Introduction
Infertility is defined as the inability to conceive after 12 months of unprotected sexual intercourse. Infertility prevalence in women is about 13%. Female infertility may have numerous etiologies among which polycystic ovary syndrome (PCOS), endometriosis, tubal and uterine pathologies, age-related infertility, lifestyle factors (smoking, obesity, etc.), and/or environmental causes are the most common. Some of these may also be developmental in origin (see below). Nevertheless, in some cases, no cause is documented or may be of developmental origin. Thus, the impact of the intrauterine environment on long-term health is a well-recognized concept. An increased risk of noncommunicable metabolic diseases has been observed in people born small for gestational age or with nonoptimal early developmental conditionsReference Hanson and Gluckman1. This concept, known as developmental origins of health and disease (DOHaD), has been extended to other health outcomes including offspring fertility (based on animal experiments and some retrospective human studies).
Undernutrition or overnutrition in pregnant sheep and cattle may induce negative effects on reproductive function of both male and female offspring. Indeed, delayed ovarian developmentReference Rae, Palassio and Kyle2, reduced follicle number in adulthood, and increased oxidative stress in the ovaryReference Bernal, Vickers, Hampton, Poynton and Sloboda3 have been reported in female sheep exposed to undernutrition in utero . In sheep exposed to overnutrition, delayed ovarian development and reduced follicle numbers in foetal ovaries have also been observedReference Da Silva, Aitken, Rhind, Racey and Wallace4. Experiments in rodents have confirmed these results, as reviewed elsewhereReference Chadio and Kotsampasi5. Moreover, a recent murine study reported that a low-protein diet during preconception, pregnancy, and lactation periods led to reduced primordial follicle numbers and increased follicular atresiaReference Winship, Gazzard, Cullen McEwen, Bertram and Hutt6.
In women, low birth weight (LBW) has been associated with an increased risk of early reproductive senescence, such as earlier menopause (FSH > 25 IU/ml)Reference Tom, Cooper and Kuh7, Reference Cresswell, Egger and Fall8. Data from the Danish National Birth Cohort (22,044 pregnancies from 21,786 women) showed that both low (<2500 g) and high (>4500 g) BWs were associated with an increase in time to pregnancy of more than 1 yearReference Nohr, Vaeth, Rasmussen, Ramlau-Hansen and Olsen9. Moreover, anovulation appears to be observed more frequently in 10 adolescent girls born small for gestational ageReference Ibanez, Potau and Ferrer10. A study comparing 37 young women with LBW to 35 controls showed that LBW women have reduced insulin sensitivity associated with an increased risk of developing PCOSReference Pandolfi, Zugaro and Lattanzio11. In a study comparing 375 controls and 368 women with endometriosis, LBW has also been associated with increased risk of endometriosis, especially deep infiltrated endometriosisReference Borghese, Sibiude and Santulli12.
The human studies detailed above all use BW as a proxy for foetal development. Nevertheless, despite having a similar BW, a long and thin infant is metabolically different from a short and chubby baby. The ponderal index (PI) assesses the ratio between weight (kg) and the cubic value of height (m3) and is usually used as a corpulence index in paediatricsReference Cooley, Donnelly and Walsh13. Birth PI can discriminate between children of the same BW and thus is a more relevant indirect indicator of foetal nutrition. PI may more accurately reflect adiposity in infantsReference Howe, Tilling and Benfield14.
Birth PI is rarely considered when assessing long-term fertility, mostly due to the fact that accurate and reliable information on both BW and birth length is usually not available. Nevertheless, it has been observed that a one unit increase in birth PI is associated with reduced risk of PCOS symptoms in 30-year-old women, whereas a 100-g increase in BW is associated with increased risk of hyperandrogenismReference Davies, March, Willson, Giles and Moore15.
