A scoping review of hormonal clinical trials in menstrual cycle-related brain disorders: Studies in premenstrual mood disorder, menstrual migraine, and catamenial epilepsy.

In this section we review peer-reviewed, published papers written in English, with human subjects who were female, of reproductive age (approximately 18–45), not pregnant or on hormonal birth control, and demonstrated a cyclical pattern of the brain disorder of interest. Our literature search began with three standardized PubMed queries, one for each disorder (See Table 4 ). After initial review of studies that met these criteria, we reviewed references found within these studies as well as several related review articles in order to identify relevant work that was not picked up in the initial queries. For this scoping review, we only include studies that directly manipulate ovarian hormones in samples with the diagnoses of interest and measure treatment effect using prospectively confirmed, cyclical outcomes. Given the wide range of methodology for measuring cyclicity, we defined two possibilities for article eligibility based on prospectively confirmed, cyclical outcomes: 1) if a study compared a single cycle phase across treatment conditions (e.g., premenstrual symptoms in the luteal phase in a treatment cycle versus baseline cycle), it was eligible for inclusion if the cyclical pattern was prospectively confirmed during at least one menstrual cycle of baseline; 2) studies that compared cyclical outcomes (e.g., a change score of luteal minus follicular phase migraine frequency during treatment) were eligible for inclusion. In line with the spirit of scoping reviews to map the existing literature on a topic ( Pham et al., 2014 ), we include any study that fits these parameters, regardless of study design (e.g., pilot studies, open-label, single-blind, or double-blind and placebo-controlled) and note where applicable that evidence from a double-blind placebo-controlled trial is considered stronger than other designs. We discuss case studies, case series, and studies where the exact method of prospective confirmation is unclear, only if the studies meet all other review parameters and are the only existing research within a category. In sum, we found N=85 studies across the three menstrual cycle related brain disorders (PMD, MM, and CE) that will be reviewed in subsequent sections ( Figure 2 ). We organize trials based on hormonal manipulation, reviewing the available literature across PMDD, MM, and CE. All trials that met our inclusion criteria are summarized in tables 5 , 6 , and 7 (separated by diagnosis, PMD, MM, and CE, respectively). Finally, at the end of each subsection, we (1) evaluate the extent and nature of existing research in each domain, (2) identify gaps, and (3) address the value of a future systematic review or meta-analysis. Combined oral contraceptives (COCs) prevent pregnancy primarily by suppressing ovulation and are generally comprised of a progestin (such as drospirenone, levonorgestrel, or norethindrone) and ethinyl estradiol (EE), in varying dosages, formulations, and schedules ( Kaunitz, 2022 ; see Kiley & Hammond, 2007 for comprehensive review). The traditional COC schedule (termed “21–7”) is 21 days of active combination treatment, followed by 7 days of a hormone-free interval that is characterized by a withdrawal bleed; the withdrawal bleed is often considered menses, but biologically differs in that menses is triggered by shedding of the endometrial lining after an ovulatory ovarian cycle, while a COC-induced bleed occurs independent of ovulation. The 7-day hormone free interval is not physiologically necessary, and many modern COCs have reduced the hormone free interval to 0–4 days ( Edelman et al., 2014 ; MacGregor & Guillebaud, 2018 ). Formulations vary mostly in their progestin content and dosage, although the dose of EE can vary as well (see Lawrie et al., 2011 for systematic review of progestins in COCs). Finally, COCs can be monophasic, biphasic, or triphasic, where the hormone dosage is constant for 21 days, 14 days (biphasic), or 7 days (triphasic). The vast majority of prescriptions are currently monophasic. Importantly, COCs share the mechanism of ovulation suppression with 21–28 days of progestin/EE dosing, with or without a hormone free interval; this section will not review hormonal contraceptive trials with similar hormonal formulations if the mechanism does not intent to suppress ovulation (e.g., intrauterine devices). Our search identified 12 trials testing COCs as a treatment for PMD, with mixed results depending on the pharmacologic formulation and dosing regimen (see Table 5 ). In general, the goal of COCs in treating PMD is to suppress ovulation, thereby removing the hormonal trigger for symptom onset ( Hammarbäck et al., 1991 ; Schmidt et al., 1991 ). The COC Yaz ® contains drospirenone, a spironolactone-derived progestin, and EE, dosed for 24 days followed by 4 days of placebo (“24–4”). This COC has been FDA-approved since 2011 for PMDD in individuals desiring hormonal contraception in addition to PMDD treatment. Randomized controlled trials (RCTs) testing 24–4 drospirenone/EE COCs show improvement in PMD symptoms for patients randomized to the COC group versus placebo ( Freeman et al., 2001 ; Freeman, 2002 ; Marr, Niknian, et al., 2011 ; Pearlstein et al., 2005 ; Yonkers et al., 2005 ); see De Berardis et al., 2007 and Lopez et al., 2012 for reviews). One article specifically analyzed functional impairment of PMD symptoms, finding that drospirenone-containing COCs also improve PMD-related impairment compared to placebo ( Marr, Heinemann, et al., 2011 ). Finally, two trials have tested drospirenone/EE for PME of depression. A preliminary, open-label trial suggested that drospirenone/EE could improve PME of depression ( Joffe et al., 2007 ); however, there is stronger evidence against drospirenone/EE’s effectiveness in PME of depression from a double blind, placebo-controlled RCT ( Peters et al., 2017 ). Additional RCTs are needed to confirm the lack of efficacy for drospirenone/EE COCs in PME of depressed mood and other chronic symptoms. While a 24–4 regimen of drospirenone/EE is effective in treating symptoms of PMD (and specifically, pure PMDD), studies have been inconclusive in determining if this effect is driven primarily by the specific progestin of drospirenone or by the short, four-day hormone-free interval (in contrast with a more traditional 21–7 dosing schedule). One three-arm RCT comparing continuous drospirenone/EE, intermittent (24–4) drospirenone/EE, and placebo found that continuous drospirenone/EE did not beat intermittent drospirenone/EE or placebo over three months ( Eisenlohr-Moul, Girdler, Johnson, et al., 2017 ). This three-arm study – like many others – was limited by a strong placebo response (see commentaries in Freeman et al., 2012 , Freeman & Rickels, 1999 ). RCTs of COCs with EE and non-drospirenone progestins (e.g., levonorgestrel, norethindrone) have found more mixed results. Some studies find continuous levonorgestrel/EE can improve mean percent improvement in PMD symptoms compared to placebo ( Bäckström et al., 1992 ; Halbreich et al., 2012 ), while triphasic norethindrone/EE does not beat placebo in RCTs ( Graham & Sherwin, 1992 ). Levonorgestrel studies have been systematically reviewed as well ( Freeman et al., 2012 ). The literature on COCs in PMDs can be summarized as follows: 1) several RCTs of drospirenone/EE 24–4 support its use in pure PMDD, but possibly not for PME of depression; 2) RCTs have examined non-drospirenone COCs, or COCs on a 21–7 schedule for PMDD, with negative or mixed evidence; and 3) there is a high placebo response in these trials. Before reviewing evidence from COC trials in MM (N=5; see Table 6 ), we must highlight that COCs are contraindicated in those who experience migraine with aura , per the U.S. Centers for Disease Control, due to elevated risk of ischemic stroke ( Curtis, 2016a ). This section thus only applies to MM without aura, and careful evaluation of focal neurological symptoms must be undergone before any research or clinical use of COCs for MM. Based on the successful PMDD RCTs using drospirenone-containing COCs, researchers ( De Leo et al., 2011 ; Sulak et al., 2007 ) have utilized similar protocols to test drospirenone-containing COCs as a treatment for MM. In an RCT comparing 24–4 dosing to 21–7 dosing of drospirenone/EE in patients with pure MM, both dosing regimens improved migraine duration and intensity, but the 24–4 dosing schedule showed significantly greater improvements ( De Leo et al., 2011 ). Similar evidence arises from a prospective open-label study, such that headache presence and severity decreased while participants used a 21–7 drospirenone/EE formulation compared to other progestins, and further decreased during a 168-day extended regimen with no hormone-free interval ( Sulak et al., 2007 ). However, the 168-day extended regimen of drospirenone-containing COCs has not been tested in double-blind RCTs. Other studies have evaluated the hormone-free interval of COCs with non-drospirenone COCs (e.g., desogestrel, levonorgestrel). One study, similar to the Sulak et al study, compared a 4-day hormone-free interval to a 168-day extended COC regimen of levonorgestrel/EE in participants with MRM ( Coffee et al., 2014 ). This study found that