The Investigation of Vitamin D and Menstrual Cycles Trial (the inVitD Trial): A clinical trial of vitamin D supplementation on the hypothalamic-pituitary-ovarian axis

The most consistent and well-established sign of a ovulation and a healthy corpus luteum is the rise in post-ovulatory progesterone 19 , 25 that culminates in a sustained, high mid-luteal peak and adequate luteal phase length 26 . Few studies have explored the influence of vitamin D levels or supplementation on menstrual cycles, only one has measured luteal phase progesterone levels. A randomized trial examining the effect of vitamin D supplementation on post-ovulatory progesterone levels found that vitamin D supplementation was associated with lower progesterone levels on cycle day 21 27 . However, this study did not center the measurements on ovulation, so a lower progesterone level on cycle day 21 could be due to a lack of ovulation, earlier or later ovulation, or a less sufficient corpus luteum. Earlier ovulation with vitamin D treatment would be consistent with our hypothesis that adequate vitamin D reduces the occurrence of delayed ovulation. While it is not direct evidence, compared with their White counterparts, Black participants were shown in two studies to have a slower increase in luteal progesterone 28 and lower overall luteal progesterone, 29 and at the same time, are consistently shown to be more likely to have lower serum levels of vitamin D 9 . It is possible that vitamin D levels underlie this racial difference in progesterone levels. This evidence suggests that mid-luteal progesterone is the culmination of the previously described menstrual cycle endpoints and provides an important summary of fertility during the menstrual cycle. We measured the urinary metabolite, pregnanediol glucuronide (PdG) to calculate mid-luteal progesterone in daily urinary samples, as done in prior work 19 , 25 , 28 . ( Table 1 ) Hypothesis : Mid-luteal progesterone will be higher in vitamin D-supplemented cycles. There are several hormonal patterns that may result in delayed ovulation or prolonged menstrual cycles. Prolonged cycles may result from delayed follicular development and the associated rise in estradiol 20 . Robust follicle development is indicated by a consistent rise in pre-ovulatory estrogen 17 , 24 . Ovulation may be delayed by diminished ovarian response to gonadotrophin stimulation 30 or lack of stimulation. While estradiol initially inhibits gonadotropin release, at levels around 200 pg/ml estradiol stimulates gonadotropin release, initiating ovulation. If estradiol levels fail to reach these peak levels, ovulation may not occur. 21 - 23 . Lower vitamin D has been associated with lower average cycle estrogen in two studies 20 , 31 , while no association was found in an additional study 32 . This limited literature indicates that estrogen may be related to vitamin D status. Given this limited literature, one secondary endpoint was the slope of urinary estrogen metabolites over the three days before the ovulatory peak 19 , 25 , 33 . ( Table 1 ) Hypothesis : Rate of estrogen rise will be higher in vitamin D-supplemented cycles. Higher levels of LH in the pre-ovulatory period have been associated with increased likelihood of conception 19 , which may also indicate that pre-ovulatory LH is a marker of healthy hypothalamic-ovarian interaction. We examined change in pre-ovulatory LH from pre- to post-supplementation, where “pre-ovulatory” is the geometric mean of urinary LH from the ovulation day and two days prior ( Table 1 ). Hypothesis : Pre-ovulatory LH will be higher in vitamin D-supplemented cycles. The inVitD Trial was a Phase II randomized, double-blind, placebo-controlled clinical trial that compared urinary metabolites of menstrual cycle hormones pre- and post-vitamin D supplementation. All study materials, including protocol, consents, and questionnaires were reviewed and approved by the NIH IRB prior to study initiation. Participants were recruited from the Detroit, MI and central North Carolina (Durham, Chapel Hill, Raleigh) with community-based methods such as flyers, public events, and radio ads. Inclusion/exclusion criteria are presented in Table 2 . Eligibility for participation was ascertained via an online screening questionnaire accessible on the inVitD Trial website. The screener was initiated 2,048 times with 1,116 people passing the screener and eligible for Phase 1. Of these 1,116, 313 finished Phase 1 ( Figure 1 ). To confirm eligibility at the first in-person clinic visit, the screening questionnaire was reviewed, and BMI was measured. Potential participants were invited to complete an online screening questionnaire and, if found eligible, were then invited to complete an online shortened consent form and baseline questionnaire. ( Figure 2 ) Participants