As we have previously observed, BW was higher in infertile men compared to fertile men included in the ALIFERT studyReference Faure, Dupont and Chavatte-Palmer16. Furthermore, the BW was inversely correlated with both total sperm count and sperm DNA integrity. The purpose of this study was to investigate whether a similar correlation exists in women of the same cohort and if PI may be an indicator of fertility later in life.
Material and methods
Ninety-nine female partners of infertile couples with idiopathic primary infertility and 100 fertile women were recruited in the ALIFERT study between September 2009 and December 2013 (N° P071224). The ALIFERT study is a prospective observational case-control study with the aim to assess the association between diet and idiopathic infertility.
Infertile women were partners of infertile couples attending four infertility centers in France (Jean Verdier hospital ART center, Bondy; Cochin hospital ART center, Paris; Hôpital Nord ART center, Saint Etienne; Polyclinique de Navarre ART center, Pau). They were eligible for the study if they presented a primary idiopathic infertility >12 months and met the following inclusion criteria: (i) they were between 18 and 38 years old, (ii) they did not present either anovulation, ovarian failure (on the basis of follicle count and hormone balance at day 3 (FSH, LH, and estradiol)) nor uterotubal pathology (assessed by hysterosalpingography), (iii) their partners did not present severe sperm alteration nor urogenital pathology, and (iv) they were in possession of their Child Health Record. Patients with current known or previous metabolic or digestive disease were excluded.
The control group consisted of fertile women volunteers recruited by advertisement (Internet advertising and word of mouth) from the general healthy population in areas of the participating centers. The criteria for eligibility were (i) age between 18 and 38 years, (ii) they had a spontaneously conceived child under 2 years of age, (iii) time to pregnancy was less than 12 months, and (iv) they were in possession of their Child Health Record.
Male partners of infertile and fertile women were between 18 and 45 years old.
Written informed consent was collected. The ethics committee (“Comité de Protection des Personnes”) approved the study as ALIFERT study (national biomedical research P071224/AOM 08180:NEudra CT 2009-A00256-51/clinical trials NCT01093378).
BW and birth length were collected from the individual Child Health Record. The Child Health Record is a booklet delivered by the national health authority to all French children at birth where all information on the child’s health is recorded by medical staff. PI was calculated as BW/length3 (kg/m3).
Anthropometric assessment
The same trained investigator measured height, weight, and waist circumference (measured at the narrowest point between the lower border of the ribs and the iliac crest) in both fertile and infertile women at the time of the inclusion visit in the ART centers. Body mass index (BMI) was calculated as weight/height2 (kg/m2).
Blood samples and analyses
Blood samples were collected after a 12-h fasting period.
High-density lipoprotein (HDL-cholesterol), low-density lipoprotein (LDL-cholesterol), triglycerides, and glucose concentrations were measured by standardized methods in the hospitals’ biology laboratory.
AMH was assayed with the AMH Gen II ELISA Kit (Beckman Coulter). All the samples were assayed by the same operator at the same time.
Blood pressure assessment
A sphygmomanometer cuff was placed on the patient arm and blood pressure was measured after 5 min of bed rest in a supine position. The systolic and diastolic blood pressures were calculated by computing the average of the right and left arm.
Tobacco consumption and exhaled carbon monoxide (CO)
Patients reported the number of cigarettes smoked per day. They have been categorized as smokers if they smoked one or more cigarettes per day and as nonsmokers if they did not smoke at all. Exhaled carbon monoxide (CO) was measured in parts per million (ppm) as a supportive indicator with the underlying assumption that exhaled CO in smokers is higher than in nonsmokersReference Deveci, Deveci, Acik and Ozan17. The assessment of exhaled CO was intended to support self-reporting of tobacco consumption. Nevertheless, exhaled CO can also be influenced by passive smoking and prolonged exposure to a polluted environmentReference Maga, Janik and Wachsmann18 that also may impact fertility. Thus, the exhaled CO level does not necessarily reflect a subject’s smoking amount, but may be more relevant to address airborne environmental exposure that may impact fertility, consequently exhaled CO levels were used instead of self-reporting.
