Abstract
Objective. We sought to identify causes for infertility in women with and without rheumatoid arthritis (RA).
Methods. Women with RA were matched to healthy controls. Differences in anti-Müllerian hormone (AMH) and anovulation were analyzed.
Results. Women with RA had lower AMH (β −1.05, 95% CI −2.09 to −0.005), but no difference was observed when AMH was log-transformed. No difference in anovulation was observed. Infertility prevalence was similar between groups, primarily attributable to polycystic ovary syndrome in healthy controls but largely unexplained in women with RA.
Conclusion. AMH was lower in women with RA, but reasons for infertility among women with RA remain unknown.
Despite decades of data suggesting infertility in women with rheumatoid arthritis (RA), we have limited understanding of why it is so common. In the Pregnancy-induced Amelioration of Rheumatoid Arthritis (PARA) study, a longer time to conception was associated with older age, high RA activity, and current nonsteroidal antiinflammatory drug (NSAID) and prednisone use1. Studies of anti–Müllerian hormone (AMH; a marker of ovarian reserve) among women with RA have yielded inconsistent results and questionable association with fertility, with some suggesting a decrease that could be associated with infertility2,3.
We aimed to identify possible causes for infertility in women with RA by comparing ovarian reserve, anovulation, and other factors between women with and without RA.
MATERIALS AND METHODS
This study was approved by Duke Health IRB (Pro00057614).
Cross-sectional study participants
Patients with RA included women aged 20–40 years diagnosed with RA by a Duke University rheumatologist and enrolled following informed consent. Healthy controls included women without autoimmune disease recruited from the Duke Clinical Research Unit Research Volunteer Registry, matched by age and current use of hormonal contraceptives.
Exclusion criteria included (1) unilateral or bilateral ovarian surgery, (2) prior exposure to known or possible ovary-toxic medications, and (3) methotrexate (MTX) treatment for ectopic pregnancy.
Ovarian reserve
AMH was analyzed using a quantitative ELISA. AMH, a serum marker of ovarian reserve, declines slowly with age as ovarian follicles diminish. Levels remain steady across the day and the menstrual cycle, with no differences in levels between the follicular and luteal phases4,5. Low AMH was defined by our laboratory as < 1.23 ng/dl (ages 20–25 yrs), < 1.03 ng/dl (ages 26–30 yrs), < 0.66 ng/dl (ages 31–35 yrs), and < 0.42 ng/dl (ages 36–40). For 90% power to detect a 0.5 ng/dl in AMH between patients with RA and controls6, we required 69 patients in each group.
Anovulation
For participants not using hormonal contraception, progesterone ≤ 3 ng/ml on days 21–23 of a menstrual cycle indicated anovulation7. Participants recorded medication use during the menstrual cycle of study. Thyroid-stimulating hormone (TSH), prolactin, and testosterone were measured in women with anovulation. Polycystic ovary syndrome (PCOS) was patient-reported along with associated symptoms, including hirsutism defined as Ferriman-Gallwey score ≥ 88,9. Obesity was defined as body mass index ≥ 30 kg/m2.
Infertility
All participants completed a reproductive health questionnaire. Women with a prior pregnancy reported time to conception. Infertility was defined as (1) inability to conceive after 12 months of trying, or (2) physician-reported diagnosis of infertility. Subfertility was defined as time to conception ≥ 12 months.
Analysis
Ovarian reserve and anovulation were compared between groups. Among women with RA, these measures were compared based on current and prior RA medication use, duration of MTX use, and current disease activity (Routine Assessment of Patient Index Data 3 and physician’s global assessment10). Differences in proportions were analyzed by Fisher’s exact test, and differences in continuous variables were analyzed by Wilcoxon rank-sum test and ANOVA. AMH was log-transformed and analyzed by multivariable linear models adjusted for age, hormonal contraceptive use, and race. Anovulation was analyzed by logistic regression models adjusted for age and race.
Logistic regression models analyzed differences in infertility, and generalized linear models estimated differences in the average number of pregnancies and live births. Models were adjusted for education, marital status, and race. SAS 9.4 (SAS Institute Inc.) was used for all analyses.
RESULTS
The study included 75 women with RA and 75 healthy controls (Table 1). While matched for age (mean 32 ± SD 5.1) and contraceptive use, women in the control group were more likely to be African American, unmarried, educated beyond college, and working full time, compared to women with RA. The average duration of RA was 9 years.
Ovarian reserve
The mean AMH in women with RA was 3.0 (SD 2.6) ng/ml and 3.9 (SD 3.9) ng/ml in controls (p = 0.1; Table 2). In adjusted models, AMH in women with RA was estimated to be lower than controls (β −1.05, 95% CI −2.09 to −0.005). Clinically low AMH was found in 12% of women with RA and 9% of controls (p = 0.4). Similar results were observed when 8 patients with PCOS were excluded.