the 4-day hormone-free interval does trigger headaches compared to removing the hormone free interval completely. In addition, during the 4-day hormone free-intervals, participants in the Coffee et al RCT were randomized to receive either placebo or frovatriptan; while the frovatripan findings are beyond the scope of the present review, headache intensity was significantly greater even under placebo during the hormone free interval compared to extended COC dosing. A single-blind, placebo-controlled study recruited patients using COCs with prospectively confirmed migraine attacks that routinely occurred during the hormone-free interval found that supplementing the hormone-free interval with transdermal estradiol reduced the number of migraine attacks, severity, vomiting, and necessity of analgesic use compared to placebo ( Nappi et al., 2005 ). Although this study did not prospectively confirm the menstrual pattern of migraine at baseline (prior to COC usage), the daily charting during several months of COC use and placebo-controlled trial of estradiol during the hormone-free interval provides evidence that abrupt estrogen withdrawal (such as during the early hormone-free interval of COC usage) may be one mechanism that triggers migraine attacks. A similar open-label trial administered conjugated equine estrogen during the hormone-free interval for participants using 21–7 COCs with the goal of reducing the degree of perimenstrual estrogen withdrawal preceding menstrual migraine; this study found that the estrogen supplementation decreased menstrual headache days by at least 50% ( Calhoun, 2004 ). However, it must be noted that this open label study did not include any randomization or placebo control. Our literature search uncovered no experimental studies of COCs in CE patients. This highlights a significant gap in the current understanding of potential CE therapies and points to a potentially important new frontier for investigation. RCTs examining COCs are abundant in the PMD literature, and fewer studies with more mixed designs have examined the role of COCs in MM. In both cases, research suggests a benefit of COCs, and these benefits appear to be maximized by shortening or eliminating the hormone-free interval. Existing trials suggest that a leading hypothesis in both fields is pathophysiology due to neurobiological sensitivity to hormone changes , and that maximizing hormonal stability may be a priority in these treatment paradigms. While these trials appear to have been motivated in part by the pharmacological development of drospirenone, a relatively novel spironolactone-derived progestin, it remains unclear whether this novel formulation plays any role in the success of these trials or if the extended hormone-free intervals (i.e., greater hormonal stability) are responsible. Systematic reviews have already been completed for COC treatment of PMD; a systematic review is likely not feasible for COC treatment of MM, due to the wide variety of study designs. However, our review suggests that a possible avenue for future research is a focus on double-blind, placebo-controlled RCTs in this arena. There appear to be several reasons that there is a gap in the literature surrounding COC and CE. First, female patients with epilepsy do remain eligible for COC use per the CDC ( Curtis, 2016a ). However, there is concern based on epidemiologic and animal work that COCs may provoke seizures, as some studies have shown up to 6.75-fold higher risk of greater seizure frequency for hormonal contraceptives versus barrier contraception ( Reddy, 2017 ). In contrast, Herzog et. al. reviewed birth control methods among females with any type of epilepsy using the Epilepsy Birth Control Registry and their results do not support this concern ( Herzog, Mandle, Cahill, Fowler, Hauser, et al., 2016 ). This work showed that the vast majority of females with epilepsy reported no change in seizure frequency regardless of contraceptive type; for those on hormonal contraception who did experience increased seizure frequency, this change was less likely for individuals using COCs compared to those using hormonal patches or progestin-only pills ( Herzog, Mandle, Cahill, Fowler, Hauser, et al., 2016 ). Additionally, there is concern that COCs can impact the efficacy of a patient’s existing anti-epileptic drug regimen through pharmacokinetic interactions via the hepatic CYP3A4 enzyme system ( Reddy, 2017 ), in addition to some evidence that certain antiseizure medications can interfere with the contraceptive efficacy of COCs ( Curtis, 2016b ; Frank & Tyson, 2020 ). However, the Epilepsy Birth Control Registry study identified that changes in seizure frequency while on hormonal contraception is highly dependent on the category of concurrent anti-epileptic drug ( Herzog, Mandle, Cahill, Fowler, & Hauser, 2016 ). Thus, it seems possible that trials of COCs in patients with existing anti-epileptic drug regimens could be ethically completed with close monitoring of drug-drug interactions and steady-state concentrations of each drug class. Together, the lack of COC trials in the context of CE is notable, particularly given the dearth of causal evidence regarding COC-related seizure risk, and given the possibility that stabilization of hormones at midfollicular levels may be therapeutic. There is notable depth of mechanistic knowledge gained from RCTs of COCs with differing formulations and dosing strategies in PMD and MM research, and thus, with appropriate monitoring and counseling, COCs may be a promising avenue of future clinical research in CE. Briefly, gonadotropin-releasing hormone agonists (GnRHa) induce a “chemical menopause” by suppressing ovulation via a negative feedback loop in the hypothalamic-pituitary-gonadal axis. GnRHa bind to the GnRH receptor, producing an initial increase in luteinizing hormone (LH) and follicle-stimulating hormone (FSH), which in turn desensitizes the pituitary gland and downregulates LH and FSH production; once the pituitary is continuously desensitized by the GnRHa, ovulation is suppressed, and the hypogonadal state is characterized by stable, low levels of ovarian hormones ( Wilson et al., 2007 ). Finally, when GnRHa are prescribed in premenopausal individuals, hormonal addback is encouraged to prevent side effects associated with a hypoestrogenic state ( Shuster et al., 2010 ). Our literature query discovered N=17 studies (see Table 5 ) of ovulation-suppressing GnRHa for treatment of PMD (one of which, Martin et al., 2006 , will be discussed in conjunction with MM). A 2004 meta-analysis of five placebo-controlled RCTs testing GnRHa for PMDD found that GnRHa are generally successful for treating PMD symptoms, and their efficacy is not reduced by stable, adjunctive hormonal addback ( Wyatt et al., 2004 ). Of the five RCTs included in their meta-analysis, the GnRHa drug varied, including low-dose intranasal daily buserelin ( Sundström et al., 1999 ), monthly depot shots of goserelin ( Leather et al., 1999 ) or leuprolide ( Brown et al., 1994 ; Freeman et al., 1997 ), or daily triptorelin injections ( Muse et al., 1984 ); regardless of the specific drug, each study found an improvement in PMD-related outcomes with GnRHa versus placebo. Additional placebo-controlled RCTs (not included in the meta-analysis) replicate these findings, with buserelin improving depressive and physical symptoms and goserelin improving anxiety and mood swings, respectively ( Hammarbäck & Bäckström, 2009 ; Hussain et al., 1992 ; West & Hillier, 1994 ). Overall, it has been found that up to 70% of individuals with treatment-resistant PMDD (e.g., that do not respond to first line treatments of SSRIs or COCs) experience symptom remission with leuprolide ( Pincus et al., 2011 ). Two mechanistic RCTs suggest that GnRHa efficacy in PMD results from prevention of abrupt changes in ovarian hormones, regardless of whether hormones are stabilized at relatively low or high levels. A landmark, double blind RCT of leuprolide showed that PMD symptoms improve when treated with leuprolide compared to placebo, and symptoms return with addback of estrogen or progesterone ( Schmidt et al., 1998 ). The same group completed a follow-up study in 2017 showing that, while PMD symptoms in patients treated with GnRHa do return to pre-treatment levels after combined estrogen and progesterone addback, this is a temporary symptom spike that dissipates after at least one month of stable hormonal addback ( Schmidt et al., 2017 ). These findings overlap with a 2009 placebo-controlled RCT that compared hormonal addback regimens after GnRHa-induced anovulation and found that three months of stable, transdermal estradiol prevented symptom resurgence, while any one-month trial of added progesterone can trigger symptoms ( Segebladh et al., 2009 ). Additional studies consistently demonstrate the safety and efficacy of stable estradiol and progestin addback in addition to GnRHa for long-term treatment of PMD ( Di Carlo et al., 2001 ; Mezrow et al., 1994 ; Mortola et al., 1991 ). While it was historically believed that PMD symptoms resulted from an abnormal sensitivity to luteal phase elevations in ovarian hormones, these trials demonstrate that it is the change in symptoms—rather than the absolute level of estradiol and progesterone—that trigger symptoms. Taken together, our literature search found an extensive