Inflammation and Ovarian Function in Reproductive-Aged Women

Most study participants were between 31 and 35 years old (57%), were in the normal range of BMI (62%), reported their race as white (78%), and reported never smoking (76%) ( Table 1 ). There were few current smokers (2%). Most study participants reported a professional or advanced degree (64%). The overall median for CRP levels was 1.02 mg/L (interquartile range (IQR) 0.39- 2.8). Median CRP was higher with a lower level of education, a higher BMI, self-reported African American race, those who were current smokers, and higher gravidity and parity ( Table 1 ). For every 20% increase in CRP, there was an associated 0.565 pg/mL significant decrease in inhibin B (95% CI: −0.838 to −0.292 pg/mL) and a 0.535% significant decrease in FSH (95% CI: −1.006 to −0.062) ( Table 2 ). For every 20% increase in CRP, there was an associated 0.866% non-significant increase in AMH (95% CI: −0.270 to 2.014). CRP was not significantly associated with DOR in the adjusted logistic model (OR: 0.969, 95% CI: 0.786 to 1.195). The significance of the associations did not change after excluding individuals who currently smoked or had diabetes ( Supplemental Table 1 ). However, the strength of association between CRP and inhibin B weakened when individuals with hypertension were excluded, but other associations were unchanged in significance level ( Supplemental Tables 2 ). The strength of the associations between CRP and both FSH and inhibin B weakened when excluding elevated CRP values ( Supplemental Table 4 ). For CRP and both AMH and DOR, the results did not change after excluding individuals with hypertension, diabetes, and elevated CRP (>20 mg/L) ( Supplemental Tables 2 - 4 ). In the exclusion of participants with high AMH (>7.75 ng/mL), the relationship between FSH and CRP weakened but other associations were unchanged ( Supplemental Table 5 ). The strength of significance in the associations remained the same when adjusting for hormonal contraception in the last one, two, or three months and excluding individuals without recent hormonal contraception data ( Supplemental Table 6 ). The inclusion of participants with blood drawn on days one and five of the self-reported menstrual cycle, along with those with blood draw on days two through four, weakened the relationship between FSH and CRP slightly but other associations remained unchanged ( Supplemental Table 7 ). There were no clear patterns between CRP and ovarian reserve biomarkers in stratifying by 25(OH)D status ( Supplemental Table 8 ).

We conducted a secondary analysis of data collected from the Time to Conceive (TTC) study. TTC was a prospective observational cohort study of women attempting pregnancy in the triangle region of North Carolina (Durham, Raleigh, and Chapel Hill) between 2008 and 2016. A detailed description of the study design has been previously published, and the study population is concordant with the sociodemographic characteristics of the Raleigh-Durham area ( Steiner et al., 2011 , 2017 ). Eligible participants were between the ages of 30 and 44 and had been trying to conceive naturally for less than 3 months (self-reported). Women who had a history of infertility, endometriosis, or polycystic ovarian syndrome, had a partner with a history of infertility, or were currently breastfeeding were excluded. All women provided informed consent, and all study activities were approved by the institutional review board (IRB) at the University of North Carolina and approval for this secondary analysis was obtained from Duke University. Eligible participants were asked to schedule a study visit at the start of their next menses. Of the original 843 participants enrolled, 790 provided a blood sample, and of them, 778 participants had serum CRP measured. ( Figure 1 ). Since CRP has been observed to vary across the ovarian cycle due to ovulatory function ( Gaskins et al., 2012 ; Wander et al., 2008 ), we only included participants with serum CRP measured on day 2, 3, or 4 of the menstrual cycle. We further excluded participants missing covariates from models, limiting our analytic sample to 703 participants. Values below the limit of detection, 0.078 ng/mL for AMH (N=13) and 9 pg/mL for inhibin B (N=49), were replaced with the limit of detection divided by the square root of 2, an estimation method for the concentration of nondetectable values tested by Horung and Reed (1990) . 