The impact of sickle cell disease and its treatment on ovarian reserve in reproductive-aged Black women

The Emory University Institutional Review Board approved this study. Women with SCD were recruited from the Grady Memorial Hospital (GMH) Comprehensive SCD Clinic in Atlanta, Georgia. Patients with SCD were ages 20–45 years, not currently pregnant, spoke English, and could provide informed consent. Recruitment was monitored to ensure adequate representation of those taking HU, those receiving chronic transfusions, and those who were neither taking HU nor receiving transfusions. We planned to recruit 50 women with SCD currently using HU who did not have iron overload, 50 with iron overload who were not using HU, and 50 who were neither taking HU nor had iron overload. Comparison women were recruited as part of the Furthering Understanding of Cancer, Health, and Survivorship in Adult (FUCHSIA) Women’s Study. The FUCHSIA Women’s Study is a population-based study examining effects of cancer on future fertility. The comparison group from the FUCHSIA Women’s Study was a group of young women with no history of cancer living in Georgia between ages 22–45 years (at the time of recruitment in 2010).( 14 ) Women from the FUCHSIA Women’s Study with at least one ovary were invited to have serum AMH measured. SCD predominately affects Black/African American patients,( 15 ) and the SCD patients seen at GMH are predominately Black/African American and living in metropolitan Atlanta. Therefore, for this analysis we restricted the comparison group from the FUCHSIA Women’s Study to Black/African American women living in Atlanta who had AMH measured. To better reflect the age-distribution of the SCD cohort, we recruited an additional 20 participants aged 20–25 years (at the time of recruitment in 2022–2023) from the GMH gynecology clinic who were attending a routine visit, did not have SCD, identified as Black/African American, and were not pregnant. Study participants completed an interview about their medical and surgical histories, obstetric and gynecological histories, history of reproductive conditions, and contraceptive use. Participants answered demographic and lifestyle questions. Women with SCD also answered questions about their SCD treatment history. For women with SCD, SCD treatment information was abstracted from electronic medical records, including current prescribed HU dose and treatment duration. GMH implemented electronic medical records in 2010 and has access to the local children’s hospital records reliably back to the early 2000s through an electronic health information exchange. All available medical records and prescriptions sent through the GMH electronic medical record prescription system were reviewed for the patients with SCD. To assess HU compliance, HbF levels,( 16 ) mean corpuscular volume (MCV), and absolute neutrophil count (ANC) were collected. HU promotes HbF production and erythrocyte volume expansion so low HbF levels and/or normal MCV in women prescribed HU indicates sub-therapeutic treatment adherence. ANC was collected to determine whether maximum tolerated dose (MTD) was reached. Receipt of blood transfusions (chronic, episodic, never), laboratory indicators of iron overload (e.g., serum ferritin), iron chelating agent use (ever use, current use), and liver iron concentration (LIC) via R2-magnetic resonance imaging (MRI) (FerriScan ® ) were also abstracted. Any discrepancies between self-report and the medical record with regards to HU or iron overload were individually adjudicated by a hematologist (RMF) and categorized. All study participants had blood drawn. AMH was measured using the MotiveBio picoAMH (sensitivity of 6.0 pg/mL). Sample size was determined using preliminary data before recruitment of women with SCD. We expected the comparison women to have a mean AMH of 2.76 ng/ml with a standard deviation of 3.07. Assuming a common standard deviation and an alpha of 0.05, we had 98% power to detect a half a standard deviation difference in mean AMH between the women who have SCD and those without. With a planned recruitment allocation of 50 women with SCD in each treatment group, we had an 83% power to detect a half a standard deviation difference in mean AMH comparing each 50-woman subgroup to those without SCD. Descriptive statistics were used to compare demographic characteristics of the women with SCD and comparison women. AMH levels below the assay limit of detection (LOD) were assigned an AMH value of LOD/√2. AMH was log-transformed given the right-skewed nature of the variable. We fit linear regression models to examine the relationship between log(AMH) and SCD, adjusting for age. We used the same age stratifications utilized by the Society for Assisted Reproductive Technology( 17 ) when reporting assisted reproductive technology (ART) outcomes given population-based AMH nomograms.