body of work contributing to the current leading hypothesis that PMD symptoms result from changes in ovarian hormone levels, rather than stable levels; clinically, these studies support GnRHa plus continuous hormonal addback of estradiol and progesterone as treatment for PMD. Of note, small trials have failed to support ovulation suppression with GnRHa as a treatment for PME of depressive symptoms, as shown in both open-label and placebo-controlled trials that compared response to treatment in patients with pure PMDD in addition to those with PME of ongoing dysphoria ( Freeman et al., 1993 , 1997 ). We describe N=4 trials of GnRHa for MM, each with a different study design (see Table 6 ). Two prospective treatment studies ( Lichten et al., 1995 ; Murray & Muse, 1997 ) found significantly reduced headache scores with GnRHa followed by hormonal addback (estradiol in Lichten et al., 1995 ; estradiol plus medroxyprogesterone in Murray & Muse, 1997 ). While the Lichten et al trial did include a placebo run-in phase before leuprolide treatment, neither study was blinded or randomized, and Murray & Muse’s work is limited by an extremely small sample (N=5). Thus, we have chosen to additionally include a double-blind RCT that recruited specifically for migraine during non-perimenstrual days ( Martin et al., 2003 ) in our scoping review, given the scarcity of GnRHa trials in MM and the mechanistic value of understanding how suppressing the cycle affects migraine outcomes in any prospectively-charted sample. This RCT in patients with non-perimenstrual migraine suppressed ovarian activity with a GnRHa for one month then randomized participants to receive either stable estradiol or placebo for two months ( Martin et al., 2003 ). The estradiol group showed a significant improvement in daily headache severity compared to placebo; headache frequency was unchanged. Together, we found three GnRHa trials for treatment of MM, each suggesting some benefit of ovarian suppression and estrogen supplementation, albeit completed in different populations with varying hormonal addback regimens. There is a clear gap in the literature that could be filled with future, double-blind RCTs testing GnRHa in MM populations. The fourth trial found in our literature search of GnRHa for MM will be discussed in the summary section, as its outcome measures included both PMD and MM ( Martin et al, 2006 ). One prospective case series has evaluated GnRHa for CE (see Table 7 ). In a small open-label, prospective trial of triptorelin (administered intramuscularly every 4 weeks), 10 patients were recruited for CE, with 90–100% of seizures occurring during the perimenstrual phase ( Bauer et al., 1992 ). Perimenstrual seizures (defined as occurring from two days prior to menses onset through the duration of menses) were improved in eight out of ten patients, with improvements ranging from reduction in seizure duration to complete remission from seizures. Of course, without placebo control, blinding, or randomization, this trial alone does not offer substantial evidence for suppressing ovulation with GnRHa as a potential treatment for CE. As with COCs, there is consistent, systematically-reviewed experimental evidence that GnRHa benefit patients with pure PMD, both during a hypogonadal chemical menopause state (low stable hormones) and during stable addback of E2 and P4 (high stable hormones), further implicating abnormal sensitivity to cyclical hormone changes as the trigger of luteal phase mood symptoms in PMDD. There is very limited evidence in patients with PME of chronic depressive symptoms, and these studies suggest that GnRHa may not benefit patients with PME of depression ( Freeman et al., 1993 , 1997 ). In the context of CE, there are no controlled trials of GnRHa, although one case series found reduced seizures during a hypogonadal chemical menopause state. This warrants further investigation in clinical trials, particularly to determine whether (similar to PMDD) patients may benefit from GnRHa plus stable addback, which increases the long-term viability of the treatment (Wagner-Schuman et al., In Press ). No RCTs of GnRHa have been undertaken in patients with MM. However, a clinical trial in a broader sample of female patients with migraine found no benefit of GnRHa alone, but reduced headache severity with GnRHa plus estrogen, while two treatment studies in pure MM found that GnRHa, with or without hormonal addback, improved symptoms of MM. These findings could indicate that estrogen withdrawal or deprivation plays some unique role in the pathophysiology of MM. Clinical trials to examine the unique effects of GnRHa alone, stable estrogen addback, and stable progesterone addback in MM and CE may provide further mechanistic and therapeutic insights. Of specific interest to our scoping search strategy, one GnRHa trial arose in both our PMD and MM queries. Although the study population was selected for a general migraine diagnosis (rather than PMD or MM specifically), the close prospective charting of both mood and headache symptoms before and during GnRHa treatment warrants discussion. This study followed patients with confirmed migraine without aura across baseline menstrual cycles, a placebo run-in, then a double-blind trial of a GnRHa (goserelin) with or without estrogen addback, and sought to identify overlap between premenstrual mood symptoms and menstrual-related headaches in each study phase ( Martin et al., 2006 ). Although the patients were recruited based on migraine diagnosis, 43% met criteria for PMD, and over half had a comorbid diagnosis of major depression. In the natural cycles, 62% of patients demonstrated parallel, cyclical patterns of premenstrual symptoms and headaches, peaking in the mid and late luteal phases. Under GnRHa treatment, premenstrual symptoms and headaches were still moderately correlated, with higher correlations in the estradiol-addback group compared to placebo. These findings suggest that a current gap in the literature may be filled by considering comorbidity in both diagnosis and treatment for PMD and MM, as the comorbidity may represent either 1) a feedback loop where the presence of a migraine triggers worsened premenstrual symptoms, and/or 2) shared hormone sensitivity in a subgroup of individuals that predisposes them to both premenstrual mood and migraine symptoms. The final category of hormonal experiments includes a variety of other direct ovarian hormone manipulations (i.e., beyond the major drug classes of COCs or GnRHa). While sections 3.2 and 3.3 focused on various methods to suppress ovulation, and tests of adjunctive or supplemental treatments within the hormone-free interval, section 3.4 includes hormonal manipulations that do not fall into those categories, including: non-contraceptive oral and vaginal progesterone; transdermal estradiol; progesterone antagonists; and the 5α-reductase inhibitor dutasteride, which prevents metabolism of progesterone into its neuroactive metabolites (in addition to other non-ovarian steroid actions). While there are many types of ovarian hormonal manipulations, not all have been tested in human observational or experimental trials for PMD, MM, or CE; for example, our literature search uncovered no trials of intrauterine devices or 28-day progestin-only contraceptives (“minipills”). Additionally, we focus this section exclusively on manipulations of ovarian hormones or their direct metabolic pathways; there are many trials of other, non-ovarian hormone manipulations that are out of the scope of this manuscript (e.g., danazol, spironolactone, and testosterone derivatives). Estrogens, progestins, and their direct metabolites have been extensively studied as potential treatments for PMD (see Table 5 ). We first discuss the range of progestin-related trials in PMD (N=21), including progesterone supplementation (N=15), progesterone antagonism (N=3), and progesterone-derived neuroactive steroid pathways (N=3). There have been many trials showing that short-term supplementation with progestins does not prevent PMD symptoms (see Ford et al., 2012 , for systematic review). In several double-blind RCTs testing vaginal progesterone suppositories, oral micronized progesterone, and oral dydrogesterone, the treatment generally fails to either 1) beat placebo or 2) reach clinical significance ( Dennerstein et al., 1986 ; Freeman et al., 1990 ; Freeman et al., 1995 ; Hoffmann et al., 1988 ; Khajehei et al., 2009 ; Maddocks et al., 1986 ; Michener et al., 1999 ; Sampson, 1979 ; Sampson et al., 1988 ; Vanselow et al., 1996 ). Other studies report inconclusive findings, such as a double-blind trial of vaginal progesterone that found no subjective overall symptom relief, but reported improvement in the subcategory of nervous symptoms ( Baker et al., 1995 ), or a non-randomized trial of 21-day progestin treatments (oral medroxyprogesterone acetate (MPA) or norethisterone), which found that only MPA improved psychological symptoms ( West, 1990 ). In the West trial, given that both treatments in their trial suppressed ovulation, yet only MPA improved symptoms, the author suggests the symptom improvement may have been due to disruption of normal menstrual cyclicity via MPA rather than a direct effect of ovulation suppression. Our literature review did identify three studies with prospective ratings that found benefits of various