Urinary hormones have been used as proxy measures for circulating serum hormones in a number of studies.1-3 However, the women studied have typically been in their teens or twenties and thirties, and no studies have examined whether urine and serum reproductive hormones maintain the same relationship into midlife. Yet, processing of circulating sex steroids and gonadotropins might be dependent upon alterations in body composition, with an increase of fat deposition, and reductions in muscle mass with midlife aging.4,5 Increased serum creatinine, attributed to a lower glomerular filtration rate, could affect apparent urinary hormone concentrations because hormones are often indexed to creatinine.6 The Study of Women's Health Across the Nation (SWAN) performed an initial validation of urine and serum hormones on a small sample (N < 30) of women aged 43 to 53 years as part of the initial Daily Hormone Study (DHS).7
Studies examining midluteal progesterone (P) as a marker for overall luteal function in women have indicated that it serves as a marker for fertility and pregnancy outcome,8,9 and also cardiovascular health and endothelial function.10 These findings make it of interest to examine the serum P and urinary pregnanediol glucuronide (uPdg) patterns in cycling perimenopausal woman. The SWAN DHS collected first-morning voided urine samples for an entire menstrual cycle in a subcohort of 848 women at the DHS baseline, and annually thereafter until women experienced their final menstrual period (FMP).11 This sample is an ideal dataset with which to examine whether serum P and uPdg retain the same relationship to each other that they have demonstrated in younger women and to see if serum P is related to luteal Pdg and demonstrates the same relationship to the FMP as does luteal Pdg: a decline with proximity to the FMP.12
The SWAN Luteal Pilot Study—a subset of the DHS—was designed to address questions about the relevance of luteal phase hormones to the menopause transition. A selected subsample of women participating in the DHS were asked to come in for a blood draw at the anticipated time of their midluteal phase and concurrent urine and serum hormones: were assessed for concordance. The association of a timed luteal serum P level with integrated luteal uPdg was also assessed. Finally, we determined whether luteal serum P demonstrated a progressive decline with proximity to the FMP. We hypothesized that the relationship of urine to serum hormones would be maintained as women aged and approached their FMPs, and that a single serum P would provide an estimate of luteal phase Pdg production and would also demonstrate a decline as women approached their FMP.
The SWAN is a community-based, multiethnic cohort study of middle-aged women from seven communities in the United States.13 The SWAN DHS has been previously described in detail.7,11,12 Briefly, the DHS was initiated 2 years after the SWAN inception cohort was enrolled; women eligible for the DHS had to meet the following criteria: intact uterus and at least one ovary; at least one menstrual period in the previous 3 months (ie, either premenopausal or early perimenopausal at the time of the first collection); no sex steroid hormone use in the previous 3 months; and not pregnant or lactating. Women collected their first morning voided urine samples daily for one complete menstrual cycle or 50 days (whichever came first). Collections were repeated once per year until the FMP or for up to 10 collections. Menopausal status was defined as: premenopausal, bleeding in previous 3 months with no past-year change in cycle predictability; early perimenopausal, bleeding in the previous 3 months with a decrease in cycle regularity in the past year; late perimenopause, between 3 and 12 months of amenorrhea; and postmenopause, at least 12 months of amenorrhea. DHS collections were initiated with the onset of menses in women who were pre or early perimenopausal; however, with progress toward the FMP, random, 50-day (or sooner, if menses occurred) collections could be initiated after 2 months of an unsuccessful attempt to initiate with a menstrual period. Hispanic and White women from the New Jersey site were not included in the luteal pilot study. The study was approved by the Institutional Review Boards affiliated with all participating sites. Written informed consent was obtained from each participant.
Luteal pilot protocol
Women who were already committed to the SWAN DHS and who provided additional informed consent were part of the luteal pilot study. Participants were asked to come in for a blood draw between days 16 and 24 since their last menstrual period to attempt to capture a luteal phase sample. The planned day for the luteal blood sampling was based upon the participant's average menstrual cycle length over the past year and timed to coincide to 7 days before the anticipated subsequent menstrual period. Blood samples were promptly centrifuged per SWAN protocol13 and frozen at −80oC until assays were performed. Urine samples were stored at −20oC in nonfrost-free freezers as previously described.7 Urinary hormone measurements were completed for all analytes reported in this study within 1 year of collection and were not previously thawed.
