Methods
This cross-sectional study was performed at the Department of Obstetrics and Gynecology, Haseki Training and Research Hospital, University of Health Sciences, Istanbul, Türkiye, between 1 November 2023 and 2 January 2024. The protocol was approved by the institutional Clinical Research Ethics Committee (approval no. 198–2023; date 1 November 2023) and conducted in accordance with the Declaration of Helsinki. All participants received verbal and written information and provided written informed consent. Reporting follows the STROBE guideline for cross-sectional studies.
Ninety healthy volunteer women aged 18–35 years attending the outpatient clinic or working at the study hospital during the recruitment period, and meeting the eligibility criteria below, were consecutively invited to participate and assigned to two groups according to their established sleep–wake pattern. Group 1 ( n = 46) comprised women with a night-wakefulness/day-sleep rhythm maintained for at least one year (night workers); Group 2 ( n = 44) comprised women with a conventional day-wakefulness/night-sleep rhythm (controls). Group assignment was operationalised from occupational schedule and self-reported habitual sleep timing: Group 1 comprised women whose principal wake period fell between approximately 21:00 and 09:00 for at least one year, most commonly physicians, nurses, and midwives working fixed night duty; Group 2 comprised women with a conventional daytime wake period and night-time sleep. Nightly sleep duration and the number of night duties per month were recorded for all participants.
Exclusion criteria were age 35 years; alcohol, tobacco, or recreational drug use; use of antidepressant, anxiolytic, or other psychotropic medication; hormone replacement therapy or oral contraceptive use; use of vitamin or mineral supplements; polycystic ovary syndrome, congenital adrenal hyperplasia, or thyroid disease; endometriosis; ovarian cyst or prior ovarian surgery; obesity; Müllerian anomaly; a diagnosed sleep disorder; acute or chronic illness or infection; malignancy; previously diagnosed diminished ovarian reserve; ongoing infertility treatment; and pregnancy or lactation.
Demographic and reproductive data (age, gravidity, parity, BMI, occupation, marital status, educational level, socioeconomic status, place of residence, menstrual cycle regularity, and daily sleep duration) were recorded. On day 3 of the menstrual cycle, all participants underwent transvaginal ultrasonography to assess the uterus and ovaries. Endometrial thickness, ovarian volume, and AFC were recorded. Ovarian volume was calculated for each ovary with the ellipsoid formula (D1 × D2 × D3 × π/6) and summed; antral follicles were defined as follicles 2–10 mm in diameter and counted as the sum of both ovaries. All examinations were performed by a single examiner (E.G.); to assess reproducibility, 25 randomly selected participants were independently re-scanned by a second examiner (N.C.), and inter-observer agreement for the antral follicle count was excellent (intraclass correlation coefficient = 0.94, 95% CI 0.88–0.97).
Day-3 venous blood was sampled for FSH, LH, estradiol, prolactin, TSH, AMH, and vitamin D, measured by standard laboratory methods. To reflect each participant’s habitual circadian phase, samples were obtained at approximately 09:00 in day-active women and approximately 21:00 in night-active women. For kisspeptin and melatonin, an additional ~ 2 mL sample was centrifuged (3000 rpm, 15 min) and the serum stored at − 80 °C until batch analysis. Serum kisspeptin and melatonin were quantified by commercial enzyme-linked immunosorbent assay (ELISA) kits (catalogue nos. E3122Hu and E1013Hu; BT Lab, China) according to the manufacturer’s instructions. According to the manufacturer, the intra- and inter-assay coefficients of variation were 3.7–6.4% and < 10% for melatonin and 2.9–6.6% and < 10% for kisspeptin. Storage time ranged from one week to two months.