progestins (oral or vaginal progesterone; oral MPA) for improving PMD symptoms ( Dennerstein et al., 1985 ; Hellberg et al., 1991 ; Magill, 1995 ). Two double-blind RCTs have tested how progesterone antagonism alters PMD symptoms. In one trial, patients with PMD were randomly assigned to receive (a) mifepristone alone, a progesterone antagonist that triggers dissolution of the corpus luteum, causing abrupt E2 and P4 withdrawal, and induction of menses, (b) mifepristone with human chorionic gonadotropin (hCG) because mifepristone triggers menstrual bleeding due to effects on the endometrium, while hCG prevents mifepristone-triggered dissolution of the corpus luteum and therefore maintains midluteal E2 and P4, or (c) placebo ( Schmidt et al., 1991 ). These three conditions were designed to test whether the late luteal E2 and P4 withdrawal trigger premenstrual mood change in those with PMD. PMD symptoms were not altered by inducing menses early with mifepristone, with or without hCG, suggesting that the hormone changes of the late luteal phase are not the proximal trigger of PMD symptoms. A smaller RCT replicated this null finding, where progesterone antagonism during the luteal phase only did not alleviate symptoms ( Chan et al., 1994 ). However, the selective progesterone antagonist ulipristal acetate, which leads to anovulation in approximately 80% of females ( Whitaker et al., 2014 ), significantly improved PMDD symptoms after three months compared to placebo, while maintaining follicular phase levels of estradiol (as compared to the hypoestrogenic effect of GnRHa) ( Comasco et al., 2020 ) Together, these progesterone antagonism trials suggest that progesterone antagonism alone does not treat PMD, unless anovulation is achieved. A final progesterone-related pathway that has been manipulated in RCTs of PMD is progesterone metabolism into neuroactive steroids, such as allopregnanolone. Inhibition of 5α-reductase, the enzyme that reduces progesterone into allopregnanolone, with dutasteride had no effect on mood in non-PMD controls or at low doses in PMD patients, but high doses of 5α-reductase in patients with PMD prevented luteal phase emergence of symptoms ( Martinez et al., 2016 ). Allopregnanolone inhibition has also been tested in an exploratory double-blind placebo-controlled RCT of the steroid antagonist isoalloprenanolone (Sepranolone), which found that in participants with pure PMDD, luteal phase dosing of Sepranolone reduced symptom severity ( Bixo et al., 2017 ). A much larger, follow-up study of Sepranolone for PMDD failed to meet its primary clinical endpoint; however, post-hoc analyses did demonstrate improvements with Sepranolone versus placebo in distress and impairment, and additionally, in an overall symptom severity score during a nine-day luteal phase window (rather than the pre-specified primary outcome of five luteal phase days) ( Bäckström et al., 2021 ). Together, these progesterone antagonism and neuroactive steroid trials strengthen our current understanding of PMD pathophysiology and represent a broad extent and range of literature relating progesterone pathways with PMD symptoms. In addition to progesterone-specific trials, estrogenic pathways have also been thoroughly tested in treatment studies of PMD (N=7). A 2017 systematic review concluded that estrogen supplementation does not improve symptoms in patients with PMDD, and in fact, may worsen symptoms when given specifically in the luteal phase ( Dhar & Pearson Murphy, 1990 ; Naheed et al., 2017 ). However, there is evidence that long-term estrogenic supplementation (continuous estradiol or tibolone for three months or longer) at dosages intended to suppress ovulation can prevent or reduce symptoms ( Magos et al., 1986 ; Smith et al., 1995 ; Taskin et al., 1998 ; Watson, 1989 ); this further supports the evidence discussed in section 3.3 that ovulation suppression, regardless of the method by which anovulation is achieved , successfully removes the trigger for PMDD. On the other hand, some evidence indicates that PME of depression may be more responsive to non-ovulation-suppressing estrogen administration. In two double-blind, crossover RCTs, we find that perimenstrual supplementation of combined estradiol and progesterone ( Eisenlohr-Moul, Bowers, et al., 2022 ) or estradiol alone ( Eisenlohr-Moul, Barone, et al., 2022 ) reduces PME of suicidal ideation and related depressive symptoms, which was common in patients during the baseline and placebo cycles. Support for this perimenstrual hormone withdrawal mechanism in PME of depression was further supported by the recapitulation of symptoms during withdrawal from exogenous hormones (but not placebo) in these trials. In sum, our literature query found evidence from many estrogen and progestin trials that falls in line with evidence from GnRHa trials for PMD, suggesting a divergent impact of steroid administration on cyclical mood changes depending on PMD subtype. Specifically, several trials of estrogen ( Ford et al., 2012 ) and progestins ( Naheed et al., 2017 ) have failed to demonstrate a robust benefit for pure PMD, whereas emerging evidence suggests that patients with PME of depressive symptoms benefit from perimenstrual administration of estradiol ( Eisenlohr-Moul, Barone, et al., 2022 ; Eisenlohr-Moul, Bowers, et al., 2022 ). This emerging work with short-term estradiol and progesterone supplementation further emphasizes that pathophysiologies may differ for pure PMDD and PME of depression; as discussed in sections 3.2 and 3.3 , it appears that the treatments that do work for pure PMDD (drospirenone-containing COCs, GnRH antagonism to induce estradiol withdrawal and a subsequent hypogonadal state) may not consistently work for PME of depression, while there is initial evidence that PME of depressive symptoms may be improved with short-term estradiol administration in the perimenstrual phase -- a treatment that has generally failed in pure PMDD ( Ford et al., 2012 ). Therefore, personalized treatment approaches for unique trajectory subtypes is an important topic for future studies across disorders. We found N=12 studies (see Table 6 ) of non-contraceptive ovarian hormone manipulations for MM, including trials of progestins (N=2), estrogens (N=9), and both estradiol and progesterone (N=1). Data regarding progestin effects on MM are extremely limited, as the only trials that were uncovered in our literature queries were completed over 50 years ago, one of which included no blinding or randomization, and one of which included MRM as a sub-analysis of a broader trial including both male and female migraineurs. In the open treatment study of 6 females with MM, perimenstrual progesterone injections did not alter menstrual migraine onset, duration, or severity ( Somerville, 1971 ). In a double-blind crossover trial of the progestin flumedroxone, the subset of females with MRM showed fewer and less severe headaches on flumedroxone versus placebo ( Bradley et al., 1968 ). However, authors note that the magnitude of frequency reduction was not clinically significant, there was no difference in headache duration, and more patients actually reported preference for placebo. We must further note that while headaches were charted daily throughout the treatment study, it is unclear if the menstrual cycle pattern was also confirmed prospectively. Thus, we include both progestin-related trials to discuss the best available data regarding progesterone and MM, acknowledging major limitations in methods; in terms of literature mapping, this presents a clear gap in evidence for progesterone’s role in MM. A final progesterone-related pathway tested in MM is isoallopregnanolone (Sepranolone), an allopregnanolone antagonist, tested in one clinical trial ( ClinicalTrials.gov identifier NCT04102995 ). Although the full results have not been published and therefore do not qualify as peer-reviewed work in our scoping literature search, the available data and overlap with allopregnanolone-related trials in PMD and CE, in addition to the limited progesterone trials in MM, warrant brief but cautious discussion. Sepranolone did not beat placebo at preventing MM attacks, which provides further support that progesterone withdrawal is not a mechanistic trigger of MM ( Nordkild, 2021 ). In contrast to trials of progestins, there is abundant literature relating estrogen and MM. An open trial by Somerville administered daily injections of progesterone or estradiol (in subsequent cycles) to 8 patients with pure MM and replicated their finding that preventing perimenstrual progesterone withdrawal had no effect on migraine symptoms (see prior paragraph), but prevention of perimenstrual estradiol withdrawal delayed expected menstrual migraine until the injections had stopped and plasma estradiol levels were below 20ng/100mL ( Somerville, 1972a ). Further research by this laboratory showed that MM symptoms could be delayed with estradiol supplementation if it is injected and long-acting, while neither short-acting estradiol or an oral formulation affected MM ( Somerville, 1972b , 1975 ); a similar open-label prospective treatment study found that at high enough doses to prevent ovulation, estradiol implants reduce MM attacks and severity ( Magos et al., 1983 ). In subsequent, larger trials of the role of estradiol in MM, results have been mixed. One non-randomized