Progesterone and contraceptive progestin actions on the brain: A systematic review of animal studies and comparison to human neuroimaging studies

In this review we systematically summarize the effects of progesterone and synthetic progestins on neurogenesis, synaptogenesis, myelination and six neurotransmitter systems. Several parallels between progesterone and older generation progestin actions emerged, suggesting actions via progesterone receptors. However, existing results suggest a general lack of knowledge regarding the effects of currently used progestins in hormonal contraception regarding these cellular and molecular brain parameters. Human neuroimaging studies were reviewed with a focus on randomized placebo-controlled trials and cross-sectional studies controlling for progestin type. The prefrontal cortex, amygdala, salience network and hippocampus were identified as regions of interest for future preclinical studies. This review proposes a series of experiments to elucidate the cellular and molecular actions of contraceptive progestins in these areas and link these actions to behavioral markers of emotional and cognitive functioning. Emotional effects of contraceptive progestins appear to be related to 1) alterations in the serotonergic system, 2) direct/indirect modulations of inhibitory GABA-ergic signalling via effects on the allopreg- nanolone content of the brain, which differ between androgenic and anti-androgenic progestins. Cognitive effects of combined oral contraceptives appear to depend on the ethinylestradiol dose.


Introduction
Hormonal contraceptives (HC) have been approved for more than 60 years now with an estimated 150 million users worldwide (UnDeAS, 2019). By mimicking the effects of endogenous progesterone, they target the hypothalamic-pituitary-gonadal axis and downregulate endogenous hormone production (e.g. Kuhl, 2011). Thus, while their primary target is in fact situated in the brain, potentially widespread effects on other neuronal and behavioral outcomes beyond the reproductive axis have only been recognized in the past decade (e.g. Pletzer & Kerschbaum, 2014). Behavioral side effects of hormonal contraceptives include both positive and negative alterations in mood (for reviews see Fruzzetti & Fidecicchi, 2020;Robakis et al., 2019;Sundstrom-Poromaa & Segebladh, 2012;Böttcher et al., 2012;Schaffir et al., 2016), and cognition (last systematically reviewed by Warren et al., 2014). Regarding mood alterations, approximately 4-10% of women develop depressive symptoms during the first few months of treatment, which is a major reason for discontinuation of contraceptive treatment (Kelly et al., 2010). In contrast, mood-stabilizing effects have been reported in long-term users of hormonal contraceptives (Jarva & Oinonen, 2007). Regarding cognitive changes, results suggest an overall improvement of verbal memory, while effects on other cognitive abilities were inconsistent and potentially dependent on the specific contraceptive formulation (Warren et al., 2014).
Particularly with regards to emotional side effects, it is of high relevance to understand, what determines women's individual sensitivity towards emotional and cognitive alterations during contraceptive treatment in order to be able to individually tailor contraceptive approaches to minimize side effects. However, the neuronal alterations underlying these behavioural changes are not well understood. Human neuroimaging studies have identified a number of brain areas relevant to emotional and cognitive processing that appear to be sensitive to hormonal contraceptive use (compare section 1.3.). For these areas studies reported changes in gray matter volume or cortical thickness, brain activation or connectivity to other brain areas (systematically reviewed by Broennick et al., 2020). But the cellular and molecular mechanisms potentially underlying these changes in neuroimaging parameters have received less attention (see Porcu et al., 2019). In order to fully understand what happens in individual women's brains during treatment with specific contraceptive progestins, it appears that a systematic interdisciplinary approach has to be the way forward. Human neuroimaging studies and preclinical animal studies have to inform each other in order to clearly delineate the brain systems responsive to contraceptive steroids. Given the replication crisis and power failures in the neurosciences due to massive multiple comparisons problems (Button et al., 2013), synchronizing approaches in terms of brain areas seems to be the logical and also the simplest first step. A second step is to focus further studies on those neural systems that most consistently respond to endogenous ovarian hormones. Accordingly, the first goal of the current review is to provide an overview of our current state of knowledge regarding cellular and molecular alterations in the brain and map out, which changes concern those brain areas that consistently respond to contraceptive treatment in human neuroimaging studies. The second goal is to obtain a systematic overview of the cellular and molecular responses to endogenous progesteronealone or in combination with estradiol. In combination, these overviews should highlight those brain systems most likely to respond to contraceptive treatment and thereby provide a roadmap for contraceptive research in the upcoming years.
To that end, we have identified four mechanisms at the cellular and/ or molecular level that may potentially underly changes in structural and functional neuroimaging parameters. These include: (i) changes in the number of neurons, i.e. neurogenesis, (ii) changes in the number of synapses between neurons, i.e. synaptogenesis, (iii) changes in the efficacy/velocity of electro-chemical information transfer between neurons, i.e. myelination, and (iv) changes in neurotransmitter signalling that may occur at multiple levels, including the synthesis, release, degradation or receptor binding of a specific neurotransmitter. All of these mechanisms have been extensively studied with respect to their modulation by endogenous sex hormones, specifically estradiol andto a lesser extentalso progesterone and testosterone. Via a systematic review, we seek to understand in which brain areas contraceptive progestin administrationalone or in combination with estrogensis related to alterations in these four mechanisms and whether similarities to human neuroimaging studies arise. We furthermore seek to identify how these mechanisms respond to endogenous progesteronealone or in combination with estrogensand compare responses between endogenous progesterone and contraceptive progestins.

