Introduction

The regular menstrual cycle is an important indicator for women’s reproductive, physical, as well as mental health, serving as an independent proxy for the overall health status among women of reproductive age1. Yet, 5 to 36% of women are affected by irregular menstrual cycles, depending on age, country of residence, or occupation2. These irregularities in the menstrual cycle can be attributed to various underlying causes, such as a dysfunction of the hypothalamic-pituitary-ovarian (HPO) axis, which influences the production of estradiol. Research findings suggest that prolonged and irregular cycle length are associated with decreased exposure to estradiol3. It should be noted that 69% of the variance in total menstrual cycle length is due to the variance of the follicular phase length when estradiol is the dominating hormone4,5.

Estrogens, including estradiol, estrone, and estriol, in conjunction with progesterone represent a major class of neuromodulatory sex steroid hormones with the brain serving as an important target for their actions6,7. Estradiol is the most biologically active estrogen in humans8. It is approximately 10 times more active than estrone and 80 times more active than estriol9. Both estrone and estriol are a precursors and metabolites of estradiol10,11. While cells in many brain regions express estrogen and progesterone receptors, an increased presence of receptors is found in the hippocampus ex vivo in humans12 and in rodents13. It is well known that the hippocampus plays a major role in memory and control of attention. Investigating the effect of endogenous estrogens and progesterone on hippocampal neuroplasticity in humans in vivo is a relatively recent scientific effort. Traditionally, data collection relies predominantly on cross-sectional designs that involve simultaneous data collection from multiple individuals, followed by mean comparisons to establish hormone-brain connections based on this aggregated data (e.g.14,15,16,17,18). Nevertheless, this cross-sectional methodology tends to disregard the rhythmic nature of hormone production in the human body. Recent years have marked a transformation in neuroimaging studies, embracing an alternative approach that entails the longitudinal monitoring of individuals over extended periods spanning weeks and months19,20. Following this, a series of recent neuroimaging studies have densely sampled women across the full menstrual cycle21,22,23,24,25,26,27,28 to enrich our understanding of hormone action in the human brain. Barth et al.21 showed in a longitudinal study of a natural cycling single woman21 a positive association between endogenous estrogen concentrations and bilateral white matter hippocampal fractional anisotropy, an indicator of the microstructural properties in white matter29. The interplay between regional brain volume and hormonal levels was also demonstrated in another longitudinal study of one woman, demonstrating no effect of estradiol, but a relation between endogenous progesterone concentrations and gray matter volume in the hippocampal subfields of CA2/3 and paralimbic structures of the parahippocampal gyrus, perirhinal and entorhinal cortex. Subsequent pharmacological suppression of progesterone eliminated these effects22. However, these studies were performed in women with a regular menstrual cycle, and it is unknown whether the relationship is present in those with irregular menstrual cycles.

The hippocampus is implicated in mental health disorders such as depression30. Women are approximately twice as likely to be diagnosed with major depressive disorder (MDD) compared to men31, and there is a recognized connection between fluctuations in ovarian hormones and depression susceptibility in women32,33. Particularly noteworthy is the association between menstrual cycle irregularities and mental disorders, including depression34,35. While the majority of individuals with regular menstrual cycles are unlikely to undergo impairing changes in affect associated with cycling hormones, a minority of individuals experience such changes36,37. An example of those who experience changes in affect with cycling hormones are individuals with premenstrual dysphoric disorder (PMDD), a severe type of premenstrual syndrome, which is observed in 2 to 8% of women in their reproductive years. PMDD is characterized by cyclic mood alterations leading to clinically marked distress and functional impairment38,39. There is a growing body of literature showing that hormonal fluctuations, rather than stable hormone levels, impact affect40,41,42, identifying mood sensitivity to fluctuating hormones43,44,45. Notably, current literature excludes individuals with irregular menstrual cycles. Despite progress in assessing menstrual cycles, there remains a dearth of approaches specifically tailored to study irregular cycles, pointing to a critical gap in the existing literature. Since menstrual cycle length varies between individuals (the average is between 21 and 37 days) and within individuals cycle-to-cycle46,47,48, self-report assessments of menstrual cycle stages are often misleading. Further, saliva-based assessments are subpar49. Recognizing the menstrual cycle as a within-person process, repeated measure designs, which are considered the state-of-the-art approach5, can capture a comprehensive picture of menstrual cycle irregularities. However, within-subject designs are challenging in terms of the demands placed on the participant. Therefore, these dense-sampling studies, as described above, are still scarce.