Women with RA who had ever taken MTX had lower AMH levels than women who had not (β −1.25, 95% CI −2.63 to 0.14; p = 0.08), adjusted for age, hormonal contraceptive use, and race. However, no dose response was observed between cumulative MTX dose and AMH level; current MTX use did not affect AMH. MTX was not associated with low AMH, although precision of the estimate was limited.
Anovulation
Among women not taking hormonal contraceptives, 19% of women with RA and 21% of controls had anovulation (Appendix 1). TSH, prolactin, and testosterone levels were normal in all women with low progesterone. In women with RA, there was no association between prior or current MTX, corticosteroid, or NSAID treatment and anovulation. Frequency of anovulation did not differ by disease activity.
Infertility
Among women who had ever been pregnant or tried to become pregnant, 19% of women with RA and 15% of controls reported infertility (Table 3). When adjusted for education, marriage, and race, no difference in infertility was found between women with RA and controls (OR 0.66, 95% CI 0.17–2.49). However, women with RA who were ever pregnant or ever tried to become pregnant had fewer pregnancies (β −0.83, 95% CI −1.60 to −0.07) and fewer live births (β −0.38, 95% CI −0.84 to 0.08) compared to controls. There was no difference in subfertility or the average number of months to conceive between the groups. The cause of infertility was known in all 5 healthy controls, but only 3 of 9 women with RA. AMH and anovulation did not vary in women with and without infertility.
Other causes of infertility
The frequency of PCOS was similar between women with RA and controls (Table 3). There were no observed differences in menstrual cycle frequency or the age when women first tried to become pregnant. Among planned pregnancies, both groups reported having intercourse, on average, 11–12 times per month when trying to conceive.
DISCUSSION
We observed no differences in self-reported infertility in women with RA compared to healthy controls and few differences in anovulation, maternal age at conception, and intercourse frequency. We did, however, find that AMH in women with RA was on average 1.05 ng/ml lower compared to controls, particularly in women with prior MTX use. There was no association between cumulative MTX dose and AMH, however, casting some doubt on the clinical significance of this finding. There have been mixed results reported in the literature concerning the effect of MTX on AMH, with 2 studies suggesting decreased ovarian reserve with increased duration of MTX use, but another study showing no change after 6 months of dosing2,11,12.
Previous data suggest an association of chronic NSAID use with anovulation and subfertility13,14,15. A previous study found that the rate of ovulatory dysfunction increased from 5% to 25–30% when celecoxib was administered around the time of ovulation16. In our study, however, RA medications, including NSAID, did not appear to be associated with anovulation, although sample sizes were limited and few women took chronic NSAID.
While the rate of infertility was similar between groups, reasons differed. PCOS or ovulatory dysfunction was the primary reported cause of infertility among controls. In contrast, the reasons for infertility were largely unknown in women with RA. Brouwer, et al13 similarly reported that infertility was largely unexplained in women with RA. Interestingly, in our study, no differences in AMH or anovulation were found in women reporting infertility.
Our study benefited from a large number of participants. However, more women were taking hormonal contraceptives than expected, leaving only 65 with measured progesterone. This limited the power to detect differences in anovulation between RA and controls, and more importantly, by RA medication use. Data were self-reported, and we were unable to confirm reports of infertility. Although matched for age, controls were more likely to be African American, unmarried, working full time, and educated beyond a college degree, compared to women with RA. While these demographics do not affect ovarian reserve and ovulation, they likely affect family planning. For this reason, we adjusted comparisons of the number of pregnancies and children for these factors.
Previous data suggest sub- and infertility in women with RA1,17. We observed different results. We hypothesize that the relatively mild level of RA activity, and limited chronic use of NSAID and prednisone may have contributed to the normal fertility experienced within this population. Additionally, some conceptions preceded RA diagnosis, perhaps avoiding disease-related infertility. We had hypothesized that infrequently assessed causes for infertility might be behind the high rate of unexplained infertility, but we found no difference in the age at attempted conception or frequency of intercourse. Our findings suggest AMH is lower in women with RA compared to controls. A history of MTX use may affect AMH, although further research is needed to confirm this association and its possible implications.
APPENDIX 1
Prevalence of anovulation in women with RA compared to healthy controls and by RA medication use.
Footnotes
Pfizer Inc. provided the funding for this project through its ASPIRE award program for investigator-initiated studies. It had no say on the design or conduct of the research, and did not alter the manuscript.
M.E. Clowse is a consultant for UCB regarding issues surrounding the use of certolizumab for RA in pregnancy. This work does not involve issues of infertility or AMH.
- Accepted for publication August 10, 2018.