pilot study in patients with prospectively confirmed pure MM at baseline found that perimenstrual administration of transdermal estradiol did not prevent headaches (perimenstrual triptan prophylaxis did prevent headaches, albeit out of the scope of our review) ( Guidotti et al., 2007 ). In double-blind, randomized, placebo-controlled crossover trials, two trials show that percutaneous estradiol prevents and reduces severity of pure MM attacks ( de Lignières et al., 1986 ; Dennerstein et al., 1988 ), while two others failed to find differences between estradiol treatment and placebo ( Almén-Christensson et al., 2011 ; Smite et al., 1994 ). However, one double-blind, placebo-controlled RCT of perimenstrual estradiol gel did find that cycles treated with estradiol gel were associated with significantly fewer migraines, and the migraines that did persist during treatment were less severe ( MacGregor, 2006 ). This study had the notable methodological strengths of monitoring LH levels to confirm ovulation, then using those data to guide individual, cycle-specific treatment days. The MacGregor et al, 2006 trial also found that baseline migraine severity and frequency was recapitulated as soon as estradiol treatment ended, further supporting estrogen withdrawal as a mechanism that triggers MM attacks. Considering these trials together, we find that the strongest evidence supports estradiol stabilization as a potential therapy for MM, if the estrogens are prescribed in the context of confirmed ovulation; the wide variety of results in this subfield is likely due to differences in dosage, formulation, and study design (see Table 6 ). Despite the range of methodology, the quantity of evidence suggests that a systematic review of estrogen’s role in the pathophysiology and treatment of MM may be warranted. Our literature query uncovered two double-blind placebo-controlled RCTs investigating the effect of supplemental progesterone on CE when taken during the luteal phase of the cycle (of a total N=7 articles on progesterone for CE; see Table 7 ). A landmark RCT (“NIH Progesterone Trial”) evaluated seizure occurrence under luteal phase progesterone treatment or placebo in participants with both catamenial and noncatamenial intractable epilepsy ( Herzog et al., 2012 ). There were no differences between progesterone or placebo groups in treatment response, defined as the proportion of individuals who experienced a 50% or greater reduction in seizure frequency during treatment. However, the relative degree of perimenstrual seizure exacerbation predicted treatment response, such that individuals with more perimenstrual seizure exacerbations (i.e., C1 subtype) were more likely to respond to progesterone treatment. Another progesterone supplementation trial measured seizure frequency in participants with perimenstrual seizure exacerbation (C1 subtype) found that in participants who received progesterone, there was a significant decrease in the average seizure occurrence during the 3 months after treatment ( Najafi et al., 2013 ). The efficacy of supplemental progesterone treatment has also been suggested in multiple case studies and non-blinded trials ( Herzog, 1995 ; Kandeepan & Shaaban, 2016 ; Kim et al., 2020 ; Mattson et al., 1984 ). One mechanistic hypothesis of progesterone’s relationship to CE based on progesterone metabolism into allopregnanolone, which is a positive modulator of the GABA-A receptor and therefore increases inhibitory neurotransmission. In a secondary analysis of the NIH Progesterone Trial, a post-hoc analysis found that changes in allopregnanolone concentration mediated reductions in seizure frequency for patients with C1-type CE who experienced successful treatment with progesterone ( Herzog & Frye, 2014 ). This preliminary finding relating neuroactive steroids to CE exacerbation suggest an area of future research where double-blind RCTs are needed. Finally, to our knowledge, there are no trials of estrogen-based supplementation for CE. Reflecting on these heterogeneous literatures leads us to several observations. The most extensive body of literature exists for non-contraceptive estrogens or progestins in PMD. Most research finds no benefit of these short-term hormone supplementations on pure PMDD symptoms, while limited evidence suggests a possible benefit of estrogens in patients with PME of depression specifically. This divergence is particularly notable given that PME of depression may be the more common phenotype, with PME of depression occurring in roughly 60% of female patients with MDD ( Hartlage et al., 2004 ) but PMDD occurring in 5.5% of the general population, and points to PME as a critical area for future treatment studies. Similarly, there is minimal evidence in MM to understand the role of progestins; however, studies show potential benefit of stable estradiol supplementation on reducing headache frequency or duration. The estrogen work in PME of depression and MM highlights a potential area of overlap in the pathophysiology of PME of depression and MM, and future studies may evaluate whether similar pathways are responsible for these phenomena (e.g., serotonergic; Aggarwal et al., 2012 ; Frokjaer, 2020 ). Finally, and in contrast, concerns over estrogenic contributions to seizure have led to limited examinations in this area. However, the most prominent evidence in CE demonstrates benefit of progestins—particularly via enhanced GABAergic metabolites such as allopregnanolone—as a possible pathophysiology and treatment mechanism in CE.

Mood disorders consist of abnormal negative affective states; these states can be primarily characterized by depressed mood, sadness, loss of interest, and/or mania or irritability ( American Psychiatric Association, 2013 ). To meet Diagnostic and Statistical Manual (DSM-5) criteria for a mood disorder, individuals must demonstrate clinically significant distress or impairment for a prolonged time period (whereby specific duration varies by diagnosis). Past year prevalence of mood disorders is approximately 10% of U.S. adults, with a lifetime prevalence of approximately 20% ( Kessler et al., 2005 ); females are two to three times as likely to experience a mood disorder than males ( Kessler et al., 1993 ). Many CNS systems have been implicated in the pathophysiology of mood disorders (see Price & Drevets, 2012 , Willner et al., 2013 for comprehensive reviews). PMD pathophysiology is broadly understood as an adverse CNS response to the normal hormonal fluctuations of the menstrual cycle, with no known abnormalities in hormonal concentrations ( Rubinow & Schmidt, 2018 ; Teatero et al., 2014 ). Premenstrual dysphoric disorder (PMDD) has been codified with diagnostic criteria in both the DSM-5 and the International Classification of Diseases (ICD-11); DSM-5 codification predates the ICD-11 and is more frequently used for diagnosis ( Reed et al., 2019 ); See Table 1 for DSM-5 diagnostic criteria for PMDD. PMDD prevalence is about 6% in females with natural menstrual cycles ( Gehlert et al., 2009 ). Of note, premenstrual syndrome (or PMS) is an extremely broad umbrella term that has historically been used colloquially and across disciplines to describe a wide range of experiences and symptoms, both with respect to symptom types (e.g., physical, emotional) and severity (e.g., ranging from mild cramping with no distress or impairment to extreme, disabling pelvic pain). Therefore, although PMS has been used at times in the past to refer to what is now officially termed PMDD, including in several trials within this scoping review, we do not use the term PMS here. Instead, we use the term Premenstrual Mood Disorder (PMD), which encompasses 1) historical iterations of terminology (including premenstrual syndrome, premenstrual tension, and late luteal phase dysphoric disorder), 2) pure DSM-5 defined PMDD, and 3) the subtypes described below. In addition to the diagnosis of PMDD, in which symptoms become minimal or absent in the week following menses, there is evidence that premenstrual (or peri-menstrual) exacerbation (PME) of chronic psychiatric disorders is common ( Kuehner & Nayman, 2021 ; Nolan & Hughes, 2022 ). In a large representative longitudinal sample of individuals with depressive disorders, roughly 60% demonstrated clinically significant cyclical worsening of at least one symptom ( Hartlage et al., 2004 ). While many small studies have demonstrated that PME can occur in nearly every psychiatric disorder ( Nolan & Hughes, 2022 ), larger epidemiologic studies with representative samples are needed to determine the prevalence of PME for specific disorders or clinically significant behaviors. The PME studies found in our scoping literature search all describe PME of underlying depressive disorders (e.g., Major Depressive Disorder). Although the PMDD diagnosis excludes those whose psychiatric symptoms are present throughout the cycle but worsen cyclically (e.g., PME), it remains unclear whether the elevated baseline symptoms that differentiate PMDD and PME reflect truly unique phenomena, contributing to our preference to use the umbrella term “PMD” to describe either cyclic psychiatric pattern. Clinically, a provisional diagnosis of PMDD can be made after a structured clinical interview during which the provider should ask for daily ratings of PMDD symptoms across at least two menstrual cycles. An