Female hormonal exposures and neuromyelitis optica symptom onset in a multicenter study.

A standardized reproductive survey was conducted between June 2011 and May 2013 for women with NMOSD (82% antibody positive) at 8 clinical centers, who were identified by chart review. These centers were as follows: Walton Centre in Liverpool, England (n = 56), Mayo Clinic-Rochester (n = 39), Washington University in St. Louis (n = 36), Charité - Universitätsmedizin Berlin (n = 26), Mayo Clinic-Scottsdale (n = 25), Johns Hopkins University (n = 19), and Massachusetts General Hospital/Brigham and Women's Hospital (n = 16). Informed consent was obtained from all participants under institutional review board policies at each site. A total of 217 women who met current published diagnostic criteria for NMOSD with or without aquaporin-4 antibodies 12 were enrolled consecutively, and all participants completed at least 1 component of the questionnaire. The reproductive survey included questions about menarche, pregnancies, menopause, and exogenous hormones. 13 It was conducted using REDCap, a secure Health Insurance Portability and Accountability Act compliant web-based survey tool. This questionnaire had previously good concordance of selected variables against the medical record in women with MS. 14 For this study, the following variables were examined: Age at menarche. Hormonal contraceptive (HC) use: Ever-use of HC was further divided by administration (oral, transdermal, IM) and duration (less than, vs at least, 12 months). The 12-month duration was selected because analyzing HC use over shorter timeframes might introduce more substantial recall bias. The formulation (estrogen, progestogen, or combined) or dosing of HC was not assessed. Number of pregnancies, including lifetime and after NMOSD first symptom onset (FS). Current reproductive status was categorized as cycling, perimenopausal (last menses in the 3–12 months prior to survey), or postmenopausal (no menses in the prior 12 months or loss of menses due to surgery). Type was categorized as (1) natural, or resulting from (2) surgical intervention (hysterectomy and/or bilateral oophorectomy) or (3) chemotherapy or radiation. Menopausal Age was defined as last menstrual period beyond which no menses occurred for 1 year (natural) or surgery date, according to Stages of Reproductive Aging Workshop +10 guidelines. 15 Menopausal hormone therapy (HT) use was categorized dichotomously for type (estrogen and/or progestogens vs other types [black cohosh, testosterone, etc]) and administration (systemic [patch + oral] vs local [gel, cream, and ring]), and for this analysis we included only systemic, homone-based HT. Age at menarche. Hormonal contraceptive (HC) use: Ever-use of HC was further divided by administration (oral, transdermal, IM) and duration (less than, vs at least, 12 months). The 12-month duration was selected because analyzing HC use over shorter timeframes might introduce more substantial recall bias. The formulation (estrogen, progestogen, or combined) or dosing of HC was not assessed. Number of pregnancies, including lifetime and after NMOSD first symptom onset (FS). Current reproductive status was categorized as cycling, perimenopausal (last menses in the 3–12 months prior to survey), or postmenopausal (no menses in the prior 12 months or loss of menses due to surgery). Type was categorized as (1) natural, or resulting from (2) surgical intervention (hysterectomy and/or bilateral oophorectomy) or (3) chemotherapy or radiation. Menopausal Age was defined as last menstrual period beyond which no menses occurred for 1 year (natural) or surgery date, according to Stages of Reproductive Aging Workshop +10 guidelines. 15 Menopausal hormone therapy (HT) use was categorized dichotomously for type (estrogen and/or progestogens vs other types [black cohosh, testosterone, etc]) and administration (systemic [patch + oral] vs local [gel, cream, and ring]), and for this analysis we included only systemic, homone-based HT. Descriptive statistics were used to summarize demographic, clinical, and reproductive characteristics. To examine the association between reproductive exposures and reported age at FS, we used multivariable regressions. Data were analyzed using the SAS-based software program JMP, version 8.0.1 (SAS Institute, Cary, NC).

