythematosus, migraine, multiple sclerosis, renal disease, rheumatic diseases, sarcoidosis, thyroid diseases. d Includes bronchial spray, heparin, insulin, pain medication, thyroid hormones. ADS = Allgemeine Depressionsskala; BMI = body mass index; PSQ = Perceived Stress Questionnaire; QoL-SF12 = Quality of Life-short form 12; PCS = Physical Component Summary.

106 RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al. Table 3. Comparison of demographic, anamnestic, biological and psychosocial parameters at time of recruitment between eligible study participants (recruited between gestation weeks 4 and 7) with normally progressing pregnancy or subsequent spontaneous abortion. Parameter

Normally Subsequent progressing spontaneous pregnancy abortion

Age, median (years) BMI, median (kg/m2) Progesterone, median (ng/ml) QoL-SF12, median (PCS) Previous miscarriage (%) Living with partner (%) University degree (%) ADS (score) Social support (score) PSQ (score) PSQ demands (score) Chronic diseased (%) Medicatione (%) Employed (%)

29.6 (297) 22.1a (288) 14.7b (298) 52.0 (286) 25.3 (150) 97.6 (292) 32.4 (296) 10.0 (290) 4.73 (295) 28.8 (292) 33.3c (295) 9.4 (298) 10.1 (298) 83.4 (296)

31.2 (38) 20.6a (37) 12.7b (37) 51.7 (38) 20.8 (24) 97.4 (38) 44.7 (38) 9.2 (38) 4.73 (38) 31.7 (37) 40.0c (37) 15.8 (38) 5.3 (38) 73.7 (38)

Values in parentheses are the number of study participants for whom data were available. Values not in parentheses are median values, unless otherwise stated. Values with the same superscript letter are significantly different: aP = 0.024, b P = 0.037, cP = 0.024. d Includes asthma, diabetes mellitus, hypertension, inflammatory bowel diseases, lupus erythematosus, migraine, multiple sclerosis, renal disease, rheumatic diseases, sarcoidosis, thyroid diseases. e Includes bronchial spray, heparin, insulin, pain medication, thyroid hormones. ADS = Allgemeine Depressionsskala; BMI = body mass index; PSQ = Perceived Stress Questionnaire; QoL-SF12 = Quality of Life-short form 12; PCS = Physical Component Summary.

(b) *

2.0

Available data

30

45

(d) PIBF (ng/ml serum × 100)

40

40

33

*

10

2.0

Available data

5

25

*

4.0 CRH (ng/ml serum)

Recruitment between weeks 4–12

(c)

*

10 PIBF (ng/ml serum × 100)

4.0 CRH (ng/ml serum)

Recruitment between weeks 4–7

(a)

5

97

44

Women with normally progressing pregnancy Women with subsequent spontaneous miscarriage

RBMOnline®

Figure 2. Serum concentrations of corticotrophin-releasing hormone (CRH) in the subgroup of women recruited between gestation weeks 4 and 7 (a) as well as in women recruited between gestation weeks 4 and 12 (c). Serum concentrations of progesterone-induced blocking factor (PIBF) in the subgroup of women recruited between gestation weeks 4 and 7 (b) as well as in women recruited between gestation weeks 4 and 12 (d). It should be noted that the samples displayed in (b) and (d) include the results displayed in (a) and (c). All data are displayed by assigning the women to two groups according to the pregnancy outcome, such as normally progressing pregnancies subsequent to recruitment/blood withdrawal and subsequent spontaneous miscarriage. Results are displayed as box plots revealing the median, and the 25th and 75th percentiles. Open circles depict outliers. Significance of differences was calculated using the Mann–Whitney U-test for unpaired samples, *P ≤ 0.05.

107

Article - Early risk factors for miscarriage - PC Arck et al.