This study seeks to narrow the gap in the current literature by focusing on menstrual cycle irregularities, aiming to further our understanding of the complex interplay between hormone fluctuations, hippocampal morphology, and effects beyond the regular menstrual cycle pattern. In a dense-sampling study of one female participant with an irregular menstrual cycle, we explored whether endogenous fluctuations in sex hormone concentrations impact hippocampal morphology and affect. First, we investigated associations of endogenous concentrations of estradiol, estrone, and progesterone with hippocampal morphology across five consecutive weeks (n = 25 testing days), covering mostly the follicular phase and ovulation. Second, we determined the association between the concentrations of circulating hormones, hippocampal morphology, and affect. This investigation aims to provide insights into brain-hormones-behavior interactions which are usually studied in regular menstrual cycles but often overlooked in cases of menstrual cycle irregularities.

Methods

Participant

A healthy female (30 years of age) participated in this dense-sampling, longitudinal study. The participant underwent testing mostly from Monday to Friday for five consecutive weeks (August 2nd–September 2nd, 2022) while freely cycling, resulting in n = 25 test sessions. The female participant was free from hormonal medication for 55 months (mean menstrual cycle length = 38.3 days, SD = 6.68 days during that time) before the assessment. The female participant had no history of psychiatric, neurological, and endocrine diagnoses, breastfeeding or pregnancy, and no history of smoking, alcohol, or drug abuse. The participant gave written informed consent, and the Friedrich Schiller University Jena Ethics Committee approved the study.

Image acquisition and postprocessing

The imaging data were acquired on a 3T Siemens PrismaFit scanner (Siemens Medical Solutions, Erlangen, Germany) with a 64-channel head coil. Structural MRI were acquired with T1-weighted (T1w) MPRAGE sequence with GRAPPA acceleration. Scan parameters were: echo time (TE) = 2.22 ms, repetition time (TR) = 2400 ms, inversion time (TI) = 1000 ms, matrix size = 320 × 320, field of view (FOV) = 256 mm, flip angle = 8°, scan orientation = sagittal, phase encoding direction = A >> P, bandwidth = 220 Hz/pixel, number of slices = 208, slice thickness = 0.80 mm, voxel size = 0.80 × 0.80 × 0.80 mm. Scans were collected every day at 7.30 am local time. The parameters used to acquire the images (e.g., sizes, space directions, space origin), and the quality of the images (e.g., motion artifacts, ringing, ghosting of the skull or eyeballs, cut-offs, signal drops, and other artifacts) were visually checked. Sequence Adaptive Multimodal SEGmentation (SAMSEG)50 was used to segment both hemispheres’ total hippocampal volumes. Initially, a subject-specific template was created by spatially co-registering all 3D T1w MRPAGE volumes through an iterative process51. The co-registered 3D volumes were then employed to implement longitudinal SAMSEG52. The final segmentations were visually quality checked and then used to extract the volume of the hippocampus structure directly for each measurement day. The numerical values of the left and right hippocampal volumes were demeaned by subtracting the hemisphere mean from each individual value, for both hemispheres respectively. Demeaning the hippocampal volume serves to center it around zero, allowing us to focus on variations around the average, which is crucial for identifying patterns in the data related to the actual variation attributed to the day of the menstrual cycle. This preprocessing step is essential for isolating variations associated with hormonal fluctuations while removing potential sources of systematic bias or shift. It helps eliminate variations caused by external factors or measurement errors. By removing these sources of variation, we can focus more precisely on changes in hippocampal volumes linked to hormonal fluctuations. As a result, this process enhances the sensitivity and specificity of our measurements. The demeaned hippocampal volumes were then added and divided by two to obtain an average demeaned hippocampal volume.

Endocrine procedure

The blood was drawn at 9.00 am. One 7.5 ml blood sample was collected in a S-Monovette® Serum-GEL (Sarstedt) with clotting activator/gel each test session. The sample clotted at room temperature and was stored at 5° Celsius until centrifugated (2500 x g for 10 minutes). Estradiol (pg/ml), progesterone (ng/ml), and luteinizing hormone (LH) serum concentrations (IU/l) were determined at the Bioscientia Laboratory in Jena, Germany. Estrone serum concentrations (pg/ml) were determined at the Bioscientia Laboratory in Ingelheim, Germany. Estradiol was assessed with the electrochemiluminescence immunoassay (ECLIA) Elecsys® Estradiol III Assay. Assay antibodies, measuring ranges (defined by the limit of detection and the maximum of the master curve), and intra-assay coefficients of variation for estradiol were the following: antibodies, two biotinylated monoclonal anti-estradiol antibodies (rabbit), 2.5 ng/ml and 4.5 ng/ml; measuring range, 18.4–11,010 pmol/l (5–3000 pg/ml), < 5% relative SD. Radioimmunoassay (RIA) was used to determine concentrations of estrone.