( 18 , 19 ) The model was then transformed back from log(AMH) to give estimates of AMH. All AMH estimates were calculated for those who were 20–34 years. We performed a second model that also adjusted for body mass index (BMI) and recent contraceptive use. Contraceptives that may decrease AMH values include combined oral contraceptives, progestin-only pills, hormonal implant, vaginal ring, and contraceptive injection (e.g., medroxyprogesterone acetate); participants were dichotomized into users and non-users.( 20 , 21 ) We performed an additional model that stratified those with SCD by SCD genotype (HbSS or HbS-Beta 0 vs. HbSC or HbS-Beta + ), adjusting for age. We categorized women with SCD into three groups: 1) those taking HU at enrollment (Current HU), 2) those previously treated with HU but not taking it at enrollment and who had not taken HU for at least 3 months preceding enrollment (Previous HU), and 3) those who had never used HU (Never HU). This approach is similar to what has been performed in prior studies investigating HU exposure.( 7 , 13 ) HU use was defined as at least 12 months of medication exposure. Since HU is primarily prescribed to those with HbSS and HbS-Beta 0 , we restricted the SCD group for the HU analysis to patients with these genotypes. Models were fit to evaluate the impact of HU on AMH, adjusting for age alone and for age, BMI, and contraceptive use. Iron overload among those with SCD was defined as an average serum ferritin level ≥1000 ng/mL or evidence of LIC >2 mg/gm by R2-MRI assessment (if available).( 22 ) All but 13 participants with SCD had ferritin levels drawn at the same time as AMH. We fit models to evaluate the impact of iron overload on AMH, adjusting for age alone and for age, BMI, and contraceptive use. We evaluated the impact that chronic transfusions may have on AMH by stratifying those with SCD into those with versus without a history of chronic transfusions. We also performed an analysis among only those with SCD, stratifying participants into 4 groups: those with iron overload who ever used HU, those with iron overload who had never used HU, those without iron overload who ever used HU, and those with no history of either iron overload or HU use. This model was adjusted for age. Additionally, an age-adjusted analysis was performed amongst those with SCD that evaluated whether baseline hemoglobin level was associated with AMH; this analysis was repeated after excluding those on chronic transfusions. Several sensitivity analyses were also performed ( Supplemental Material (Methods) ).

There were 152 women with SCD and 128 comparison women without SCD included in this study ( Table 1 ). The women with SCD were younger (mean age: 30.1 years, standard deviation [SD] 7.4) than comparison women (mean age: 34.8 years, SD 6.8). Women with SCD were less likely to be overweight or obese than comparison women. Among women with SCD, 25.0% had iron overload and 53.9% had a history of current or previous HU use ( Table 2 ). Most women with SCD were HbSS (67.8%). For the women who had LIC measured by R2-MRI, there were no discrepancies in categorization of iron overload status using LIC versus serum ferritin. The median AMH was lower among women with SCD in each age stratification compared to comparison women ( Figure 1 ). The predicted age-adjusted AMH among women 20–34 years with SCD was 46% (95% confidence interval [CI] 27–60%) lower than that for 20–34 years old comparison women (2.23, 95% CI 1.80–2.76 vs. 4.12, 95% CI 3.11–5.45, respectively) ( Table 3 ). These estimates remained similar but less precise after also adjusting for BMI and contraceptive use. When women with SCD were stratified by genotype, age-adjusted AMH estimates were lowest amongst those with HbSS or HbS-Beta 0 genotype (1.98, 95% CI 1.55–2.53), compared to estimates for those with HbSC or HbS-Beta + genotype (2.93, 95% CI 2.05–4.20) and estimates for comparison women (4.12, 95% CI 3.12–5.45). Of the 82 women with SCD with a history of HU use, 58 were on HU during study participation (Current HU) and 24 had previously taken it but were not taking it at enrollment (Previous HU). All but 2 in the Previous HU group had discontinued HU more than 1 year before enrollment. Over half (45/82) of women with a history of HU use had over 5-years of exposure to the medication and over one-quarter (22/82) had over 10 years of exposure. Of those currently taking HU, the majority (82.8%) were taking 1000–1500 mg per day. The mean HU dose was 18.2 (SD 5.7) mg/kg body weight. Aggregate lab values among the Current HU group suggested treatment adherence (mean HbF: 13.7 [SD 8.1], mean MCV: 101.4 [SD 15.4], mean ANC: 4.8 [SD 2.3]). After restricting the SCD group to those with HbSS or HbS-Beta 0 genotype, there were 52 in the Current HU group, 21 in the Previous HU group, and 33 in the Never HU