Table 4. Univariate analysis of risk factors for spontaneous abortion. Parameter Number Reference Odds 95% P-value variable ratio confidence interval All eligible study participants Age (years) 862 0–<33/>33 1.76 1.00–3.09 0.051 BMI (kg/m2) 837 >20/0–<20 2.33 1.30–4.19 0.005 Progesterone (ng/ml) 862 >12/0–<12 2.24 1.26–4.00 0.006 Gestational age (weeks) 864 8–12/4–7 3.83 2.13–6.91 <0.0001 Eligible study participants recruited between gestation week 4 and 7 Age (years) 335 0–< 33 / > 33 2.35 1.18–4.70 0.015 BMI (kg/m2) 325 0–< 20 / > 20 2.89 1.41–5.93 0.004 Progesterone (ng/ml) 335 0–< 12 / > 12 2.32 1.16–4.64 0.018 BMI = body mass index.

Table 5. Multivariate analysis of risk factors for spontaneous abortion. All eligible study participants. Parameter Reference variable

Regression Standard Wald chi P-value parameter error square estimate-

All eligible study participants Intercept – –3.47 0.34 101.12 Age (years) 0–<33/>33 –0.91 0.69 1.74 BMI (kg/m2) 0–<20/>20 0.34 0.59 0.32 Progesterone (ng/ml) 0–<12/>12 –0.05 0.43 0.01 Gestational age (weeks) 4–7/8–12 0.48 0.47 1.04 Age/progesterone – 1.89 0.67 7.84 Age/gestational age – 1.48 0.72 4.07

<0.0001 NS NS NS NS 0.005 0.04

Eligible study participants recruited between gestation weeks 4 and 7 Intercept – –3.033 0.369 67.51 <0.0001 Age (years) 0–<33/>33 0.66 0.52 1.63 NS BMI (kg/m2) 0–<20/>20 1.49 0.42 12.9 0.0003 Progesterone (ng/ml) 0–<12/>12 0.11 0.53 0.04 NS Age/progesterone – 1.62 0.79 4.21 0.04 BMI = body mass index; NS = not statistically significant.

108 RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al.

Woman with normally progressing pregnancy Woman with subsequent spontaneous miscarriage

90 Progesterone (ng/ml serum)

80 70 60 50 40 30 20 10 4

5

6

7

8

9

10

11

Gestational age at time of recruitment

12

Figure 3. Serum progesterone concentration in relation to gestational age at recruitment in women with normally progressing pregnancy and those with subsequent spontaneous abortion. The light grey circles represent individual results of progesterone analysis in women with normally progressing pregnancies (light grey horizontal lines indicate the median of progesterone in these women in each gestational week). The dark grey circles show the individual results of progesterone analysis in women with subsequent spontaneous abortion (dark grey horizontal lines indicate the median of progesterone in these women in each gestational week). The Spearman rank correlation between gestational age and concentrations of progesterone was r = 0.14, P < 0.0001.

Table 6. Influence of progesterone concentration and BMI on risk of spontaneous abortion in subgroups of women defined by age and week of gestation (interaction model). Subgroup Parameter Reference Odds 95% variable ratio confidence interval

P-value

Younger women at early gestationa Older women at early gestationb Younger women at late gestationc Older women at late gestationd

NS 0.031 0.003 0.001 NS NS 0.050 NA

Progesterone (ng/ml) BMI (kg/m2) Progesterone (ng/ml) BMI (kg/m2) Progesterone (ng/ml) BMI (kg/m2) Progesterone (ng/ml) BMI (kg/m2)

>12/0–≤12 >20/0–≤20 >12/0–≤12 >20/0–≤20 >12/0–≤12 >20/0–≤20 >12/0–≤12 >20/0–≤20

1.12 2.86 7.13 14.15 0.73 1.97 9.97 NA

0.40–3.10 1.10–7.42 1.96–25.93 2.81–71.21 0.16–3.40 0.58–6.74 1.00–99.29 NA

232 subjects, 19 spontaneous abortions. 91 subjects, 17 spontaneous abortions. 373 subjects, 12 spontaneous abortions. d 137 subjects, 4 spontaneous abortions. BMI = body mass index; NA = not available (data on BMI, age or progesterone were not available for all subjects); NS = not statistically significant. a

b c

109 RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al. Table 7. Mean number of weeks between recruitment and onset of spontaneous abortion.