Progesterone was assessed with the ECLIA Elecsys® Progesterone III Assay. Assay antibodies, measuring ranges, and intra-assay coefficients of variation for progesterone were the following: antibodies, biotinylated monoclonal anti-progesterone antibody (recombinant sheep), 30 ng/ml; measuring range, 0.159–191 nmol/l (0.05–60 ng/ml), <5% relative SD.

Ovulation was confirmed through ovulation tests and LH blood concentrations. LH was assessed with the ECLIA Elecsys® LH Assay. Assay antibodies, measuring ranges, and intra-assay coefficients of variation for LH were the following: antibodies, biotinylated monoclonal anti-LH antibody (mouse), 2.0 mg/l; measuring range, 0.3–200 mIU/ml (0.3–200 IU/l); intra-assay precision, ≤ 2.2% variation coefficient.

All assays were determined on the cobas® e 801 analyzer (Roche Diagnostics GmbH, Mannheim, Germany) and were used according to the manufacturer’s instructions.

To evaluate hormonal fluctuations over the five-week testing period, we used a centering approach for absolute hormonal concentrations of estradiol, estrone, and progesterone. This involved subtracting the overall mean hormonal concentration from each individual hormonal value (e.g. hormonal concentration on test day 1 minus the mean overall hormonal concentration). This method allows us to account for hormonal fluctuations relative to the average hormonal level across the five-week duration.

Psychological measures

Positive and negative affect was assessed for each test session separately using the Positive and Negative Affect Schedule (PANAS)53. It is a widely used and well-established instrument for assessing affective states which can be employed in daily assessments. The PANAS is a 20 item self-reporting questionnaire assessing positive emotions such as joy, interest, and alertness, and negative emotions such as sadness, distress, and irritability. Each item on the PANAS is rated on a 5-point scale, ranging between 1 and 5, with 1 indicating low agreement to the specific item (not at all) and 5 indicating a high agreement (very much). The positive affect score was calculated as the average of the 10 positive items. The negative affect score was calculated as the average of the 10 negative items. Hence, positive and negative affect scores can range from 1 to 5. Lower scores represent lower levels of positive and negative affect, whereas higher scores represent higher levels of positive and negative affect, respectively.

Statistical approach

Statistical analyses were performed using R software (https://www.r-project.org), Statistical Package for Social Sciences (SPSS) version 27, and GraphPad Prism 8. First, cubic regression curve estimations were used as the data followed a cubic curve to check whether hormones, hippocampal volume, and affect changed significantly across the 25 testing sessions.

Second, Shapiro-Wilk’s test was used to check for the normal distribution of the variables. As hormonal concentrations were not normally distributed, Spearman correlations were performed between hippocampal volumes, hormones, and positive and negative affect. False Discovery Rate (FDR) correction was used to correct for multiple comparisons in all analyses54.

Third, in case of a significant correlation, we used post-hoc mediation regression analyses to investigate whether changes in positive and negative affect were a direct effect of fluctuations in hormonal concentrations across the 25 test sessions or an indirect effect mediated by fluctuations in left and right hippocampal volume. Post-hoc mediation regression analyses were only performed in case prior Spearman correlations were significant, ensuring a more targeted exploration of the relationship between variables. Mediation regression models were calculated with positive and negative affect as dependent variables and hormonal levels as independent variables. Hippocampal volumes were added as a mediator variable to the model. Due to high multicollinearity among the independent variables the analyses were conducted separately for each hormone, hippocampal hemisphere, and positive and negative affect. In our mediation regression models, path a is the linear effect of the hormonal levels (independent variable) on hippocampal volume. Path b is the effect of hippocampal volume (mediator) on positive and negative affect (outcome variables). The indirect effect a*b measures the amount of mediation, and the direct effect c’ is the effect of the hormonal levels on positive and negative affect after controlling for hippocampal volume. The total effect c is the sum of direct and indirect effects. Results were based on 5000 bootstrapped samples. Residuals of the regressions were normally distributed.