unacceptably high false-positive rate ( Eisenlohr-Moul, Girdler, Schmalenberger, et al., 2017 ; Gehlert & Hartlage, 2009 ), possibly due to cultural stereotypes suggesting that all females experience PMS, led to the unprecedented decision to require the use of daily symptom ratings across two months in order to make the DSM-5 diagnosis of PMDD. Before an official diagnosis can be made, daily ratings must be evaluated by a provider to determine whether there is evidence of consistent menstrual cycle-related affective symptoms. Specifically, the provider must observe a cyclical symptom pattern across at least two cycles in which clinically significant affective symptoms (substantially distressing and/or impairing) are present in the premenstrual week and remit in the week following menses. For the diagnosis of PME, prospective identification of this subgroup (which is not yet an official diagnosis or specifier) in research studies is achieved using a similar method that identifies a similar degree of symptom change but does not include the PMDD requirement that symptoms become “minimal or absent” in the week following menses ( Hartlage & Gehlert, 2001 ). It should be noted that, in the case of PME of chronic symptoms, at least one study has reported an under-reporting of cyclical mood change, suggesting the possibility that false negatives are also of concern when retrospective self-report is used as a screening method ( Eisenlohr-Moul et al., 2018 ). While some evidence suggests that healthcare providers rarely collect or consider daily symptom ratings in clinical practice ( Craner et al., 2014 ), the majority of high-profile clinical research (especially clinical trials) in PMD have utilized daily symptom ratings to evaluate both baseline diagnosis and response to treatment. Despite progress in the field of PMD toward prospective confirmation of the diagnosis, particularly in clinical trials, more work may be needed to standardize the diagnostic process. Even when prospective daily ratings are used, inter-investigator variability appears to have a large impact on who is diagnosed with PMDD ( Eisenlohr-Moul, Girdler, Schmalenberger, et al., 2017 ). Prominent examples include (1) differences in the cycle day demarcations for high- and low-risk phases, (2) differences in the denominator chosen for calculating “percent change” in mood across the cycle (e.g., low-risk phase score, high-risk phase score, absolute range of scale, range of scale used by participant), and (3) differences in the threshold selected as the cutoff for diagnosis (e.g., 30%, 50%, 75%). In our prior study of 200 participants seeking a diagnosis of PMDD, we found that when all combinations of denominator and cutoff options were calculated on the same daily ratings from this sample, the prevalence of PMDD ranged from 1% to 27% depending on the combination of methods used ( Eisenlohr-Moul, Girdler, Schmalenberger, et al., 2017 ). While strict numerical guidelines for diagnosis may not be optimal in clinical settings, our findings led us to recommend that clinical research in PMDD utilize algorithmic scoring of daily ratings to make the diagnosis as standardized as possible and allow for shared meaning between laboratories; we further suggested (1) premenstrual week (−7 to −1 prior to menses onset) versus postmenstrual week (days 4 to 10 post-menses onset) as our phase demarcations, (2) participant range of scale used for the denominator, and (3) 30% change for the cutoff, as these each represent middle-of-the-road options that corresponded well to expert diagnosis. Migraine is a genetically influenced complex disorder characterized by episodes of moderate-to-severe headache, most often unilateral and generally associated with nausea and increased sensitivity to light and sound ( “The International Classification of Headache Disorders, 3rd Edition,” 2018 ). Subtypes of migraine include migraine without aura, migraine with aura, and chronic migraine. Migraine auras are fully reversible visual and/or sensory and/or speech/language symptoms, without motor weakness. Chronic migraine is diagnosed when attacks occur on 15 or more days in a month for more than three months, with migraine features on at least eight or more days in a month. Migraine point prevalence is approximately 10% in U.S. males and 20% in U.S. females ( Burch et al., 2018 ). Elucidating mechanisms of migraine pathophysiology is an area of active research (see Ferrari et al., 2022 ; Puledda et al., 2023 for comprehensive reviews). MM is defined as a pattern in which migraine attacks occur predominantly in the perimenstrual frame in at least two of three menstrual cycles ( “The International Classification of Headache Disorders, 3rd Edition,” 2018 ). MM is currently located in the Appendix of the International Classification of Headache Disorders (ICHD-3) due to uncertainty over its subtypes. See Table 2 for ICHD-3 diagnostic criteria of MM. For our scoping review purposes, we do not follow ICHD-3 guidelines to consider hormonal contraceptive-induced withdrawal bleeds as menstruation, as this bleeding is not a product of the natural menstrual cycle (i.e., ovulation-induced hormone flux and related bleeding). MM can also be found in the ICD coding system, with specifications for Menstrual Migraine, menstrual migraine with status migrainosus, and menstrual migraine: intractable with status migrainosus (where status migrainosus refers to the headache part of the migraine lasting more than 72 hours). Studies of prevalence range widely in MM, suggesting a similar concern over false positive rates using retrospective self-report as discussed in PMD. Some sources suggest MM affects as few as 3% of cycling females ( Couturier et al., 2003 ), while other studies include prevalence rates up to 50% ( Pinkerman & Holroyd, 2010 ). This discrepancy has been highlighted previously, with researchers noting the wide range in prevalence based on study population, recruitment method (e.g., community versus migraine center patients), and diagnostic criteria ( Pavlović et al., 2015 ). Beyond acknowledging the lack of clear prevalence in MM, studies have utilized statistical modeling to compare retrospective report, prospective report, and a novel, probability-based algorithm intended to identify if the probability of observing a menstrual pattern was statistically greater than chance ( Marcus et al., 2010 ). Unsurprisingly, they found that a third of retrospective reports of MM were false positives after prospective daily charting; in addition, they found that the probability model beat ICHD criteria at diagnosing severe menstrual cycle-related migraine, while the ICHD was generally more sensitive. Given that the ICHD criteria are sensitive as a screening tool but not diagnostically specific ( Verhagen et al., 2022 ), additional studies have developed statistical models that are able to robustly exclude chance associations between menstruation and migraine ( Barra et al., 2015 , 2019 ). This body of work stresses the importance of greater standardization and quantification of these diagnoses based on quantitative scoring of daily headache ratings— particularly in the context of clinical trials, where diagnostic accuracy is critical to ensure that hypothesized pathophysiologies (e.g., estrogen withdrawal) are truly relevant to all participants (i.e., since false positive reports of cyclicity would make this mechanism irrelevant and ineffective). The ICHD-3 Appendix includes diagnostic codes for two general subtypes of this phenomenon, pure MM and Menstrually Related Migraine (MRM) (See Table 2 ). To our knowledge, no evidence has been provided to support an etiologic distinction between these two presentations, similar to the current understanding of PMDD versus PME. The primary clinical diagnosis of migraine combines a patient’s history and a physical exam. Once the diagnosis of migraine is established, additional questions determine if the migraine is temporally related to changes in estrogen. Specific topics that can identify hormonal triggers include: association with ovulation or menses; if a hormonal contraceptive affected migraines; how pregnancy affected migraine; and if fertility therapy affected migraine. Exogenous hormone use, such as with COCs, can lead to a rapid decline in estrogen during placebo weeks and subsequently unveil menstrual migraine ( O’Neal, 2022 ). If the attacks meet criteria for migraine and occur exclusively in the perimenstrual frame (days −2 to +3 surrounding menses onset) in at least two of three consecutive menstrual cycles and no other times of the cycle, a diagnosis of PMM can be made. If the attacks meet these criteria and occur beyond just the perimenstrual frame, a diagnosis of MRM can be made. Migraine attacks are recorded as binary variables in daily charting (e.g., a patient did or did not report a migraine today), which limits the issue of inter-investigator variability in calculating percent change that is seen in the PMD literature; however, there is no published guideline on the relative risk ratio between the perimenstrual and lower-risk phases that should be used as a diagnostic threshold. Epilepsy is a disease of recurrent seizures, which result from abnormal electrical activity in the brain whereby large populations of neurons fire together and disrupt normal neurologic functioning resulting in