Synthetic progestins currently in use in HC
Hormonal contraceptives consist of either a synthetic progestin (progestin only), or a synthetic progestin in combination with an estrogen (see Table 1). Figure 1 provides an overview of synthetic progestins predominantly used for hormonal contraception. They are devided in four main groups, the Pregnanes (progesterone derivates), Gonanes (19-nortestosterone derivates), the spironolactone derivate Drospirenone and the fourth-generation progestin Dienogest, an aromatized 19-nortestosterone derivative. Derivates of progesterone i.e. medroxyprogesterone, chlormadinon and cyproterone, synthesized via 17-hydroxy-progesterone, were already used 60 years ago in firstgeneration contraceptives and are nowadays experiencing a revival due to their anti-androgenic properties (e.g. Davtyan, 2012). The most recent progestin nomegestrol, is derived from progesterone via 19-norprogesterone. All Pregnanes are administered as acetate and are known for their anti-androgenic properties. The 19-nortestosterone derivative levonorgestrel is the key progestin contained in secondgeneration contraceptives and the standard of care in many countries due to its low risk for thrombo-embolic events. Derivates of levonorgestrel (gestoden, desogestrel, etonorgestrel, norelgestromin and norgestimate) are used in third-generation contraceptives. All Gonanes have androgenic properties, though the latter are lower in thirdgeneration progestins compared to Levonorgestrel.
Progestin only contraceptives include 1) progestin only pills (POPs), which contain either desogestrel or drospirenone 2) IUDs, which contain levonorgestrel 3) contraceptive implants, which contain etonorgestrel 4) or injections, which contain medroxyprogesterone acetate. In combined oral contraceptives (COCs), as well as rings and patches, 1-4 and various other progestins are administered in combination with an estrogen. The most commonly used estrogen is ethinylestradiol, although newer COCs contain the bio-identical estrogens estetrol or estradiol-valerate, which is quickly metabolized to estradiol upon ingestion.

Mechanisms of action of synthetic progestins
Synthetic progestins have been thouroughly studied and characterized regarding their pharmacological profile and binding affinity to various nuclear steroid hormone receptors (compare Table 2). All progestins are designed to activate nuclear progesterone receptors and are anti-estrogenic due to the downregulation of estrogen receptor expression and endogenous estradiol production (e.g. Kuhl, 2011). This is relevant, since endogenously, progesterone actions are primed by estrogens, given that estrogen receptors upregulate the expression of progesterone receptor genes (Quadros et al., 2008). Accordingly, important differences may arise between progestin only formulations and combined estrogen-progestin formulations that supplement for the reduced endogenous estradiol production.
However, while synthetic progestins mimic the actions of endogenous progesterone via their binding to intracellular progesterone receptors, their actions may still differ from endogenous progesterone in various aspects. First, synthetic progestins differ in their binding affinities as well as agonistic and antagonistic potential towards androgen receptors, glucocorticoid receptors and mineralocorticoid receptors (compare Table 2; see Kuhl, 2011 for a review). Accordingly, synthetic progestin actions may also arise from their androgenic or antiandrogenic activities, as well as glucocorticoid or mineralocorticoid activities, and these actions may differ between different synthetic progestins.
Second, progesterone actions on the brain are not only excerted via nuclear progesterone receptors. The first, and best documented alternative pathway concerns progesterone metabolites, in particular the 5αreduced metabolite allopregnanolone. Allopregnanolone is a positive allosteric modulator of the GABA-A receptor, which is potentially responsible for the anxiolytic, anti-depressive and sedative properties of progesterone (Pinna, 2020). However, synthetic progestins are not metabolized to allopregnanolone and differentially affect the levels of allopregnanolone in the brain (systematically reviewed in the current manuscript, compare Table 3). Nevertheless, some progestins, Table 1 Hormonal contraceptives currently in use. 1 currently the standard of care The table lists only synthetic steroid combinationsvariations in concentrations and recommended intake regimen (21+7, 24+4, 28+0) are available. Oral administration often requires a pro-drug, i.e. the physiologically active progestin is listed in brackets.
Finally, both membrane-associated progesterone receptors (mPR) and progesterone receptor membrane components (PGMRC) have been identified and are likely responsible for some fast-acting cellular changes in response to progesterone (reviewed by Zhu et al., 2008). In particular, some of the neuroprotective effects of progesterone have been attributed to its interaction with mPRs and PGMRCs (e.g., Sun et al., 2016). To the best of our knowledge, it is yet unclear, whether the contracptive progestins listed in Table 2 also interact with mPRs and PGMRCs.
Accordingly, results from progesterone administration studies do provide a promising starting point for contraceptive progestin administration studies, but differences may occur when progesterone actions are not excerted via intracellular progesterone receptors. Thus, when assessing contraceptive progestin actions on the brain, it is important to (i) distinguish between the different types of progestins, (ii) distinguish between synthetic progestins administered alone or in combination with estrogens, and (iii) compare the synthetic progestin actions to progesterone actions to distinguish between actions dependent on nuclear progesterone receptors and actions dependent on other mechanisms.