Results

Analysis I: Fluctuations across the 25 test sessions

Absolute hormonal concentrations (estradiol: F(3,21) = 6.698, p = 0.002; estrone: F(3,21) = 14.728, p < 0.001; progesterone: F(3,21) = 46.306, p < 0.001), hippocampal volume (F(3,21) = 5.574, p = 0.006), and affect (positive affect: F(3,21) = 17.604, p < 0.001; negative affect: F(3,21) = 13.986, p < 0.001) changed significantly across the 25 testing days covering mainly the follicular phase and ovulation using cubic regression curve estimations (see Fig. 1). The menstrual cycle at the time of the scan lasted 53 days, which represented a longer irregular menstrual cycle than usual (M = 38.3 days, SD = 6.36 days during the 55 hormone-medication-free months prior to the study). Ovulation occurred on testing days 21 and 22, representing menstrual cycle days 37 and 38. Following this, the study covered 20 days of the follicular phase, 2 days of ovulation, and 3 days of the luteal phase. The luteal phase of this menstrual cycle covered 15 to 16 days in total.

Fig. 1: Hormonal concentrations, hippocampal volume, positive and negative affect across 25 testing days.
figure 1

Estradiol, estrone, and progesterone concentrations are displayed across the 25-day experiment. Changes in hippocampal volume, positive and negative affect across the experiment are displayed. Note that ‘Test Day 1’ refers to cycle day 10. Ovulation occurred on testing days 21 and 22 which refer to cycle days 37 and 38. Hormone icon pictogram, source: iStock. Licensed under the standard license.

Analysis II: Fluctuating hormonal concentrations in association with hippocampal volume and affect

Next, we tested whether fluctuations in hormonal concentrations, derived from the five-week period average, were associated with both hippocampal volume and affect using Spearman correlations. Bilateral hippocampal volume correlated significantly with fluctuating estradiol (r = 0.637, p = 0.001, pFDR = 0.002) and fluctuating estrone (r = 0.745, p < 0.001, pFDR < 0.001) but not with fluctuating progesterone (r = −0.036, p = 0.863, pFDR = 0.919). Both fluctuating estradiol and estrone correlated significantly with positive (estradiol: r = −0.469, p = 0.018, pFDR = 0.026; estrone: r = −0.427, p = 0.033, pFDR = 0.040) as well as negative affect (estradiol: r = 0.773, p < 0.001, pFDR < 0.001; estrone: r = 0.661, p < 0.001, pFDR < 0.001), suggesting that increasing concentrations of estrogens were associated with decreasing positive but with increasing negative affect. Fluctuating progesterone correlated significantly with positive (r = 0.464, p = 0.019, pFDR = 0.026) but not with negative affect (r = −0.021, p = 0.919, pFDR = 0.919), suggesting that increasing progesterone concentrations were associated with increasing positive affect.

Furthermore, positive affect was significantly inversely associated with bilateral hippocampal volume (r = −0.681, p < 0.001, pFDR < 0.001), whereas negative affect was significantly associated with bilateral hippocampal volume (r = 0.485, p = 0.014, pFDR = 0.026).

Analysis III: Estrogens are linked to negative affect, hippocampal volume is linked to positive affect

We used post-hoc mediation regression analyses to investigate whether positive and negative effects were a direct effect of hormonal concentrations across the 25 test sessions, or an indirect effect mediated by fluctuations in hippocampal volume. These analyses were selectively performed in case prior Spearman correlations were significant, ensuring a focused exploration of the relationship between variables. Consequently, post-hoc mediation regression analyses were specifically applied to fluctuations in estradiol and estrone across the five-week duration.

For the outcome variable positive affect, model 1.a included estradiol as the independent variable and hippocampal volume as mediator. For model 1.b, estradiol was replaced by estrone as independent variable. Neither estradiol nor estrone were identified as significant predictors since total effects c and direct effects c’ were insignificant. The indirect effect a*b for hippocampal volume was significant in both models, suggesting that bilateral hippocampal volumes were related to positive affect rather than estrogens. Figure 2 shows the detailed results for mediation analysis model 1.a and 1.b.

