The purpose of this study is to assess whether the natural form of estrogen (17-beta estradiol) given as a patch so that it is absorbed through your skin, is better at improving bone strength over 1 year than natural estrogen (17-beta estradiol) taken by mouth, or a synthetic form oestrogen (ethinyl estradiol) given as a patch that also provides birth control.

Participants will:

1. Take estrogen for 1 year either (i) in its natural form as a patch twice a week (and progesterone by mouth for 12 days of each month), or (ii) in its natural form as a pill daily (and progesterone by mouth for 12 days of each month), or (iii) in a synthetic form as a birth control patch weekly for 3 weeks with 1 week off the patch. You will not be able to choose which form of estrogen you will receive as this will be assigned to you based on a pre-existing randomization sequence (like the flip of a coin)

2. Take provided calcium and vitamin D supplements

3. Attend 4 study visits over 12 months with two at the beginning and then every 6 months that include:

   • History and Physical Exams
   • Lab Work
   • Imaging studies
   • Questionnaires
   • Dietary recalls

Low bone mass is a major co-morbid complication of functional hypothalamic amenorrhea (FHA) in adolescents and young adult women, including those with anorexia nervosa (AN) and exercise-induced amenorrhea (EIA), and the prevalence of fractures is markedly higher than in normal-weight controls (43% in EIA, 38% in AN vs. 22% in controls).1 Adolescence and young adulthood are a critical time for bone accrual. Peak bone mass, a major determinant of bone mineral density (BMD) and fracture risk in adult life, is established between 20-25 years of age in women. Insults to bone accrual during the adolescent and young adult years could result in permanent bone mass deficits, leading to increases in fracture risk. Despite weight regain and menses recovery in some individuals, BMD remains lower than in normal-weight peers, likely because weight recovery is often partial, relapses are common, and not all hormonal alterations contributing to low BMD in FHA completely normalize.2, 3 Thus, additional intervention is necessary to optimize skeletal health.

While many factors contribute to impaired skeletal health in FHA, a key contributor is the associated hypogonadism. Estrogen has anti-resorptive effects on bone through increases in osteoprotegerin and decreases in RANKL and inflammatory cytokines, and bone anabolic effects through inhibition of sclerostin. Yet, estrogen administration as the oral combined estrogen-progestin contraceptive pill is not effective in improving skeletal health4-6 because of hepatic first pass effects resulting in reduction in IGF-1 (a key bone trophic hormone, particularly during adolescence), and increases in SHBG (binding protein of the sex steroids), with a reduction in bioavailable estrogen.6, 7 In contrast, Investigators have demonstrated that transdermal physiologic 17β-estradiol replacement (17β-E2) (with cyclic progestin), which bypasses hepatic first pass metabolism and does not decrease IGF-1 or increase SHBG, increases bone accrual in adolescents and young adults with AN8 and EIA,6 resulting in its incorporation into Endocrine Society Guidelines for FHA management9.

However, transdermal 17β-E2 (with cyclic progesterone) does not have contraceptive efficacy and is not suitable for sexually active young women, unless administered with another mode of contraception (e.g. a progestin releasing IUD), and not always desirable to young women, leading to reduced uptake. The transdermal contraceptive patch containing a non-physiologic form of estrogen (ethinyl estradiol, EE) with a progestin (levonorgestrel, LNG) offers a systemic route of estrogen administration with avoidance of hepatic first pass metabolism. However, it is not known whether this patch is effective in improving skeletal health in FHA. Limited available data suggest otherwise,10 possibly because of an increase in SHBG with EE, despite transdermal administration.11-13 Further, many women prefer an oral pill because of skin irritation with the patch, cosmetic issues, and difficulty remembering to change the patch once or twice a week vs. taking a pill daily. While combined oral contraceptive pills containing EE are not effective in improving bone outcomes, it is not clear whether oral 17β-E2 (a physiologic form of estrogen) could be as effective as transdermal 17β-E2 despite hepatic first pass metabolism. Importantly, oral and transdermal 17β-E2 are often used interchangeably in other hypogonadal states,14 despite some evidence that oral 17β-E2 suppresses IGF-1 and increases SHBG.15, 16 The investigators thus propose a randomized controlled trial (RCT) of 2 mg oral 17β-E2 (with cyclic progesterone) vs. 100 mcg transdermal 17β-E2 (with cyclic progesterone) vs. 30 mcg transdermal EE (with LNG) in young women 14-25 years old with FHA (who are in the process of accruing peak bone mass) to determine whether transdermal 17β-E2 is more effective than oral 17β-E2 or transdermal EE in improving bone outcomes, or whether one or both of the latter two modes of estrogen administration are equivalent to transdermal 17β-E2 in this context. Results from this study could markedly impact clinical management of young women with FHA and potentially allow for alternative forms of estrogen administration in those with FHA.

Aim 1: To characterize the impact of transdermal 17β-E2 (with cyclic progesterone) vs. oral 17β-E2 (with cyclic progesterone) and transdermal EE (with LNG) on bone geometry, structure, density and strength in young women 14-25 years old with functional hypothalamic amenorrhea. Investigators hypothesize that transdermal 17β-E2 will be superior to oral 17β-E2 and transdermal EE in achieving optimal improvement in:

Hypothesis 1A: Areal BMD at the spine, hip and femoral neck; spine trabecular bone score (DXA)

Hypothesis 1B: Volumetric BMD, bone geometry and structure at the distal radius and tibia (HRpQCT)

Hypothesis 1C: Bone strength estimates (assessed using microfinite element analysis)

Aim 2: To determine effects of transdermal 17β-E2 (with cyclic progesterone) vs. oral 17β-E2 (with cyclic progesterone) and transdermal EE+LNG on levels of IGF-1 and SHBG, in relation to markers of bone turnover. Investigators hypothesize that:

Hypothesis 2A: Compared to transdermal 17β-E2, oral 17β-E2 will cause greater decreases in IGF-1, and both oral 17β-E2 and transdermal EE will cause greater increases in SHBG (with decreases in bioavailable estradiol).