Gestational age at time of recruitment (weeks) 4–5 6 7 8 9 10 11

Number of women Spontaneous abortion (weeks)a

11 10 18 6 4 5 2 4.8 ± 4.0 4.1 ± 3.2 3.3 ± 3.2 2.7 ± 1.5 10.8 ± 7.9 4.6 ± 7.6 8.0 ± 4.2

Mean ± SD number of weeks until spontaneous abortion occurred.

a

Discussion Among the risks investigated for miscarriage in relation to the medical and psychosocial factors investigated in the present study, the strongest associations were found for a low BMI, high demands accompanied by increased concentrations of CRH, low progesterone (low PIBF) and a higher maternal age. Further, the risk for a subsequent miscarriage due to low concentrations of progesterone, associated with low concentrations of PIBF, was particularly present in women at an age above 33 years and at a very early gestational age (weeks 4–7). While not yet fully understood, it is critical to pursue the impact of age on the fertility potential of women, given the global trend to delay childbearing. It is widely accepted that a woman’s fertility declines gradually after the age of 20, and precipitously after the age of 35 (te Velde and Pearson, 2002). A less receptive endometrium and/or a decrease in oocyte quality may be responsible for the increased frequency of miscarriage in waning reproductive years. Higher maternal age may also account for insufficient concentrations of progesterone, and it has been proposed that fertilized oocytes and/or the corpus luteum produce an insufficient amount of progesterone to sustain implantation during senescence. The interaction model generated in the present study revealed that the influence of progesterone on risk of miscarriage was dependent on the age of the woman, and was particularly profound during the first seven weeks of gestation. Thus, low concentrations of progesterone in older women with subsequent miscarriage may be the result of poor oocyte/corpus luteum quality, but could also indicate an impaired luteo-placental shift of progesterone production in older women (Itskovitz and Hodgen, 1988).

110

Inadequate concentrations of progesterone during early pregnancy may also be linked to stress-mediated activation of the hypothalamic–pituitary–adrenal axis and subsequent CRH and glucocorticoid concentrations (Padgett and Glaser, 2003). Interestingly, elevated concentrations of cortisol have been associated with very early miscarriage (Nepomnaschy, 2006). CRH is released from the anterior pituitary to regulate the stress hormone axis, such as cortisol secretion, but is also synthesized and secreted by the placenta. High concentrations of glucocorticoids have adverse effects on the uterus and fetus, and inhibit pituitary luteinizing hormone and ovarian oestrogen and progesterone secretion (Magiakou et al., 1997). Such inhibitory effects of stress hormones on the female reproductive

system are responsible for the ‘hypothalamic’ amenorrhoea of stress, and – as shown in mice – may also account for inadequate concentrations of progesterone during pregnancy, subsequently resulting in miscarriage (Magiakou et al., 1997). Besides such inhibition of progesterone synthesis and subsequent PIBF secretion, it should also be considered that progesterone may be catabolized more rapidly, thereby resulting in inadequate concentrations seen in subsequently failing pregnancies. The enzyme 20α-hydroxysteroid dehydrogenase (20α-HSD) converts progesterone to 20α-hydroxyprogesterone (Piekorz et al., 2005); however, an effect of stress perception/stress hormones on 20α-HSD is yet to be confirmed. Significantly elevated serum progesterone concentrations from as early as 4 weeks’ gestation have successfully predicted women destined to have viable intrauterine pregnancies (al Sebai et al., 1995; Ionnadis et al., 2005). However, although progesterone is a known adjunctive marker for prediction of early pregnancy outcome (Buletti et al., 1996), serum concentration testing, while widely available and inexpensive, is rarely included in the clinical blood work done upon confirmation of a pregnancy. This finding of this study suggests that serum progesterone concentration is a highly informative indicator of risk in early pregnancy. Also, given that low progesterone concentrations may be therapeutically approached by pharmacological interventions, it is crucial to determine the time window available to identify women who may be at risk of miscarriage. The time gap between the recruitment visit to the physician and the onset of a miscarriage ranged from 2.7 to 10.8 weeks in this study. Whether pharmacological interventions during this time window may suffice to prevent miscarriage is an important question for future studies. Perceived stress/high demands could be identified as a risk factor by the psychometric tools used in this study. However, these effects are subtle and future research should aim at establishing a stress questionnaire allowing stress perception to be captured among a distinct population, such as pregnant women. Regardless of this limitation, previous studies have also described an independent stress effect in retrospective study designs (Neugebauer et al., 1996; Boyles et al., 2000; Maconochie et al., 2007). Consequently, the present prospective study can be considered as an evaluation of the suitability of psychometric instruments for capturing the stress effect in

RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al. ‘low risk’ early pregnancies. In animal studies, the biological mechanisms linking stress to progesterone depression are fairly clear. Exposure to stress in the form of restraint (Wiebold et al., 1986) or sound (Joachim et al., 2003) has induced an increase in fetal losses in pregnant mice via a significant reduction in progesterone concentrations accompanied by diminished expression of progesterone receptors at the fetomaternal interface (Joachim et al., 2003). Administration of the progesterone derivative dydrogesterone has been shown to restore the pregnancy-protective immune milieu in the mouse model of stress-triggered fetal loss, as well as in humans with threatened abortion (Joachim et al., 2003; Blois et al., 2004; Gruber and Huber, 2005; Kalinka and Szekeres-Bartho, 2005). Further, progesterone supplementation has been shown to be efficacious when continuation of pregnancy is hampered by immunological factors, luteinic and neuroendocrine deficiencies, and myometrial hypercontractility (Di Renzo et al., 2005). However, other trials failed to detect a significant benefit of progesterone supplementation during pregnancy in women without a history of recurrent miscarriage (Di Renzo et al., 2005; Nardo and Sallam, 2006). Thus, additional randomized controlled trials are needed to increase the power of the currently available meta-analyses to further evaluate progesterone supplementation (Oates-Whitehead et al., 2003; Wahabi et al., 2007). Here, the interaction model displayed in Table 6 may facilitate the identification of women who could benefit from progesterone supplementation. The present study also detected increased concentrations of the stress hormone CRH in the serum of a subcohort of women with failing pregnancies and low concentrations of progesterone. Here, a causal relationship between high concentrations of CRH triggered by high stress perception and subsequent inhibition of progesterone secretion remains to be proven. Clearly, it should be taken into consideration that CRH is produced in large quantities in most female reproductive tissue, including trophoblast and the ovary (Bamberger et al., 2007). Hence, it could be argued that the CRH concentrations obtained in the present study may not reflect psychological stress. However, since groups of women who were matched based on gestational age were compared, the differences observed with regard to CRH concentrations in serum analysis may result from a higher stress perception. However, one should refrain from extrapolating the CRH values to nonpregnant cohorts, since the baseline concentrations of CRH in pregnant women are increased due to the CRH production by reproductive tissues such as the trophoblast and the ovary. Over the past two decades, both obesity and underweight have been extensively studied in the context of reproduction, predominately with respect to success rates after IVF (Helgstrand and Andersen, 2005). In the present cohort an increased risk of miscarriage, especially during the first 7 weeks of pregnancy, was identified for women who were quite lean, with a BMI of less than 20 kg/m2, irrespective of their age. It is well-known that the BMI is significantly correlated to serum leptin concentrations (Baptista and Beaulieu, 2002). Leptin is a hormone secreted mainly by adipose cells, and has a primary role in the regulation of body weight by establishing a feedback loop between the energy reserves and the hypothalamic centres that control food intake. Thus, it is proposed that leptin acts as a permissive factor that allows the onset of energy-demanding situations, such as pregnancy, only when the amount of the energy reserve is sufficient enough to guarantee its success (Casabiell et al., 2001). RBMOnline®