Continuous variables are summarised as mean ± standard deviation when normally distributed and as median (interquartile range) otherwise, and categorical variables as counts and percentages. Distribution was assessed with the Shapiro–Wilk test. Between-group comparisons used the independent-samples t test for normally distributed variables and the Mann–Whitney U test otherwise, and the chi-square or Fisher exact test for categorical variables. Associations were examined with Spearman correlation. Because gravidity and parity differed between groups, the primary outcomes (AMH, AFC, ovarian volume, kisspeptin, melatonin, TSH) were additionally analysed by analysis of covariance (ANCOVA) with age, BMI, and gravidity as covariates; adjustment did not alter the direction or significance of any between-group comparison. A two-tailed p < 0.05 was considered significant. No a priori sample-size calculation was performed, because no comparable prior study provided an expected effect size for kisspeptin or melatonin in this population on which to base one. As a sensitivity analysis (independent-samples t test, two-sided α = 0.05, 80% power, G*Power 3.1), with 46 and 44 participants per group the study was powered to detect a standardised between-group difference (Cohen’s d) of 0.59 or larger; smaller true differences of d = 0.5, 0.3, or 0.2 would have required approximately 64, 175, and 393 participants per group, respectively, so smaller effects could not be excluded. For the primary ovarian-reserve outcome (AMH), the effect size actually observed in this cohort was d = − 0.10 (pooled SD 2.31 ng/mL); reliably detecting an effect of this magnitude would require approximately 1600 participants per group, several orders of magnitude beyond the feasible sample for a single-centre study. Kisspeptin and melatonin, by contrast, differed with large effect sizes (d = 0.84 and 0.90; see Results), well within the range this sample could detect. The null finding for ovarian reserve should therefore be interpreted as showing no clinically large difference, not as excluding a small one. Analyses were performed with SPSS version 28.0 (IBM Corp., Armonk, NY, USA).
Results
All 90 enrolled women completed the study, with no missing data for any variable. The two groups were well matched for the principal determinants of ovarian reserve: age (Group 1, 27.5 ± 2.6 years; Group 2, 27.0 ± 5.0 years; p = 0.392) and BMI (23.2 ± 4.3 vs. 23.5 ± 4.0 kg/m², p = 0.677) were nearly identical, and the cohort as a whole was young (median 28 years; range 18–35) with normal body habitus (median BMI 22.7 kg/m²). As expected from the recruitment frame, the groups differed in occupational background: physicians and nurses/midwives were over-represented in Group 1 ( p < 0.001). Marital status and socioeconomic status did not differ between groups ( p = 0.301 and p = 0.068), but educational attainment did ( p < 0.001), tracking the same occupational composition: night-active women were more often physicians or held a master’s degree, whereas day-active women more often worked as nursing aides or had completed only high school. All participants resided in an urban setting, so place of residence did not vary and could not confound the comparison. Habitual sleep was approximately 28 min shorter in night-active women (6.9 ± 0.7 vs. 7.3 ± 1.2 h; p = 0.019; Cohen’s d = − 0.48), a small-to-moderate difference consistent with the curtailed daytime sleep typical of night work (Table 1 ).
Table 1 Sociodemographic and occupational characteristics of the two groups Variable Group 1 ( n = 46) Group 2 ( n = 44)
p
Age (years) 27.5 ± 2.6 27.0 ± 5.0 0.392 BMI (kg/m²) 23.2 ± 4.3 23.5 ± 4.0 0.677 Marital status (married), n (%) 15 (32.6) 20 (45.5) 0.301 Education, n (%) 0.001* High school 1 (2.2) 11 (25.0) Bachelor’s degree 34 (73.9) 30 (68.2) Master’s degree 11 (23.9) 3 (6.8) Occupation, n (%) 0.001* Physician 11 (23.9) 3 (6.8) Nurse/midwife 34 (73.9) 30 (68.2) Nursing aide 1 (2.2) 11 (25.0) Socioeconomic status: high, n (%) 21 (45.7) 11 (25.0) 0.068 Place of residence: urban, n (%) 46 (100) 44 (100) — Sleep duration (h) 6.9 ± 0.7 7.3 ± 1.2 0.019 Data are mean ± SD unless stated. Group 1, night-wakefulness/day-sleep; Group 2, day-wakefulness/night-sleep. Continuous variables compared by Mann–Whitney U test; categorical variables by chi-square or Fisher exact test. Values in parentheses for categorical variables are percentages within that group unless otherwise stated. Education and occupation categories are mutually exclusive within each group; the omnibus chi-square p value for Education and Occupation is shown on the category header row; indented rows below each header give the breakdown by category and are not tested individually *Omnibus chi-square p value across all categories of that variable BMI body mass index, SES socioeconomic status