Hypothesis 2B: Those with greater decreases in IGF-1 and greater increases in SHBG will demonstrate greater decreases in a bone formation marker (N-terminal propeptide of Type 1 procollagen, P1NP); changes in P1NP will be associated positively with changes in bone variables.

Data resulting from the study will be instrumental in driving recommendations for estrogen replacement in FHA.

Design: 114 young women 14-30 years old with FHA will be randomized in a 1:1:1 fashion to 1 of 3 groups for 12 months (1) 100 mcg transdermal 17β-E2 (with cyclic progesterone) OR (2) 2 mg oral 17β-E2 (with cyclic progesterone) OR (3) transdermal EE+LNG, to determine whether transdermal 17β-E2 is more effective than oral 17β-E2 or transdermal EE in improving bone outcomes, or whether one or both of the latter two modes of estrogen administration are equivalent to transdermal 17β-E2 in this context. All participants with FHA will also receive at least 600 IU of vitamin D and 1000 mg of calcium daily to avoid effects of suboptimal vitamin D concentrations and calcium status on bone endpoints. The study will be randomized, but not blinded (i) because there is a risk of participants assuming they are receiving contraception (with a blinded design) resulting in accidental pregnancies, and (ii) because bone outcomes (unlike e.g. psychological outcomes) are not expected to be impacted by study blinding/unblinding.

Analytical Plan:

The study will be done as intention-to-treat, analyzing changes in bone health at 12 months among participants randomized to the treatment groups. In general, the safety and efficacy populations will be the same, namely, all participants randomized to a treatment.

Participant characteristics at the time of randomization will be tabulated by group. Categorical variables will be summarized by percentages; continuous variables will be summarized by medians and interquartile ranges. No inferential statistical are planned.

Analysis of Aim 1:

Repeated measures models will be used to compare changes in areal BMD at the spine, hip and femoral neck as well as spine trabecular bone score from DXA. Based on published results,6, 69 investigators anticipate the use of a spatial power covariance structure, with equal variances over time and within-subject correlations that are a decreasing function of the time between two measurements within an individual. They will consider other covariance structures, such as those induced by random effects models, if the spatial power covariance assumptions do not seem to hold. Contrasts within the repeated measures model will be used to make comparisons of the primary endpoints both overall and at specific time points. The analyses will be adjusted for age, height, race and ethnicity by including terms for these characteristics in the repeated measures model. Similar analyses will be done for the endpoints in hypotheses 1B and 1C. The analyses will be carried out primarily in SAS 9.4 PROC MIXED. Analysis of the Secondary Endpoint(s): The analyses for hypotheses 2A and 2B are similar to those of the primary endpoint. Contrasts will be used to make specific two-group comparisons of the changes in IGF-1 and SHBG between groups.

Inclusion Criteria:

• Females, age 14-30 years, skeletally mature with bone age ≥ 14 years (only 2% of growth left)
• Women of reproductive age: use of an effective non-hormonal contraceptive method or a progestin releasing intrauterine device (no systemic skeletal effects) for study duration if sexually active. Note: Women who receive a progestin implant for contraception after study enrollment will be allowed to continue and will not be excluded from study.
• Biochemical criteria: negative βHCG (pregnancy test), TSH within 2x the upper limit of normal, prolactin <10 ng/mL above upper limit of normal, potassium between 3.0-5.0, ALT ≤3 times upper limit of normal, LDL ≤190 mg/dl.
• Patients with known hypothyroidism will be included if appropriately treated with levothyroxine and have a TSH within 2x the upper limit of normal for at least a month preceding the baseline study visit.
• Menstrual criteria: < 3 menses in the preceding 6 months.

Exclusion Criteria:

• Disease other than FHA known to affect bone, including untreated thyroid dysfunction, Cushing's disease, renal failure, diabetes mellitus

  1. Primary thyroid dysfunction will be defined as a TSH level more than 2X the upper limit of normal per given reference range with unknown thyroid antibody status, or an abnormal TSH if known positive antibodies.
  2. Patients with hypothyroidism will be excluded if not appropriately treated with levothyroxine and if they do not have a TSH level within 2X the upper limit of normal for at least a month preceding the baseline study visit, given possible effects on the reproductive axis and bone.

• Use of other medications known to affect bone metabolism within 3 months of the study (other than calcium and vitamin D supplementation)

• Substance use disorder; current smoker (>10 cigarettes per day)

• Pregnant, planning to become pregnant within 12 months of the end of treatment and/or breastfeeding

• Hypertension or use of anti-hypertensive medications

• Other conditions causing oligo-amenorrhea such as PCOS, premature ovarian insufficiency

• Known sensitivity or absolute contraindication to any component of study medications (high risk thromboembolic disease, breast cancer or other estrogen- or progestin-sensitive cancer, liver tumors, acute viral hepatitis, decompensated cirrhosis, undiagnosed abnormal uterine bleeding

• BMI ≥ 25 kg/m2 (efficacy of the contraceptive patch being used in the study is lower at higher BMIs)

Primary and secondary endpoint data, including all bone endpoints, will be submitted to the Harvard Dataverse repository to enable other researchers to analyze our study data independently. The privacy, rights, and confidentiality of human research participants will be protected by sharing only de-identified data.

Data sharing will occur no later than the end of performance period of the extramural award that generated the data. There is no end date planned at this time.

Study investigators will make access to deidentified data available via repository without restriction to access. Data available will include deidentified demographic data and those related to primary and secondary endpoints.