This hypothesis is supported by published evidence indicating that abnormally low serum leptin concentrations are detectable in women suffering from miscarriage (Lage et al., 1999), but is challenged by observations that elevated leptin concentrations result in impaired oocyte quality and/or early embryo development (Brannian et al., 2001). Based on recent insights, it is proposed that weight gain may be a therapeutic approach in women with a BMI of less than 20 kg/m2 and possibly with a history of miscarriage, in order to improve the prerequisites for a subsequent pregnancy. However, since a BMI of less than 20 kg/m2 may be linked to stress (Kivimaki et al., 2006) or to psychiatric disorders such as anorexia nervosa, a proposal to gain weight may require psychological intervention strategies in some patients. Since genetic abnormalities of the fetus are often quoted to be the main cause for pregnancy losses (Stern et al., 1996; Iyer et al., 2007), it would have been desirable to obtain insights regarding the karyotype of the fetus in the cases of miscarriage in the present study. However, due to the multi-centred study design, fresh fetal tissue needed to perform chromosomal analysis, such as standard G banding, was largely unavailable. In conclusion, a subtle, but nevertheless significant risk of miscarriage during the first trimester is a serious concern for pregnant women and their physicians. This study provides valuable information for clinicians by redefining groups of ostensibly low risk women with respect to increased but typically ignored risks of miscarriage. Taking into consideration the insights from the present study and currently available data from basic science and epidemiology, it is postulated that miscarriage in humans is not a single entity event, but the result of complex interdependencies between demographic, anamnestic, physiological and psychological risk factors. Independent effects were shown for advancing age, low BMI and low serum progesterone, all of which are easily detectable in a regular office visit. These insights will help clinicians to identify pregnant women who require extra monitoring or therapeutic interventions such as progestogen supplementation during early pregnancy.

Acknowledgements The authors are indebted to the support of the 99 clinicians in private practice in Berlin and their staff. Without their enthusiasm, it would have been impossible to perform the present study. The authors are grateful to Jane Irons for her support in editing the manuscript and Frauke Riller and David Clark for the stimulating discussions and constructive comments. This work was made possible by research grants from the Charité (KKS) to PCA, MR and BFK. This study was integrated in EMBIC, a Network of Excellence co-financed by the European Commission throughout the FP6 framework program ‘Life Science, Genomics and Biotechnology for Health’. MP is supported by the Rahel Hirsch Program and SMB by the Habilitation Program, both awarded by the Charité.

References Al-Sebai MA, CR Kingsland, M Diver et al. 1995 The role of a single progesterone measurement in the diagnosis of early pregnancy failure and the prognosis of fetal viability. British Journal of

111

Article - Early risk factors for miscarriage - PC Arck et al.

112

Obstetrics and Gynaecology 102, 364–369. Arck P, Hansen PJ, Mulac Jericevic B et al. 2007 Progesterone during pregnancy: endocrine-immune cross talk in mammalian species and the role of stress. American Journal of Reproductive Immunology 58, 268–279. Bamberger CM, Minas V, Bamberger AM et al. 2007 Expression of urocortin in the extravillous human trophoblast at the implantation site. Placenta 28, 127–132. Baptista T, Beaulieu S 2002 Are leptin and cytokines involved in body weight gain during treatment with antipsychotic drugs? Canadian Journal of Psychiatry 47, 742–749. Baulieu EE 1997 RU 486 (mifepristone). A short overview of its mechanisms of action and clinical uses at the end of 1996. Annals of the New York Academy of Sciences 828, 47–58. Blois SM, Ilarregui JM, Tometten M et al. 2007 A pivotal role for galectin-1 in fetomaternal tolerance. Nature Medicine 13, 1450– 1457. Blois SM, Joachim R, Kandil J et al. 2004 Depletion of CD8+ cells abolishes the pregnancy protective effect of progesterone substitution with dydrogesterone in mice by altering the Th1/Th2 cytokine profile. Journal of Immunology 172, 5893–5899. Boyles SH, Ness RB, Grisso JAA et al. 2000 Life event stress and the association with spontaneous abortion in gravid women at an urban emergency department. Health Psychology 19, 510–514. Brannian JD, Schmidt SM, Kreger DO et al. 2001 Baseline nonfasting serum leptin concentration to body mass index ratio is predictive of IVF outcomes. Human Reproduction 16, 1819–1826. Bulletti C, Flamigni C, Giacomucci E 1996 Reproductive failure due to spontaneous abortion and recurrent miscarriage. Human Reproduction Update 2, 118–136 Casabiell X, Pineiro V, Vega F et al. 2001 Leptin, reproduction and sex steroids. Pituitary 4, 93–99. Christiansen OB, Nielsen HS, Kolte AM 2006 Future directions of failed implantation and recurrent miscarriage research. Reproductive BioMedicine Online 13, 71–83. Cnattingius S, Signorello LB, Anneren G et al. 2000 Caffeine intake and the risk of first-trimester spontaneous abortion. New England Journal of Medicine 343, 1839–1845. Csapo AI, Pinto-Dantas CA 1965 The effect of progesterone on the human uterus. Proceedings of the National Academy of Sciences of the USA 54, 1069–1076. Daya S 1996 Evaluation and management of recurrent spontaneous abortion. Current Opinion in Obstetrics and Gynecology 8, 188–192. Di Renzo GC, Mattei A, Gojnic M et al. 2005 Progesterone and pregnancy. Current Opinion in Obstetrics and Gynecology 17, 598–600. Elsenbruch S, Benson S, Rucke M et al. 2007 Social support during pregnancy: effects on maternal depressive symptoms, smoking and pregnancy outcome. Human Reproduction 22, 869–877. Fliege H, Rose M, Arck P et al. 2005 The Perceived Stress Questionnaire (PSQ) reconsidered: validation and reference values from different clinical and healthy adult samples. Psychosomatic Medicine 67, 78–88. Fydrich T, Geyer, M, Hessel A et al. 1999 Fragebogen zur sozialen Unterstützung (F-SOZU): Neue Ergebnisse zur Testgüte und Normierung an einer repräsentativen Stichprobe. Diagnostica 45, 212–216. Gruber CJ, Huber JC 2005 The role of dydrogesterone in recurrent (habitual) abortion. Journal of Steroid Biochemistry and Molecular Biology 97, 426–430. Hannes, M, J Englert, W Gotlieb et al. 1992 Recurrent spontaneous miscarriage. Revue Médicale de Bruxelles 13, 103–106. Hautzinger M, Bailer M 1993 Allgemeine Depressionsskala. Beltz Test GmbH, Weinheim, Germany. Heffner LJ 2004 Advanced maternal age – how old is too old? New England Journal of Medicine 351, 1927–1929. Helgstrand S, Andersen AM 2005 Maternal underweight and the risk of spontaneous abortion. Acta Obstetrica et Gynecologica Scandinavica 84, 1197–1201. Hosmer D, Lemeshow S 1989 Applied logistic regression. John Wiley

