Duration, Recency, and Type of Hormonal Contraceptive Use and Anti-Müllerian Hormone Concentrations

Abstract

Objective:

To assess whether duration, recency or type of hormonal contraceptive used is associated with anti-Müllerian hormone (AMH) concentration given that the existing literature regarding the association between hormonal contraceptive use and AMH concentrations is inconsistent.

Design:

Cross-sectional study.

Setting:

Baseline data from the Study of the Environment, Lifestyle and Fibroids Study (SELF), a five-year longitudinal study of African American women.

Patients:

1,643 African American women aged 23–35 years at time of blood draw (2010–2012).

Interventions:

None.

Main Outcome Measures:

Serum AMH measured using an ultrasensitive enzyme linked immunosorbent assay (ELISA). Linear regression models were used to estimate percent differences in mean AMH concentrations and 95% confidence intervals (CIs) by hormonal contraceptive use, with adjustment for potential confounders.

Results:

In multivariable-adjusted analyses, current users of hormonal contraceptives had 25.2% lower mean AMH concentrations than never users (95% CI −35.3%, −13.6%). AMH concentrations showed little difference among previous users of hormonal contraceptives compared with never users (−4.4%, 95% CI −16.3%, 9.0%). AMH concentrations were not appreciably associated with cumulative duration of use among previous users or time since last use among non-current users. Current users of combined oral contraceptives (−24.0%, 95% CI −36.6%, −8.9%), vaginal ring (−64.8%, 95% CI −75.4%, −49.6%), and depot medroxyprogesterone acetate (−26.7%, 95% CI −41.0%, −8.9%) had lower mean AMH concentrations than never users.

Conclusion:

The present data suggest that AMH levels are significantly lower among current users of most forms of hormonal contraceptives, but that the suppressive effect of hormonal contraceptives on AMH levels is reversible.

Capsule:

Anti-Müllerian hormone (AMH) levels were significantly lower among women currently using most forms of hormonal contraceptives, but the suppressive effect of hormonal contraceptives on AMH levels appeared reversible.

INTRODUCTION

Ovarian reserve refers to a woman’s reproductive potential as a function of the number of her remaining oocytes. Though there are several known serum biomarkers of ovarian reserve (follicle stimulating hormone, estradiol, and inhibin B), growing evidence over the last decade has suggested that anti-Müllerian hormone (AMH) more accurately reflects the follicular pool (1–4). AMH production begins in the granulosa cells of primary follicles, though the majority of circulating AMH is derived from pre-antral and early antral follicles (5, 6). As such, serum AMH is proportionally related to the size of the primordial follicle pool and represents ovarian reserve (2).

Relative to conventional biomarkers, AMH concentrations have low inter- and intra-cycle variability (7, 8). Based on these findings, AMH is increasingly used for diagnostic and prognostic purposes (9–13). To aid in the clinical interpretation of AMH concentrations, several nomograms have been published describing the range of normal AMH levels in various populations (14–19). The ‘utility of these nomograms has been questioned, however, as AMH concentrations can vary with numerous intrinsic, iatrogenic, and environmental factors (20–22).

One factor that may affect AMH concentration is hormonal contraceptive use. The effect of hormonal contraceptive use on AMH concentration has broad clinical implications given that more than 60% of reproductive-age women currently use contraceptives, the majority of which are hormonal methods (23). While combined oral contraceptives (COC) have traditionally been the most commonly used hormonal contraceptives, other hormonal methods (including depot medroxyprogesterone acetate [DMPA], the patch, hormonal intrauterine device [LNG-IUD], and ring) have increased in popularity in recent years (24).

Existing literature regarding the association between hormonal contraceptive use and AMH concentration is inconsistent. While some studies suggest hormonal contraceptive use lowers AMH concentration (20, 25–34), others have found no effect (35–39). The larger studies to date concluded that women actively using hormonal contraceptives have significantly lower AMH levels (20, 25, 27). A systematic review reported that AMH concentration does not seem to be altered in those using hormonal contraceptives during the first six months, but there does appear to be a decline in AMH concentration in long term users. This reduction in AMH concentration in long term users appears reversible after discontinuation of hormonal contraceptives (40). As previous studies have been largely limited to women using COC, it is unclear whether these findings can be generalized to other types of hormonal contraceptives. Few studies have investigated the effect of specific types of hormonal contraceptives on AMH concentrations (27, 29–31, 36, 39), and many of these studies have small sample sizes. The largest study examining the relationship of various types of hormonal contraceptives and AMH concentration concluded that AMH levels measured concurrent with use were lower in users of some, but not all, of the types of contraceptives studied (25). Users of COC, the progestin-only pill, and the levonorgestrel-releasing intrauterine system had lower AMH levels (25).