Serum was measured for serum-luteinizing hormone (LH), follicle-stimulating hormone (FSH), estradiol (E2) and P, and urine was measured for LH, FSH, estrone conjugates (E1c), and uPdg using established, previously described methods.7,14 Both serum and urine hormones were measured in singlicate using an ACS-180 automated analyzer (Bayer Corp., Norwood, MA). Serum E2 concentrations were measured with an ACS-180 platform-based immunoassay in which the software was adapted to accommodate the E2-6 antibody used. For this adapted assay using the unique E2-6 antibody, interassay and intra-assay coefficients of variation averaged 10.6% and 6.4%, respectively, over the assay range, and the lower limit of detection (LLD) was 1 pg/mL.15 Serum FSH, LH, and P concentrations were measured with a two-site chemiluminometric immunoassay using constant amounts of two monoclonal antibodies provided by Bayer Corp. Interassay and intra-assay coefficients of variation and LLD were as follows: for FSH 10.9% and 3.9% with an LLD of 1.18 IU/L14; for LH 10.7% and 4.8% with an LLD of 0.1 IU/L; for P 7.3% and 3.0% with an LLD of 0.1 ng/mL.
As previously described, urine samples were collected in 7% glycerol to preserve the integrity of the molecules and make the specimens suitable for long-term storage, and all urinary hormone levels were normalized to creatinine before any analysis.7 Interassay and intra-assay coefficients of variation and LLDs were as follows: for FSH: 11.4% and 3.8%, LLD 0.3 mIU/mL; for LH 10.9% and 4.6%, LLD 0.1 mIU/mL; for E1c 11.5% and 8.1%, LLD 0.1 ng/mL; and for Pdg 17.8% and 7.7%, LLD 0.1 pg/mL.7
Objective, validated algorithms were applied to detect a sustained, robust rise in Pdg consistent with ovulation,16 and the timing of this luteal shift was determined by a shift in the ratio of E1c to Pdg, and designated as the day of luteal transition (DLT); data were centered to this date, which was set to 0.17 Cycles in this study were divided into those that qualified as having robust luteal function (evidence of luteal activity [ELA]) versus those that did not qualify as meeting this criterion (non-ELA). Area-under-the curve methods, as previously described,11 were applied to Pdg excretion curves to evaluate luteal function using urinary Pdg (integrated uPdg).12 For ELA cycles, hormones were organized around the DLT, which was designated day 0. Integrated uPdg was measured from day 0 through the end of the collection for ELA cycles.
Because each of the study outcomes involved slightly variable sample availability, Supplemental Table 1 (http://links.lww.com/MENO/A500) provides details on the exact sample sizes for each outcome. All cycles in the luteal pilot study that had a serum collection within the 16 to 24-day window after the last menstrual period were used to assess whether the relationship between urine and serum hormones are maintained as women enter the sixth decade of life. However, for hypotheses examining whether a single luteal phase serum P level is reflective of integrated-luteal urinary Pdg, only ELA cycles in which the serum sample was correctly drawn during the luteal phase were used (N = 125); these cycles were retrospectively defined. To test the association between a single luteal phase serum P and time relative to the FMP, the analytical sample only included women for whom we had FMP date and luteal phase serum P available. Hormone values were log-transformed due to skewness in the distributions of values; after log transformation, the data were normally distributed. Correlations between concurrent serum and urine hormones were assessed using unadjusted Pearson's correlation. Linear regression was used to determine adjusted proportion of variability in serum hormones that were explained by their urinary metabolites while accounting for race/ethnicity,12 age and log-body mass index (BMI),14 and smoking status at urine collection. Linear regression was also used to assess association between single luteal phase serum P and time relative to the FMP. Time to the FMP was calculated as the difference in years between date of serum collection and date of FMP. The significance level was adjusted to account for multiple testing per evaluated hormone as denoted in footnotes to Tables 1–3. Most of the women were sampled only once in the luteal pilot; however, 77 women provided a concurrent serum and urine sample in more than 1 year, and those additional time points were not included in the primary analysis. Secondary analyses using linear mixed model to adjust for correlated data provided the same findings (data not shown).
