Laparoscopically confirmed endometriosis and anti-Müllerian hormone levels: Findings from the Nurses' Health Study II

The NHSII is a prospective cohort of 116,429 women that began in 1989 when female registered nurses between the ages of 25–42 years returned a mailed questionnaire [ 15 ]. Participants responded to mailed questionnaires every two years that collected detailed information on a variety of health, lifestyle, diseases, and risk factors. Between 1996 and 1999, NHSII participants who had not been diagnosed with cancer (other than non-melanoma skin cancer), were premenopausal, and who had not been pregnant in the past 6 months were asked to provide a blood sample. Participants who were not using exogenous hormonal medications (menopausal hormone therapy (HT) or oral contraceptives (OC)s) were asked to collect timed samples during the follicular phase (day 3 – 5) and the luteal phase (an estimated 7 – 9 days before the anticipated start date of next menses). Women with irregular cycles were asked to collect their luteal phase sample 22 days after last menses. Women who were using exogenous hormonal medications were asked to provide a single untimed sample. To determine the luteal phase sample timing, the start date of the next menses was confirmed by a postcard. The NHSII protocol was approved by the Institutional Review Board (IRB) of the Partners Health Care System, Boston, MA, USA and this analysis was reviewed by the IRB of the University of Arizona (Protocol Number: 2001288334). AMH was measured among a subset of plasma samples previously assayed as part of four sub-studies. Sub-studies consisted of three nested case-control studies for breast cancer (n=794), ovarian cancer (n=46), and early menopause (n=820), and one study of risk factors for AMH decline (n=800). We excluded breast cancer cases (n=399), ovarian cancer cases (n=23), participants who were not premenopausal (n=1), and participants who contributed to multiple studies (n=48). As had been done previously, for participants who were selected as controls for more than one study (n=48), we randomly selected one AMH measure [ 16 ]. On each questionnaire in the NHSII from 1993 onwards, participants were asked whether they had physician-diagnosed endometriosis. If participants responded ‘yes,’ they were asked to report the year of diagnosis and whether the endometriosis diagnosis had been confirmed by laparoscopy, the clinical gold standard for endometriosis diagnosis [ 17 – 19 ]. Among a subgroup of participants, self-reported endometriosis was validated in 1994 (n=200) and 2011 (n=711). A diagnosis of endometriosis was confirmed via medical records for the vast majority of participants who reported laparoscopically confirmed endometriosis (95%–100%), but only in 56% of women who reported endometriosis without laparoscopic confirmation [ 20 ]. Therefore, in order to reduce misclassification of our exposure, we restricted our endometriosis definition to those who reported laparoscopic confirmation of their endometriosis. Endometriosis diagnosis status was updated over time, but once a woman reported laparoscopically confirmed endometriosis she was categorized as having a history of endometriosis for the remainder of follow-up. For our primary exposure definition, participants were defined as having endometriosis if they had reported laparoscopically confirmed endometriosis on the questionnaire prior to blood draw. For the primary analysis we excluded those with non-laparoscopically confirmed endometriosis diagnosis (n=52), leaving 1,961participants. AMH (ng/mL) was assayed using the ultra-sensitive ELISA assay from ANSH Labs (picoAMH, Webster, TX) employing a quantitative sandwich enzyme immunoassay technique at Massachusetts General Hospital (ovarian cancer case-control study), Boston Children’s Hospital (early menopause case-control study, age study) and Ansh Labs (breast cancer case-control study). All samples were assayed randomly independent of case/control status. Coefficients of variation ranged from 0.6% and 14.5% [ 16 ]. Samples of AMH that were determined to be below the limit of the detection for the assay were replaced by a value equal to the limit of detection divided by the square root of two [ 21 ]. Covariates were a priori selected to be potential confounders of the association between endometriosis and AMH [ 22 ]. Information on participant height, weight at age 18 years, age at menarche, and number of years until menstrual cycles became regular after menarche was collected at cohort baseline in 1989. Information on weight, oral contraceptive (OC) use, current exogenous hormone use, smoking status, fasting status, and time of blood collection was collected at the time of blood collection. Information on history of OC use, infertility history, smoking history, parity history, and total breastfeeding was measured using the most proximal questionnaire to the date of blood collection. To account for the non-normal distribution of AMH, continuous values of AMH were log transformed. Linear regression was used to estimate the association between history of laparoscopically confirmed endometriosis and log AMH values. We calculated percent difference and corresponding 95% confidence intervals in AMH ([exp (β)-1 × 100]). Primary models were adjusted for participant age (age, age 2 ), and blood sample characteristics including fasting status at blood collection (yes/no), time of day of blood collection (4 categories), luteal day at blood collection (4 