were then invited to schedule an enrollment clinic visit within one month of baseline completion. ( Figure 2 ) Participants with a greater than 60-day gap between completion of the baseline questionnaire and the enrollment clinic visit retook the baseline questionnaire to ensure their information was current (N=31). At the 60-90 minute enrollment visit, after obtaining written, informed consent, eligibility was confirmed by reviewing the screening questionnaire and measuring BMI, and determining nonpregnant status with a pregnancy test. Research staff also confirmed that the participant would be willing and able to complete study activities during the next 4-6 months. Then anthropometric measures, including blood pressure and body fat percentage, and biospecimens were collected. If a participant used a mobile phone menstrual cycle tracking application, start and end dates for up to six previous menstrual cycles were manually extracted by study staff. Eligible, enrolled participants took part in up to two phases of the study: Phase 1 followed participants for at least one menstrual cycle prior to supplementation (pre-supplementation cycle). ( Figure 2 ) Daily urine collection began with the first day of menstrual bleeding following the first clinic visit. Participants were withdrawn from the study if a cycle was longer than 60 days as they no longer met eligibility requirements (N=2). If a participant was menstruating during the first clinic visit, they were asked to start urine collection with the first day of the following menstrual cycle. During Phase 1 participants completed daily and weekly diaries and were invited to participate in an Actiwatch sleep study and a menstrual effluent collection study, which were exploratory components of the protocol not directly associated with the hypotheses. ( Figure 2 ) Participants were deemed noncompliant in Phase 1 if any of the following occurred: they completed fewer than 80% of their urine collection samples, didn’t complete both dietary surveys, completed fewer than four daily diaries per week, or completed fewer than two weekly diaries. Participants determined to be noncompliant due to any of the above conditions were ineligible to participate in Phase 2. Blood drawn at the enrollment visit, prior to Phase 1, was assayed for 25OHD concentration. ( Figure 2 ) Participants with concentrations less than 20 ng/ml who were defined as “compliant” in Phase 1 were invited to participate in Phase 2. To maintain blinding, a random sample of participants with 25OHD levels at least 20 ng/ml were also invited to participate in Phase 2. For all other participants with vitamin D levels over 20 ng/ml, participation ended after Phase 1. In Phase 2, participants with 25OHD levels less than 20 ng/ml were randomized to receive either 4,200 IU per week (the IOM recommended daily allowance 8 ) or 50,000 IU per week (recommended at the time by the Endocrine Society 34 ) of vitamin D (cholecalciferol) supplementation for up to 17 weeks. If a cycle continued longer than 17 weeks, participants were instructed to request additional supplement, but no one made such a request. Phase 2 participants with 25OHD levels of at least 20 ng/ml received placebo study capsules (cellulose) that were identical in appearance to the vitamin D study capsules. Cholecalciferol (vitamin D3) was selected over other forms of vitamin D, such as ergocalciferol (vitamin D2), because it is readily absorbed and bioavailable in humans. Although ergocalciferol has equivalent efficacy in raising 25(OH)D long-term 1 , it is less effective at raising 25(OH)D quickly 2 , which is important for this trial. Phase 2 participants were not informed of their vitamin D levels, or if they were receiving the vitamin D supplement or the placebo, until participation in the inVitD Trial was completed. Participants in Phase 2 were followed for three menstrual cycles beyond Phase 1. During Phase 2, participants took their assigned capsule each week, completed daily diaries and weekly diaries, and took at-home urinary ovulation tests for two menstrual cycles. In the third menstrual cycle they again collected daily urine specimens. After approximately 8 weeks of supplementation (range: 8-11 weeks), participants attended a clinic visit for blood collection and the measurement of serum calcium levels. Henry Ford Health (HFH) is a Detroit-based health system with a racially diverse patient population. To identify potential participants, we queried the HFH electronic health record system (Epic) using study inclusion parameters and exclusion criteria ( Table 2 ). The generated list was sorted by race to ensure diversity in recruitment and recruitment letters were mailed to more than 29,000 potentially eligible females. Additionally, 32,000 emails were sent to local residents