Sociodemographic and occupational characteristics of the two groups
Data are mean ± SD unless stated. Group 1, night-wakefulness/day-sleep; Group 2, day-wakefulness/night-sleep. Continuous variables compared by Mann–Whitney U test; categorical variables by chi-square or Fisher exact test. Values in parentheses for categorical variables are percentages within that group unless otherwise stated. Education and occupation categories are mutually exclusive within each group; the omnibus chi-square p value for Education and Occupation is shown on the category header row; indented rows below each header give the breakdown by category and are not tested individually
*Omnibus chi-square p value across all categories of that variable
BMI body mass index, SES socioeconomic status
Reproductive history also differed between the groups: gravidity and parity were both lower in night-active women (gravidity present: 9 [19.6%] vs. 20 [45.5%], p = 0.009; parity present: 8 [17.4%] vs. 20 [45.5%], p = 0.004), whereas abortion rate, mode of delivery, and menstruation duration did not differ (Table 2 ).
Table 2 Reproductive and menstrual characteristics of the two groups Variable Group 1 ( n = 46) Group 2 ( n = 44)
p
Gravidity present, n (%) 9 (19.6) 20 (45.5) 0.009 Parity present, n (%) 8 (17.4) 20 (45.5) 0.004 Abortion present, n (%) 2 (4.4) 2 (4.5) 1.000 Menstruation duration (days) 5.8 ± 1.3 6.1 ± 1.6 0.187 Cycle length (days) 28.0 (28.0–30.0) 28.0 (28.0–30.0) 0.549 Data are mean ± SD unless stated. Group 1, night-wakefulness/day-sleep; Group 2, day-wakefulness/night-sleep. Continuous variables compared by Mann–Whitney U test; categorical variables by chi-square or Fisher exact test. Gravidity present denotes at least one prior pregnancy of any outcome; parity present denotes at least one prior delivery at ≥ 20–24 weeks’ gestation, live or stillborn. Menstruation duration (bleeding days per cycle) is distinguished from cycle length (interval between cycles); neither differed between groups. Values in parentheses for gravidity, parity, and abortion are percentages within that group; the value in parentheses for cycle length is the interquartile range (menstruation duration is reported as mean ± SD, with no parenthetical value)
Reproductive and menstrual characteristics of the two groups
Data are mean ± SD unless stated. Group 1, night-wakefulness/day-sleep; Group 2, day-wakefulness/night-sleep. Continuous variables compared by Mann–Whitney U test; categorical variables by chi-square or Fisher exact test. Gravidity present denotes at least one prior pregnancy of any outcome; parity present denotes at least one prior delivery at ≥ 20–24 weeks’ gestation, live or stillborn. Menstruation duration (bleeding days per cycle) is distinguished from cycle length (interval between cycles); neither differed between groups. Values in parentheses for gravidity, parity, and abortion are percentages within that group; the value in parentheses for cycle length is the interquartile range (menstruation duration is reported as mean ± SD, with no parenthetical value)
Neither the sonographic nor the biochemical indices of ovarian reserve distinguished the two groups, and the magnitude of every between-group difference was negligible. Ovarian volume (14.3 ± 4.9 vs. 13.8 ± 4.8 cm³; p = 0.623; d = 0.08) and AFC (median 6.0 [IQR 4.0–9.0] vs. 6.0 [5.0–8.0]; p = 0.791; d = 0.01) were essentially superimposable, as was serum AMH (2.7 [1.8–4.3] vs. 3.1 [1.4–4.7] ng/mL; p = 0.862; d = − 0.09). The gonadotropic axis showed the same pattern: day-3 FSH (5.6 [4.9–7.1] vs. 6.2 [5.2–7.2] U/L), LH (6.3 ± 2.3 vs. 6.7 ± 2.8 U/L), the FSH/LH ratio (0.94 [0.71–1.29] vs. 0.93 [0.79–1.25]), estradiol, and prolactin were all comparable (all p > 0.05), and menstrual cycle length was similar (28.0 [28.0–30.0] vs. 28.0 [28.0–30.0] days; p = 0.549). Serum vitamin D was numerically higher in night-active women but did not reach significance (14.3 [8.4–19.7] vs. 10.1 [6.6–16.7] µg/L; p = 0.065); both groups fell within the deficient range, in keeping with mid-winter recruitment. Thus, across six complementary markers spanning follicle number, follicle reserve, and gonadotropic drive, circadian pattern was not associated with any detectable difference (Table 3 ).
