Preconception ovarian reserve and placenta-mediated pregnancy complications among infertile women

Abstract

Research question:

Are preconception ovarian reserve markers, such as Anti-Mullerian hormone and antral follicle count, associated with preeclampsia and placenta mediated pregnancy complications among women with unexplained infertility who conceive with superovulation?

Design:

This is a secondary analysis of women with unexplained infertility who had a singleton live birth after enrollment in the Analysis of Multiple Intrauterine Gestations after Ovarian Stimulation (AMIGOS) trial that randomized couples to superovulation with letrozole, clomiphene, or gonadotropins with insemination for up to 4 cycles

Results:

Compared to controls (N=156), women who developed preeclampsia (N=17) had lower Anti-Mullerian hormone levels (2.24±1.20 vs. 2.89±2.32, p=0.07) and lower antral follicle count (18±7.67 vs. 21±11.43, p=0.16); though these differences were not statistically significant. There was no relationship between Anti-Mullerian hormone (OR: 1.00, 95% CI: 0.76–1.25) or antral follicle count (OR: 0.98, 95% CI 0.93–1.04) with preeclampsia and between Anti-Mullerian hormone (OR: 1.00, 95% CI: 0.83–1.17) and antral follicle count (OR: 1.00, 95% CI: 0.97–1.04) with placenta medicated pregnancy complications after adjusting for age, BMI and race.

Conclusions:

Preconception ovarian reserve markers are not associated with preeclampsia and placenta mediated pregnancy complications among women with unexplained infertility who conceive with superovulation with insemination.

Introduction

Placenta-mediated pregnancy complications (PMPCs), such as preeclampsia, small for gestational age neonate and stillbirth, are risk factors for subsequent cardiovascular disease and cerebrovascular events^1,2. While shared cardiovascular risk factors, such as dyslipidemia, diabetes, hypertension and abdominal obesity, may account for the association, persistent preeclampsia-induced endothelial dysfunction may contribute to the pathogenesis as well.

Premature ovarian aging has been associated with preeclampsia. Women with diminished ovarian reserve (DOR), defined as poor response to ovarian hyperstimulation in in vitro fertilization (IVF) and women with premature ovarian failure who conceive through donor oocyte IVF are at increased risk for preeclampsia^3,4. Women with lower Anti-Mullerian hormone (AMH) levels in the first and third trimesters appear to be at increased risk of developing gestational hypertension and preeclampsia when compared to controls, particularly if women exhibited an AMH less than the 10^th centile^5,6. Of note, pregnancy is associated with a reduction in AMH levels and preconception AMH levels were not reported. Finally, women who have a history of preeclampsia exhibit significantly lower AMH levels a decade after the incident pregnancies than women who remained normotensive during pregnancy⁷. Collectively these data suggest ovarian aging may precede the onset of preeclampsia and possibly predict long-term consequences of preeclampsia.

Studies of preconception ovarian reserve markers, such as AMH and antral follicle count (AFC), and the risk of developing PMPCs are lacking. We sought to examine this relationship among women with infertility who conceived through superovulation with insemination. We hypothesize that lower ovarian reserve among infertile women will be associated with an increased risk of PMPCs.

Materials and Methods

This is a secondary analysis of data obtained from Analysis of Multiple Intrauterine Gestations after Ovarian Stimulation (AMIGOS), a multicenter randomized clinical trial comparing live-birth rate and multiple pregnancy rates after superovulation with letrozole, clomiphene or gonadotropins with insemination for couples with unexplained infertility. The trial was registered on ClinicalTrials.gov (#NCT01044862). The protocol was reviewed and approved by an Advisory Board and Data Safety Monitoring Committee appointed by the National Institutes of Health/National Institute of Child Health and Human Development. Each infertility center of the Reproductive Medicine Network obtained approval from their Institutional Review Board and all participants provided written informed consent. This research received exempt status from the University of Rochester Institutional Review Board.

The study design of AMIGOS, methods, inclusion and exclusion criteria have been described in detail in prior publications⁸. In brief, 900 women between ages 18 and 40 years of age with unexplained infertility were randomized to superovulation with letrozole, clomiphene or gonadotropins with insemination for up to 4 cycles. Women had regular menses (nine or more cycles per year), a normal uterine cavity with at least one patent fallopian tube, and a male partner with a semen specimen of at least 5 million sperm per milliliter. Demographics, reproductive history, medical history were obtained using standardized forms during the AMIGOS trial. Adverse events in pregnancy were obtained from review of maternal medical records and newborn weight was obtained from the birth record.