A total of 274 serum samples were collected concurrent with the DHS urine collection within a postmenstrual window of 16 to 24 days (see Supplemental Table 1, http://links.lww.com/MENO/A500 for flow chart). Six of these samples were excluded due to either missing serum or urine hormones for the date of collection, or missing creatinine values, which precluded accurate reporting of urinary hormone concentrations. Thus, a total of 268 concurrently collected urine and serum samples were available across multiple time points. Serum and urine hormone data from first available collection were used for analysis from 170 women for E1c/E and for Pdg/P pairs, but because two urine samples were missing for FSH and 11 serum samples were missing for LH, a total of 168 and 159 women had FSH and LH paired for serum and urine hormone assessments. Out of the 170 women with first available matching samples, 125 women had ELA cycles in which the serum sample was correctly drawn during the luteal phase and thus included in integrated luteal uPdg analysis; of those, 77 women additionally had FMP date available and were thus included in analysis relative to time since FMP (Supplemental Table 1, http://links.lww.com/MENO/A500). Participants were distributed across the racial and ethnic groups represented by SWAN, with the exception of Hispanic women who were under-represented (Table 4). Overall, participants were slightly younger and earlier in their menopause transition than the rest of the SWAN cohort, consistent with their eligibility for the DHS. Most were in the early perimenopause, and premenopausal women and those in the early perimenopause were similar in characteristics (Table 4). Serum and urine LH and FSH were significantly higher in early perimenopausal women compared with premenopausal women (Table 5).
Associations between serum and urine hormones
For LH, FSH, estrogen, and P, the positive relationship between serum and urine levels was statistically significant (Fig. 1, Table 1).
- Pearson's correlation: Pearson's r varied from a low for LH of 0.573 to 0.845 for FSH.
- Multivariable linear regression: After adjustment for race/ethnicity, BMI, and smoking status, β-coefficients ranged from a low of 0.44 to 1.1 for LH and P/uPdg, respectively. Corresponding adjusted R2 values ranged from 0.37 for E/E1c, to 0.42 for LH, 0.74 for P/Pdg, and 0.76 for FSH (Table 1).
Relationship between luteal P and urinary Pdg in luteal-phase-only samples
Although blood sample collection was timed to coincide with the midluteal window (days 5-9 post-DLT), this level of precision was only achieved in 66 or 52.8% of first available cycles with evidence of luteal activity in the luteal pilot (n = 125). Early luteal (days 0-4 postday of luteal transition [DLT]) serum samples constituted 29.6% (n = 37) and late luteal (≥10 days post-DLT) samples 17.6% (n = 22). The correlation of serum P to urinary Pdg was unchanged when only luteal samples were assessed (data not shown).
Relationship between luteal P and integrated luteal uPdg, and luteal serum P with approach of the FMP
Adjusted linear regression for log-transformed serum P and integrated luteal uPdg indicated a statistically significant β-coefficient of 0.4 and R2 of 0.09 (Table 2). Pearson's r was also signficant at 0.26 (Table 2). When examined by the stage of the luteal phase when the blood sample was drawn, the mid-luteal phase (days 5-9 after the DLT) and late luteal phase (≥10 days after the DLT) relationships remained significant, but the early luteal phase (days 0-4 after the DLT) did not demonstrate a signficant relationship between serum P and integrated luteal uPdg (Table 3).
Finally, concurrent serum P, urinary Pdg, and integrated luteal uPdg were examined in relationship to the FMP. While both same-day urinary Pdg and integrated luteal uPdg were related to the time to the FMP, luteal serum P was not related to the timing of the FMP in the unadjusted analysis. Results were similar after adjusting for age, race/ethnicity, BMI, and smoking status, although luteal serum P became marginally significantly associated with time to FMP (Table 6).
Herein we have demonstrated that the robust relationships between urinary and serum hormones that have been previously reported in mid-to-late reproductive aged women1,2,7 are maintained in women as they approached their FMP and entered the sixth decade of life. These data provide reassurance for future research using urinary hormone analyses in populations of aging women and help further validate the analyses of the SWAN DHS. Moreover, we demonstrated the feasibility of a planned luteal phase blood sampling paradigm, despite the relative irregularity of menstrual cycles during that time. However, unlike in midreproductive life, luteal serum P seems less associated with overall P output across the luteal phase.
Urinary hormone assays have demonstrated great usefulness for epidemiologic and animal studies, as they can provide a great deal of information over long periods of time, are noninvasive, and require minimal preparation and handling to obtain a reliable sample. The ability to follow women over months to years has been useful in elucidating the reproductive endocrine processes surrounding the onset of menarche,3 the processes of premature menopause,18 and diminished ovarian reserve,19 in addition to a number of studies of the menopause transition.20 Urinary hormones are assumed to be reflective of serum hormones, as urine is an ultrafiltrate of plasma, and, absent degradation in the circulation, urinary hormones should be an effective proxy for serum. It is therefore important to assure the integrity of urinary hormone assays, especially so in a study such as SWAN, which has 10 years’ worth of longitudinal urinary hormone data. We found that, overall, the strongest relationships were seen between serum and urine for FSH and P/Pdg, and less strong, although highly statistically significant relationships were observed for LH and for E/E1c. These findings are somewhat expected, because both urine and serum FSH and P/Pdg represent the same molecular species, with Pdg, or pregnanediol glucuronide, being the principal metabolite of serum P. However, the less strong relationship and greater overall variation between serum and urine LH was less expected, because correlations between serum and urine LH in younger women have been more robust.2,7 Finally, E1c, which represents a mixture of E2, estrone, and both glucuronide and sulfated conjugates, undergoes the most metbolism of all four reproductive hormones measured herein, and would therefore be more likely to have the weakest association with serum E2.