Contraceptive actions on the human brainneuroimaging studies
A recent meta-analysis from 2020 identified 33 neuroimaging studies on hormonal contraceptive actions on the brain, though some reported on the same samples (Broennick et al., 2020). Since then, five additional neuroimaging studies (Petersen et al., 2021;Chen et al., 2021;Wen et al., 2021;Menting-Henry et al., 2022;Noachtar et al., 2022), and one PET (Larsen et al., 2022) study have been published on this subject. Results of those studies are mixed, which is most likely due to the following reasons: Most importantly, 33 out of the 38 total neuroimaging studies are cross-sectional studies comparing current users of hormonal contraceptives (mostly COC) to current non-users of hormonal contraceptives. It has been extensively discussed that those designs underly a variety of sampling biases (Pletzer & Kerschbaum, 2014;Broennick et al., 2020;Schuster et al., 2022). Even if demographic differences in age, education and socio-economic status are controlled for, long-term users usually show little to no side effects or beneficial effects on their contraceptive, while non-users usually have a history of adverse side effects on various hormonal contraceptives. Accordingly, crosssectional studies are just as likely to capture pre-existing differences in the brain that contribute to the vulnerability to develop side effects or changes attributable to the previous use of hormonal contraceptives, as they are to capture changes attributable to the current contraceptive use. Accordingly, the results of cross-sectional studies should be discussed separately from the results of longitudinal or placebo-controlled studies. Comparisons may be cautiously drawn, if previous contraceptive use or previous adverse effects during contraceptive use are taken into consideration by cross-sectional studies.
Furthermore, different progestins may differ in their neuronal outcomes and should therefore be studied separately regarding their effects on the brain. Unfortunately, 24 out of 33 cross-sectional studies and one out of five longitudinal studies available did not assess or report the progestin used in hormonal contraceptives. Accordingly, results are hard to interpret. Therefore, we will only summarize results of longitudinal neuroimaging studies and cross-reference them with those crosssectional studies, that specify the type of progestin used.