In this secondary analysis, we identified cases based on the diagnosis of preeclampsia, early onset preeclampsia (<34 weeks gestation) and a composite outcome of placenta-mediated pregnancy complications which included preeclampsia, intrauterine growth restriction, small for gestational age (SGA) neonate, placental abruption and stillbirth. SGA status of the neonate was assigned based on percentile birth weights⁹. Controls were identified as women who conceived and did not experience preeclampsia, intrauterine growth restrictions, small for gestational age neonate, placental abruption or stillbirth. We excluded subjects with multiple gestations and limited our analysis to singleton birth.

Ovarian reserve was assessed by pretreatment AMH and AFC which were collected in the early follicular phase prior to the start of letrozole, clomiphene or gonadotropins. AMH samples were batched and analyzed at the Ligand Assay and Analysis Core Laboratory at the University of Virginia. AMH was analyzed from samples stored at – 80°C, using the Beckman-Coulter Gen 2 (Webster, TX) ELISA assay with no pre-dilution. The sensitivity of the assay was to 0.25 ng/ml, with a range of 0.25–15 ng/mL. The inter-assay coefficient of variation was 7.0%, and the intra-assay coefficient of variation was 3.0%. Antral follicle count was performed on cycle day 3 and was assessed during transvaginal ultrasound (using a 5–9 mHz vaginal probe, most centers used 5 mHz) of both the right and left ovary, using standardized procedures as outlined previously⁸.

Descriptive statistics were used to compare demographics characteristics, medical history and pregnancy characteristics among cases with PMPCs and controls. The chi-squared test was used to compare independent sample proportions and the 2-sample t-test was used to compare independent sample means.

Multivariable logistic regression was used to evaluate the relationship between the probability of pre-eclampsia, early preeclampsia, or PMPC in logistic regression models that adjust for either AMH or AFC, and also adjust for age, BMI and race. In these models, AMH, AFC, age, and BMI were treated as continuous variables, and race was treated as a binary variable (white versus other). Additional models were also fit that did not include BMI, and an unadjusted model was also fit for each outcome /ovarian reserve marker combination. Odd ratios and their 95% confidence intervals were calculated, and the Hosmer-Lemeshow lack of fit test checked for model violations. Covariates associated with preeclampsia risk were pre-selected based on existing literature¹⁰. All P-values < 0.05 were considered statistically significant. Analyses were conducted using R version 3.3.2.

Results

A total of 174 women achieved singleton births. The demographic and medical characteristics of women with preeclampsia (N=17) and controls (N=156) are shown in Table 1. Eight women had early preeclampsia, and 39 women had PMPC. Among women with preeclampsia, a greater proportion were Black / African American (23.5%) than among controls (5.1%). Women with preeclampsia were much more likely to have chronic hypertension than controls (23.5% versus 3.2%, p = 0.002). Although not statistically significant, women who developed preeclampsia had somewhat lower levels of AMH (2.24±1.20 vs. 2.89±2.32, p=0.07) and also lower AFC (18±7.67 vs. 21±11.43, p=0.16) than controls.

Table 1.

Characteristics of women with preeclampsia with control women with singleton pregnancies

	Preeclampsia (N=17)	Controls (N=156)	P-value*
Age ± SD (years)	33.8±4.51	31.5±4.26	0.054
BMI (kg/m2), N (%)			0.297
<18.5	0 (0)	2 (1.3)
18.5–24.9	5 (29.4)	64 (41.3)
25–29.9	3 (17.7)	42 (27.1)
≥30	9 (52.9)	47 (30.3)
Race, N (%)			0.034
White	12 (70.6)	133 (85.3)
Black or African American	4 (23.5)	8 (5.1)
Asian	1 (5.9)	8 (5.2)
Mixed race	0	7 (4.5)
Ethnicity, N (%)			0.98
Hispanic or Latino	2 (11.8)	13 (8.3)
Not Hispanic or Latino	15 (88.3)	143 (91.7)
Nulliparous, N (%)	2 (33.3)	33 (47.1)	0.82
Chronic hypertension, N (%)	4 (23.5)	5 (3.2)	0.002
Anti-Mullerian hormone (ng/dl), mean ± SD	2.24±1.20	2.89±2.32	0.06
Antral follicle count, mean ± SD	18±7.67	21±11.43	0.16

Note – no patients had preexisting diabetes or thrombophilia. Percentages may not add up to a 100% because of rounding.