A single, prospectively timed luteal phase serum collection was successful in at least targeting some time point at or after the DLT for 52% of the evaluated women. This finding indicates that targeting the luteal phase for blood sampling may be feasible for women in the early menopause transition, that is, before cycle irregularity becomes too great. This is of interest, because luteal P production may be a predictor of cycles that are more likely to be fertile and to result in pregnancy, which is true for midreproductive aged women.8 Luteal sampling may also help identify women with anovulatory bleeding, which has been linked to endometrial hyperplasia risk in perimenopausal women.21 It may also be desirable to time a luteal blood hormone collection to measure corpus luteum hormones that are not secreted into urine, such as inhibin A or relaxin.
A single, midluteal serum P level has been used to determine the probability of pregnancy8 in 2,376 cycles of infertile women who were undergoing ovulation stimulation with clomiphene, letrozole, or gonadotropins. These investigators found that a P level above the 10th centile for each treatment group was associated with more than twice the probability of pregnancy than lower midluteal P. This finding implies that midluteal P reflects the overall robustness of the menstrual cycle. Others have examined P levels in unstimulated pregnancy cycles of 297 women and compared them to 406 nonpregnancy cycles to determine the lower limit of P associated with a pregnancy. They identified a fifth percentile of P of 5.6 ng/mL, and no pregnancies were observed in women with a midluteal P below 2.3 ng/mL.9 Our findings imply that the reproductive competency of the menstrual cycle of perimenopausal women may be evaluable with a timed midluteal P level. However, hormonal output of the menstrual cycle in reproductively aging women is only one part of predicting fertility; oocyte quality and quantity are also critical predictors of reproductive outcome. Early, mid, and late luteal serum P levels (7.6 [4.6,10.1], 14.7 [10.8, 18.9], and 6.6 [4.2, 10.4] ng/mL, respectively) indicate that an expected pattern of luteal P secreton was likely achieved—an approximately bell-shaped curve rising 1 day after ovulation and peaking in the midluteal phase at or around 7 days postovulation.
Progesterone may be of importance in the menopause transition, apart from its role in ovulation and fertility. Ambient P has been associated with decreased arterial stiffness in 42 midreproductive-aged women who were studied during the early and late follicular phase and again in the luteal phase during confirmed ovulatory cycles.10
We have previously shown that integrated luteal uPdg declines with proximity to the FMP.12 In this study, we observed that Pdg was related to timing of the FMP, but observed only a marginally significant decrease in luteal serum P with proximity to the FMP. Larger studies are needed to confirm this observation. This finding stands to reason, in as much as uPdg represents multiple measurements over the course of the entire luteal phase, and therefore likely reflects most accurately the totality of corpus luteum function. However, urinary measurements are generally believed to be more subject to within and between-woman variation and possibly less reflective of hormone production than are circulating serum hormones. The current study highlights the value of more comprehensive luteal phase sampling in detecting change over time and justifies the use of urinary measurements for this purpose.
This study had some strengths and weaknesses. We know of no other study that has been performed to assess luteal function in perimenopausal women. Because our serum hormones were collected in the context of a menstrual cycle in which women were collecting daily urine samples, it was possible to place each sample within a definite point in the menstrual cycle, with strong confirmation that we observed truly luteal blood samples. On the contrary, our ability to collect a sample within the midluteal window was only successful in about 50% of women, which likely limited our sample to the more regularly cycling women in the sample. Thus, this group may be less representative of the entire pool of women within the luteal pilot sample. Moreover, in breaking down the luteal samples to early, mid, and late luteal phases, sample sizes became small, with cell sizes as small as 22, which may have provided inadequate statistical power to detect some modest but meaningful associations (eg, the relation of serum P to timing of the FMP) as statistically significant. The multiple statistical testing performed in this study may also have led to some spurious findings of statistical significance. Adjusting the P level of significance for some of the statistical comparisons hopefully minimized this type of potential error.