alterations in sensory function, motor function, and/or consciousness ( Devinsky et al., 2018 ). There are various epilepsy syndromes, defined by seizure type (focal vs generalized), epilepsy type (focal, generalized, or combined), and etiological (e.g., structural, genetic, infectious, metabolic, immune, or unknown), as defined by the International League Against Epilepsy ( Scheffer et al., 2017 ). Point prevalence of active epilepsy in the U.S (defined as one or more seizures in the past year or clinically diagnosed epilepsy currently controlled with medication) is estimated to be 1.2% ( Zack & Kobau, 2017 ). CE is generally defined as an increase in seizure activity associated with particular phases of the menstrual cycle. Despite the increasingly widespread recognition of CE as a distinct neuroendocrine disorder, there is little consensus on well-defined diagnostic criteria; there is no official diagnostic system that recognizes CE, unlike the DSM-5 recognition of PMD and the ICHD-3 Appendix diagnoses of MM. Studies often include prevalence rates ranging from 10% to 70% ( Reddy, 2004 , 2009 ). Interestingly, one study found that prospective charting of seizures and menses dates in 40 females with refractory epilepsy led to a prevalence of 12.5%, despite 78% of participants retrospectively reporting an association between menstruation and seizure exacerbation ( Duncan et al., 1993 ). Given the clear inconsistency between retrospective and prospective self-report, one large, prospective study of females with medication-resistant epilepsy defined CE as a participant’s average daily seizure frequency within the “exacerbation phase” being twice as high compared to other specified cycle phases ( Herzog et al., 1997 ). Of note, the proposed threshold of a “two-fold” or greater difference in seizure frequency between specifically defined menstrual cycle phases was created to simplify and standardize prior concern regarding inter-investigator variability in cycle phase designations, mathematical analysis, and definition of cyclical exacerbation ( Herzog et al., 1997 ; Newmark & Penry, 1980 ). These concerns parallel discussions referenced above in PMD and MM, where shared definitions of cyclical symptom change are a crucial step in accurate diagnosis of menstrual cycle-related brain disorders. Researchers have found that defining CE based on a twofold change in seizure frequency across cycle phases diagnoses a catamenial pattern in approximately one-third of women with medication-resistant focal epilepsy ( Herzog, 2008 ). However, as these diagnostic criteria are not codified in any professional manual or official guideline, there is a persistent lack of clarity about the true prevalence of CE ( Herzog, 2015 , 2022 ). Three distinct clinical subtypes of CE (see Table 3 ) have been identified based on the exacerbation phase for seizures: perimenstrual (C1), periovulatory (C2), and inadequate luteal (C3) ( Herzog et al., 1997 ). To our knowledge, there are no standardized guidelines for clinical diagnosis of CE put forth by any major professional organization or manual. Several review articles suggest that CE should be diagnosed with careful charting of menses and seizure diaries ( Foldvary-Schaefer & Falcone, 2003 ; Verrotti et al., 2012 ). However, the duration of charting and degree of change in seizure frequency required to make the diagnosis are not established. Furthermore, we did not account for psychogenic non-epileptic seizures (PNES) in our search terms, which is known to be co-morbid in 10–30% of patients with epilepsy ( Liampas et al., 2021 ). Not addressing the possible presence of PNES in patients with CE can make the determination of CE prevalence challenging and highlights the importance of standardizing the diagnostic assessment and documentation of seizure activity, particularly in this patient population. After taking stock of definitions, subtypes, and assessment practices across these menstrual cycle-related brain disorders, we provide the following reflections. First, significant limitations in diagnostic codification remain. In PMD, PMDD has the clearest codification of criteria as an official disorder within the DSM-5 and ICD, although no codification of PME has been established. MM is a codified subtype of migraine in the ICD, but it lacks specific detail regarding diagnostic methods; it is more specifically defined in the Appendix of the ICHD-3. In contrast, CE has specific detail regarding diagnostic thresholds and cycle phasing, including proposed hormone-related subtypes, but this appears to have resulted in no official diagnostic status or criteria. The next critical area for standardization surrounds the issue of what defines a menstrual cycle. In the DSM-5 diagnosis of PMDD, criterion G specifies “symptoms are not attributable to the physiological effects of a substance.” If an individual experiences adverse mood reactions to fluctuations in exogenous hormonal preparations, such as during the placebo days of a COC regimen (which can cause uterine bleeding due to withdrawal from exogenous hormones but does not indicate a preceding ovulation), they do not meet criteria for PMDD. This criterion is essential in PMDD, as current best evidence suggests that ovulation is the proximal trigger for PMDD symptoms (see section 3 ); if an individual is not ovulating (as is usually the case when taking COCs), then – by definition – their cyclical mood symptoms are not PMDD. In contrast, the ICHD-3 Appendix defining diagnostic criteria for MM specifies “menstruation is considered to be endometrial bleeding resulting either from the normal menstrual cycle or from the withdrawal of exogenous progestogens, as in the use of combined oral contraceptives or cyclical hormone replacement therapy.” Finally, the C3 subtype of CE refers specifically to anovulatory cycles. This presents a clear discrepancy between PMDD, MM, and CE definitions of a menstrual cycle, and whether the abnormal neurobehavioral sensitivity observed in these patients is due to ovulation and related ovarian hormone fluctuations, or a sensitivity to any acute hormone change. Across disorders, retrospective self-report of cyclical symptom pattern appears to massively inflate prevalence (i.e., patients are not able to accurately report cyclical pattern via interview or single-time-point survey). Therefore it appears that there may be a general cognitive bias among patients leading to over-attribution of a wide range of physical and emotional symptoms to the menstrual cycle. At the same time, retrospective underreporting of cyclical mood change has also been reported in some clinical populations ( Eisenlohr-Moul et al., 2018 ). To mitigate these biases, prospective, daily self-report of symptoms appears to be required for valid diagnosis. However, few studies are available to determine the most appropriate methods for quantifying degree of cyclical symptom change and selecting thresholds that represent clinically meaningful change. These efforts at quantification and standardization of cyclical outcomes may prove especially critical in the context of clinical trials and other research endeavors, where accidental recruitment of a sample contaminated by false positives (i.e., those reporting cyclical symptom change who do not demonstrate cyclical change in daily ratings) or false negatives (self-reported “controls” who do in fact demonstrate cyclical change) could massively undermine statistical power and the ability to understand and treat cycle-specific etiologies. Moreover, notable differences in the methods of subtyping have led to a very different meaning of “subtype” across these disorders. In the cases of PMD and MM, subtyping has predominantly taken the form of sequestering exacerbation subtypes (PME, MRM), whose symptoms worsen significantly around menses but also persist to some degree in other phases, excluding these individuals from the “pure” phenotypes of PMDD and PMM. In contrast, subtyping of CE has taken a different form, with a focus on temporal, mechanistic subtyping in which unique biotypes are hypothesized to be triggered by specific hormonal events across the cycle. Upon reflection, we expect the latter approach to be more useful. While separation of exacerbation phenotypes (PME, MRM) may seem a reasonable choice in clinical studies in order to reduce “noise” related to chronicity and comorbidity, it is not clear why they would represent fundamentally different entities with unique mechanisms, since they are characterized by similar cycling of symptoms at a higher average level. Therefore, from a diagnostic standpoint, we consider that the burden of proof remains on those who would separate the exacerbation syndromes from the “pure” syndromes to demonstrate the divergent validity of these proposed diagnostic groups (i.e., to demonstrate that they truly differ in mechanism or need for treatment). In contrast, the proposed subtyping of CE focuses on different mechanisms of seizure at different cycle phases, with periovulatory (C1) representing potentially unique mechanistic subtype from perimenstrual (C2) type CE. This approach provides useful, testable mechanistic insights which could lead to personalized medicine approaches. Therefore, it may be useful to apply similar temporal or mechanistic grouping approaches when hypothesizing possible subtypes of PMD and MM.