25-hydroxyvitamin (25(OH)D) was examined in a sensitivity analysis to determine if stratifying by 25(OH)D status modified results based on previous findings within this dataset ( Jukic et al., 2018 ). Final sample sizes for each biomarker were 703 participants for AMH and diminished ovarian reserve (DOR), 654 for FSH, and 652 for inhibin B ( Figure 1 ). AMH, FSH, and inhibin B were measured from serum samples that were stored at −30°C until analysis. Samples were shipped frozen and analyzed at the University of Southern California Reproductive Endocrinology Laboratory. The assays used to assess ovarian reserve concentrations were as follows: FSH (Immulite Analyzer, Siemens, Deerfield, IL), inhibin B (ELISA, Ansh Labs, Webster, TX), and AMH (Ultrasensitive AMH ELISA, Ansh Labs, Webster, TX). Interassay coefficients of variation were 4% to 5% for FSH, 5% to 8% for inhibin B, and 9% to 11% for AMH. The DOR was dichotomized and defined as an AMH concentration below 0.7 ng/mL, based on previous literature( Brodin et al., 2013 ; Penzias et al., 2020 ; Pfister et al., 2019 ). The interassay coefficients are within the acceptable range of variation for AMH based on previous validation studies ( Wallace et al., 2011 ). CRP was measured from serum samples stored at −30°C until analysis. CRP was measured using a high sensitivity immunoturbidimetric at Duke University. Interassay coefficients of variation were 5.6% and 8.9%. The limit of detection was 0.0013 ng/mL. Information on the laboratory analyses for CRP have been described previously ( Doumatey et al., 2014 ; Jukic et al., 2022 ). For a sensitivity analysis examining Vitamin D, 25(OH)D was extracted from stored blood spots using previously described methods ( Jukic et al., 2018 ). Briefly, 25(OH)D3 and 25(OH)D2 were measured through liquid chromatography-tandem mass spectrometry ( Jukic et al., 2018 ). We selected covariates based on previous associations with inflammation or ovarian reserve ( Dólleman et al., 2013 ; Jukic et al., 2022 ; Plante et al., 2010 ; Richardson et al., 2014 ). For each ovarian reserve biomarker, covariates were selected using a directed acyclic graph to identify confounders a priori ( Divani et al., 2015 ; Dólleman et al., 2013 ; Jukic et al., 2018 , 2022 ; Landersoe et al., 2020 ; Plante et al., 2010 ; Richardson et al., 2014 ). Participant demographics, reproductive and contraceptive history, smoking history, and other lifestyle and behavioral factors were reported on self-administered questionnaires at baseline. We examined the distribution of biomarkers for the following characteristics: age, race/ethnicity, highest level of education, body mass index (BMI), smoking history, parity, and gravidity. Age was calculated at the time of the blood drawn from the participant date of birth and the date of the blood drawn and used continuously. Race and ethnicity were not collected separately in the original questionnaire, and responses were self-reported based on the following categorization, with participants selecting one: African American, Asian American, Hispanic, Other or Mixed, and White. The categories of “Native American” and “Other or Mixed” were combined for the present analysis due to the sparse representation of those selecting “Native American”. Therefore, the group “Other or Mixed” may represent individuals who immigrated from other places not included in the questionnaire or individuals that identified as multiple categories. Self-described race is conceptualized as a social construct that represents unmeasured and uncontrolled factors in an individual’s environment that may be associated with CRP and ovarian biomarkers, and therefore it is included it in this analysis ( Gravlee, 2009 ). Previous research has demonstrated disparities among racial and ethnic groups for ovarian reserve markers, though the causal pathway between the social construct of race and ovarian reserve remains uncertain ( Bleil et al., 2014 ). The highest level of education was divided into four categories for descriptive analysis: some college or less, four-year degree, some graduate education, and graduate or professional degree. BMI was calculated based on the participants’ reported weight and height. BMI (kg/m 2 ) was categorized as underweight (<18.5), normal (18.5 – 30) for descriptive analysis, however, it was measured continuously within the statistical models ( Flegal et al., 2013 ). Smoking history was categorized as current smoker, never smoker, and former smoker. Gravidity and parity were defined as the total number of reported pregnancies and births, respectively. Our directed acyclic graph suggested the following as the minimal adjustment set for participants: BMI, age, smoking status and history, and self-described race. All models were therefore adjusted for these variables. Histograms and quartile-quartile plots demonstrated that AMH, FSH and CRP were not normally distributed, and these biomarkers were natural log-transformed. Inhibin B was approximately normally distributed, so