at addresses that were purchased by the National Institute of Environmental Health Sciences (NIEHS). Other recruitment activities at HFH clinics and in the surrounding community included distributing flyers by mail and through organizations and businesses, local events, and brochures. Participants in the inVitD Trial were screened via the described online methods and then completed enrollment activities at HFH outpatient clinics in Detroit and the Greater Detroit, Michigan metropolitan area. The study site at HFH discontinued enrollment in May 2023. To increase enrollment, in February, 2023, a second study site was opened in North Carolina. Participants were recruited to the Clinical Research Unit (CRU) at the NIEHS in Durham, North Carolina, via flyers at local businesses and universities, attendance at local events to advertise the inVitD Trial, visits to local businesses, places of worship, and community organizations, and advertisements in local newsletters, newspapers (online and print), radio stations, advertisement space inside local buses, and internet sites such as Craigslist. Participants with a vitamin D level <20 ng/ml (low) who opted to continue into Phase 2 of the trial were randomized to receive either 4,200 IU/week or 50,000 IU/week of vitamin D supplements. Randomly selected participants with vitamin D levels higher than 20 ng/ml, who opted into Phase 2, received placebo capsules. All capsules (i.e., placebo, 4,200 IU, and 50,000 IU) were identical in appearance. Randomization assignment was determined by a random number sequence generated by the trial biostatistician. Block randomization was used to ensure that within each group of 5 or 10 people, at least some were randomized to each dosage. Permuted blocks of size 5 and 10 were generated with random assignment to supplementation with either 4,200 IU per week or 50,000 IU per week. Within each block, the ratio of participants assigned to 4,200 IU and 50,000 IU was 2:3. For each successive participant enrolled into Phase 2, the next randomization number within the appropriate block in the schedule was assigned. Randomization was initially stratified by BMI measured at the first clinic visit (<25 or 25-35 kg/m 2 ) and self-identified race/ethnicity from the trial screening questionnaire resulting in three strata (Non-Hispanic White, Non-Hispanic Black, and Another category). The latter category included participants who reported one of the following race groups: American Indian/Alaskan Native, Asian, Middle Eastern or North African, Native Hawaiian or Pacific Islander, or “another option”, or if they reported their race as White or Black and their ethnicity as Hispanic. Participants were invited to select all that apply and those that selected more than one racial group were also included in “Another category”. Due to slow overall enrollment, all stratification by BMI and race was eliminated so that all those who successfully completed study activities during Phase 1 and had vitamin D concentrations less than 20 ng/ml were invited to continue in the study. One in 10 participants with 25OHD levels of at least 20 ng/ml were randomly selected to participate in Phase 2 as part of the placebo group. Eight months into the 27-month trial, the 4,200 IU arm of the study was discontinued to reduce costs and improve efficiency between the two study sites and compounding pharmacy. Moving forward, the rate of placebo selection was two in every 10 participants with sufficient vitamin D levels (per stratum). Sixteen participants were randomized to the 4,200 IU arm prior to it being discontinued. Study data for women in this arm was retained for research and analysis. As the study power was calculated based on the 50,000 IU supplementation group, we did not make any sample size changes after discontinuing the 4,200 IU arm. All participants with a 25OHD level less than 20 ng/ml received 50,000 IU of vitamin D in Phase 2. The trial statistician (DLH), site investigational pharmacy staff (HFH and CRU), vitamin D vendor (Keystone Pharmacy), study physician (HFH and CRU), and data managers (N=2, DLH) were unblinded to 25OHD levels and assigned vitamin D doses or placebo. Keystone Compounding Pharmacy was subcontracted to compound the cholecalciferol supplement into 4,200 IU and 50,000 IU capsules, as well as placebo capsules. See Appendix for more details. Keystone shipped the study capsules to HFH Pharmacy Investigational Drug Services where they were stored in a secure location for distribution to study participants. The pharmacist notified the unblinded Investigational Pharmacy staff who separated the study capsule packages by content (4,200 IU cholecalciferol, 50,000 IU cholecalciferol, or placebo) and repackaged them into bottles of 17 pills each, putting each dosage in three distinct