It is also unclear whether AMH levels are permanently influenced by hormonal contraceptive use. Although a prospective study of long-term users of COCs reported that after discontinuation, AMH concentration increased for the subsequent two months (41). To our knowledge, no study has investigated the association of current and former use of hormonal contraceptives including long-acting reversible contraceptives (LARCs) and AMH concentration. There is also a paucity of data in the literature on whether the age of initiation of use or cumulative use influences AMH concentration. To address these deficits in the literature, we evaluated the association between AMH concentration and history, type, duration and recency of hormonal contraceptive use among a large population of reproductive-aged African American women.

MATERIALS AND METHODS

Study participants

We analyzed baseline data from the Study of Environment, Lifestyle, and Fibroids (SELF). From November 2010 through December 2012, 1,693 premenopausal African American women aged 23–34 were recruited to participate. Exclusion criteria included: hysterectomy, a previous clinical diagnosis of uterine fibroids, use of medication to treat an autoimmune disorder (including: multiple sclerosis, Grave’s disease, scleroderma, lupus, or Sjogren’s), or use of chemotherapy or radiation treatment for a previous cancer diagnosis. Details about the study design, methods, and recruitment have been described previously (42) and additional information is included as supplemental material. The study was approved by the institutional review boards of the participating institutions.

Assays

Women were recruited from 2010–2012. At the time of enrollment, up to 55 mL of blood was drawn from each participant. Until analysis was performed serum was stored at −80° C. ln 2018, frozen samples were shipped to Ansh labs, (Webster, Texas, USA) where AMH assays were performed using the picoAMH assay which is an ultrasensitive enzyme linked immunosorbent assay (ELISA). AMH values are presented in concentration of ng/mL. The lower limit of detection of the test was 1.3 pg/mL and the limit of blank was 0.5 pg/mL. The measuring range was 6.0 to 1,150 pg/mL (0.043 pmol/L to 8.21 pmol/L). If necessary, specimens were diluted up to 20-fold prior to assay, thus extending the measuring range up to 23,000 pg/mL (23 ng/mL, 164 pmol/L). Intra-assay and inter-assay coefficients of variation were <5%.

Hormonal Contraceptive Variables

We restricted this analysis to include women who were using the following types of hormonal contraceptives: combined oral contraceptives (COC), etonogestrel/ethinyl estradiol vaginal ring (ring), depot medroxyprogesterone acetate (DMPA), etonogestrel implant (implant), ethinyl estradiol/norelgestromin transdermal patch (patch), and levonorgestrel intrauterine device (LNG-IUD). Due to limited information on details of use, progestin-only pill (POP) users were not analyzed as a separate group of hormonal contraceptives. The association between AMH concentration and use of hormonal contraceptives, age at initiation, cumulative duration of use, recency of use and type of hormonal contraceptive was examined. Data on hormonal contraceptive use were self-reported by participants.

Age at first use of each hormonal contraceptive was reported at interview. For current and previous users of hormonal contraceptives, cumulative duration of use before enrollment was reported for each method separately in years and months, with the exception of COCs. Cumulative duration of COCs for current and previous users was calculated by first subtracting the age at which the participant reported initiating COCs from her age at enrollment (current user) or the age at which she discontinued use. This value was then weighted based on the participant’s response to how much of the time between those ages they used COCs (very little of that time, less than half of that time, about half of that time, more than half of that time, most of that time or the whole time). Among former users of hormonal contraceptives we calculated the duration of time since last hormonal contraceptive use (recency) by subtracting the age at last hormonal contraceptive use from the participant’s current age. Type-specific duration of use captured the entire history of use and was not designed to capture the most recent contiguous period of use. Total cumulative duration of all hormonal contraceptive use was calculated for current and former users of any of the queried hormonal contraceptive types by adding the individual durations of each type.