Effects of COC-treatment on the frontal cortex
A Swedish double-blind randomized placebo-controlled trial included a task on emotional face matching (Gingnell et al., 2013), response inhibition (Gingnell et al., 2016), as well as a resting state scan (Engman et al., 2018). 34 women with previous negative affect during COC-use were recruited and received either placebo or a COC containing 0.30 μg EE and 0.15 mg LNG for one month. Alongside this study, Petersen et al. (2021) recruited 26 women with previous negative affect during COC-use to receive one month of placebo and one month of 0.30 μg EE and 0.15 mg LNG double-blinded and in randomized order and assessed cortical thickness as well as mood scores. Both studies reported a significant deterioration during COC-use compared to placebo. Furthermore, imaging studies emerged the prefrontal cortex as area of interest in the context of levonorgestrel treatment. During the emotional face matching task, activity in the left middle frontal and bilateral inferior frontal gyri was significantly reduced during COC-use compared to placebo (Gingnell et al., 2013). During response inhibition, activity in the right middle frontal gyrus was significantly reduced during COC-use compared to placebo (Gingnell et al., 2016). Finally, cortical thickness was significantly reduced in the right inferior frontal gyrus during COCuse compared to placebo (Petersen et al., 2021). In accordance with these longitudinal observations, reduced gray matter volumes in the bilateral middle frontal gyri and reduced cortical thickness in the orbitofrontal gyri were observed in cross-sectional studies comparing users of androgenic COC to naturally cycling women (Pletzer et al., 2015;Petersen et al., 2015a,b). However, no differences between users of Table 2 Pharmacokinetics and Pharmacodynamics of various progestins. Abbrev. = Abbreviation, 1 Binding affinities to the respective nuclear receptors in percent relative to the endogenous hormones, 2 percent of the hormone bound to the respective binding globulins, PR = Progesterone receptor (binding affinities were re-calculated from binding affinities reported in the literature as assessed in competition to promegestone, which has a 2x higher binding affinity to the PR than progesterone), ER = estrogen receptor (binding affinities were assessed relative to estradiol; studies did not differentiate between ER α and β), AR = androgen receptor (binding affinities were assessed relative to dihydrotestosterone, which has a 2x higher binding affinity to the AR than testosterone), GR = glucocorticoid receptor (binding affinities were re-calculated from binding affinities reported in the literature as assessed in competition to dexamethasone, which has a 10x higher binding affinity to the GR than cortisol), MR = mineralocorticoid receptor (binding affinities were assessed relative to aldosterone), Bioavail. = Bioavailability, SHBG = sex hormone binding globuline. Italics indicate antagonistic binding. A-A = antiandrogenic potency relative to cyproterone acetate. Data were compiled and re-calculated from Kuhl (1990), Kuhl (2005) and Kuhl (2011 Table 3 Progestin effects on allopregnanolone levels in various brain areas. OVX = ovariectomy, E2 = estradiol, EV = estradiol valerate, MPA = medroxyprogesterone acetate, DSP = drospirenone, DDG = dydrogesterone, NOMAC = nomegestrol acetate, CMA = chlormadinone acetate, LNG = levonorgestrel, andro. = androgenicity, AA = anti-androgenic, A = androgenic, ctx = cortex, HIPPO = hippocampus. androgenic COCs and non-users were observed upon viewing of food stimuli, though samples size in this study was very small (Arnoni-Bauer et al., 2017). Taken together, the results suggest, that several prefrontal areas underly a down-regulation during treatment with androgenic COC, specifically LNG. These results may be related to decreased emotional responsiveness and mood worsening observed during treatment with LNG. However, it is yet unclear, whether these results are restricted to, or more pronounced in women with previous negative affect during COC use, as they may be particularly sensitive to the emotional effects of LNG. Accordingly, it is of particular interest for animal models, whether LNG affects neurogenesis, synaptogenesis or neurotransmitter content in frontal brain areas. Finally, it is an interesting question, whether different progestins have different effects on frontal cortex volume or reactivity. For instance, frontal cortex reactivity appears to be increased upon injection with the anti-androgenic progestin MPA (Basu et al., 2016). An observation that is in line with results from a cross-sectional study in which the majority of women were using anti-androgenic COC (Bonenberger et al., 2013). Likewise, our restrospective study revealed that frontal cortex volumes were increased in a group of COC users, that were majorily treated with drospirenone containing COC (Pletzer et al., 2010). However, no differences between groups of mostly antiandrogenic COC use and non-users were observed in the prefrontal cortex upon viewing of erotic stimuli  or during a stress task (Chung et al., 2016). However, it has to be kept in mind that all of the above-mentioned cross-sectional studies included lass than 15 subjects per group. The findings of increased frontal cortex reactivity with anti-androgenic OC use are in accordance with an animal study demonstrating increased synaptogenesis in this area upon MPA treatment (Chisholm et al., 2012a,b). Indeed, future research should be conducted focusing on the effect of anti-androgenic synthetic progestins contained in COC like chlormadinone acetate or drospirenone on neurogenesis or synaptogenesis in the frontal cortex. Lisofsky et al. (2016) recruited 28 women, who were planning to start COC use, and collected structural and resting state scans prior to their COC use and after three months of COC use. 28 naturally cycling women, who did not start COC use served as a control group. As in Gingnell et al. (2013) and Petersen et al. (2021), mood was significantly worsened during COC-treatment. A significant reduction in gray matter volumes of the left amygdala was observed during COC-treatment and accompanied by a significant reduction in resting state connectivity between the amygdala and the dorsolateral prefrontal cortex. In line with this observation, reduced amygdala reactivity to emotional stimuli in COC-users was observed in a cross-sectional study (Petersen & Cahill, 2015). Since neither study reported information on the specific contraceptive formulation, it is unclear whether these findings hold across contraceptive formulations. Reduced amygdala connectivity to cortical areas during combined EE/LNG use was observed by Engman et al. (2018) during the resting state. Furthermore, a recent cross-sectional study demonstrates differential modulation of amygdala-behavior relationships by androgenic and anti-androgenic COC (Menting-Henry et al., 2022). In line with contraceptive studies, reduced amygdala reactivity to emotional stimuli was also observed in phases of high endogenous progesterone (Derntl et al., 2008).

Effects of COC-treatment on the salience network
The salience network is a large-scale brain network comprised of the anterior cingulate cortex (ACC) and bilateral Insula (Uddin et al., 2016). The ACC demonstrates increased connectivity to the precuneus and dorsolateral prefrontal cortex during EE/LNG treatment (Engman et al., 2018), while the insula showed decreased reactivity to emotional stimuli (Gingnell et al., 2013). The latter effect appeared to be related to mood deterioration. In contrast, a cross-sectional study reported heightened reactivity to traumatic film viewing in the insula and ACC of COC-users compared to non-users (Miedl et al., 2018). Furthermore, increased gray matter volumes of the ACC and insula were reported during the active compared to the inactive pill phase by De Bondt et al. (2013), while Petersen et al. (2015b) observed decreased cortical thickness of the ACC and insula in COC-users compared to non-users. Neither study did report information on the specific COC-formulations used, however given the timing and place of the studies, it may be speculated that majority of participants in De Bondt et al. (2013) and Miedl et al. (2018) used anti-androgenic COCs, while majority of participants in Petersen et al. (2015b) used androgenic COCs. In addition, reactivity of the ACC and amygdala to endogenous estradiol was abolished in COC-users (De Bondt et al., 2013;De Bondt et al., 2016). Accordingly, it would be of great interest to explore, whether androgenic and anti-androgenic COC differentially modulate the salience network. However, in a sample of mostly anti-androgenic COC-users compared to non-users, a study by Abler and colleagues observed decreased insula reactivity upon erotic stimulation , but increased insula reactivity during monetary reward processing (Bonenberger et al., 2013). Accordingly, it is possible that COCs alter the type of stimuli selected by the salience network, rather than the reactivity of the salience network per se.