categories), and season of blood collection (winter, spring, summer, fall). Multivariable models were adjusted for potential confounders between endometriosis and AMH [ 22 ] including smoking status (never, current, former), pack-years of smoking (continuous), BMI at blood draw (continuous), current exogenous hormone use (yes/no), duration of OC use (never, 1–23, 24–71, 72–119, ≥120 months), years until cycle became regular (never, < 1, 1–2, 3–4, ≥5 years), and age at menarche (≤11, 12, ≥13 years). To investigate effect modification, models were stratified by BMI at blood draw (<25, ≥ 25 kg/m 2 ), history of infertility at blood draw (yes, no), and parity history at blood draw (parous, nulliparous). Likelihood ratio tests were used to test for statistically significant differences between groups [ 22 ]. Sensitivity analyses were performed for our endometriosis definition, blood draw characteristics, and participant characteristics. Given the known diagnostic delay between endometriosis symptom onset and diagnosis [ 23 , 24 ], we pre-dated endometriosis diagnosis date by 4, 6, and 8 years. The majority of participants who provided a blood sample were premenopausal and not taking exogenous hormones. However, there were some participants who were taking exogenous hormones (either OCs or menopausal hormonal treatment) (n=30) who provided a blood sample not timed with the menstrual cycle. In sensitivity analyses we excluded these participants. Polycystic ovary syndrome (PCOS) is a benign gynecologic condition that is associated with higher AMH levels. Therefore, in sensitivity analyses we excluded participants who had an AMH level ≥ 10 ng/mL which may be suggestive of PCOS [ 25 ] (n=61). We additionally adjusted for covariates that may be influenced by endometriosis [ 26 ] and therefore could be on the causal pathway between endometriosis and AMH [ 27 ] including total breastfeeding duration [ 28 ] (nulliparous or breastfed 12 months) and parity [ 5 ] (0,1,2,3+). We expanded our endometriosis definition to include anyone diagnosed with self-reported endometriosis irrespective of laparoscopic confirmation (n=52). To investigate the influence of surgery timing on AMH, we restricted our endometriosis definition to those who had been diagnosed via laparoscopy within 4 years prior to blood draw. Lastly, quantile regression was utilized to compare median AMH levels for women with and without a history of endometriosis.

At the time of blood draw, 119 participants reported having been diagnosed with endometriosis prior to blood collection; 1842 participants reported no prior diagnosis of endometriosis. Participants were on average 40 years old at the time of blood collection ( Table 1 ). Participants with a history of endometriosis were less likely to have a BMI of ≥ 30 kg/m 2 at the time of blood collection (13% vs. 15%) and at age 18 years (1% vs. 2%), to be a never smoker (69% vs. 72%), and to have breastfed for more than 12 months (30% vs. 47%). Women with a history of endometriosis were more likely to report a longer history of OC use (≥ 120 months: 13% vs. 11%), have a history of infertility (70% vs. 20%), and be nulliparous at blood collection (41% vs. 18%). Participants with a history of endometriosis had 34.7 % lower AMH levels (95% CI: −49.8, −15.2) compared to participants without a history of endometriosis in models adjusted for age and blood collection characteristics ( Table 2 ). In multivariable models that incorporated information on blood collection and potential confounding factors, participants with a history of endometriosis had 29.6% lower AMH levels (95% CI: −45.4, −9.2) than participants without a history of endometriosis. In sensitivity analyses where we further adjusted for parity and breastfeeding, participants with a history of endometriosis had 24.5% lower AMH levels compared to women without a history of endometriosis (95% CI: −41.6, −2.5). We did not observe statistically significant differences in the association between endometriosis and AMH levels by BMI at blood draw (p-value test for heterogeneity: 0.15), history of infertility at blood draw (p-value test for heterogeneity: 0.56), or parity at blood draw (p-value test for heterogeneity: 0.73) ( Table 3 ). However, the association between endometriosis and AMH appeared greater among those who had a BMI ≥ 25 kg/m 2 (−44.0%, 95% CI: −63.7, −13.8) compared to those who had a BMI < 25kg/m 2 (−19.8%, 95% CI: −41.7, 10.4). In sensitivity analyses, predating endometriosis by 4, 6, and 8 years did not meaningfully change results. In results excluding those taking exogenous hormonal medication (n=30) (−28.4%, 95% CI: −44.7, −7.1) and in separate analyses excluding those (n=61) with AMH ≥10 ng/mL (−29.2%, 95% CI: −45.1, −8.7) results did not appreciably change. When we expanded our definition of endometriosis to include all who self-reported endometriosis diagnosis, irrespective of laparoscopic confirmation, we observed similar results (−27.2%, 95% CI: −42.1, −8.58). When we restricted our analysis to women whose endometriosis was diagnosed within four years prior to blood collection, the magnitude of association increased (−38.5%, 95 CI: −64.5, 6.80); however, given the decrease in sample size and its influence on statistical power, the findings were no longer statistically significant. When utilizing quantile regression, women with a history of endometriosis had 0.48 ng/mL lower median AMH compared to women with no history of endometriosis ( Supplemental Table 1 ).