Covariates

We controlled for both age and a quadratic term for age because of the strong non-linear association between age and AMH levels. We also controlled for BMI (calculated from height and weight data measured at the clinic by trained study staff), and self-reported history of abnormal menstrual bleeding, any thyroid condition, and seeking care for difficulty conceiving as these factors were found to be associated with AMH levels in previous analyses of this cohort (22). Although menstrual cycle length was also found to be significantly associated with AMH levels in the previous analyses, this variable could not be included in this multivariable model because cycle length can only be assessed accurately off hormonal contraceptives. Therefore, we controlled for a self-reported diagnosis of PCOS given the association with cycle length. Given that PCOS is often under-ascertained among women, we additionally performed sensitivity analyses that controlled for reported menstrual cycle length when participants were aged 18–22 and off hormonal contraceptives. We reasoned that long menstrual cycles during this age range may serve as a proxy for PCOS which is often undiagnosed (43). Although we did not find an association between cigarette smoking and AMH levels in this cohort previously (44), given that other studies have demonstrated a relationship between cigarette smoking and both AMH and earlier time to menopause (20, 45, 46) we also conducted sensitivity analyses that controlled for current or previous cigarette smoking.

Statistical Analysis

We assessed the distribution of AMH levels and the explanatory variables of interest (use of hormonal contraceptives, cumulative duration of use, age at initiation, recency of use and type of hormonal contraceptives used). We calculated means or medians (with standard deviations [SD] or interquartile ranges [IQR]) for continuous variables, and proportions for categorical variables. As AMH levels were not normally distributed, for sample description we compared median AMH levels using Dwass, Steel, Critchlow-Fligner (DSCF) multiple comparison analysis based on two-sample Wilcoxon rank-sum tests.

We used a multivariable linear regression model to estimate the percent difference in mean AMH levels (β) and 95% confidence intervals (CI) for hormonal contraceptive users vs non-users. The covariates included in the multivariable model included age, age², BMI, and self-reported history of abnormal menstrual bleeding, any thyroid condition, seeking care for difficulty conceiving or PCOS. We investigated hormonal contraceptives as a group (COC, ring, DMPA, implant, patch and LNG-IUD together). We also examined individual types of contraceptives separately if we had adequate sample size (COC, ring, DMPA and LNG-IUD) to determine differences in AMH levels between current and never users. To further investigate the association between AMH levels and duration of use and recency of use in current and previous users where the sample size was large enough, we also examined users of COC, DMPA, and LNG-IUD separately. For all analyses, the AMH concentrations were log-transformed to account for the non-normality.

For the full analysis that grouped all hormonal contraceptive types, all participants were included in the analysis. For the subgroup analyses by type (COC, DMPA, and LNG-IUD), participants were excluded if they were pregnant within 3 months of the baseline visit (n=29) or had used a hormonal contraceptive other than the type of interest in the analysis within 12 months of the baseline visit and within 24 months for DMPA use (COC Analysis: n=465 excluded; DMPA analysis: n=517 excluded; LNG-IUD analysis: n=588 excluded). The COC subgroup analysis further excluded participants who had ever used POPs (n=76).

All analyses were carried out using Statistical Analysis Software version 9.4 (SAS Institute, Cary, NC).

RESULTS

Baseline Characteristics

1,693 African-American women were enrolled in SELF, and serum AMH was available for 1,643 individuals. Demographic details for this cohort of women are included in Table 1. The mean age of participants (±SD) considered in this analysis was 29.2 ± 3.4 years. The median AMH concentration was 4.07 (IQR 2.29–6.70). 460 participants (28.0%) were currently using some form of a hormonal contraceptive and a further 951 (57.9%) endorsed a history of previous hormonal contraceptive use.