Effects of COC-treatment on the hippocampus/parahippocampus
No longitudinal study observed changes in hippocampal volumes or activation due to COC use, although the cluster of reduced gray matter in the left amygdala observed by Lisofsky et al. (2016) extended in the the parahippocampal gyrus. However, given that the hippocampus is the most plastic area of the human brain and appears to be particularly receptive to endogenous hormonal fluctuations in both human and animal studies (see Hillerer et al., 2019 for a review), it seems worthwhile to investigate the effects of different contraceptive formulations on the hippocampus. Unfortunately, results regarding these areas are rather mixed. While some studies report increased volumes of the hippocampus/parahippocampus with anti-androgenic COC use (Pletzer et al., 2010;Pletzer et al., 2015), the opposite effect was observed in studies using mixed samples of androgenic and anti-androgenic COC-users (deBondt et al., 2013;Lisofsky et al., 2016;Pletzer et al., 2019a,b). However, it should be taken into account that the most pronounced hormonal effects in the hippocampus were observed in response to endogenous estradiol, which are reversed by endogenous progesterone (Hillerer et al., 2019). Accordingly, COC-effects in the hippocampus may be the result of how specific progestins modulate the effects of ethinylestradiol.

Objectives of the current review
The goal of the current review was to address the effects of (i) endogenous progesterone and (ii) the contraceptive progestins listed in Table 1 alone or in combination with estrogens on the cellular and molecular mechanisms, potentially underlying changes in brain structure and function identified by human neuroimaging studies on contraceptive use. As outlined in the introduction potential mechanisms include (i) neurogenesis, (ii) synaptogenesis, (iii) myelination and (iv) neurotransmitter signalling. With respect to the latter, we restricted our search to six major neurotransmitter systems, i.e., acetylcholine, glutamate, GABA, serotonine, dopamine, and norepinephrine, which are related to cognitive and emotional functioning. Beyond the scope of this review, indirect actions of contraceptive progestins on neurotransmission may arise via their interactions with various endocrine and neuromodulatory systems, e.g., thyroid hormones, oxytocin or betaendorphin. Results on endogenous progesterone are included to (i) compare the actions of synthetic progestins to endogenous progesterone and (ii) generate hypotheses for future studies on contraceptive progestin actions on the brain. While results on the whole brain will be extracted, a particular focus lies on cognitively relevant brain areas, as identified as responsive to contraceptive progestins in human neuroimaging studies.

Methods
Literature research was performed according PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines (Moher et al., 2009) and last updated on 06.01.2023.

Search strategy
We performed a PubMed/Medline search using key words combining the administered agent (progesterone OR "hormonal contraceptive" OR "hormonal contraception" OR "oral contraceptive" OR "synthetic progestin" OR levonorgestrel OR chlormadinone OR medroxyprogesterone OR cyproterone OR desogestrel OR etonogestrel OR gestodene OR dienogest OR nomegestrol OR drospirenone) with synonyms for the respective outcome variables of interest (neurogenesis OR "neuronal proliferation" OR "neuronal survival" OR BDNF; synaptogenesis OR "synaptic spine" OR "dendritic spine" OR "mushroom spine" OR synapse; myelin OR myelination; acetylcholine OR glutamate OR GABA OR serotonine OR dopamine OR norepinephrine) andif necessary because not included in the outcome variables of interestthe target area of interest (brain). All searches were filtered for language of the article ("English" or "German"), species ("other animals") and sex ("female"). The specific search terms for each outcome measure are listed in the respective paragraphs of the results section. Separate searches were performed for neurogenesis, synaptogenesis, myelination and neurotransmitter systems and duplicate articles were removed prior to record screening. Additionally, the reference lists of relevant articles were screened for additional relevant articles by two reviewers (KH and BP). The number of articles found, and the number of articles included are listed in the respective paragraphs of the result section. The detailed search strategy is presented in a flow chart in Figure 2.