We observed that women with a history of endometriosis had lower AMH levels than women without a history of endometriosis in fully-adjusted multivariable models. This finding was robust to sensitivity analyses and did not vary by infertility or parity history. The majority of prior research on endometriosis and AMH has focused on the impact of endometriosis surgery, particularly surgery to remove endometriomas that impact the ovary. Surgery for endometriosis was associated with lower levels of AMH at one month [ 14 , 29 ], three months, and six months [ 30 ] post-surgery compared to baseline AMH. Additionally, research in fertility clinics has observed that women previously diagnosed surgically with endometriosis had lower AMH levels compared to other patients undergoing fertility treatment [ 11 , 13 , 31 ] and this pattern was consistent across all stages of endometriosis [ 11 , 31 ]. The influence of ovarian surgery, endometriosis lesions affecting the ovary, and endometriosis that is not affecting the ovary on AMH levels is challenging to disentangle given the complexity of diagnosing endometriosis and finding an appropriate comparison group in research. Endometriosis may influence AMH levels independent of surgery. AMH levels were studied in women with endometriosis undergoing in vitro fertilization at a tertiary care center (n=671). Both women with prior ovarian surgery and women without prior ovarian surgery had lower AMH levels (AMH 2.5 ng/mL) and a higher probability of decreased ovarian reserve (OR: 2.39 and 2.67 respectively) compared to women treated for male factor infertility (AMH 3.6 ng/mL) [ 11 ]. This finding suggests that endometriosis surgery even in the absence of ovarian involvement may contribute to lower AMH levels. In separate research, individuals with endometriomas diagnosed by ultrasound and not needing surgery (n=40) had greater declines in AMH levels over six months compared to age-matched females who did not have pain symptoms or ovarian cysts (n=40) [ 32 ] (26.4% decrease vs. 7.4% decrease). Additionally, in a case-control study, women undergoing diagnostic surgery for endometriosis based on pain and ultrasound visualization (unilateral endometriomas, nodules in the rectovaginal septum or in the utero- sacral ligaments, ovary adhesion to the uterus, kissing ovaries), had lower AMH levels prior to surgery than fertile control women matched on age, BMI, and ethnicity who did not have symptoms of endometriosis and did not have visualization of endometriosis on ultrasound (p-value:0.001) [ 33 ]. However, other studies have observed no association with serum AMH levels in women with endometriosis prior to surgery [ 12 , 34 ], but have observed lower AMH levels in the follicular fluid of women with endometriosis compared to women without endometriosis undergoing laparoscopic surgery [ 12 ]. Our analysis included all individuals with surgically confirmed endometriosis irrespective of lesion location, stage, severity, number of surgeries, or surgical treatments. We would expect that between 30–75% of women with laparoscopically confirmed with endometriosis have endometriomas [ 35 , 36 ]. In sensitivity analyses, we expanded our endometriosis definition to include participants with and without surgical confirmation. Endometriosis was still associated with lower AMH (−27.2%), but our results did attenuate slightly from the main results among women with surgically diagnosed endometriosis (−29.6%). The reason for this attenuation may be two-fold. As observed in prior research in the NHSII, there is a greater probability of misclassification of endometriosis when the definition of endometriosis includes all women, not just those with surgical confirmation [ 20 ]. This attenuation may also reflect the influence of gynecologic surgery on AMH levels. Disentangling the contribution of surgical intervention separate from endometriosis on AMH levels is complex. Our findings of lower AMH levels among women with endometriosis are consistent with previous research which observed that women with endometriosis may have earlier natural menopause compared to women without endometriosis [ 8 – 10 ]. Indeed, findings from the NHSII observed that women with a history of endometriosis had nearly a 30% greater risk of early natural menopause in models adjusted for confounding and reproductive factors [ 8 ]. All research on endometriosis and premature menopause to date has been conducted among large observational cohorts of women and did not take into account endometriosis lesion location or surgery type. This consistent pattern of association between endometriosis and risk of premature menopause suggests that endometriosis lesions-- irrespective of ovary involvement and surgery-- may contribute to earlier age at menopause. In addition to the influence of surgery, there are other potential pathways through which endometriosis may be associated with lower AMH. Women with endometriosis may have higher levels of inflammatory markers and oxidative stress [ 33 ] which in turn may incite damage to the ovary, leading to lower ovarian reserve. Emerging research has also suggested that AMH may contribute to endometriosis etiology and growth. The endometrium and the endometriotic lesions of women with endometriosis have high levels of AMH and AMHRII mRNA expression [ 34 ], with specimens from deep endometriosis and ovarian endometriosis exhibiting the highest