In summary, we have demonstrated that the excellent correspondence between urine and serum hormones is maintained among women who are undergoing the menopause transition, and that urinary hormone tracking of menstrual cycles remains a valid strategy for elucidating the reproductive endocrinology of this time period of a woman's life. Moreover, we have observed that a timed, midluteal P level is reflective of the Pdg output of the entire luteal phase, although the overall strength of the correation was weak. These findings make it possible to study the reproductive endocrinology of the menopause transition in greater detail, with an ability to focus on the corpus luteum.
We thank the study staff at each site and all the women who participated in SWAN.
1. Munro CJ, Stabenfeldt GH, Cragun JR, Addiego LA, Overstreet JW, Lasley BL. Relationship of serum estradiol
concentrations to the excretion profiles of their major urinary metabolites as measured by enzyme immunoassay and radioimmunoassay. Clin Chem
2. Santoro N, Brown JR, Adel T, Skurnick JH. Characterization of reproductive hormonal dynamics in the perimenopause. J Clin Endocrinol Metab
3. Zhang K, Pollack S, Ghods A, et al. Onset of ovulation after menarche in girls: a longitudinal study. J Clin Endocrinol Metab
4. Kravitz HM, Kazlauskaite R, Joffe H. Sleep, health, and metabolism in midlife women and menopause: food for thought. Obstet Gynecol Clin North Am
5. Dugan SA, Gabriel KP, Lange-Maia BS, Karvonen-Gutierrez C. Physical activity and physical function: moving and aging. Obstet Gynecol Clin North Am
6. Verma M, Khadapkar R, Sahu PS, Das BR. Comparing age-wise reference intervals for serum creatinine concentration in a “Reality check” of the recommended cut-off. Indian J Clin Biochem
7. Santoro N, Crawford SL, Allsworth JE, et al. Assessing menstrual cycles with urinary hormone assays. Am J Physiol Endocrinol Metab
8. Hansen KR, Eisenberg E, Baker V, et al. Midluteal progesterone
: a marker of treatment outcomes in couples with unexplained infertility. J Clin Endocrinol Metab
9. Takaya Y, Matsubayashi H, Kitaya K, et al. Minimum values for midluteal plasma progesterone
concentrations in patients who achieved pregnancy with timed intercourse or intrauterine insemination without a human menopausal gonadotropin. BMC Res Notes
10. Spaczynski RZ, Mitkowska A, Florczak M, et al. Decreased large-artery stiffness in midluteal phase of the menstrual cycle
in healthy women of reproductive age. Ginekol Pol
11. Santoro N, Lasley B, McConnell D, et al. Body size and ethnicity are associated with menstrual cycle
alterations in women in the early menopausal transition: The Study of Women's Health across the Nation (SWAN) Daily Hormone Study. J Clin Endocrinol Metab
12. Santoro N, Crawford SL, El Khoudary SR, et al. Menstrual cycle
hormone changes in women traversing menopause: Study of Women's Health Across the Nation. J Clin Endocrinol Metab
13. Sowers MF, Crawford SL, Sternfeld B. Lobo R, Kelsey J, Marcus R. SWAN: a multicenter, multiethnic, community-based cohort study of women and the menopausal transition. Menopause: Biology and Pathobiology
. New York: Academic Press; 2000. 175–188.
14. Randolph JF Jr, Sowers M, Gold EB, et al. Reproductive hormones in the early menopausal transition: relationship to ethnicity, body size, and menopausal status. J Clin Endocrinol Metab
15. England BG, Parsons GH, Possley RM, McConnell DS, Midgley AR. Ultrasensitive semiautomated chemiluminescent immunoassay for estradiol
. Clin Chem
16. Kassam A, Overstreet JW, Snow-Harter C, De Souza MJ, Gold EB, Lasley BL. Identification of anovulation and transient luteal function using a urinary pregnanediol-3-glucuronide ratio algorithm. Environ Health Perspect
17. Waller K, Swan SH, Windham GC, Fenster L, Elkin EP, Lasley BL. Use of urine biomarkers to evaluate menstrual function in healthy premenopausal women. Am J Epidemiol
18. Brown JR, Skurnick JH, Sharma N, Adel T, Santoro N. Frequent intermittent ovarian function in women with premature menopause: a longitudinal study. Endocrine
19. Pal L, Zhang K, Zeitlian G, Santoro N. Characterizing the reproductive hormone milieu in infertile women with diminished ovarian reserve. Fertil Steril
20. Shideler SE, DeVane GW, Kalra PS, Benirschke K, Lasley BL. Ovarian-pituitary hormone interactions during the perimenopause. Maturitas