and Sons, New York. Ioannidis G, Sacks G, Reddy N et al. 2005 Day 14 maternal serum progesterone levels predict pregnancy outcome in IVF/ICSI treatment cycles: a prospective study. Human Reproduction 20, 741–746. Itskovitz J, Hodgen GD 1988 Endocrine basis for the initiation, maintenance and termination of pregnancy in humans. Psychoneuroendocrinology 13, 155–170. Iyer P, Wani L, Joshi S et al. 2007 Cytogenetic investigations in couples with repeated miscarriages and malformed children: report of a novel insertion. Reproductive BioMedicine Online 14, 314–321. Joachim R, Zenclussen AC, Polgar B et al. 2003 The progesterone derivative dydrogesterone abrogates murine stress-triggered abortion by inducing a Th2 biased local immune response. Steroids 68, 931–940. Johns J, Muttukrishna S, Lygnos M et al. 2007 Maternal serum hormone levels for prediction of adverse outcome in threatened miscarriage. Reproductive BioMedicine Online 15, 413–421. Kalinka J, Szekeres-Bartho J 2005 The impact of dydrogesterone supplementation on hormonal profile and progesterone-induced blocking factor concentrations in women with threatened abortion. American Journal of Reproductive Immunolology 53, 166–171. Kivimaki M, Head J, Ferrie JE et al. 2006 Work stress, weight gain and weight loss: evidence for bidirectional effects of job strain on body mass index in the Whitehall II study. International Journal of Obesity 30, 982–987. Lage M, Garcia-Mayor RV, Tome MA et al. 1999 Serum leptin levels in women throughout pregnancy and the postpartum period and in women suffering spontaneous abortion. Clinical Endocrinology 50, 211–216. Levenstein S, Prantera C, Varvo V et al. 1993 Development of the Perceived Stress Questionnaire: a new tool for psychosomatic research. Journal of Psychosomatic Research 37, 19–32. Maconochie N, Doyle P, Prior S et al. 2007 Risk factors for first trimester miscarriage--results from a UK-population-based case– control study. BJOG: An International Journal of Obstetrics and Gynaecology 114, 170–186. Magiakou MA, Mastorakos G, Webster E et al. 1997 The hypothalamic–pituitary–adrenal axis and the female reproductive system. Annals of the New York Academy of Sciences 816, 42–56. Nardo LG, Sallam HN 2006 Progesterone supplementation to prevent recurrent miscarriage and to reduce implantation failure in assisted reproduction cycles. Reproductive BioMedicine Online 13, 47–57. Nelson DB, Grisso JA, Joffe MM et al. 2003 Does stress influence early pregnancy loss? Annals of Epidemiology 13, 223–229. Nepomnaschy PA, Sheiner E, Mastorakos G et al. 2007 Stress, immune function and women’s reproduction. Annals of the New York Academy of Sciences 1113, 350–364. Nepomnaschy PA, Welch KB, McConnell DS et al. 2006 Cortisol levels and very early pregnancy loss in humans. Proceedings of the National Academy of Sciences of the USA 103, 3938–3942. Neugebauer R, Kline J, Stein Z et al. 1996 Association of stressful life events with chromosomally normal spontaneous abortion. American Journal of Epidemiology 143, 588–596. Norwitz ER, Schust DJ, Fisher SJ 2001 Implantation and the survival of early pregnancy. New England Journal of Medicine 345, 1400–1408. Oates-Whitehead RM, Haas DM, Carrier JA 2003 Progestogen for preventing miscarriage. Cochrane Database of Systematic Reviews 4, CD003511. Padgett DA, Glaser R 2003 How stress influences the immune response. Trends in Immunology 24, 444–448. Piccinni MP 2006 T cells in normal pregnancy and recurrent pregnancy loss. Reproductive BioMedicine Online 13, 840–844. Piekorz RP, Gingras S, Hoffmeyer A et al. 2003 Regulation of progesterone levels during pregnancy and parturition by signal transducer and activator of transcription 5 and 20alphahydroxysteroid dehydrogenase. Molecular Endocrinology 19, 431–440. Red-Horse K, Zhou Y, Genbacev O et al. 2004 Trophoblast