Abbreviations: BMI, body mass index

^*P-values for tests of differences for categorical variables were calculated from a chi-squared test, and for continuous variables were calculated from a 2-sample t-test.

There was no association between AMH and preeclampsia (adjusted OR: 1.00, 95% CI: 0.76–1.25, p=0.98) after adjusting for age, BMI and race (Table 2). Similarly, the associations between AMH and early preeclampsia (adjusted OR: 0.88, 95% CI: 0.50–1.31, p=0.62) or PMPCs (adjusted OR: 1.00, 95% CI: 0.83–1.17, p=0.96) were extremely weak. There were essentially no associations between AFC and preeclampsia (adjusted OR: 0.98, 95% CI; 0.93–1.04, p=0.56), early preeclampsia (adjusted OR: 1.01, 95% CI: 0.94–1.09, p=0.73) or PMPCs (adjusted OR: 1.00, 95% CI: 0.97–1.04, p=0.92) after adjusting for age, BMI and race (Table 3). In these models, BMI was a significant predictor of preeclampsia and early pre-eclampsia, but not PMPC, regardless of the ovarian reserve marker used in the model. In models that did not adjust for BMI, and in models that did not adjust for any additional covariates, neither AMH nor AFC were statistically significant predictors of any of the outcomes considered.

Table 2.

Adjusted odds ratios for outcomes in relation to AMH among women with a live birth (N=174).

Outcome	OR (95% CI)	p-value
Preeclampsia	1.00 (0.76–1.25)	0.98
Early preeclampsia*	0.88 (0.50–1.31)	0.62
Placenta mediated pregnancy complications**	1.00 (0.83–1.17)	0.96

^*Early preeclampsia is defined as diagnosis prior to 34-week gestation.

^*Placenta-mediated pregnancy complications is a composite outcome including pre-eclampsia, placental abruption, intrauterine growth restriction and small for gestational age.

All models adjusted for BMI, age and race (white versus other).

Abbreviations: AMH, Anti-Mullerian hormone, BMI, body mass index

Table 3.

Adjusted odds ratios for outcomes in relation to Antral Follicle Count (AFC) among women with a live birth (N=174).

Outcome	OR (95% CI)	p-value
Preeclampsia	0.98 (0.93–1.04)	0.56
Early preeclampsia*	1.01 (0.94–1.09)	0.73
Placenta mediated pregnancy complications**	1.00 (0.97–1.04)	0.92

^*Early preeclampsia is defined as diagnosis prior to 34-week gestation

^**Placenta-mediated pregnancy complications is a composite outcome including pre-eclampsia, placental abruption, intrauterine growth restriction and small for gestational age

All models adjusted for BMI, age and race (white versus other).

Abbreviations: BMI, body mass index

The Hosmer-Lemeshow lack of fit test did not show any model violations for the unadjusted models, or for the models that adjusted for age, BMI, and race. The only model for which the Hosmer-Lemeshow lack of fit test revealed a possible violation was the model for pre-eclampsia with AMH (but not with AFC) that adjusted for age and race, but not BMI. The nature of this lack of fit suggested that there were more pre-eclampsia cases at both the lowest and highest predicted values, which could have been driven by the small sample size, or by lack of adjustment for BMI.

Discussion

The relationship between lower AMH and PMPCs is of interest as these findings may represent early manifestations of vascular disease. While there was some indication of lower AMH and AFC in women with preeclampsia compared to controls, the relationship between preconception ovarian reserve markers and PMPCs was not statistically significant once we controlled for covariates. Whereas studies that report an association between low AMH and preeclampsia evaluated AMH in pregnancy or decades later, we found no relationship when analyzing preconception AMH and AFC and PMPC among women with infertility who conceived with superovulation and insemination.

Our finding adds to studies that found no relationship between ovarian reserve and PMPCs, but our study differs based on the type and timing of the ovarian reserve marker that were analyzed. For example, Van Disseldorp et al. 2010 performed a case control study of women with diminished ovarian reserve based on poor response to ovarian hyperstimulation, defined as <3 oocytes retrieved, and women with normal response, defined as 8–12 oocyte retrieved, and found no difference in the risk of gestational hypertension and preeclampsia among those that conceived, questioning the claim of a vascular etiology linking these two outcomes¹¹. Birdir et al focused on AMH as a marker of ovarian reserve and found that AMH values at 11–13 weeks’ gestation were higher among women with preeclampsia than controls, though the median multiple of the expected median was no different between the groups, leading them to conclude that first trimester AMH levels are not an effective early predictor for preeclampsia¹². Finally, Bhide et al. 2017 attempted to replicate Yarde’s finding that women who experience preeclampsia have lower AMH levels after pregnancy and did not see a difference in AMH levels between women with severe preeclampsia and women who had an uncomplicated pregnancy, though the size of the study was smaller and the duration of follow up was shorter^7,13.