Fig. 2: Mediation analysis of hormonal levels, positive affect, and hippocampal volumes.
figure 2

Path a is the linear effect of the hormonal levels (independent variable) on hippocampal volume. Path b is the effect of hippocampal volume (mediator) on positive affect (outcome variable). The indirect effect a*b measures the amount of mediation, and the direct effect c’ is the effect of the hormonal levels on positive affect after controlling for hippocampal volume. The total effect c is the sum of direct and indirect effects. All paths’ estimates are depicted as regression coefficients, respective p-values and 95% confidence interval (95%CI). Significant results are indicated in bold. n.s. = non-significant. Hormone icon pictogram, source: iStock. Licensed under the standard license.

For negative affect, models 2.a included estradiol as the independent variable and hippocampal volume as mediators. Since total effects c and direct effects c’ were significant, estradiol was identified as a significant predictor for negative affect. Model 2.b included estrone as the independent variable, negative affect as the outcome variable, and hippocampal volume as mediator. Similar to estradiol, the total effects c and direct effects c’ of estrone predicted negative affect. The indirect effect a*b for hippocampal volume was not significant in both models, suggesting that negative affect was directly related to fluctuating concentrations of estrogens and was not mediated by bilateral hippocampal volumes. Figure 3 shows the detailed results for mediation analysis 2.a and 2.b.

Fig. 3: Mediation analysis of hormonal levels, negative affect, and hippocampal volumes.
figure 3

Path a is the linear effect of the hormonal levels (independent variable) on hippocampal volume. Path b is the effect of hippocampal volume (mediator) on negative affect (outcome variable). The indirect effect a*b measures the amount of mediation, and the direct effect c’ is the effect of the hormonal levels on negative affect after controlling for hippocampal volume. The total effect c is the sum of direct and indirect effects. All paths’ estimates are depicted as regression coefficients, respective p-values and 95% confidence interval (95%CI). Significant results are indicated in bold. n.s. = non-significant. Hormone icon pictogram, source: iStock. Licensed under the standard license.

Discussion

In our series of dense-sampling venipuncture and brain imaging, fluctuating concentrations of estrogens were positively associated with bilateral hippocampal volume during a prolonged follicular phase. The results are consistent with Barth et al.21 reporting increased volumetric fractional anisotropy (FA) in the bilateral hippocampus associated with increased estradiol across the full menstrual cycle21. Moreover, our results are consistent with cross-sectional data reporting lower hippocampal volumes in the early follicular phase and increased hippocampal volume during the late follicular phase of the menstrual cycle55,56. As was the case for our study, Barth et al.21 did not report hippocampal volume associations with progesterone21. This is inconsistent with the reported results of associations between hippocampal volume and progesterone by Taylor et al.22. One explanation for the missing associations between progesterone and hippocampal volume in our study could be that the 25 test sessions in the current study only covered three days of the luteal phase of the menstrual cycle when progesterone concentrations are dominant. The majority of test sessions covered the follicular phase as well as ovulation when progesterone is, overall, low. Another explanation would be that additional brain regions than the hippocampus may underly relationships with hormones and affect. Taylor et al.22 performed correlations between progesterone concentrations and volumes of hippocampal subfields, such as CA1, CA2/3, dentate gyrus, and the medial temporal gyrus. The hippocampus has been investigated in more details by Barth et al.21 who performed whole-volume hippocampal correlations20. The dentate gyrus of the hippocampal formation is one of the few brain areas that may exhibit adult neurogenesis. Therefore, the background of the observed structural changes in the hippocampal formation might be linked to alterations in dendritic branching or neuronal cell growth57. The underlying mechanisms by which sex hormones and hippocampal morphology are linked still need to be elaborated. Published results to date indicate that hormonal fluctuations across the menstrual cycle as well as during an irregular prolonged follicular phase impact hippocampal morphology. While regular menstrual cycles of ~28 days are associated with an increased estradiol exposure, longer irregular menstrual cycles are associated with an decreased estradiol exposure as estradiol concentrations remain lower for a longer period of time3. In the context of the current study, this implies that decreased hippocampal volumes are associated with prolonged irregular cycles.