Article screening
Screening for inclusion criteria was performed in separate steps on title, abstract and full-text by two independent reviewers (KH and BP). Discrepancies were resolved by discussion. Articles were included if they fulfilled the following ranked inclusion criteria: 1. publication type: original research article 2. language: Englisch or German 3. species: non-human 4. sex: female (studies on males or mixed-sex groups were excluded) 5. age: adolescent-adult (developmental studies were excluded) 6. target area for outcome assessment: brain/neuronal tissue 7. administered agent: progesterone or synthetic progestin, alone or in combination with any estrogen; no restriction regarding dosage, timing or frequency of administration; no restriction regarding control: any placebo, vehicle, oil without treatment agent. However, articles that lacked an adequate control group were excluded. 8. outcome variable of interest Inclusion criteria regarding age and sex were chosen in order to reflect the target population for hormonal contraceptive use in humans. Studies in which progesterone was not administered, but elevated through a natural model (estrous cycle, pregnancy) were compared to the administration studies via narrative synthesis.
Outcome variables of interest included: • neurogenesis: cell proliferation as assessed via number of BrDU or Ki67 positive cells, cell survival or apoptosis as assessed via the number of cells surviving in neurodegeneration models, or, indirectly, brain derived neurotrophic factor (BDNF) • synaptogenesis: synapse/spine density or number as assessed via Golgi staining or synaptic protein expression (e.g. synaptophysin, syntaxin) • myelination: myelin content or thickness, axon diameter • neurotransmitters (ACH, glutamate, GABA, serotonine, dopamine, norepinephrine): synthesis (as assessed vie synthesizing enzyme content or expression), levels, release (i.e. levels upon stimulation), current in reaction to application of the neurotransmitter or receptor agonists, turnover (as assessed via the levels of metabolites), degradation (as assessed via degrading enzyme content or expression), transport (as assessed via transporter content or expression), receptor density, receptor binding

Data extraction
The following data were extracted from article texts and graphs by two reviewers (KH and BP): (i) species, (ii) age, (iii) ovariectomy (yes/ no), (iv) type of progestin administered, (v) type of estrogen administered, if any, (vi) brain area, (vii) outcome measure, (viii) result (increase, decrease or no change). In case of missing information, authors of eligible studies were contacted via email. Articles, which did not provide details on the inclusion criteria listed above, and for which this information could not be obtained by contacting the authors, were excluded. Given the limited number of eligible studies for the majority of outcome measures, outcome measures weren't quantified. However, it was listed in Tables 3-15, whether an increase, decrease or no change was observed, followed by a narrative synthesis of the tabulated results. Tables 3-15 are organized first by outcome, second by administered agent, third by the brain area in which the outcome was assessed, and fourth by publication date. Given that this review focuses on cognitively relevant brain areas, results on hypothalamic nuclei are reported in Supplementary Material 1. All data types were continuous.

Bias assessment
Bias assessment was performed using SCYRCLES risk of bias assessment tool for animal studies (Hooijmans et al., 2014), which consists of 10 items addressing the randomization (Item 1) and blinding (Item 3, Item 5 and Item 7) of treatment allocation, matching of treatment groups (Item 2 and Item 4), randomization of outcome assessment (Item 6), as well as the completeness of outcome reporting (Item 8 and Item 9). Item 1 ("Was the allocation sequence adequately generated and applied") was answered YES if researchers noted that animals were randomly assigned to the treatment groups, even if no details on the allocation sequence were provided. Item 2 ("Were the groups similar at baseline or were they adjusted for confounders in the analysis?") was YES if the researchers noted that the groups were matched even if no details on the criteria for matching were given or if all animals were so similar that no matching seems necessary. Item 4 ("Were the animals randomly housed during the experiment?") and Item 6 ("Were animals selected at random for outcome assessment?") was answered YES if researchers noted that animals were randomly housed or selected for outcome assessment, even if no details on the randomization procedure were provided. Items 1-7 were answered Unclear, if no information on randomization, matching or blinding was provided. The answer NO was only given if the information provided clearly indicated that no randomization, matching or blinding had taken place. Item 8 ("Were incomplete outcome data adequately addressed?") was answered Unclear, if it could not be determined how many animals were analyzed with respect to how many animals were tested. This was usually the case when the number of animals tested was not mentioned in the methods section. Item 9 ("Are reports of the study free of selective outcome reporting?") was answered Unclear, if the outcomes were not fully listed in the methods section. This was usually the case when an outcome was assessed in multiple brain areas and it was unclear whether the areas listed in the results section were all the areas assessed or just a selection. For Item 10 ("Was the study apparently free of other problems that could result in high risk of bias?") we focused our assessment on the statistical analysis applied. A risk of bias was assumed, if multiple outcomes were assessed, but no multiple comparisons correction was performed (or only multiple comparisons correction for the number of groups, but not for the number of outcomes). Detailed results for each item and each study are presented in Supplementary Material 2 and a summary is provided in Figure 3. Furthermore, a sum score of SCYRCLE points was obtained by counting the number of YES answers for each study (compare Tables 3-15). Average number of SCYRCLE points for each outcome area are reported in the respective paragraphs of the results section to provide a rough overall confidence level for the results presented in each section.