Table 3 Ovarian reserve markers and neuroendocrine parameters Parameter Group 1 ( n = 46) Group 2 ( n = 44)
p
Ovarian volume (cm³) 14.3 ± 4.9 13.8 ± 4.8 0.623 Antral follicle count 6.0 (4.0–9.0) 6.0 (5.0–8.0) 0.791 FSH (U/L) 5.6 (4.9–7.1) 6.2 (5.2–7.2) 0.586 LH (U/L) 6.3 ± 2.3 6.7 ± 2.8 0.554 FSH/LH ratio 0.94 (0.71–1.29) 0.93 (0.79–1.25) 0.910 Estradiol (ng/L) 37.6 (26.5–52.6) 35.0 (27.1–46.8) 0.473 Prolactin (µg/L) 15.6 (9.9–22.8) 13.9 (9.4–18.9) 0.252 AMH (ng/mL) 2.7 (1.8–4.3) 3.1 (1.4–4.7) 0.862 Vitamin D (µg/L) 14.3 (8.4–19.7) 10.1 (6.6–16.7) 0.065 TSH (mU/L) 2.0 (1.4–3.0) 1.5 (1.1–2.3) 0.020 Kisspeptin (ng/L) 889.5 (404.4–1080.6) 336.2 (184.3–689.1) < 0.001 Melatonin (ng/L) 699.9 (379.3–1037.4) 225.1 (147.9–459.7) < 0.001 Data are median (interquartile range), except ovarian volume and LH, which were normally distributed in both groups and are presented as mean ± SD. Normality was assessed with the Shapiro–Wilk test. Bold rows indicate p < 0.05; p -values are from the independent-samples t test for ovarian volume and LH and from the Mann–Whitney U test for all other variables. Group differences in kisspeptin and melatonin remained significant after ANCOVA adjustment for age, BMI, and gravidity; the TSH difference did not persist after adjustment AMH anti-Müllerian hormone, FSH follicle-stimulating hormone, LH luteinizing hormone, TSH thyroid-stimulating hormone
Ovarian reserve markers and neuroendocrine parameters
Data are median (interquartile range), except ovarian volume and LH, which were normally distributed in both groups and are presented as mean ± SD. Normality was assessed with the Shapiro–Wilk test. Bold rows indicate p < 0.05; p -values are from the independent-samples t test for ovarian volume and LH and from the Mann–Whitney U test for all other variables. Group differences in kisspeptin and melatonin remained significant after ANCOVA adjustment for age, BMI, and gravidity; the TSH difference did not persist after adjustment
AMH anti-Müllerian hormone, FSH follicle-stimulating hormone, LH luteinizing hormone, TSH thyroid-stimulating hormone
In sharp contrast to the uniform ovarian-reserve findings, the circadian-sensitive markers separated the groups with large effect sizes. Serum kisspeptin was markedly higher in night-active women (median 889.5 [IQR 404.4–1080.6] vs. 336.2 [184.3–689.1] ng/L; p < 0.001; d = 0.84), as was serum melatonin (699.9 [379.3–1037.4] vs. 225.1 [147.9–459.7] ng/L; p < 0.001; d = 0.90). Both distributions were non-normal (right-skewed in day-active women but approximately symmetric to mildly left-skewed in night-active women) and wide, but the group separation was robust to this dispersion. TSH was also higher in Group 1 (median 2.0 [1.4–3.0] vs. 1.5 [1.1–2.3] mU/L; p = 0.020; d = 0.35), a small effect with both groups remaining comfortably within the euthyroid reference range. The contrast between the large kisspeptin and melatonin effects (d ≈ 0.8–0.9) and the near-zero ovarian-reserve effects (d ≤ 0.10) is the central quantitative finding of the study. Importantly, the elevations in kisspeptin and melatonin each remained significant in ANCOVA after adjustment for age, BMI, and gravidity (and after log-transformation), indicating that they were not artefacts of the groups’ differing reproductive background; the TSH difference, by contrast, did not survive covariate adjustment (ANCOVA p = 0.21), so it is best regarded as a weak and unstable signal (Table 3 ).