group. The Current HU group had age-adjusted AMH values 63% lower than comparison women. Those in the Previous HU and Never HU groups also had lower age-adjusted AMH values (35% and 44% lower, respectively) than comparison women, but these differences were not as marked ( Figure 2 ). The AMH estimate among SCD patients with iron overload (1.89, 95% CI 1.27–2.81) was slightly lower than the estimate among those with SCD without iron overload (2.35, 95% CI 1.85–2.98); both of these AMH estimates were lower than those of comparison women (4.11, 95% CI 3.10–5.44). The relationship between AMH and iron overload status was unchanged after also adjusting for BMI and contraceptive use ( Figure 3 ). The results of several supplemental analyses were similar to those in the primary analyses ( Supplemental Material (Results) ). In a sensitivity analysis among only women with SCD, there were 26 with iron overload who ever used HU, 12 with iron overload who never used HU, 56 without iron overload who ever used HU, and 58 with no history of either iron overload or HU use. Those without iron overload and no history of HU use had higher estimated AMH values than the other groups ( Figure 4 ). There was little variation in predicted AMH by baseline hemoglobin among women with SCD. This remained true after excluding those on chronic transfusions ( Supplemental Table 6 ).

Women with SCD had almost 50% lower age-adjusted AMH levels than those of comparison women without SCD. This difference remained after adjusting for potential confounders, namely BMI, contraceptive use, smoking status, and reproductive conditions that affect AMH. There was a notable difference in AMH estimates between those with HbSS/HbS-Beta 0 and those with HbSC/HbS-Beta + genotypes. Despite generally having milder symptoms,( 23 ) individuals with HbSC or HbS-Beta + still had lower AMH values than comparison women without SCD. SCD genotype and disease severity often correlate with the degree of baseline anemia, with HbSC and HbS-Beta + classically having milder anemia.( 24 , 25 ) We did not find an association between hemoglobin level and AMH among the women with SCD, suggesting that anemia severity alone cannot explain differences in AMH levels between those with SCD and comparison women. It is possible that those with more severe disease experience more frequent VOEs.( 24 , 26 ) Increased VOE frequency may lead to repetitive ischemic episodes within the ovaries, which may affect the number of remaining viable ovarian follicles. One small study evaluating follicle number in the ovaries of 4 girls with SCD (ages 9–16 years) after unilateral oophorectomy found that ovarian follicle density was similar to that of girls without SCD.( 27 ) Our study found that AMH, a surrogate marker for the number of remaining ovarian follicles, was lower among those with SCD after adjusting for age. Our study included only women of reproductive age (20–45 years); it is possible that the repeated insults of SCD-related VOEs generate a cumulative impact over time on ovarian reserve. Consequently, there may be an optimal window for fertility preservation for females with SCD in the immediate post-puberty timeframe prior to when AMH decline begins. While SCD genotype may have an influence on the degree of cumulative VOE damage on ovarian reserve, it should be emphasized that those with HbSC and HbS-Beta + may also still benefit from fertility preservation, as we found that their AMH estimates were also lower than age-matched comparison women. Additional research is needed to elucidate whether fertility preservation outcomes among those with SCD differ from age-similar women without SCD and whether differences vary based on the age at which fertility preservation is undertaken. In our study, women who were taking HU had lower AMH than women with SCD who had never taken HU. Our results are similar to those found in prior smaller studies.( 12 , 13 ) Additionally, we were able to separate current from prior HU use and found that AMH estimates were similar between those who had previously (but not currently) used HU and those with SCD who had never used HU. Thus, HU may have a transient effect on AMH. AMH is a glycoprotein secreted from granulosa cells of pre-antral and antral ovarian follicles.( 28 ) While AMH has been shown to reflect the primoradial follicle pool in healthy women,( 29 ) it is possible that HU affects follicular development without causing actual follicular toxicity.( 30 ) If the impact of HU on AMH is transient, it is unclear how long it would take for AMH levels to rebound to levels of those of women with SCD with no history of HU use. These results should be interpreted with caution as it is unclear what they may mean for AMH values among those with long-term HU use with subsequent drug discontinuation. Additionally, these results should not be used to suggest that disease-modifying therapy for sickle cell disease, such as HU, should be stopped or avoided altogether. It is possible that those who discontinue HU have milder disease and perceive the medication ineffective at reducing symptoms.