Table 1:

Demographic Data for Study Participants (n=1643)

	Hormonal contraceptive usea
	Current (n=460)	Former (n=951)	Never (n=232)
Age at enrollment, y (mean ± SD; range)b	28.8 ± 3.3; 23–35	29.6 ± 3.5; 23–35	28.4 ± 3.4; 23–35
Body mass index, kg/m2 (mean ± SD; range)	32.2 ± 8.6; 17.8–68.3	34.1 ± 9.7; 15.9–79.4	35.0 ± 10.9; 17.5–72.9
AMH, ng/mL (median, IQR)	3.56 (3.8)	4.23 (4.8)	4.71(4.3)
AMH, ng/mL (range)	<0.002–26.0	<0.002–55.7	0.04–22.2
Education (%)
High school/GED or less	15.7	23.2	29.7
Some college, associate’s, or technical degree	50.7	50.9	44.8
Bachelor’s or graduate degree	33.7	25.8	25.4
Income (%)
Less than $20,000	37.6	48.1	52.6
$20,000–50,000	40.4	36.3	29.7
More than $50,000	21.7	14.8	16.4
Current smoker (%)	11.5	22.0	23.3
History of polycystic ovarian syndrome (%)	2.2	4.1	1.3
History of abnormal menstrual bleeding (%)	9.6	12.5	10.3
History of thyroid condition (%)	2.4	3.5	1.7
History of seeking care for difficulty conceiving (%)	2.6	7.5	5.6
History of previous birth (%)	67.6	64.1	34.9
Number of pregnancies (mean ± SD; range)c	2.9 ± 1.8; 1–15	3.3 ± 2.1; 1–13	2.5 ± 1.9; 1–13

SD=standard deviation, IQR=interquartile range

^aIncludes users of COC, ring, DMPA, implant, patch and LNG-IUD

^bWomen ages 23–34 were recruited, but some women had turned 35 by the time that all baseline activities and enrollment were completed.

^cAmong gravid women (n=356 current users, n=738 former users, n=112 never users)

AMH and Hormonal Contraceptive Status

Median AMH levels among current hormonal contraceptive users (3.56, IQR 2.03–5.81) was lower than those who had previously used hormonal contraceptives (4.23, IQR 2.30–7.13; p<.001), and those who had never used hormonal contraceptives (4.71, IQR 2.92–7.19; p<.001) (Figure 1). As shown in Table 2, multivariable adjusted analysis demonstrated that AMH levels in current users of hormonal contraceptives were 25.2% lower than AMH levels in those who had never used hormonal contraceptives. AMH levels were not significantly different comparing previous with never users of hormonal contraceptives.

Figure 1:

Box plot demonstrating median AMH concentrations of current hormonal contraceptive users compared to former and never hormonal contraceptive users.

* denotes statistically significant difference in AMH compared to other groups.

Table 2:

Multivariable Adjusted Associations of AMH with Hormonal Contraceptive Use^a, Cumulative Duration of Use, Age at Start, and Time Since Last Use Compared to Never Users^b

	N	Estimate of % change in AMH (95% CI)
Status of hormonal contraceptive use
Never	232	Reference
Former	951	−4.4% (−16.3%, 9.0%)
Current	460	−25.2% (−35.3%, −13.6%)
Total duration of hormonal contraceptive use (ever users)
Current and cumulative use <1 year	20	−33.5% (−56.1%, 0.8%)
Current and cumulative use 1–3 years	92	−21.1% (−36.9%, −1.4%)
Current and cumulative use ≥4 years	328	−26.9% (−37.4%, −14.7%)
Former and cumulative use <1 year	183	−2.2% (−18.1%, 16.8%)
Former and cumulative use 1–3 years	354	−3.4% (−17.0%, 12.5%)
Former and cumulative use ≥4 years	414	−6.5% (−19.5%, 8.6%)
Age at start of hormonal contraceptive (ever users)
<15 years	145	−12.2% (−27.5%, 6.3%)
15–20 years	968	−11.3% (−22.3%, 1.3%)
21–25 years	236	−14.0% (−27.3%, 1.6%)
≥26 years	47	−19.7% (−39.9%, 7.2%)
Time since last use of hormonal contraceptive (ever users)
Current	460	−26.1% (−36.2%, −14.5%)
Within last year	124	−5.1% (−22.2%, 15.8%)
> 1 year ago	827	−4.4% (−16.4%, 9.4%)

^aIncludes users of COC, ring, DMPA, implant, patch and LNG-IUD

^bThe reference group is “never users” (n= 232)

The multivariable model adjusted for age, age², BMI, and reported histories of PCOS, abnormal menstrual bleeding, thyroid condition, and seeking care for difficulty conceiving.