We calculated the AMH/kisspeptin, AMH/melatonin, and kisspeptin/melatonin ratios as an internal-consistency check: if AMH were confounded by the same circadian or reproductive-background factors driving the kisspeptin and melatonin differences, the ratios would be expected to shift as those numerators or denominators changed. Because AMH was unchanged while its denominators kisspeptin and melatonin rose, the derived ratios shifted in the expected direction: the AMH/kisspeptin and AMH/melatonin ratios were lower in night-active women ( p = 0.008 and p = 0.005), and the kisspeptin/melatonin ratio was higher ( p = 0.011). These ratios are deterministic functions of their components rather than independent measurements, and we report them only for completeness; they are interpreted no further (Supplementary Table S1). The divergent pattern between the circadian-sensitive markers and the unchanged ovarian-reserve indices is summarised in Fig. 1 .
Fig. 1 Between-group comparison of serum neuroendocrine markers and anti-Müllerian hormone. Boxes show the median and interquartile range (whiskers, 1.5×IQR; circles, individual participants) for Group 1 (night-wakefulness/day-sleep; n = 46) and Group 2 (day-wakefulness/night-sleep; n = 44). A Kisspeptin and ( B ) melatonin were markedly higher in night-active women, ( C ) TSH was modestly higher, whereas ( D ) AMH did not differ. p values are from the Mann–Whitney U test; * p < 0.05, *** p < 0.001; ns, not significant. AMH, anti-Müllerian hormone; TSH, thyroid-stimulating hormone
Between-group comparison of serum neuroendocrine markers and anti-Müllerian hormone. Boxes show the median and interquartile range (whiskers, 1.5×IQR; circles, individual participants) for Group 1 (night-wakefulness/day-sleep; n = 46) and Group 2 (day-wakefulness/night-sleep; n = 44). A Kisspeptin and ( B ) melatonin were markedly higher in night-active women, ( C ) TSH was modestly higher, whereas ( D ) AMH did not differ. p values are from the Mann–Whitney U test; * p < 0.05, *** p < 0.001; ns, not significant. AMH, anti-Müllerian hormone; TSH, thyroid-stimulating hormone
Across the whole cohort, neither age nor BMI correlated with any ovarian-reserve marker, the sole exception being a weak inverse association between BMI and LH ( r = − 0.231; p = 0.028). Menstruation duration correlated weakly and negatively with vitamin D, kisspeptin, and melatonin (r between − 0.23 and − 0.24; all p < 0.05). Within night-active women, two work-exposure metrics showed weak associations: monthly night-shift count correlated inversely with FSH ( r = − 0.317; p = 0.032) and positively with AMH ( r = 0.363; p = 0.013), and the number of years of night-shift employment correlated positively with ovarian volume ( r = 0.355; p = 0.016) and AFC ( r = 0.301; p = 0.042). All coefficients were modest (|r|≤0.36), the FSH and AMH correlations point in opposing directions, and no correction was made for the many comparisons performed; these associations are therefore hypothesis-generating only and should not be read as evidence of a dose-response relationship between night-shift exposure and ovarian reserve (Table 4 ).