A relationship between corpus luteum function and hypertensive disorders in women with infertility has become apparent through compelling observation data¹⁴. The corpus luteum secretes estradiol, progesterone, relaxin and other factor during the luteal phase to support implantation. Programmed or medicated frozen embryo transfer protocols utilize a GnRH agonist and/or estradiol to suppress development of a dominant follicle and corpus luteum. Implantation is supported through replacement of estradiol and progesterone, but not relaxin or other factors. Programmed frozen embryo transfer protocols have been strongly associated with preeclampsia when compared to natural cycle frozen embryo transfer protocols, which depend on a corpus luteum for luteal support¹⁵. Conception in the absence of a corpus luteum is associated with a lack of the natural decline in mean arterial blood pressure (MAP), a lower reactive hyperemia index and a lower number of circulating endothelial progenitor cells. In addition, conception in the presence of 3 or more corpus luteum, which would be typical an in vitro fertilization cycle, is associated with a similar lack of the natural decline in MAP and higher baseline pulse wave amplitude. The luteal phase dysfunction that is observed with DOR has been hypothesized as a mechanism of action which could account for the association of DOR and hypertensive disorders of pregnancy. In this analysis, our cases and controls underwent ovulation induction with letrozole, clomiphene or gonaodtropins which would ensure that 1 or more corpora lutea were present in the luteal phase. We limited our analysis to singleton pregnancies which is often associated with a single corpus luteum.

The strength of this study is that the cases and controls were identified from the AMIGOS dataset, which is a well-characterized cohort of couples with unexplained infertility¹⁶. Women with infertility and PCOS were excluded from AMIGOS, which is important to highlight as women with PCOS are at increased risk for preeclampsia and PCOS is associated with elevated AMH^17,18. No other studies that have evaluated AMH and preeclampsia have clearly excluded women with PCOS from their analyses or controlled for this diagnosis. Another strength of our study is that AMH was collected during the preconception period but analyzed using the same assay for all subjects eliminating concerns regarding the inter assay variability that has challenged the validity of AMH research¹⁹. We were also able to analyze preconception antral follicle count as another measure of ovarian reserve, which has not been previously reported. Finally, we were able to expand our outcome to include not just preeclampsia, but also early preeclampsia and additional PMPCs including IUGR, SGA and abruption.

A limitation of this study is that women with a day 3 FSH level greater than 12 IU/mL were excluded from the AMIGOS study, though women were not excluded based on AMH or AFC. There is potential that the findings would differ if women with elevated FSH were included in the cohort. Another limitation of the study is the small number of subjects with preeclampsia or early preeclampsia in the cohort. The prevalence of preeclampsia ranges from 2–8% of pregnancies and the prevalence of preeclampsia in this cohort was 10%. We observed significant differences in notable risk factors such as chronic hypertension and Black race observed between cases and controls and that BMI was a significant predictor of preeclampsia and early preeclampsia in the adjusted models²⁰. Given the narrow confidence intervals we observed for AMH and AFC and the outcome. We were unable to analyze AMH levels in pregnancy as this data was not available.

In conclusion, preconception AMH and AFC are not association with the risk of preeclampsia or placenta mediated pregnancy complications in women with infertility undergoing ovulation induction. These findings may be reassuring to women with diminished ovarian reserve and infertility who conceive and for providers who care for this population. We feel that the findings of this study, using a very well validated cohort, is an important contribution to the conflicting literature regarding the relationship between ovarian reserve and PMPCs.

Highlights:

• Lower Anti-Mullerian hormone in pregnancy is associated with preeclampsia.

• Preconception ovarian reserve markers in women with unexplained infertility who conceived a singleton live birth with superovulation and insemination were evaluated for a relationship with placenta mediated pregnancy complications.

• There was no relationship between Anti-Mullerian hormone or antral follicle count with preeclampsia or placenta medicated pregnancy complications.