Cumulative Duration, Age at Start, and Recency of Using Hormonal Contraceptive Use (Combining Use of COC, Ring, DMPA, Implant, Patch and LNG-IUD)

The mean lifetime cumulative duration of hormonal contraceptive use among ever users was 61.0 months (SD 49.58, range 0–236.6) and among current hormonal contraceptive users was 62.0 months (SD 49.6; range 1–205 months). Within current and previous users, there was no appreciable difference by cumulative duration of use. Among participants who ever used hormonal contraceptives, the mean age at first use was 18.1 years (SD 3.34, range 8–33 years). As shown in Table 2, compared to never users, AMH levels did not show differences by age of initiation among ever users. The mean amount of time since last use of hormonal contraceptives in those who ever used was 42.8 months (SD 51.07, range 0–251). As shown in Table 2, AMH levels were lower only among current users of hormonal contraceptives (−26.1%, 95% CI −36.2%, −14.5%). Those who last used within the year or those who last used more than one year prior did not have meaningfully lower AMH concentrations.

AMH and Hormonal Contraceptive Type

When type of hormonal contraceptive was assessed among current hormonal contraceptive users, the most commonly used types were COC (39%), LNG-IUD (25%), and DMPA (22%). Median AMH levels among users of the various types of hormonal contraceptives compared to those who never used hormonal contraceptives are shown in Table 3. Current use of COC, vaginal ring, and DMPA were inversely associated with AMH levels in the multivariable analysis. Compared with never users of hormonal contraceptives, AMH levels were 24.0% lower in COC users. The dose of ethinyl estradiol (EE) in the COC was not associated with AMH levels. When the lowest dose of EE (10–20 mcg) was compared to 25–30 mcg and 35 mcg of EE, there was no difference in AMH concentration. AMH levels were 64.8% lower in vaginal ring users and 26.7% lower in DMPA users compared to never users of hormonal contraceptives. There was a small, non-significant reduction in AMH levels for current LNG-IUD users.

Table 3:

Multivariable Adjusted Associations of Hormonal Contraceptive Type with AMH Among Current Users Compared to Never Users

	N	AMH, ng/mL Median (IQR)	Estimate of % change in AMH (95% CI)
Never use	232	4.71 (2.92, 7.19)	Reference
COC	180	3.54 (2.01, 5.97)	−24.0% (−36.6%, −8.9%)
Ring	28	1.89 (1.38, 3.91)	−64.8% (−75.4%, −49.6%)
DMPA	103	3.23 (1.92, 5.72)	−26.7% (−41.0%, −8.9%)
LNG-IUD	116	3.89 (2.46, 6.15)	−8.7% (−25.7%, 12.2%)

COC=combined oral contraceptives, DMPA=depot medroxyprogesterone acetate, LNG-IUD=levonorgestrel IUD The multivariable model adjusted for age, age², BMI, and reported histories of PCOS, abnormal menstrual bleeding, thyroid condition, and seeking care for difficulty conceiving

The association between AMH levels and lifetime cumulative duration of use in current and previous users was examined by hormonal contraceptive type. Participants who were current users of COC (−25.8%, 95% CI −37.6%, −11.7%) or DMPA (−34.4%, 95% CI −47.2%, −18.5%) with cumulative use of one year or more had lower AMH levels than never users. There were very few current users of COC or DMPA with less than a year of cumulative use (n=10 and n=8 respectively). While AMH concentrations were lower for current COC users of less than one year of cumulative use compared to never users the estimate was quite imprecise (−36.8%, 95% CI −63.4%, 9.0%). AMH levels were not significantly lower in previous COC or DMPA users compared to never users, regardless of the duration of COC or DMPA use. There was a trend toward lower AMH levels in those who were current LNG-IUD users for one or more years (−13.8%, 95% CI −29.0, 4.6%), but there was no significant association between AMH levels and duration of use in current or former LNG-IUD users when compared to never users, regardless of length of use.