Table 4 Selected statistically significant correlations (Spearman) Association Group
r
p
BMI – LH All −0.231 0.028 Menstruation duration – vitamin D All −0.234 0.026 Menstruation duration – kisspeptin All −0.233 0.027 Menstruation duration – melatonin All −0.243 0.021 Night-shift count – FSH Group 1 −0.317 0.032 Night-shift count – AMH Group 1 0.363 0.013 Night-shift duration – ovarian volume Group 1 0.355 0.016 Night-shift duration – AFC Group 1 0.301 0.042 Only associations reaching p <0.05 are shown. Given the large number of correlations tested and the absence of correction for multiple comparisons, these results are exploratory AFC antral follicle count, AMH anti-Müllerian hormone, FSH follicle-stimulating hormone, LH luteinizing hormone
Selected statistically significant correlations (Spearman)
Only associations reaching p <0.05 are shown. Given the large number of correlations tested and the absence of correction for multiple comparisons, these results are exploratory
AFC antral follicle count, AMH anti-Müllerian hormone, FSH follicle-stimulating hormone, LH luteinizing hormone
Conclusion
The primary effect of a night-active circadian rhythm identified in this study is neuroendocrine, not gonadal: it is associated with a higher circadian-sensitive kisspeptin and melatonin signal, while ovarian reserve itself remains unaffected within the timeframe studied. In young, normal-BMI women, a night-wakefulness/day-sleep circadian rhythm was not associated with altered ovarian reserve, as ovarian volume, AFC, AMH, gonadotropins, estradiol, and prolactin were comparable to those of day-active women. The same rhythm was associated with higher serum kisspeptin and melatonin that persisted after adjustment for age, BMI, and gravidity, and with a modestly higher TSH that did not survive adjustment. Because the two groups were sampled at opposite points of the daily melatonin curve (~ 21:00 vs. ~ 09:00), these elevations cannot be distinguished from a phase-of-sampling effect and should be regarded as biomarkers of circadian state rather than as evidence of increased neuroendocrine output; we therefore caution against treating these markers, or ratios derived from them, as surrogates of ovarian reserve. Longitudinal studies with serial, phase-controlled sampling are required to establish the long-term reproductive significance of these neuroendocrine differences.
Discussion
In this cross-sectional study of young, normal-BMI women, a night-wakefulness/day-sleep rhythm was not associated with any difference in ovarian reserve, whether assessed sonographically (ovarian volume, AFC) or biochemically (AMH), or in the day-3 gonadotropins, estradiol, or prolactin. The same women, however, had substantially higher serum kisspeptin and melatonin and modestly higher TSH. The primary and methodologically cleanest finding is the preservation of ovarian reserve: AMH, FSH, LH, and estradiol were measured on the hospital’s routine, validated automated platform and are therefore unaffected by the difference in sampling time that complicates interpretation of the neuroendocrine markers. The higher kisspeptin, melatonin, and TSH are accordingly treated throughout as secondary, hypothesis-generating observations, interpreted as biomarkers of circadian state rather than as evidence of altered hormone output.
The absence of a difference in ovarian reserve is consistent with the best available occupational evidence. Johnson et al. found no overall association between night-shift work and AMH in a large prospective cohort of nurses [ 16 ], and Mínguez-Alarcón et al. reported no association of shift work with AMH or AFC among women attending a fertility centre [ 17 ]. Our cohort, restricted to women younger than 35 years with normal BMI and normal baseline reserve, represents a population in whom age- and adiposity-related effects on the follicle pool are minimal, which probably explains why no between-group difference emerged. It is important not to over-interpret a null finding: a cross-sectional comparison cannot exclude a small or slowly accruing effect, and AMH and AFC capture the quantity rather than the reproductive competence of the follicle pool.
The higher serum melatonin in night-active women is the opposite of the suppression usually described in shift workers, in whom nocturnal light typically lowers melatonin [ 18 ]. The most likely explanation is the timing of sampling relative to each woman’s habitual phase: blood was drawn near the physiological melatonin rise (~ 21:00) in night-active women and during the daytime trough (~ 09:00) in day-active women, so the contrast reflects phase-aligned sampling as much as any change in total output. Because melatonin is also an intrafollicular antioxidant correlated with AFC and AMH [ 19 , 20 ], the preserved ovarian reserve despite higher circulating melatonin is best read as concordant with normal ovarian physiology rather than as evidence of benefit [ 21 ].