it was not log-transformed. Frequencies, medians, and interquartile ranges (IQRs) were used to describe the concentrations of CRP, AMH, and FSH respectively. Frequencies, means, and standard deviations were used to describe concentrations of inhibin B. Multivariable linear regression was used to assess the association between CRP and each biomarker of ovarian reserve (AMH, FSH, and inhibin B). The residuals and fitted values were examined to confirm these were normally distributed. We determined the best parameterization of CRP by comparing several models (linear, linear splines, and restricted cubic splines) with Akaike’s Information Criterion (AIC). We also visually examined the associations estimated with restricted cubic splines for evidence of non-linearity. Based on the AIC and visual interpretation of the spline models based on the recommendation by Harrell, the linear parameterization of CRP was chosen (2015). Models estimated associations between CRP and AMH, FSH and inhibin B. Estimates and 95% confidence intervals from each model were converted to estimate effects for a 20% increase in CRP, providing a percentage change for AMH and FSH, due to the natural log-transformation, and unit change for inhibin B, due to the lack of log transformation. Multivariable logistic regression was used to assess the relationship between CRP and DOR. In a sensitivity analysis, to determine the strength of the relationship between ovarian reserve biomarkers and chronic inflammation, models described were repeated independently with the following exclusions: individuals who smoked (N=13), individuals with hypertension (N=35), individuals with diabetes (N=7), individuals with CRP above 20 mg/L (N=19), individuals with AMH above 7.75 mg/L (N=86), and individuals without recent hormonal contraception data (N=86). The effects of smoking status, diabetes, and hypertension were examined to determine if there were observed differences on the biomarkers by disease status and smoking exposure status. High CRP may indicate an acute elevated immune response ( Boots & Jungheim, 2015 ). High AMH may be an indicator of polycystic ovarian syndrome (PCOS), but this is not yet well understood ( Homburg & Crawford, 2014 ). Due to previous work indicating 25(OH)D status may impact inflammatory cytokines 25(OH)D status was included in a sensitivity analysis to determine if it modified the relationship between CRP and ovarian reserve biomarkers ( Jukic et al., 2018 ; Mellenthin et al., 2014 ). Therefore, the association between CRP and each biomarker was stratified by 25(OH)D status based on the Endocrine Society guidelines indicating 25(OH)D below 30 ng/mL represents insufficiency ( Jukic et al., 2018 ). All statistical analyses were conducted in Stata/IC, version 16.1 (StataCorp, College Station, TX).

While we did not observe a significant association between CRP and AMH or DOR, we did observe a significant association between higher CRP and lower inhibin B and lower early follicular phase FSH which indicates a possible relationship between inflammation and ovarian function. This relationship supports previous work that chronic inflammation may influence the ovarian cycle in animal models ( Lliberos et al., 2021 ; Z. Zhang et al., 2020 ). Lower follicular phase inhibin B has been previously associated with increased risk of a short luteal phase, although CRP has not been found to be associated with luteal phase length ( Harris et al., 2021 , 2023 ; Pfister et al., 2019 ). Inhibin B is a product of pre-antral follicles and appears to have an inverse relationship to FSH during the follicular phase of the menstrual cycle ( Honour, 2018 ; Robertson, 2012 ). High CRP during the early follicular phase may contribute to lower inhibin B levels which subsequently affect the luteal phase ( Hall, 2015 ; Harris et al., 2021 ; Honour, 2018 ; Klein et al., 2004 ; Robertson, 2012 ). However, the connection between inhibin B and successful conception remains unclear ( Harris et al., 2021 ; Honour, 2018 ; Pfister et al., 2019 ; Robertson, 2012 ). Higher early follicular phase FSH has been associated with age ( Hall, 2015 ). The number of follicles decrease with age, which is followed by lower inhibin B production and a premature rise in FSH prior to the final menstrual period ( Hall, 2015 ; Honour, 2018 ). If CRP were associated with decreased ovarian reserve, we would expect to see an association between CRP and higher early follicular phase FSH. However, here we must acknowledge the limitation of a cross-sectional analysis: it is possible that CRP causes an increase in FSH over time, which we did not detect here. Instead, it is possible that inflammation is causing an acute suppression of early