Contraceptive progestin actions on allopregnanolone content
As progesterone asserts its effect on the brain not only via binding to progesterone receptors, but also via conversion to allopregnanolone, we first compiled evidence on the contraceptive progestin actions on allopregnanolone levels in the brain. Articles on the progestin actions on allopregnanolone were identified using the following search terms: ("hormonal contraceptive" OR "hormonal contraception" OR "oral contraceptive" OR "synthetic progestin" OR levonorgestrel OR chlormadinone OR medroxyprogesterone OR cyproterone OR desogestrel OR etonogestrel OR gestodene OR dienogest OR nomegestrol OR drospirenone) AND allopregnanolone AND brain. The search returned 23 articles, three of which were reviews, one did not focus on the brain, one was a developmental study, three did not administer a synthetic progestin, and five did not measure allopregnanolone as an outcome variable. Accordingly, nine relevant articles remained, which are listed in Table 3. The average number of SCYRCLE points for these studies was 2, the average number of animals per group was 10 and the average number of dependent variables assessed was 12.
Anti-androgenic progestins with the exception of drospirenone appear to increase allopregnanolone levels in the brain. Effects appear to be weaker in the hypothalamus than in other brain areas and enhanced when simultaneously administered with an estrogen. The androgenic progestin levonorgestrel decreases allopregnanolone levels in the brain. However, it has to be kept in mind that all studies on anti-androgenic progestins used ovariectomized animals, while studies on levonorgestrel used gonadally intact animals.

Progesterone and contraceptive progestin actions on neurogenesis
86 articles on progesterone or contraceptive progestin actions on neurogenesis were identified using the following search terms: (progesterone OR "hormonal contraceptive" OR "hormonal contraception" OR "oral contraceptive" OR "synthetic progestin" OR levonorgestrel OR Table 4 Progesterone-and contraceptive progestin-administration studies in which neurogenesis or BDNF was the outcome variable. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol benzoate, PROG = progestogen, P4 = progesterone, MPA = medroxyprogesterone acetate, LNG = levonorgestrel, DSG = desogestrel, NETA = norethisterone acetate, VNO = vomeronasal organ, HYPO = Hypothalamus, HIPPO = hippocampus, DG = dentate gyrus, CA = cornu ammunis, OB = olfactory bulb, ctx. = cortex, BrdU = Bromodeoxyuridine, n. = neuron, BDNF = brain derived neurotrophic factor. # ketamine treated rats. *colchizin-treated rats, + excitotoxity. Studies on contraceptive progestins are marked in italics. chlormadinone OR medroxyprogesterone OR cyproterone OR desogestrel OR etonogestrel OR gestodene OR dienogest OR nomegestrol OR drospirenone) AND (neurogenesis OR "neuronal proliferation" OR "neuronal survival" OR BDNF). After record screening, 15 articles remained that assessed neurogenesis via BrDU or Ki67 labelling of cells or BDNF levels in cognitively relevant brain areas. Screening of reference lists returned three additional relevant articles. Accordingly, results of 18 articles were evaluated, which are listed in Table 4. The average number of SCYRCLE points for these studies was 3, the average number of animals assessed was 9 and the average number of dependent variables assessed was 8. Regarding progesterone, there appears to be a lack of studies focusing on progesterone action on neurogenesis per se in healthy adult animals. In fact, there are only four studies focusing on the hippocampus, while other brain areas relevant to cognition or emotion, like the frontal cortex or amygdala haven't been systematically investigated with regard to neurogenesis. Accordingly, results regarding progesterone actions on neurogenesis remain inconclusive. The majority of studies reports no change in neuronal proliferation following progesterone administration either alone or in combination with estradiol. Given that the positive effects of estradiol on neurogenesis have been repeatedly established (for reviews see Wan et al., 2021;Mahmoud et al., 2016;Galea et al., 2006), it seems that after estrogen-priming, progesterone counteracts the positive effects of estradiol on neurogenesis. This idea is in accordance with studies demonstrating a reduction in neurogenesis during pregnancy (Ziegler-Waldkirch et al., 2018). In the absence of estradiol, two studies reported positive effects of progesterone on neuronal proliferation in the dentate gyrus (Liu et al., 2010;Bali et al., 2012). Likewise, the majority of studies report no effect of progesterone administration on BDNF mRNA or protein in brain areas known for their importance in cognition, though some increase in the hippocampus has been observed when progesterone was administered after estrogen-priming. Neuroprotective effects of high doses of progesterone on cell survival after traumatic brain injuries have mostly been studied in male animals (e.g. Jones et al., 2005;Djebaili et al., 2005;Barha et al., 2011; for a review see Stein, 2011) or non-sexed neuronal cell cultures (Jodhka et al., 2009;Liu et al, 2009) and thus, may not be applicable for females under physiological conditions. Two studies Ciriza et al., 2006) report positive effects of progesterone on neuronal survival under cytotoxic conditions.
Regarding contraceptive progestin actions, Chan et al. (2014) found no changes in the rates of cell proliferation in response to MPA alone, but reduced rates of cell proliferation when MPA was combined with estradiol benzoate (EB). In contrast, Liu et al. (2010) found increased rates of cell proliferation in response to MPA, as well as NET and LNG, particularly when combined with estradiol. Likewise, Garay et al. (2014) administered the novel progestin Nestorone in a model of experimental autoimmune encephalomyelitis and found increased rates of cell proliferation. For most third and fourth generation progestins, as well as the Table 5 Progesterone-and contraceptive progestin administration studies in which synaptogenesis was the outcome variable. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol benzoate, PROG = progestogen, P4 = progesterone, MPA = medroxyprogesterone actetate LNG = levonorgestrel, HYPO = Hypothalamus, AN = arcuate nucleus, VMH = ventromedial hypothalamus, POA = preoptic area, HIPPO = hippocampus, DG = dentate gyrus, CA = cornu ammunis, dors. = dorsal, mPFC = medial prefrontal cortex, # = number of, OXT = oxytocin, pos. = positive, dens. = density, bt. = buttons, c. = cells, clust = clusters. Studies on contraceptive progestins are marked in italics. anti-androgenic progestins CMA and CPA, information about effects on cell proliferation and cell survival are largely missing to date. Regarding BDNF, a reduction was reported after progestin administration by four studies Simone et al., 2015;Boi et al., 2022;Li et al., 2019). Frye et al. (2013) administered MPA without an estrogen to ovariectomized mice and observed BDNF reductions in the prefrontal cortex, but not in the hippocampus. Simone et al. (2015) and Boi et al. (2022) administered LNG to gonadally intact rats with and without ethinylestradiol and observed BDNF reductions in the hippocampus under both conditions.
In summary, the majority of studies report no effects of progesterone on neurogenesis, BDNF or neuronal survival, though the small number of studies that do find significant effects, report positive effects. For synthetic progestins results regarding neurogenesis are inconclusive, while mostly negative results have been reported regarding BDNF.