The association between AMH levels and time since last use was also examined by hormonal contraceptive type. In previous COC users, including those who had used within the last year, there was no significant association between AMH concentration and time since last use when compared to never users. Although AMH levels were not different for previous users of DMPA as a group compared to never users, there were lower AMH levels for those who had used DMPA within the last year (−15.6%, 95% CI −39.3%, 17.4%), or one to two years ago (−12.6%, 95% CI −31.9%, 12.1%). Last use of DMPA more than two years ago was not appreciably associated with current AMH concentrations. Compared to never users, AMH levels were not significantly lower in those who previously used LNG-IUD regardless of recency of use.

Sensitivity Analyses

Sensitivity analyses in which we further controlled for cigarette smoking (current, previous, never) or usual menstrual cycle length at age 18–22 years among non-users of hormonal contraceptives showed largely similar results.

DISCUSSION

In this large cohort study, AMH levels were lower among the group of current hormonal contraceptives users, while past use was not associated with AMH levels. Neither cumulative duration of hormonal contraceptive use among previous users nor recent cessation of hormonal contraceptive use were appreciably associated with AMH concentrations. Among past users of hormonal contraceptives, AMH levels appear to rebound within a year of last use. Type of hormonal contraceptive used appears to play a role in the observed association. Current use of COC, ring, and DMPA was inversely associated with AMH levels. Together our findings suggest that current users of many types of hormonal contraceptives have lower AMH levels, and this effect is reversible. AMH levels in previous COC users may return to baseline more quickly after ceasing use than AMH levels in previous DMPA users.

The present findings contribute to a conflicting body of literature regarding the association between hormonal contraceptives and AMH concentration. Although our findings concur with three large studies (20, 25, 27) that demonstrated reductions in AMH levels with hormonal contraceptive use, several other studies have reported that AMH concentrations are not influenced by current hormonal contraceptive use (35–39). One study showed no difference in the AMH levels of participants who had been using COC for more than a year compared with age-matched women who had not used hormonal contraceptives within the last year (35). Another study found little appreciable change in AMH levels after six cycles of COC use (37). Two other studies of short term hormonal contraceptive use of up to two cycles also found no association with AMH levels. (38, 39). Notably, these were relatively small studies (20–45 women) and this may have contributed to findings that were discordant from our study. Further, in two of these studies the women used COC for a short duration which may not have been long enough to lead to a decrease in AMH levels.

Many studies that have reported lower AMH levels among hormonal contraceptive users have solely considered women using COC (20, 26, 28, 32, 33). While several studies have included individuals using other types of hormonal contraceptives along with COC, (27, 29, 31) studies that have specifically examined the relationship between AMH concentrations and individual types of hormonal contraceptives other than COCs are limited. In one of the largest studies examining multiple types of contraceptives, AMH levels were 17.1% lower in 90 IUD users (95% CI −31.4, 0.002), but this decline was not statistically significant. In the same study there was not a significant decrease in AMH levels in 16 ring users (−12.2%, 95% CI −42.7, 34.3) (25). Another study reported significantly lower AMH levels in COC, patch, and ring users (n=18 in each group) (30). Conversely, a study that included users of COC (n=23), IUD (n=20), and DMPA (n=20) concluded that AMH levels were unaffected by hormonal contraceptive use. Notably, although there was not a statistically significant decline in AMH levels, there was a relatively large difference in pre- and post-treatment AMH levels in the COC group (27.2 pmol/L vs 17.1 pmol/L) (36). Further, the analyses were not age adjusted and the median age for DMPA users was 37 years, making a comparison to our data challenging. In our study AMH levels of current LNG-IUD users were not significantly lower than AMH levels of never users. While there was a trend toward lower AMH levels (−13.4%) in those who were currently using the LNG-IUD for one year or more, the association with AMH concentrations was not significant. Further investigation is warranted to examination the association between AMH levels and LNG-IUD use.