In addition to variations in hormonal patterns and bilateral hippocampal volumes, we report significant fluctuations in positive and negative affect across the 25 test sessions. Both positive and negative affect were significantly associated with estrogen levels and hippocampal volume. Decreasing positive emotions were significantly associated with increasing concentrations of estrogens, suggesting that positive emotions were low during phases when estrogens were high, such as during the late follicular phase and ovulation. However, the mediation analysis revealed that hippocampal volume indirectly influenced this relationship, suggesting that positive affect fluctuations are more closely tied to hippocampal morphology than directly to estrogen levels. On the other hand, increased negative emotions, such as sadness and irritability, were associated with increasing concentrations of estrogens, indicating elevated negative affect during estrogen peaks. Unlike positive affect, the association between increased negative affect and estrogens was not mediated by hippocampal volume. The findings suggest that fluctuations in negative emotions were better explained by estrogens than by fluctuations in hippocampal morphology. It should be noted that, in our study, negative affect scores were non-pathological and generally low, given the healthy participant involved.

Contrary to common assumptions that negative affect is more prevalent during the premenstrual phase and positive affect is more common around ovulation58,59,60,61, our results may suggest an opposing mechanism. Furthermore, there is also evidence that ovarian hormones make little or no contribution to daily mood and affective variability in naturally cycling women62,63,64,65,66. However, the peripubertal phase and the onset of menopause, characterized by irregular menstrual cycles, appear to be a time of mood sensitivity to hormone changes, indicating an increased susceptibility to mood variations43,44,45. Specifically, the mood sensitivity to estradiol predicts risk to perimenopausal depression, particularly in women who are otherwise considered at low risk45. Additionally, irregular menstrual cycle variability before pregnancy was reported to be associated with depression during pregnancy67. Focusing on menstrual cycle irregularities in an otherwise healthy individual, our study may provide insight into the endocrine factors that underlie increased susceptibility and prevalence of depression in women32,33. Women are twice as likely to be diagnosed with depression compared to men31 and this increased susceptibility is seen only during the reproductive years while the prevalence of depression in prepuberty and after the age of 55 is almost the same in men68. The menstrual cycle serves as an important indicator for reproductive, general, and mental well-being in women69. Notably, it plays a pivotal role in ensuring fertility and functioning of the female reproductive system70. Additionally, the hormonal fluctuations of estrogens and progesterone during the menstrual cycle influence various bodily functions beyond reproduction, impacting aspects such as bone health71,72 and cardiovascular function73,74, with implications that extend into later stages of life. Menstrual cycle irregularities have been associated with a greater risk of premature mortality1 and mental health conditions34. The findings of our study highlight the role of estradiol and estrone in affect sensitivity in an irregular menstrual cycle. Moreover, the results suggest that short-term changes in hippocampal volume within 25 test sessions influence the perception of less positive emotions. Prior research has linked a decrease in positive affect with depression75, as well as changes in hippocampal volume with depressive symptoms57. We cannot provide conclusive interpretations for these associations given that our findings are based on one healthy participant with menstrual cycle irregularities. However, the question arises as to what function these short-term changes serve across the menstrual cycle and whether fluctuations in estrogens and hippocampal volume act as a protective factor or risk factor for developing depressive disorders in some women. The results of this study suggest that we need to continue to investigate the influence of irregular menstrual cycles on the brain and mental health.

It should be noted that estrone is a precursor to estradiol. Our results indicate its potential role, besides estradiol, on affect during the (irregular) menstrual cycle. It remains unclear whether the role of estrone in hippocampal morphology and affect is uniquely in women with an irregular cycle or whether it has the same impact in women with a regular menstrual cycle of ~28 days.

Several limitations to our study must be noted. First, although no endocrine condition or diagnosis was known before scanning, the participant had a 53-day menstrual cycle during the 25 testing sessions. A local gynecologist ruled out a diagnosis of Polycystic Ovarian Syndrome. Further hormonal analyses, ordered by the gynecologist, revealed increased prolactin levels of unknown origin after scanning. Hyperprolactinemia is the most common pituitary hormone hypersecretion syndrome, which most commonly affects women between the ages of 25 and 3476, with women often reporting irregular menstrual cycles77. Given that stress influences prolactin secretion in humans78, it is possible that prolactin serum concentrations were elevated due to the stressful procedures of daily MRI scans and blood draw. Following this, the results could also be explained by increased or fluctuating prolactin concentration, which was not assessed during this study. It is, furthermore, noteworthy to consider potential interactions with the hypothalamic-pituitary-adrenal (HPA) axis. The HPA axis regulates cortisol release in response to stress and hormonal variations. Hormonal fluctuations, particularly estradiol and estrone, can influence the sensitivity of the HPA axis79, potentially impacting cortisol levels80. Our findings, which include associations between affective states and hormonal changes, align with the complex relationship between the HPA axis, emotional well-being, and stress-related conditions, including depression80. While our study did not measure the HPA axis or cortisol, these results highlight the importance of their inclusion in future research.