AMH and AMHRII mRNA expression. Our study has many strengths, including being nested within a large prospective cohort study, having a more representative comparison group of women from the population, and having a long duration of follow-up after surgical endometriosis diagnosis. There are also important limitations which should be considered. Information on whether individuals with endometriosis had endometrioma(s) or had surgery affecting the ovary were not available [ 37 ], and therefore we could not explore endometrioma presence or excision as a potential mediating pathway by which AMH may be impacted in those with endometriosis compared to those without. In our sample, we are unable to determine the exact timing of onset of endometriosis; the time from endometriosis symptom onset to surgical diagnosis can be between 4 to 7 years in our cohort and in the general population, respectively [ 23 ]. To reduce potential misclassification due to this delay, we restricted the definition of endometriosis in our analyses to laparoscopically confirmed endometriosis which has been shown to have high concordance with medical records (≥96%)[ 20 ]. In sensitivity analyses to investigate the robustness of our endometriosis definition, we predated our endometriosis diagnoses by 4, 6, and 8 years, and we expanded our definition to include all women with any self-reported endometriosis, not just those surgically confirmed, and our results did not appreciably change. Unfortunately, data on endometriosis lesion location and grade, were not collected and may also influence AMH levels. As with all research, there may be misclassification of self-reported characteristics, such as menstrual cycle length and time to regularity. However, prior validation research from our cohorts has shown very high concordance [ 38 ]. Additionally, our study utilized existing data on AMH assayed for other studies within the Nurses’ Health Study II. For our study, we analyzed data from serum samples that had been stored between 14–22 years prior to being assayed for AMH. The NHSII has safeguards in place to ensure reliable storage of samples [ 39 ] and prior research has observed robust long-term stability of AMH when stored at very cold temperatures [ 40 ], as these samples have been. Coefficients of variation for AMH ranged from 0.6% and 14.5%; however, any misclassification of AMH resulting from measurement error would be non-differential with respect to endometriosis, suggesting that differences observed are a conservative underestimation. To our knowledge, this is the largest study of endometriosis and AMH with longest duration of follow-up between endometriosis diagnosis and AMH measurement. We observed that women with a history of laparoscopically-confirmed endometriosis had lower AMH levels compared to women without a history of endometriosis. Lower AMH levels may be one mechanism through which endometriosis may influence fertility and menopause timing, as well as cardiometabolic outcomes.

Anti-müllerian hormone (AMH) is secreted by granulosa cells in the ovaries in proportion to the number of developing follicles [ 1 ]. As a woman ages, her AMH steadily declines which mirrors the decline in the number of available oocytes for ovulation [ 2 ]. While AMH does not predict pregnancy success in women trying to conceive, it is associated with oocyte yield, infertility diagnoses, and time to menopause. AMH has become a commonly used proxy for clinically monitoring declines in ovarian function [ 2 , 3 ]. Endometriosis is a chronic, non-malignant gynecologic condition that burdens approximately 10% of women [ 4 ]. Women with endometriosis may experience dysmenorrhea, cyclic and acyclic pelvic pain, dyspareunia, dysuria, dyschezia, and infertility [ 5 ]. Prior research from the prospective Nurses’ Health Study II (NHSII) cohort has suggested that women with laparoscopically-confirmed endometriosis had a twofold increased risk of infertility compared to women without endometriosis [ 6 ]. However, the exact mechanism between endometriosis and risk of infertility is not fully understood, with research suggesting poor oocyte quality, endometrial dysfunction, and chronic inflammation contribute to differences [ 7 ]. Women with endometriosis may also have earlier natural menopause compared to women without endometriosis [ 8 – 10 ]. Despite these prior associations between endometriosis with risk of infertility and earlier age at menopause, there has been limited research on the association between endometriosis and AMH. Endometriosis may be associated with lower AMH levels [ 11 – 13 ]. However, the current research has been cross-sectional or restricted to small samples of women undergoing fertility treatments [ 11 ] and/or women from one clinical practice undergoing gynecologic surgery [ 14 ], and no study has been able to follow women for more than a few years after their endometriosis diagnosis. Therefore, to overcome the limitations in prior research, we investigated the association between laparoscopically confirmed endometriosis and AMH among participants who provided a blood sample in the NHSII. This cohort allowed for the inclusion of all of women with surgically-diagnosed endometriosis, not just those undergoing infertility treatment, and for investigating the association between endometriosis and AMH midlife (average age 40 years), not just proximal to endometriosis diagnosis.