To our knowledge, this is the first study to demonstrate that DMPA use is inversely associated with AMH levels. Although only current users of DMPA had significantly lower AMH levels, AMH levels were lower in more recent past users of DMPA which suggests that AMH levels may take time to recover after cessation of DMPA use. We hypothesize that the association observed among users of DMPA (a high-dose progestin-only injection administered every three months) is best explained by previous findings that suggested luteinization has an adverse effect on AMH production by granulosa cells (47). Conversely, the effect of the combined hormonal contraceptive methods on AMH concentrations is likely the result of down-regulation of the hypothalamic-pituitary-ovarian (HPO) axis, resulting in decreased production of FSH and LH, and altered antral follicle development (26). Our results demonstrate significantly lower AMH levels in current DMPA users with cumulative use of more than one year. However, given our small numbers for current DMPA users with cumulative use less than one year, we are unable to determine if short term use of DMPA is sufficient to result in lower AMH concentrations. A larger sample of women using DMPA for less than one year is needed.

AMH levels were lower across all durations of use categories for current hormonal contraceptive use suggesting that even short-term use of hormonal contraceptives may lower AMH concentrations. Previous users of hormonal contraceptives did not have AMH levels that were materially different from never users, nor was length of time using hormonal contraceptives associated with AMH levels in previous users. These findings suggest that the decline in AMH levels due to hormonal contraceptive use is reversible within one year of cessation of use and agree with several previous studies. In a study of 68 women with a history of long term COC, there was a 53% improvement in AMH after discontinuation of COC and AMH levels increased from baseline until two months after COC were discontinued (41). Further supporting the notion that AMH concentration recovers after cessation of hormonal contraceptives, in women who were off of hormonal contraceptives for one year but that had previously used for ten or more years, AMH levels were similar to those who never used hormonal contraceptives(48). The reversible nature of the observed reductions in AMH levels, demonstrated in our study as well as others (20, 33), seems to indicate the decline in AMH concentration associated with hormonal contraceptive use does not reflect a change in ovarian reserve. More likely, the reduction in AMH levels among current hormonal contraceptive users represents alterations in follicular development resulting from down-regulation of the HPO axis (26).

There are several strengths of this study. First, this study examines a large population with access to and wide uptake of a number of different forms of hormonal contraceptives that may have been utilized less in older aged cohorts. Further, over 1,600 study participants were recruited directly from the community, rather than from fertility clinics or other medical settings. In addition, this is the first study to evaluate the relationship between use of hormonal contraceptives and AMH levels in a population of young African-American women.

One study limitation is that because only African American women were studied, the generalizability of these findings may be limited, as several studies have suggested racial/ethnic differences in AMH levels (21, 49–52). However we would not expect hormonal contraceptives to influence AMH concentrations differently by race or ethnicity, although confirmation of this requires further investigation. Additionally, the smaller sample sizes of some of the individual types of hormonal contraceptives (patch n=5, implant n=13, and ring n=28) precluded evaluating these methods individually. Sample size may also be considered in the interpretation of the more detailed analyses of the associations between AMH levels and COC, DMPA and LNG-IUD use given the small number of users in some of the subgroups. Another study limitation is that the duration variable was calculated using cumulative duration, which could alter results in those who had breaks in their use. An additional limitation is that the samples were stored for 6–8 years before the assays were run and there are limited data on the reliability of AMH levels after long-term storage of serum. The samples were not run in duplicate due to a limited budget. Lastly, all variables related to contraceptive use were self-reported and there was no feasible way to validate participants’ hormonal contraceptive use using pharmacy or medical records. Given the in-depth nature of the survey questions (participants answered up to 41 questions on current and previous contraceptive use), and research documenting generally high accuracy of self-reported data on oral contraceptive use relative to pharmacy or medical records (53–55), bias due to misclassification is likely to be small. Furthermore, the prevalence of hormonal contraceptive use in the SELF cohort (28%) is highly consistent with national data on reproductive-aged women (56). We would expect any misclassification of current hormonal contraceptive use to be non-differential, which would bias the results towards the null.

CONCLUSIONS

The present data suggest that AMH levels are significantly lower among women currently using most forms of hormonal contraceptives, but that the suppressive effect of hormonal contraceptives on AMH levels is reversible. Consequently, when measured as a marker of ovarian reserve, AMH should be interpreted with caution in women using hormonal contraceptives. Although this study demonstrates that previous hormonal contraceptive users ultimately achieve similar AMH levels to those who have never used, the precise duration of time required for AMH levels to return to normal after discontinuation of hormonal contraceptives requires further investigation.