Exhaled CO measurement in parts per million (ppm) was performed by having subjects exhale completely then inhale fully in open air, withhold their breath for 10 s, and then exhale completely into a portable CO monitor (Tabataba analyser-FIM medical, Villeurbanne 69625 France).
Statistical analysis
Missing Data : Seventy-four women (26 fertile and 48 infertile) out of 199 had missing observations, representing a total of 8.1% of the dataset. Consequently, only women with complete data were included in the analysis. Data of 51 infertile women and 74 fertile women were analyzed.
Baseline characteristics: Baseline characteristics of the women were described by fertility status (mean and standard deviation). The Wilcoxon test was used for comparing fertile versus infertile women.
Logistic regression: Associations between PI and fertility status were investigated using logistic regression models, first considering the following variables (i.e., age, BMI, waist circumference, fasting blood glucose, exhaled CO, AMH, BW, birth height, and PI). Waist circumference, BW, and birth height were not included in the models due to collinearity with BMI and PI, respectively. Akaike information criterion (AIC) was used to select the best-fitting model and removing nonsignificant variables afterward. The final regression model was adjusted for five variables: age, BMI, fasting blood glucose, exhaled CO, and PI. Odds ratios (ORs) and 95% confidence intervals (CIs) are reported.
Moreover, the association between PI and BW was investigated and the correlation coefficient was calculated. To compare the relative significance of the two parameters, logistic regression analysis was also performed using BW instead of PI.
R 3.5.1 software (https://www.r-project.org) was used for all statistical analyses. p < 0.05 was considered significant.
Results
Baseline characteristics
Characteristics of fertile and infertile women are described in Table 1. Infertile women BMI and waist circumference were significantly higher compared to fertile women. Infertile women were more often smoker (12.2% versus 7%) and they had higher exhaled CO levels. There was a correlation between exhaled CO levels and smoking status in both infertile and fertile women (infertile: r 2 = 0.52, fertile: r 2 = 0.58, all: r 2 = 0.54). Higher fasting blood glucose and lower triglycerides were observed in infertile women, but no difference could be observed concerning cholesterol (HDL and LDL), plasma AMH concentrations, or blood pressure. Moreover, BW was higher in infertile compared to fertile women whereas birth length did not differ between the two groups. Consequently, PI was significantly higher in infertile women compared to fertile women. Only patients born at term (gestational age was between 37 and 41 weeks of amenorrhea) were included in the study.
Table 1. Baseline characteristics of infertile and fertile women
Data are means ± standard deviations.
Associations between PI and fertility status
PI, anthropometric, metabolic factors, and age according to the fertility status are presented in Table 2. Increased PI at birth, increased BMI, increased glycemia, and high exhaled CO were identified as significant risk factors for infertility (Table 2). Glycemia seems to have a strong effect on the risk of infertility with an OR of 2.73 [1.23–6.07], p = 0.014. PI was associated to infertility with an OR of 1.27 [95% CI, 1.06–1.52], p = 0.009. PI and BW were correlated (correlation coefficient: 0.587) (Table 3). In addition, BW was also associated to infertility but the association (1.002 [95% CI, 1.00–1.003]; p = 0.023) was not as strong as with the PI.
Table 2. Factors associated with fertility and infertility (multivariate logistic regression)
OR, odds ratio; CI, confidence interval; BMI, body mass index.
Table 3. Correlation between birth weight, birth height, and ponderal index
Discussion
We have previously observed that infertile men have a significantly higher BW than fertile men and that BW is associated with semen parametersReference Faure, Dupont and Chavatte-Palmer16. In the present study, we also observed higher BW in infertile compared to fertile women. These results seem to be contradictory compared to other studies that have shown a negative impact for LBW on fertilityReference Ibanez, Potau and Ferrer10–Reference Borghese, Sibiude and Santulli12. However, none of the patients in the present study had a very LBW (<1500 g), and only one had a BW below 2000 g. Therefore, as hypothesized previouslyReference Nohr, Vaeth, Rasmussen, Ramlau-Hansen and Olsen9, in addition to low and very low BWs, a heavy BW may be a risk factor for the development of infertility in adulthood.