Second, due to a longer menstrual cycle of overall 53 days during the scanning sessions, we were not able to scan the participant across a complete (irregular) menstrual cycle which usually consists of menses, follicular phase, ovulation, and luteal phase. Future studies should continue dense-sampling studies with complete menstrual cycles. However, when conducting in vivo research with humans, unexpected events like a prolonged menstrual cycle are not always predictable or avoidable. Despite striking differences in cycle length between densely-sampling participants, the consistency of our findings with previous reports underscores the robustness of these associations.

Third, one of the authors was the participant in this study and it was not possible to conduct a blinded study. As a result, responses to the questions on the positive and negative affect scores may have been influenced by the knowledge that individuals, for example, might experience premenstrual symptoms such as stress anxiety, fatigue, mood swings, anxiety, or depression shortly before the end of the menstrual cycle81. However, this is unlikely given that the participant experienced an irregular menstrual cycle with a total length of 53 days at the time of scanning. Therefore, the participant was not fully aware of which phase she was in, since hormonal levels were not revealed until after completion of the test sessions.

Fourth, we made rigorous efforts to minimize systematic bias or shift in the neuroimaging data. This was accomplished by consistently using the same scanner at the same time of the day for all scans, using above-standard spatial resolution, employing a longitudinal segmentation pipeline, and applying volume normalization techniques, resulting in a highly sensitive measurement. The observed hippocampal volume changes, which are approximately 1%, reflect a sufficient level of bias reduction. It is worth noting that due to the high sensitivity of the measurement, even small variations can be meaningful. In contrast, a cross-sectional study reported a test-retest reliability of hippocampal volumes at a magnitude of 3% in both men and women not controlling for the menstrual cycle phase82.

Lastly, since this study is a longitudinal study with a dense single-subject design, interpretations and explanations of the reported relations should be made with caution as no causal effects can be generalized to larger populations. Given that our study is the first to report the interplay of estrogen fluctuations with hippocampal volumes and positive and negative affect across a dense-sampling study of five weeks in an irregular menstrual cycle, further studies are needed to replicate and extend these results. Furthermore, scores on positive and negative affect were non-pathological since the study was conducted with a healthy participant without a history of mental health diagnoses. Future studies could include female participants diagnosed with MDD to clarify whether the influence of the menstrual cycle and its hormonal fluctuations is different in individuals with and without clinical depression.

The strength of this study was its dense daily measurement time resolution over a total of five weeks to investigate macrostructural changes in hippocampal volume in the brain under the influence of hormones of the female menstrual cycle (irregularities). Compared to this study, Barth et al.21 acquired MRI scans every second or third day in two separate scanning sessions covering two full menstrual cycles. Because our study revealed changes in the hippocampus across female menstrual cycle irregularities and Taylor et al.22 reported changes across a regular menstrual cycle, it would be beneficial to examine male participants over five weeks and determine sex hormones, such as estradiol, progesterone, and testosterone, to clarify whether these changes are unique to women. Furthermore, investigating female participants with a diagnosis of MDD would shed light on whether fluctuating hormones and hippocampal volumes are associated with an increased susceptibility to depressive symptoms in women. Lastly, while our study focused on general affect, future investigations could benefit from a more nuanced examination using instruments like the Daily Record of Severity of Problems (DRSP)83, which may provide a more detailed understanding of specific and more severe affective disturbances associated with the menstrual cycle.

In conclusion, this dense-sampling study provides valuable insights into the complex interplay between endogenous hormone fluctuations, hippocampal morphology, and affect in a participant with an irregular menstrual cycle. The findings highlight the significant associations of estradiol and estrone with bilateral hippocampal volume, suggesting potential hormonal contributions to brain structure. Moreover, fluctuating concentrations of estrogens were linked to affect, revealing their role in influencing positive and negative emotions. The study’s focus on an irregular menstrual cycle emphasizes the importance of investigating hormone-brain relationships beyond regular cycles, shedding light on potential implications for mental health disorders prevalent in women. However, these results are based on a single participant, warranting caution in generalizing findings to the broader population. Further research with larger and diverse samples is necessary to validate and expand these findings, elucidating the mechanisms underlying hormonal influences on brain health and affect regulation in women.