locations. After a participant was randomized, the HFH study manager was notified to enter a blinded prescription in the respective participant’s electronic medical record, which the blinded study physician signed. Upon physician signature, the HFH study manager notified the unblinded HFH Investigational Pharmacy staff to fill the prescription. The HFH Investigational Pharmacy staff viewed the participant’s information to see their randomization code and consent. They then located the correct bottle containing 17 study capsules to ship to the participant. A barcoded label with the participant’s study ID was printed to replace any other labels on the bottle. The Investigational Pharmacy staff then packaged and shipped the bottle directly to the participant’s home via FedEx. Investigational Pharmacy staff also included ovulation testing instructions and 60 prepackaged ovulation strips. Each delivery was tracked, and participants confirmed receipt of the study capsule package to study staff. Keystone also shipped the 50,000 IU cholecalciferol or placebo study capsules to the National Institutes of Health (NIH) Clinical Center (CC) Investigational Pharmacy in real time once each prescription was received. The pharmacy staff received study capsules, re-labelled to blind the bottle, and sent the bottle to the participant. The bottles were pseudo-randomized participant numbers (sequentially numbered) instead of the participant’s inVitD study ID per the official randomization process, best used by date, and dosing instructions. Unlike the HFH Investigational Pharmacy, the NIH CC Investigational Pharmacy was unable to send the ovulation strips to the participant since they were not allowed to dispense medical devices. Therefore, the CRU study personnel mailed 60 prepackaged ovulation strips to each Phase 2 participant after their prescription was signed. Each delivery was tracked, and participants confirmed receipt of the study capsules and ovulation strips to study staff through the study management system. Participants were asked in their daily diary if they took their study provided supplement that day. At the final in-person study visit, participants were requested to bring any unused supplement. Study staff then assessed intervention compliance by capsule count and disposed of any unused supplements. If participants did not bring their remaining supplements to the final visit, they were asked to take a photo of the remaining capsules and send it to the study staff. In addition, at the end of study participation, 25OHD and cholecalciferol were assayed in blood samples to assess adherence. The change in 25OHD with supplementation is the recommended biomarker for assessing adherence. 8 A detailed description of data collection materials and methods is presented in Table 3 . Sample size estimates were based on testing the average hormone levels in post- and pre-supplementation cycles. Literature indicates that the within-person correlation in hormone levels is high, 0.5-0.8 for pre-ovulatory estrogen 35 , 36 , 0.80 for mid-luteal progesterone 35 , 36 , and 0.8 for ovulatory LH 36 . For those with a significant change in 25OHD levels as expected in this trial, we expected a large change in hormone levels (and lower within-woman correlations). Using previously published hormone levels and standard errors 29 , 37 , 38 , for a pre-/post-supplementation difference in mid-luteal progesterone of 0.8 ug/mg creatinine, a standard deviation of 2ug/mg creatinine, and within-person correlation of 0.5, a sample size of 66 is needed for 90% power and an alpha of 0.05 based on a paired t-test. Thus, the target sample size was a minimum of 66 for the 50,000 IU supplementation arm. Within-person analysis: we will compare hormonal values between the pre- supplementation cycle and the last supplemented cycle, within-woman. The repeated measures design allows each participant to act as their own control and provides an overall test of the effect of “treatment” on the hormonal endpoints. This test is an “intent-to-treat” analysis, and participants and cycles will be included regardless of compliance with study supplementation. We will analyze all log-transformed hormonal measures (mid-luteal progesterone, rate of estrogen rise, and pre-ovulatory LH) as paired differences with a paired t-test, with an alpha of 0.05 considered significant. Between-person analysis: a linear mixed model will be used to quantify the association between 25OHD levels and hormone levels in the two cycles for each participant, using a random effect to account for the within-person correlation (e.g., with a compound symmetric correlation structure). Because this is not a paired analysis, participants who contributed one urine kit, rather than two, can be included. The structure of this association will