The higher kisspeptin in night-active women is biologically coherent with the position of arcuate kisspeptin neurons within the circadian and metabolic timing network, and with the documented, largely indirect, crosstalk between melatonin and kisspeptin signalling [ 22 ]. Notably, the elevation in kisspeptin was not accompanied by any change in downstream FSH, LH, or the FSH/LH ratio. This dissociation suggests that, in healthy young women, the hypothalamic–pituitary–gonadal axis buffers moderate changes in upstream kisspeptin, leaving gonadotropin output and ovarian reserve unchanged over the period studied. Our finding that kisspeptin did not track AMH between groups contrasts with the positive kisspeptin–AMH correlation reported in PCOS cohorts a difference plausibly attributable to our exclusion of PCOS and to the narrow, normal range of reserve in our participants [ 23 , 24 ].
The parallel rise in kisspeptin and melatonin is consistent with both systems tracking the altered light–dark and sleep–wake schedule rather than acting in opposition. Our cross-sectional design cannot separate a phase-of-sampling effect from a genuine, sustained re-setting of these rhythms; that gonadotropins, AMH, and antral follicle count were unchanged indicates that, either way, any upstream change had not reached the ovary over the period studied.
TSH was modestly higher in night-active women while remaining within the euthyroid range. This accords with evidence that circadian disruption and shift work can raise TSH [ 25 , 26 ] and may reflect stress- or sleep-related modulation of the hypothalamic–pituitary–thyroid axis; the small magnitude argues against clinical thyroid dysfunction. The derived ratios involving kisspeptin and melatonin differed between groups, but because AMH was unchanged these ratios are arithmetic by-products of their numerators and denominators rather than independent markers, and we caution against interpreting them as novel indices of ovarian reserve.
These findings carry a measured message for counselling women in night-shift occupations, who are over-represented in healthcare. Our data give no basis for warning young, healthy night-active women that their schedule is depleting their ovarian reserve: ovarian volume, antral follicle count, and AMH were indistinguishable from those of day-active peers. This reassurance is bounded by the cross-sectional design and the young age of the cohort and does not extend to the established links between long-term shift work and menstrual disturbance, subfertility, or earlier menopause. The neuroendocrine differences are best framed as biomarkers of circadian state rather than indicators of reproductive harm; the more relevant counselling points remain sleep sufficiency and mitigation of circadian misalignment [ 27 ].
The strengths of this study include the sampling timed to each participant’s habitual active period, the rigorous exclusion of conditions that confound ovarian reserve (PCOS, thyroid disease, obesity, smoking), and the simultaneous assessment of sonographic, gonadotropic, and neuroendocrine markers. Restriction to young, normal-BMI women reduces confounding and isolates the effect of circadian pattern in a relatively homogeneous group.
Several limitations temper the conclusions. The single-centre design, two-month recruitment window, and modest sample size limited power: as detailed in Methods, the study could reliably exclude only moderate-to-large differences (d ≥ ~ 0.59), and the ovarian-reserve null finding (observed d = − 0.10 for AMH) should be read as showing no large difference rather than definitive equivalence, whereas the kisspeptin and melatonin effects (d ≈ 0.8–0.9) were well within the detectable range; the exploratory correlations were uncorrected for multiple comparisons. Most importantly, blood was sampled at different clock times in the two groups (~ 21:00 vs. ~ 09:00), on opposite limbs of the melatonin rhythm, so the higher kisspeptin, melatonin, and TSH cannot be separated from a time-of-sampling effect [ 28 ]. Circadian group was assigned from occupation and self-reported habits, without actigraphy, light dosimetry, or a validated chronotype instrument; ambient light exposure, sleep quality, and cortisol were unmeasured [ 29 ]. Night-active women were also disproportionately physicians, nurses, or midwives with higher educational attainment, and recruitment spanned only the autumn–winter months, so occupational, educational, and seasonal factors are incompletely separable from circadian pattern itself and should be borne in mind when interpreting the neuroendocrine findings. Finally, kisspeptin and melatonin were measured by commercial ELISA, which over-reads melatonin through cross-reactivity with related indoleamines [ 30 , 31 ], so absolute values should be read as relative rather than absolute; the markers underpinning our principal conclusion (AMH, FSH, LH, estradiol) were measured on the hospital’s validated automated platform and are unaffected.