RBMOnline®

Article - Early risk factors for miscarriage - PC Arck et al. differentiation during embryo implantation and formation of the maternal–fetal interface. Journal of Clinical Investigation 114, 744–754. Schindler AE 2005 Endocrinology of pregnancy: consequences for the diagnosis and treatment of pregnancy disorders. Journal of Steroid Biochemistry and Molecular Biology 97, 386–388. Stern JJ, Dorfmann AD, Gutierrez-Najar AJ et al. 1996 Frequency of abnormal karyotypes among abortuses from women with and without a history of recurrent spontaneous abortion. Fertility and Sterility 65,250–253. Sthoeger ZM, Mozes E, Tartakovsky B 1993 Anti-cardiolipin antibodies induce pregnancy failure by impairing embryonic implantation. Proceedings of the National Academy of Sciences of the USA 90, 6464–6467. Stray-Pedersen BS, Stray-Pedersen PS 1984 Etiologic factors and subsequent reproductive performance in 195 couples with a prior history of habitual abortion. American Journal of Obstetrics and Gynecology 148, 140–146. Szekeres-Bartho J, Barakonyi A, Par G et al. 2001. Progesterone as an immunomodulatory molecule. International Immunopharmacology 1, 1037–1048. te Velde ER, Pearson PL 2002 The variability of female reproductive ageing. Human Reproduction 8, 141–154. Wahabi H, Abed Althagafi N, Elawad M. 2007 Progestogen for treating threatened miscarriage. Cochrane Database of Systematic

Reviews 18, CD005943. Ware JE, Jr. 2003 Conceptualization and measurement of healthrelated quality of life: comments on an evolving field. Archives of Physical Medicine and Rehabilitation 84, 43–51. Wiebold JL, Stanfield PH, Becker WC et al. 1986 The effect of restraint stress in early pregnancy in mice. Journal of Reproduction and Fertility 78, 185–192. Wilcox AJ, Baird D, Weinberg CR 1999 Time of implantation of the conceptus and loss of pregnancy. New England Journal of Medicine 340, 1796–1799.

Declaration: The authors report no financial or commercial conflicts of interest. Received 13 July 2007; revised and resubmitted 5 September 2007; refereed 20 December 2007; accepted 25 February 2008.

113 RBMOnline®