be investigated using different parameterizations of 25OHD. For example, a linear fit of 25OHD would suggest that mid-luteal progesterone increases linearly across levels of 25OHD. We will also explore the importance of a quadratic term (with a likelihood ratio test). We can compare non-nested models, for example the linear fit with a dichotomous fit to 25OHD (of < or ≥20 ng/ml), using Akaike’s Information Criterion (AIC). We will investigate potential confounders such as age, BMI, season, and physical activity in adjusted models. Missing data may occur at the level of an entire urine-collection kit, at the level of individual daily samples, or due to inability to identify the day of ovulation. Because the hormonal endpoints are defined using specific cycle days, missingness affects calculability of endpoints rather than simply availability of assays. Missing entire urine kits: Participants missing a complete pre- or post-supplementation cycle cannot contribute to the paired primary analysis. These participants will remain eligible for analyses that use all available cycles (e.g., mixed models). Missing daily hormone measurements within a cycle: Required days for endpoint definitions (e.g., luteal days 5–6 for mid-luteal progesterone; three days before the estrogen peak for estrogen slope) must be present for the endpoint to be calculated. If one day of a multi-day endpoint is missing, we will calculate the endpoint using available days when scientifically reasonable (e.g., using day 6 only), with sensitivity analyses excluding partially observed cycles. If all required days are missing, the endpoint will be missing. We do not plan to apply multiple imputation because missingness in these specific cycle days reflects biological timing and participant behavior rather than random loss. Missing ovulation day: Ovulation will be determined using the validated E1c:PdG ratio and, if needed, the LH peak. If ovulation cannot be identified, cycle-anchored endpoints cannot be calculated. These cycles will be classified as anovulatory or unclassifiable and excluded from analyses requiring ovulation-centric endpoints but retained for analyses where ovulation timing is not required. For mixed models relating 25OHD to hormone levels, all available cycles will be included under a missing-at-random assumption, which is appropriate for intermittent missing daily samples 39 , 40 . Sensitivity analyses will compare results restricted to cycles with complete endpoint data. AEs were monitored and documented by study staff per 21 CFR 312.32 safety reporting regulations. Only AEs that were potentially linked to the study supplement had the potential to lead to discontinuation of the study supplementation. The diagnosis of a medical condition that violated the eligibility criteria, for example a new diagnosis of diabetes, kidney disease, sarcoidosis, or the use of a new medication (i.e., anticonvulsants), led to withdrawal from the study. Any medical condition that was present at the time that the participant was screened was considered part of their baseline and was not reported as an AE. However, if a study participant’s condition worsened at any time during the study, it would have been recorded as an AE; however, there were no instances of this occurring. At each visit, participants were asked about any AEs they had experienced. Participants also shared AEs with study staff when communicating between clinic visits. Any AEs and SAEs were reported to the NIH IRB in accordance with their reporting guidelines. The inVitD Trial was a two-site clinical trial of the effects of vitamin D supplementation on menstrual cycle hormones. Interventions to improve menstrual cyclicity are important to those who menstruate and are particularly important for those who do not wish to use hormonal contraception or are attempting to become pregnant. If vitamin D supplementation is effective at increasing estrogen, progesterone or LH levels during the menstrual cycle, this would suggest that it should be further explored as a low-cost intervention for menstrual cycle health. Moreover, it identifies a biological pathway for the observed effects of vitamin D on menstrual cycle length and regularity, and on the probability of conception. The inVitD Trial has limitations. First, the trial protocol was intensive, requiring daily urine collections for the assessment of hormones. When participants missed urine collection, this led to missing hormone measures. When participants travelled, they were offered supplies to continue collection. If they could not continue, they were excluded from further participation. Participants withdrew from the study, also leading to missing data and samples. Second, the final sample size was smaller than originally targeted. In conclusion, the inVitD Trial sought to determine the effect of vitamin D supplementation on menstrual cycle hormones. This trial is important because of the few interventions that are known to improve menstrual cycle health. The results of this trial will provide a potential treatment option for people seeking to improve their cycle function and regularity without hormonal interventions. Participation in the inVitD Trial ended in June 2024. Biospecimens are currently being analyzed for hormone and vitamin D levels.