We observed that infertile women have a higher PI at birth than fertile women. Although PI may not be a better indicator of foetal morbidityReference Cooley, Donnelly and Walsh13 than BW, it reflects potential asymmetric growth due to a nonoptimal periconceptional environment as well as neonatal fat massReference Chen, Tint and Fortier19. Additionally, variations in PI have been associated with adverse long-term consequencesReference Crusell, Damm and Hansen20. The multivariate analyses performed here show that a high PI at birth is an infertility risk factor in adulthood. In addition, infertility correlates with increased BMI, glycaemia, and exhaled CO.
Fetal programming of metabolic diseases is a well-established concept known as DOHaD. This concept recognizes that an unfavorable maternal environment can lead to placental abnormalities, which may impact fetal development through various mechanisms such as hormonal imbalance, oxidative stress, and epigenetic changesReference Chavatte-Palmer, Rousseau-Ralliard, Tarrade and Menezo21, Reference Tarrade, Panchenko, Junien and Gabory22. In the long term, an increased risk of developing overweight or metabolic diseases has been observed in cases of intrauterine growth retardation or macrosomiaReference Hanson and Gluckman1. Thus, an increased risk of infertility may also be a consequence of an unfavorable gestational environment.
In this study, infertile women were recruited if they were part of a couple with unexplained infertility. Consequently, we did not observe any difference in AMH levels between fertile and infertile groups. In this population, we can assume that the long-term consequences of an inadequate in utero environment may not impact ovarian reserve but possibly oocyte quality. Nevertheless, developmental origin cannot be excluded from women presenting ovarian failure. In order to confirm this hypothesis, a new study with patients presenting ovarian failure would need to be carried out. Moreover, we observed worse anthropometric and metabolic parameters in infertile women. Weight gain, abdominal obesity, or metabolic disorders are associated with systemic inflammation and oxidative stress, which plays a critical role in female reproductive function and oocyte qualityReference Michalakis, Mintziori, Kaprara, Tarlatzis and Goulis23. Mechanisms are undoubtedly complex and multifactorial. Based on current knowledge, it is difficult to assess whether prenatal development has a direct effect on adult fertility or if it leads to the development of long-term metabolic disorders that will negatively impact fertility. Studies in sheep have shown that excess maternal nutrition during pregnancy affected offspring ovaries at the fetal stageReference Da Silva, Aitken, Rhind, Racey and Wallace4. Fewer follicles were observed in the ovary, demonstrating that a direct effect of the maternal environment cannot be excluded. Finally, it is difficult to explain why fertile women had slightly higher triglyceride levels than infertile women, but the average value remains below a pathological value.
Epigenetic modifications are widely involved in metabolic programming, but little is known about epigenetic programming of oocytes or ovariesReference Puttabyatappa and Padmanabhan24. Nevertheless, transgenerational programming through two or three generations may suggest epigenetic changes in oocytes are transferred to the offspringReference Miska and Ferguson-Smith25.
Many studies in animal model have shown that maternal overnutrition or undernutrition during gestation increases oocyte apoptosisReference Puttabyatappa and Padmanabhan24. Thus, in a rabbit model exposed in utero to a high-fat diet, we previously observed increased follicular atresia, suggesting apoptotic mechanisms during folliculogenesis. In this rabbit model, no difference was found for the primordial, primary, secondary, or tertiary follicle counts. Oxidative stress may be involved in this phenomenonReference Leveille, Tarrade and Dupont26.