FSH secretion ( Lliberos et al., 2021 ; Z. Zhang et al., 2020 ). This is supported by the finding of a longer follicular phase with higher levels of CRP ( Harris et al., 2023 ). It is possible that inflammation impairs follicular development or function ( Boots & Jungheim, 2015 ; Clancy et al., 2013 ) In a mouse model, higher levels of ovarian inflammatory markers were associated with decreased follicle development ( Z. Zhang et al., 2020 ). Previous research has found an association between chronic autoimmune conditions and lower AMH but have not measured CRP directly ( Cui et al., 2016 ; Kitajima et al., 2011 ; Zhao et al., 2020 ). Pelvic inflammatory disease has been previously associated with lower AMH levels ( Cui et al., 2016 ). Crohn’s disease has been associated with lower AMH levels, with older women exhibiting a negative association between AMH and Crohn’s status ( Zhao et al., 2020 ). Further, in mouse models, an association between inflammatory markers in the ovary and follicle depletion has been documented, as well as an association between increased age and signs of localized ovarian inflammation ( Lliberos et al., 2021 ; Z. Zhang et al., 2020 ). However, it remains possible that within this study population of healthy individuals the relationship between CRP and AMH may differ from previous research among those with autoimmune conditions. The lack of a strong association between CRP and AMH may provide evidence that CRP levels do not affect localized inflammation. For DOR assessment using AMH values, clinical definitions range from AMH below 0.4 to below 1.0 ng/mL ( Pastore et al., 2018 ). Using the 0.7 ng/mL cutoff consistent with all previous analyses within this cohort we did not identify association between CRP and DOR ( Pfister et al., 2019 ). Within this study population, higher AMH has been associated with long menstrual cycles, and long follicular phases ( Harris et al., 2021 ). It is possible that high AMH is associated with anovulation or prolonged cycles, as is seen with polycystic ovarian syndrome (PCOS), which is also associated with higher CRP. The strengths of the present study include the sample size, community-based design, and the collection of three key ovarian biomarkers. Nonetheless, the findings from the present study may be limited in generalizability given the sample characteristics, as they largely represent a college-educated, white population between the ages of 30 and 44 based on the socio-demographics of the Raleigh-Durham region. Further, given that the study design was predicated on time to conception, it was limited to females attempting pregnancy or soon to be attempting pregnancy at the time of enrollment, which has the potential to create bias ( Weinberg et al., 1994 ). The significance of association between CRP, both FSH and inhibin B should be examined in future studies ( Gaskins et al., 2012 ; Yang et al., 2020 ).

Increased inflammation, as measured by C-reactive protein, was associated with lower early follicular phase follicle-stimulating hormone and inhibin B. Inflammation may exert an effect ovarian function.

Human ovarian function can be understood to be deeply intertwined with the evolution of our species given its follicle dynamics evolved in response to evolutionary pressures and processes ( Ellison, 1990 ; Lorenz et al., 2015 ; Vitzthum, 2009 ; Voland, 1998 ). Ovarian reserve, which refers to the quantity of oocytes an individual’s ovaries contain, is an integral aspect of human ovarian function ( Steiner et al., 2017 ). Normative follicular dynamics may be impaired by inflammation, which signals activation of the immune system( Boots & Jungheim, 2015 ). Differences in the biomarker Anti-Mullerian Hormone (AMH), a measure of ovarian reserve, have been documented across self-identified race and ethnicity that remain inadequately understood ( Bleil et al., 2014 ; Richardson et al., 2014 ). Decline in ovarian reserve over the life course is thought to represent a key feature of ovarian aging, which may occur at differential paces despite the relatively constrained age of menopause in humans ( Gold, 2011 ; Murphy et al., 2013 ; Sievert, 2014 ).. Inflammation may represent a biological pathway by which ovarian function may be disrupted ( Weiss et al., 2009 ). Ovulation requires the ovarian tissue to rupture in order for the mature oocyte to be expelled from the ovary ( Boots & Jungheim, 2015 ). Inflammation plays a key role in weakening the wall of the follicle, allowing for the eventual separation of the oocyte from the ovary ( Boots & Jungheim, 2015 ). However, it has been demonstrated that excessive inflammation over the menstrual cycle may interfere with follicle microenvironment and development, which