Taken together, these findings indicate that, in young women with normal BMI, differences in circadian rhythm are reflected in circadian-sensitive neuroendocrine markers, kisspeptin, melatonin, and TSH, rather than in static measures of ovarian reserve. Longitudinal studies with phase-controlled, repeated sampling, objective light and sleep monitoring, and measurement of intrafollicular melatonin are needed to determine whether sustained circadian disruption affects reproductive outcomes over time [ 32 ].
Introduction
Shift work has become an integral feature of modern economies, requiring a substantial proportion of the workforce, including many women of reproductive age, to remain awake at night and sleep during the day. This inversion of the sleep–wake cycle conflicts with the endogenous circadian system governed by the hypothalamic suprachiasmatic nucleus, which times the secretion of numerous hormones, including melatonin, cortisol, and the reproductive hormones [ 1 ]. Epidemiological evidence links shift work to menstrual disturbances and reduced fecundity: a meta-analysis reported higher odds of menstrual irregularity among shift workers [ 2 , 3 ], and prospective cohort data have associated rotating night work with menstrual dysfunction and impaired fecundity, as well as with earlier age at menopause [ 4 , 5 ].
Ovarian reserve, the size and quality of the remaining follicle pool, is most commonly estimated in clinical practice by AMH, AFC, and ovarian volume [ 6 , 7 ]. Because circadian disruption alters the hormonal milieu in which follicles develop, it is plausible that chronic night work could accelerate the decline of ovarian reserve. The available evidence, however, is mixed: two well-conducted occupational studies found no association between night-shift work and AMH or AFC, suggesting that any effect, if present, is small in young women [ 8 , 9 ].
Two circadian-sensitive signalling molecules are of particular interest. Kisspeptin, the product of the KISS1 gene, is the principal upstream activator of gonadotropin-releasing hormone (GnRH) neurons and is essential for pubertal onset, gonadotropin secretion, and ovulation [ 10 , 11 ]; exogenous kisspeptin can even trigger oocyte maturation in assisted reproduction [ 12 ]; kisspeptin-secreting neurons of the arcuate nucleus also participate in the circadian and metabolic timing network [ 13 , 14 ]. Melatonin, secreted by the pineal gland during darkness, is a potent intrafollicular antioxidant that supports oocyte maturation and may protect against age-related ovarian decline [ 15 ]. Both molecules are therefore biologically positioned at the intersection of circadian timing and ovarian function, yet their serum behaviour in women with an inverted sleep–wake schedule has rarely been examined together with ovarian reserve markers. Assessing them alongside AMH and AFC has both mechanistic and clinical rationale: because kisspeptin acts upstream of the hypothalamic–pituitary–gonadal axis while melatonin acts locally within the follicle, a joint panel can indicate whether a circadian signal detected upstream (kisspeptin, melatonin) has already propagated downstream to measurable ovarian reserve, or remains confined to the neuroendocrine level, information that neither marker class provides on its own.
We therefore conducted a cross-sectional study comparing women with a night-wakefulness/day-sleep rhythm and women with a conventional day-wakefulness/night-sleep rhythm. Our research question was whether an established night-active circadian rhythm is associated with a detectably diminished ovarian reserve, and, in parallel, whether it is associated with altered circulating kisspeptin and melatonin. We hypothesised that, in young women with normal BMI, ovarian reserve would be preserved across the two rhythms, whereas circadian-sensitive neuroendocrine markers, specifically kisspeptin and melatonin, would differ. The objective, directly aligned with this question, was to evaluate ovarian reserve and these neuroendocrine markers in parallel within the same cohort, in order to characterise the reproductive endocrine signature of circadian rhythm difference rather than to infer a change in fertility potential.
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.