The demographic and disease characteristics of the 217 respondents are provided in table 1 . Mean FS age was 40.1 years (SD 14.2, median 39.8, and interquartile range 28.0–51.6). There was no relationship between FS age and NMO antibody positivity (χ 2 , p = 0.68). With respect to hormonal exposures, the mean age at menarche was 12.8 years (SD 1.7), and 61% of women reported systemic oral HC use for at least 12-month duration. The median number of total pregnancies was 2 (SD 1.6) and occurring after NMOSD FS was 0 (SD 0.7). Thirteen percentage of respondents (26/193) reported delay in achieving pregnancy of 12 months or longer, and in the 11 in whom a reason for their infertility was identified, the causes were as follows: ovarian cysts (n = 3), structural problems (n = 3, including 1 endometriosis), hormonal dysregulation (n = 3), and male factors (n = 2). Twelve participants (6%) reported receiving infertility treatment. Demographic, clinical, and reproductive characteristics of 217 women with NMOSD from 8 participating clinical centers Among the 117 postmenopausal participants, 70% (n = 82) reported natural menopause (mean age: 48.9 years [SD 3.9]); only 25% (n = 29) reported systemic HT use. Given prior reports that women with neurologic disabilities may access fewer components of general health promotion (such as cancer screening, 16 or HT, which was often “standard of care” during the epoch corresponding to this cohort's menopausal transition 17 ), we compared HT use in postmenopausal participants with FS occurring before vs after their final menstrual period. Women with FS after their menopause reported greater HT use (37%) than women with FS prior to menopause (16%; χ 2 test = 5.68, p = 0.016). Association between reproductive exposures and age at disease onset is summarized in table 2 : Summary of results Ten participants (4.6%) reported FS prior to age of 18, 7 of whom reported FS within 2 years prior to (n = 2) or after (n = 5) menarche. There was no association between FS age and menarcheal age (linear regression, estimate = 0.91, p = 0.12). Ever-use of systemic HC was associated with earlier FS (39.3 vs 43.4 years), but this was marginally significant (linear regression, estimate = 2.1, p = 0.05; figure ). NMOSD = neuromyelitis optica spectrum disorder. A later FS was also associated with a greater number of preceding pregnancies (estimate = 3.3, p < 0.0001). Because participant age confounded the interpretation (i.e., parity increases with age), we categorized individuals according to FS age (before 30, 30–39, 40–49, and 50 years and older). When we included this categorical term in the linear regression model, the association between FS age and number of preceding pregnancies was not significant (linear regression, estimate = 0.30, p = 0.16). Because HC use might result in lower parity, we included both parity and HC use in a linear regression model, and the association between HC use and FS age became more significant (linear regression, estimate = 2.7, p = 0.007); the number of preonset pregnancies remained significant (linear regression, estimate = 3.3, p < 0.0001). Among the 106 women who were postmenopausal at the time of the survey and who provided a menopausal age, there was no relationship between age at menopause and FS age (linear regression, estimate = 0.30, p = 0.10). Among these women, 59 (56%) reported FS after their final menstrual period and 25 (24%) reported FS within 2 years of their final menstrual period (n = 18 before and n = 7 after). In women with postmenopausal onset of FS, there was no relationship between systemic HT use and FS age (linear regression, estimate = 0.95, p = 0.66). In sensitivity analyses, the significance of the findings reported was unchanged when we included race, ethnicity, and NMO antibody positivity as covariates in the models.

This study, conducted in a large international survey of women with NMOSD, represents a preliminary analysis of the relationship from NMOSD onset to hormonal factors beyond pregnancy. With respect to endogenous exposures, unlike in MS, we saw no association between menarcheal age and NMOSD onset. 18 – 21 As in other cohorts, 5% of our participants reported FS before the age of 18. 2 , 22 The association between a greater number of preceding pregnancies and later NMOSD FS ( p < 0.0001) was expected, as parity increases with age. Future studies should examine the association between age at first offspring's birth (not available in our cohort) and FS age. It is important that the number of pregnancies after disease onset was far outnumbered by those prior to disease onset. While this may partly reflect the older age of women at FS, it may also reflect reproductive choices relating to uncertainty of disease course, need for new medications that have uncertain effects on a fetus, and concerns about the future ability to care for a child, as has been reported in MS. 23 Inclusion of NMO antibody positivity in the sensitivity analyses did not change the significance (or nonsignificance) of the relationship between endogenous hormonal exposures and FS age. In a separate manuscript, we describe the association between pregnancy and relapses in this cohort (Klawiter, personal communication). It is notable that 6% of the respondents reported use of fertility treatments. Larger case-control studies adjusting for participant age are needed to determine whether there is an association between NMOSD and infertility. With respect to exogenous exposures, we observed a marginally significant association between ever-use of HC and earlier age at NMOSD onset. From a broader perspective, evidence for an association between HC and autoimmune diseases is mixed (including lower, 24 – 26 neutral, 27 , 28 and higher 29 risk of MS) and likely is confounded by changing HC formulations across epochs. Furthermore, the confounding effect of delayed (or no) childbearing that is achieved through HC use has not been adequately assessed, nor has the effect of participant age on recall of HC use. Follow-up studies are required to confirm this association, and, if this is confirmed, to assess HC dosing effects and whether the association is causative. Approximately, one-quarter of all participants reported a postmenopausal onset of FS, which contrasts with MS, where fewer than 10% develop MS FS after the age of 50. 30 The detection rate of serum aquaporin-4 autoantibodies has been reported to increase in women relative to men particularly after the age of 50, suggesting an effect of reproductive immunosenescence on NMOSD features. 22 We observed no relationship between FS age and reported NMO antibody positivity. The primary strengths of this multicenter study include an international design with a relatively large sample size, the use of specific diagnostic criteria for NMOSD, and the use of a standardized fairly comprehensive reproductive survey. Still, numbers were overall low, participants were mostly Caucasians from European and North American centers, possibly limiting generalizability of findings, occasionally data were missing, and reliance on patient-reported hormonal exposures may have introduced recall bias. In addition, we were not adequately powered to detect epochal factors potentially associated with both exogenous hormone use and NMOSD susceptibility or disease course, or to detect differences across clinical centers. We also identified further refinements needed to the reproductive survey (e.g., age at first birth and HC formulations). While we selected age at NMOSD as our initial clinical measure of interest, further studies should evaluate the association between endogenous and exogenous hormonal exposures and other NMOSD characteristics of interest, including clinical severity and progression, and severity of inflammation on MRI. Studies that include detailed hormonal analysis and immunopathogenetic measures are required to replicate these results and further elucidate the role of endogenous and exogenous hormonal exposures on this female predominant autoimmune disease, as well as important pathophysiologic implications.