Menstrual cycles are a vital sign 1 – a marker of both general and reproductive health. Moreover, women with menstrual cycle irregularities are often treated with hormonal contraception 2 which has potential health consequences, including cardiovascular (e.g., hypertension) and mental health (e.g. anxiety; depression) effects 3 , 4 . There currently are no alternative treatments for women who desire non-hormonal options to regulate their cycles or are planning to become pregnant. While menstrual cycles are experienced by half of the world’s 20- to 50-year-old population, there are few clinical trials focused on menstrual cycle health. While the role of vitamin D in bone health is well-known, its role in reproduction is unclear and a flourishing area of research 5 . Low levels of 25-hydroxyvitamin D (25OHD), the accepted clinical biomarker of vitamin D status, (<20 or <30 ng/ml) have been associated with a lower probability of spontaneous conception 6 , 7 . Total 25OHD is the sum of 25OHD 3 and 25OHD 2 8 and The Institute of Medicine suggests that almost all people achieve bone health with a total 25OHD level of at least 20 ng/ml. 8 However, levels lower than this are common in the U.S., especially among reproductive-aged non-Hispanic Black females. In the most recent data available (2015-2018), the median 25OHD levels for this group were 15.4 ng/ml for ages 12-19 and for 15.6 ng/ml for ages 20-39. 9 In laboratory animals, low vitamin D has been associated with anovulation and prolonged estrus cycles 10 , 11 . In humans, two cross-sectional studies 12 , 13 and one prospective study 14 reported that lower levels of 25OHD were associated with increased odds of irregular or long menstrual cycles. Importantly, all 13 or a majority of 12 the participants in these studies were African-American, and all three studies were large, ranging from 435 to 1100 participants. The prospective study also reported that participants with low vitamin D (40 ng/ml) had the lowest odds of a long follicular phase. This suggests that ovulation is an important facet of the underlying biological pathway modulated by vitamin D levels. This is further supported by the lack of follicular development, lack of ovulation and prolonged estrus cycles seen in mice and rats 10 , 11 , 15 , 16 . Several studies have also linked vitamin D deficiency to lower fertility 6 , 7 . In summary, low levels of vitamin D are associated with long menstrual cycles, delayed ovulation, short luteal phases, and lower fertility, but the biological pathways underlying these associations are unknown. Short luteal phases (<10 days compared with the usual 12 or more days) are considered subfertile, and are often accompanied by low luteal phase progesterone concentrations 17 and lower follicular estradiol 18 . Cycles with delayed ovulation may also be subfertile 19 and may result from reduced estrogen production 20 - 23 . Delayed ovulation may indicate faltering follicle development. Robust follicle development is indicated by a consistent rise in pre-ovulatory estrogen 17 , 24 . Higher levels of pre-ovulatory luteinizing hormone (LH) have been associated with increased likelihood of conception 19 . The most well-established sign of ovulation and corpus luteum health is a rapid rise in luteal progesterone 19 , 25 . More work is needed to determine whether these hormonal changes underlie the associations between vitamin D and menstrual cycle characteristics. A clinical trial of vitamin D supplementation in participants with low vitamin D levels ensures that the supplementation is given to those who can benefit from it. A trial can identify the effects of vitamin D supplementation, and a pre/post design can control for within-person confounders such as genetics. The primary objective of The Investigation of Vitamin D and Menstrual Cycles Trial: A Phase II Clinical Trial (inVitD Trial) was to investigate vitamin D’s influence on the hypothalamic-pituitary-ovarian axis by careful evaluation of hormonal and ovulatory changes that occur with vitamin D supplementation ( Supplemental Table 1 ). We examined the effects of vitamin D supplementation in those with low levels of vitamin D, on the hypothalamic-pituitary-ovarian axis in cycling females as indicated by mid-luteal progesterone, estrogen rise, and pre-ovulatory luteinizing hormone.