Oocyte and follicular atresia may be early signs of ovarian aging. Indeed, the environment at birth may also be involved in the age at which menopause occurs. Perinatal exposure to famine has been associated with advanced menopauseReference Yarde, Broekmans and van der Pal-de Bruin27. Some studies have also related a high BWReference Tom, Cooper and Kuh7 or PI to early menopause (40–42 years)Reference Cresswell, Egger and Fall8. Indeed, a suboptimal perinatal environment also affects longevity and accelerates aging. Oxidative stress and reduced telomere length may be key mechanisms involved in this phenomenonReference Tarry-Adkins, Martin-Gronert, Chen, Cripps and Ozanne28, Reference Krisher29.
More epidemiologic and experimental animal studies are necessary to fully understand the mechanisms of this phenomenon.
This study presents both strengths and limitations. One of the strengths of the study is the recruitment of fertile and infertile women under the same conditions, which is rare. Furthermore, BW and birth length data were directly obtained from official childhood records, which limit self-reporting bias. Anthropometric assessments were performed by the same trained investigator using the same calibrated devices, thus minimizing observation bias. Limitations of this study include the small sample size. Moreover, BW, birth height, and PI are proxies of the in utero environment and do not represent all of the events occurring during pregnancy. Furthermore, no women born with a very LBW were recruited for the study; therefore, it was not possible to assess the consequences of very LBW on fertility. Finally, infertile couples (unlike fertile couples) in this study were selected by medical services, which could have introduced a selection bias. Indeed, fertile women were slightly older than infertile, which could be explained by the fact that they were recruited after the birth of their child and they were not usually included immediately after childbirth.
Furthermore, socioeconomic status and physical activity may impact fertility. Since these two variables may both impact BMI, we chose to use BMI for the logistic regression model. Nevertheless, no difference in physical activity between fertile and infertile women was observed in a study including a similar populationReference Foucaut, Faure and Julia30.
In order to explore further the role of in utero environmental conditions on fertility, retrospective or even prospective cohorts of individuals whose pregnancy history is well known and who would be monitored over the long term would be necessary.
Conclusion
This study is unique in that two groups of similar size and characteristics were recruited and assessed. The results confirm the importance of the intrauterine environment, not only for long-term metabolic health but also for fertility. We can consider the risk of infertility to be part of the adverse outcomes related to an inadequate periconceptional environment in addition to metabolic disorders, cancer, autoimmune diseases, and neurodegenerative diseasesReference Barouki, Gluckman, Grandjean, Hanson and Heindel31. These are additional arguments for setting up preventive programs and interventions to improve gestational health.
Acknowledgments
The authors want to acknowledge all the participants involved in the study.
Financial support
This study was supported by national biomedical research P071224 ALIFERT.
Ethical Standards
The ethics committee (“Comité de Protection des Personnes”) approved the ALIFERT study (national biomedical research P071224/AOM 08180:NEudra CT 2009-A00256-51/clinical trials NCT01093378). The study was conducted according to the protocol, to the law of December 20, 1988, as amended by Act 2004-806 of August 9, 2004, to the ethical principles established by the 18th World Medical Assembly, and to French Good Clinical Practice. Before starting the research, an authorization file (as defined in Article L 1123-12) was approved by Agence Française de Sécurité Sanitaire des Produits de Santé (AFSSAPS), the favorable opinion of the Committee for Protection of Persons of Ile de France was obtained, and the Commission Nationale de l’In-formatique et des Libertés (CNIL) declaration was performed. All patients signed consent forms.
Authors contribution
C.D. participated in the study conception and design, in patient’s recruitment, data acquisition, interpretation and analysis, and drafting of the manuscript. A.H., F.J., and P.C.P. participated in study design, performed statistical analysis, and participated in critical revision of the manuscript for intellectual content. C.F. participated in the study conception and design, in patient’s recruitment and data acquisition. S.C. participated in the study conception and design. R.L. participated in the study conception and design, interpretation of data, critical revision of the manuscript for intellectual content, and supervised the study. The collaborators of the ALIFERT collaborative group participated in the study design and were involved in patients’ recruitment.