( 31 ). While we did restrict our HU analysis to those with HbSS and HbS-Beta 0 genotypes, it is possible that there remained differences in disease severity given the heterogeneity seen even within each SCD genotype. Additionally, despite HU most often being prescribed to those with HbSS and HbS-Beta 0 , there were some women in our cohort with HbSC and HbS-Beta + who had a history of HU exposure. Joseph et al.( 7 ) also included those with compound heterozygous genotypes and reported HU exposure amongst these less common genotypes. While those with HbSC and HbS-Beta + genotypes were excluded from our HU analysis to try to better control for disease severity, it should be noted that HU may affect ovarian reserve among treated women with HbSC and HbS-Beta + genotypes, so they may also benefit from a discussion with a fertility specialist prior to initiating treatment. We did not find an appreciable difference in AMH estimates between those with SCD with and without a history of chronic transfusions, or between those who did and did not have iron overload. Data from the thalassemia literature have suggested that there exists a weak negative correlation between ferritin levels and AMH, supporting the notion of a deleterious effect of iron overload on the ovaries.( 32 ) We defined iron overload by current ferritin values and R2-MRI LIC. The proposed mechanism of ovarian damage from iron overload is ROS generation that causes DNA damage.( 33 , 34 ) Data from mouse models exposed to excess iron have shown iron deposition in the ovaries and higher levels of expression of genes implicated in cell cycle arrest and inhibition of granulosa cell proliferation.( 9 ) While AMH can act as a surrogate marker for oocyte quantity, it does not measure oocyte quality. Despite not seeing a difference in AMH between those with SCD who did and did not have iron overload, there may be a detrimental impact of iron overload and generation of ROS on oocyte quality, which is more difficult to assess clinically. Furthermore, a history of prior iron overload would theoretically still leave those with SCD with a history of exposure to ROS that may have caused permanent damage. It is possible that some women were classified as not having current iron overload who had ovaries affected by a prior history of iron overload. Additionally, iron overload may have an accumulating effect on ovarian damage, but we were unable to assess length of time of iron overload. However, there were only 8 women with SCD who reported ever taking iron chelators who were classified as not having iron overload so the potential for misclassification despite this limitation is likely small. The use of AMH as a proxy for ovarian reserve has been extensively researched in the context of fertility treatment and has been shown to be helpful for estimating outcomes in oocyte cryopreservation and ART.( 35 – 39 ) However, AMH does not predict natural fecundity( 40 ) and cannot be used as a marker of infertility. Some studies have suggested that women with SCD are at higher risk of infertility than the general population.( 41 ) Additional research is needed to better elucidate the mechanisms by which SCD may increase infertility risk and whether there may be an optimal time for intervention (e.g. fertility preservation) given the lower-than-expected AMH values in those with SCD and potentially shorter reproductive window. Furthermore, additional research is needed to provide better guidance on SCD treatment management around the time of fertility preservation, particularly if there is concern for a transient effect of HU on AMH. Care must be taken to balance optimizing fertility preservation cycle outcomes with optimizing SCD management. This study is the largest of its kind to evaluate ovarian reserve in a cohort of reproductive-aged women with SCD, and the first to have comprehensive medical record data to verify SCD treatment exposures. We were able to adjust for several potential confounders of ovarian reserve. Despite this, it is possible that there is residual confounding affecting our analyses. Some of our confounders, such as contraceptive use, were based on self-report, which may be subject to misclassification. AMH is a single measure of ovarian reserve; collecting longitudinal data on AMH may be more informative. Additionally, it may be helpful for future research to incorporate antral follicle counts by ultrasound or additional hormonal measurements. While we tried to evaluate the impact of SCD severity using SCD genotype, we did not look at number of VOEs requiring healthcare utilization, which may better assess