In this article, we present a scoping review of existing human research studies with exogenous hormone manipulations across three menstrual cycle-related brain disorders: premenstrual mood disorder (PMD), menstrual migraine (MM), and catamenial epilepsy (CE). These three CNS disorders share the menstrual cycle – a repeating series of predictable, measurable biological events – as their symptom trigger, suggesting that these disorders may represent different phenotypic manifestations of shared neuroendocrine sensitivity ( Figure 3 ). However, these three fields typically operate independent of each other, as mood disorders, migraine, and seizures have unique proximal pathophysiologies and clinical management. Per the expectations and best practices in scoping studies, we map the extant literature across varying study designs and quality of research in order to assess the volume, sources, and types of literature within the field ( Arksey & O’Malley, 2005 ; Pham et al., 2014 ). The present work fulfills multiple purposes within the scoping review framework: 1) we examine a field of literature that has been previously hard to conceptualize; 2) we summarize findings for future researchers; 3) we identify gaps in the literature; and 4) we consider where systematic reviews may be warranted. Below, we outline major impressions reached from the scoping review. First, it is clear that all fields studying the menstrual cycle could benefit from improved measurement methods and diagnostic standardization, and a systematic review of work in this area is warranted. There are seemingly infinite ways to divide the cycle for measurement, including but not limited to: luteal versus follicular; luteal, menstrual, follicular, ovulatory; premenstrual versus post menstrual; including a peri menstrual; measuring serum estradiol and progesterone; urine LH testing; saliva hormone monitoring; menses onset counting; and so on ( Schmalenberger et al., 2021 ). Further, several studies we screened did not include any prospective charting of outcomes, and others did not differentiate between individuals with ovulatory menstrual cycles and those with withdrawal bleeding related to exogenous hormone flux (e.g., using cycle-altering COCs); we excluded these studies with less well-defined methodologies. Second, we identified interesting contrasts between subtyping practices in PMD and MM versus CE. Both PMD and MM fields define two subtypes of their respective disorders: (1) a “pure” form of the disorder, in which symptoms arise only before and during menses, then remit completely for the rest of the cycle (e.g., pure premenstrual dysphoric disorder and menstrual migraine); and (2) an “exacerbation” form of the disorder, whereby symptoms can be chronic, but show a clear worsening around the onset of menses (e.g., premenstrual exacerbation of an underlying psychiatric disorder and menstrually related migraine). In contrast, epilepsy researchers recognize three subtypes of CE defined by their proximity to different biological events of the cycle: C1 type, where seizure frequency peaks around menses (when estradiol and progesterone rapidly fall); C2 type, where seizure frequency peaks around ovulation (following the LH surge and estradiol peak); and C3 type, where seizure frequency appears to occur before menses despite the absence of ovulation. These CE biotypes provide ground for clear, testable hypotheses that could elucidate unique underlying triggers for each subset of participants. We envision a future where neuroendocrine researchers in psychiatry, neurology, and related fields can diagnose patterns of cyclical brain disorders based on their biotype. Finally, we conclude that researchers across these fields have much to gain from understanding the experimental trials and mechanistic insights from each menstrual cycle-related brain disorder subfield. One example is utilizing dosing and medication information from existing trials of supplemental estrogens to predict findings. Since several studies include extensive hormonal assays to identify effects of transdermal, injected, or oral estrogens on ovulation, future work in any of the three subfields should lean heavily on this information when designing study hypotheses. Further, improved communication between researchers could lead to more collaborative studies; if a cycle manipulation trial is being undertaken in either population, both MM and PMD fields could benefit from adding measures of either mood or headache (respectively) to identify comorbidities and symptom overlap. We also find that all three disorders have recently focused on neuroactive steroids, such as allopregnanolone and isoallopregnanolone, as potential mediators of neuroendocrine relationships. Specifically, experimental work in PMD that shows blocking neuroactive steroid synthesis prevents mood symptoms ( Martinez et al., 2016 ), and, despite null findings for certain primary endpoints, clinical trials have safely tested isoallopregnanolone as a treatment for both PMD and MM, providing useful future directions ( Bäckström et al., 2021 ; Bixo et al., 2017 ; Nordkild, 2021 ). In contrast, this relationship has been tested only in post-hoc mediation analyses ( Herzog & Frye, 2014 ) and case studies ( Herzog & Frye, 2003 ) of CE. Thus, neuroactive steroid research presents an example of how study design and methods from one field could be carried over to expedite research and improve predictions in another. Of course, there are limitations to the scoping review process. First, we summarize literature across broad diagnostic categories, which do not necessarily have similar needs. While we do see strong value in comparing across menstrual cycle related brain disorders, we acknowledge that what works for one disorder may not be feasible or relevant in another. Second, the scoping review process focuses on mapping a breadth of literature, with a less critical eye towards study quality. Future work could take a more specific approach, only including the best and least biased experimental designs (such as randomized, double-blind, placebo-controlled clinical trials). Additionally, a scoping review is inherently non-systematic and undertakes no meta-analytic processes. Thus, we cannot speak to or compare effect sizes, sample sizes, or draw any evidence-based conclusions. Finally, our scoping review search query was for hormonal manipulation studies in humans, which inherently excluded a broad basis of preclinical literature. There is much knowledge to be gained from animal models and basic science methodologies to understand both pathophysiology and treatment mechanisms. We propose that future systematic or meta-analytic reviews take a translational approach to better understand mechanisms and neurobiology of these three menstrual cycle related brain disorders. Across PMD, MM, and CE, various exogenous hormone manipulations have been tested to treat symptoms, including altering the cycle with combined oral contraceptives, suppressing ovarian hormone production with gonadotropin releasing hormone antagonism, and a wide variety of additional hormone manipulations, such as supplemental estrogens and progestins. While no individual treatment or study design treats all three menstrual cycle-linked CNS disorders, it is clear that there is overlap in study design and neuroendocrine-based hormone sensitivity. In the current scientific landscape that promotes open science, reproducibility, and translational collaboration, our scoping review finds that there are massive opportunities for collaboration and education among menstrual cycle researchers across psychiatric and neurologic fields.

Ovarian steroid hormones exert myriad effects on the brain throughout the lifespan (see McEwen, 2002 ; Pletzer et al., 2023 ; Schiller et al., 2016 for reviews). While these reproductive hormone fluctuations (such as puberty, pregnancy, the menstrual cycle, or menopause) do not lead to clinically significant symptoms in most cases, specific menstrual cycle-linked disorders are triggered by aberrant effects of ovarian hormones on the central nervous system (CNS). Independent lines of research have established that an abnormal sensitivity to the normal hormone flux of the menstrual cycle drives clinical subtypes of three CNS-mediated disorders: premenstrual mood disorder, (PMD; Epperson et al., 2012 ); menstrual migraine, (MM; MacGregor, 2007 ); and catamenial epilepsy (CE; Foldvary-Schaefer & Falcone, 2003 ). As cyclical hormone levels and patterns do not generally differ in case-control studies between patients of each respective disorder (PMD, MM, CE) and healthy controls, the current best evidence points to some form of abnormal neuroendocrine sensitivity to the menstrual cycle—that is, they may be described as “hormone-sensitive” disorders, whereby individuals have heightened responsiveness or reactions to normal hormone changes. Given the diverse effects of hormones in the brain, the mechanisms of such hormone sensitivities are likely to differ both between and within these traditional diagnostic categories, and are probably best studied dimensionally rather than categorically ( Insel et al., 2010 ). Prior work has reviewed possible shared sex steroid-based etiologies of PMD, CE, and MM, with a focus on the link between sex steroids and the GABAergic neurotransmitter system that modulates CNS excitability ( Guille et al., 2008 ; Smith et al., 2007 ). Therefore, although PMD, MM, and CE are characterized by unique symptoms and pathophysiologies, they share the menstrual cycle as a temporal, within-person trigger that links reproductive hormones to their CNS disturbances. Considering this fundamental overlap, we conducted a scoping review of experimental hormone research studies in PMD, MM, and CE in order to identify similarities and differences in research methodology used across these disorders. To our knowledge, while there is a strong body of literature reviewing the potential mechanisms that link menstrual cycle-linked neuroendocrine disorders ( Bäckström et al., 2003 ; Finocchi & Ferrari, 2011 ; Guille et al., 2008 ; Reddy, 2022 ; Smith et al., 2007 ), there are no existing reviews to date summarizing the hormonal experiments that have been conducted across these topics. We utilize the scoping review format, as the broad goals of scoping reviews include identifying types of available evidence, examining the extent, range, and nature of how research is conducted across fields, identifying gaps in the literature, and assessing if a systematic review is warranted ( Arksey & O’Malley, 2005 ). A scoping review allows us to chart the current state of the experimental clinical literature to inform future research goals, unlike a systematic review, which generally focuses on answering a specific research question and reviewing the quality of included research ( Munn et al., 2018 ). The structure of this scoping review is as follows: we briefly review clinical diagnosis and assessment practices of each disorder ( Section 2 ), which is a necessary piece of the scoping review given the heterogeneity of methods used for studying menstrual cycle-based neuroendocrine disorders. The bulk of the scoping review ( Section 3 ) will describe the hormone manipulation studies conducted in human subjects with PMD, MM, and CE, with the goal of understanding each field’s primary methods and leading hormone-related hypotheses. Where appropriate, we will identify methodological gaps in the literature (e.g., hormonal manipulations examined in one disorder but not another), and comment on areas where future systematic reviews or meta-analyses are warranted ( Arksey & O’Malley, 2005 ; Pham et al., 2014 ; Tricco et al., 2016 ). Section 3 is organized by type of experimental manipulation: (1) trials of combined oral contraceptive (COC) agents, (2) trials of ovulation prevention with GnRH agonists (chemical menopause agents), and (3) trials of miscellaneous hormone manipulations. Finally, we provide a summary ( Section 4 ). In line with the scoping review format, we focus primarily on describing the designs and number of experiments that have been presented in the literature, and do not provide a systematic discussion of the potential mechanisms or specific findings. In summary, our scoping review aims to assess the nature of the current literature in menstrual cycle-linked CNS disorders, highlighting methodological strengths and gaps within disciplines and areas for potential cross-disciplinary collaboration and future work. The monthly female reproductive cycle ( Figure 1 ), lasting an average of 28 days, is structured around ovulation, when an egg is released from the ovary for the purposes of possible fertilization (i.e., pregnancy), and menstrual bleeding, the shedding of the uterine lining that begins a new cycle ( Welt et al., 2022 ). Hormonal feedback loops between the ovaries and the brain coordinate these recurring events.