could in turn influence ovulatory function to disrupt ovulation, luteal function and, embryo implantation ( Bertone-Johnson et al., 2019 ; Robker et al., 2011 ; Yang et al., 2020 ). Inflammation has been hypothesized to disrupt ovarian function through oxidative stress, lipid accumulation in non-adipose tissue cells, and the stress response of the endoplasmic reticulum ( Capobianco et al., 2010 ; Robker et al., 2011 ), but the relationship between chronic inflammation and ovarian function remains unclear ( Chow et al., 2017 ; Clancy et al., 2013 ; Ferrucci & Fabbri, 2018 ; Kushner & Antonelli, 2015 ; Ridker, 2003 ). C-reactive protein (CRP), a marker of chronic inflammation, is an acute-phase reactant synthesized in the liver in response to tissue damage ( Espey, 1994 ; Pepys & Hirschfield, 2003 ) and is involved in the activation of the innate immune response ( Pepys & Hirschfield, 2003 ). CRP may influence ovarian function both indirectly and directly through its local actions in the ovary. Inflammatory cytokines, including CRP, have been hypothesized to play a key role in the induction of apoptosis of cells within the ovary, which has also been proposed as a mechanism by which endocrine disruptors affect ovarian reserve ( F.-L. Zhang et al., 2021 ). Although the successful development of follicles requires a careful, regulated inflammatory response, the follicular environment may be disrupted by elevated or sustained inflammation ( Pacella et al., 2012 ; Yang et al., 2020 ). Evidence supporting the association between measures of ovarian reserve and inflammation primarily come from studies on women with specific medical conditions ( Cui et al., 2016 ; Kitajima et al., 2011 ; Zhao et al., 2020 ). For instance, endometriosis and pelvic inflammatory disease have both been associated with reduced ovarian reserve measures ( Cui et al., 2016 ; Kitajima et al., 2011 ). CRP has also been examined in relation to altered follicular wave dynamics, with higher inflammation associated with a greater number of follicular waves ( Clancy et al., 2013 ). Elevated CRP may represent tissue repair caused by the process of follicle development and ovulation, however, the boundaries of normative inflammatory response during the ovarian cycle are not yet well-understood( Jilma et al., 1997 ). For instance, there is evidence that a sustained pro-inflammatory follicular environment may lead to premature follicle death ( Lliberos et al., 2021 ). Higher inflammation may also represent dysregulated resolution of inflammatory processes in the ovary over time, which may be in part captured through the levels of Anti-Müllerian hormone (AMH), follicle-stimulating hormone (FSH) and inhibin B levels, which are three key biomarkers of ovarian function ( Gleicher et al., 2011 ; Hutchinson et al., 2011 ; Jabbour et al., 2009 ). AMH has been associated with age at menopause, consistent with an understanding that chronological age is associated with declining ovarian reserve and follicle numbers ( Depmann et al., 2016 ). FSH is secreted by the pituitary gland and is essential for follicle growth and selection for ovulation ( Rose et al., 2000 ). Inhibin B is the product of pre-antral and small antral follicles, and exerts an inhibitory action on FSH ( Robertson, 2012 ). Both AMH and inhibin B decline as individuals approach menopause, while FSH increases ( Honour, 2018 ). However, over the life course, the precise mechanism of declining follicle numbers associated with aging in humans is not well understood ( Depmann et al., 2016 ; Weiss et al., 2009 ). In murine models of reproductive aging, chronic inflammation has been associated with a decline in follicle numbers, suggesting that inflammation could be one mechanism of reproductive aging ( Broekmans et al., 2007 ; Lliberos et al., 2021 ; Z. Zhang et al., 2020 ). Women with inflammatory conditions, including pelvic inflammatory disease, endometriosis, and Crohn’s disease, have lower levels of ovarian reserve as measured by AMH ( Cui et al., 2016 ; Kitajima et al., 2011 ; Zhao et al., 2020 ). It is plausible that inflammation is related to follicle depletion over the life course, but these associations are not well-understood ( Cui et al., 2016 ; Huang et al., 2020 ; Kitajima et al., 2011 ; Zhao et al., 2020 ). Understanding how inflammation is associated with biomarkers of ovarian function will build upon existing research to elucidate the relationship between inflammation and hormones associated with ovarian function, which is essential to both the normative ovarian cycle and early pregnancy. Our objective was to examine and explore associations between chronic inflammation and three biomarkers which capture signals of follicle development.