the true impact of disease severity. While we did look at chronic kidney disease (CKD) as a marker for disease severity and found no difference in AMH between those with SCD who did and did not have CKD (results not shown), we were not able to evaluate all potential SCD-related comorbidities. During recruitment, we found substantial overlap between women with SCD taking HU and those with iron overload. Thus, our ability to recruit 50 women into each exposure group was limited and we may have been underpowered to detect differences between the different subgroups. We did not have adequate sample size to consider interaction between our 3-level HU variable (i.e., current, previous, never use) and iron overload. Additionally, we only included reproductive-age women and therefore are unable to evaluate potential early AMH changes that may result from SCD and its management during childhood and adolescence. We were also unable to evaluate the impact of other therapies for SCD on AMH, such as bone marrow transplantation, which has been previously shown to have an extremely detrimental impact on ovarian reserve,( 12 , 42 – 44 ) as this was outside the scope of this current project. While fewer data exist on the impact of gene therapy on AMH reserve, this newly-FDA approved therapy( 45 ) presumably has a severely detrimental effect on ovarian reserve given the need for myeloablative conditioning. Additional research is needed on this topic as this therapy becomes more available to patients. Our findings suggest that both SCD and its treatments, particularly HU, may impact ovarian reserve. While not a predictor of natural fertility, AMH does predict response to fertility treatment and may help predict time to menopause.( 46 ) Women with SCD who have low AMH may have a shorter reproductive window and a lower response to fertility treatment than similar-aged women without SCD. Additional research is needed to better determine which patients would benefit most from referral to a reproductive endocrinologist to discuss reproductive health and family-building plans in the setting of SCD.

As the population of reproductive-aged adults with sickle cell disease (SCD) has grown,( 1 – 3 ) new concerns have arisen regarding how SCD affects their adult lives. There is a critical knowledge gap regarding the effects of SCD, its treatments, and its complications on reproductive potential, especially among women. Ischemic infarcts to the ovaries may affect their function or accelerate follicular pool depletion. However, there are scant data on how repeated vaso-occlusive events (VOE) of SCD affect the ovaries. Prior studies have suggested that women with SCD have lower anti-Mullerian hormone (AMH), a marker of ovarian reserve that estimates the remaining follicle pool, than age-matched women without SCD.( 4 – 7 ) SCD treatments may also exacerbate follicular loss. Blood or exchange transfusions are often used to treat and/or prevent severe complications from SCD.( 8 ) Iron overload from frequent transfusions may lead to iron deposition in various organs, including the ovaries. Iron deposition has been shown in the ovaries of mouse models, and the resultant generation of reactive oxygen species (ROS) is hypothesized to cause cell cycle arrest and inhibit granulosa cell proliferation.( 9 ) Hydroxyurea (HU) is a SCD-modifying agent that decreases the frequency of acute VOE complications, prevents end-organ disease, and increases survival( 10 ) by increasing intra-erythrocyte fetal hemoglobin (HbF) and decreasing the secondary inflammatory effects of SCD.( 11 ) A prior study, which only included 33 female participants up to age 21 years and which did not follow them into adulthood, suggested that a longer exposure to HU may be associated with increased risk of diminished ovarian reserve.( 12 ) A separate study suggested that HU use was associated with lower average AMH levels; however, the comparison group only included 7 participants with SCD with no prior HU exposure.( 13 ) More recently, Joseph et al. published a study on the effect of HU on ovarian reserve with a larger cohort of women that included 23 women with SCD who were unexposed to HU; they found that women who had previously been on HU but discontinued it had similar ovarian reserve as those with SCD who were HU-naive.( 7 ) The impact of SCD and its treatments on reproductive potential remains critically understudied. The primary objective of this study was to evaluate the impact of SCD, HU, and transfusion-related iron overload on AMH in women with SCD compared to a similar cohort of women without SCD. We hypothesized that women with SCD would have lower AMH than women without SCD. Additionally, those with SCD with a history of HU use or a history of iron overload from chronic transfusions would have lower AMH than women with SCD who did not receive these treatments.