Progesterone and contraceptive progestin actions on synaptogenesis
116 articles on progesterone or contraceptive progestin actions on synaptogenesis were identified using the following search terms: (progesterone OR "hormonal contraceptive" OR "hormonal contraception" OR "oral contraceptive" OR "synthetic progestin" OR levonorgestrel OR chlormadinone OR medroxyprogesterone OR cyproterone OR desogestrel OR etonogestrel OR gestodene OR dienogest OR nomegestrol OR drospirenone) AND (synaptogenesis OR "synaptic spine" OR "dendritic spine" OR "mushroom spine" OR synapse) AND brain. After record screening, 21 articles remained that assessed synapse number or density via Golgi staining or synaptic protein markers in cognitively relevant brain areas. Screening of reference lists returned two additional relevant articles. Accordingly, results of 23 articles were evaluated, which are listed in Table 5. Results on synaptogenesis in hypothalamic nuclei are summarized in the Supplementary Material. The average number of SCYRCLE points for these studies was 2, the average number of animals per group was 6 and the average number of dependent variables assessed was also 6.
Overall progesterone appears to increase synaptogenesis in some hypothalamic nuclei (compare Supplementary Material), the CA1 region of the hippocampus, the dorsal raphe nucleus and the prefrontal cortex. Likewise, progesterone seems to upregulate genes relevant for synapse formation (Bethea & Reddy, 2010). However, non-significant effects Table 6 Progesterone-and contraceptive progestin administration studies in which myelination was the outcome variable. All studies used adult animals. OVX = ovariectomy, EST = estrogen, E2 = estradiol, PROG = progestogen, P4 = progesterone, NET = Nestorone, MS = multiple sclerosis, MOG = myelin oligodendrocyte glycoprotein, EAE = experimental autoimmune enzephalomyelitis, cupriz = cuprizone, spinal c. = spinal cord, CC = Corpus Callosum, MBP = myelin basic protein, PLP = proteolipid protein, demyelin. = demyelination, MAG = myelin associated glycoprotein, MAL = myelin and lymphocyte protein, ODC = oligodendrocyte. Studies on contraceptive progestins are marked in italics.  have also been reported for those brain areas. Insterestingly, there is only one report about a decrease in synaptogenesis following progesterone administration in the absence of estrogen in the pineal gland in vitro. Beneficial effects of progesterone occur mostly in combination with estrogen, though positive effects have also been observed in the absence of estrogen. Likewise, studies on endogenous progesterone fluctuations reveal either no effect on dendritic spines in various brain areas along the estrous cycle (Morisette et al., 1992) or an increase in synapse density and synaptic protein-expression during proestrus (Langub et al., 1994;Prevot et al., 2000;Spencer et al., 2008) and with increasing number of pregnancies in various brain areas (Flores-Vivaldo et al., 2019). Regarding contraceptive progestins, only MPA and levonorgestrel have been evaluated with respect to their effects on synaptogenesis. Results on other progestins are lacking. As for progesterone, positive effects of MPA have been observed with and without estradiol on various measures of synaptogenesis with focus on the hippocampus. In contrast, one study reports non-significant effects of LNG on synaptogenesis (Li et al., 2019).
In summary, progesterone and MPA appear to increase synaptogenesis in cognitively relevant brain areas, while other progestins have yet to be evaluated regarding their effects on synaptogenesis.

Progesterone and contraceptive-progestin actions on myelination
74 articles on progesterone or contraceptive progestin actions on myelination were identified using the following search terms: (progesterone OR "hormonal contraceptive" OR "hormonal contraception" OR "oral contraceptive" OR "synthetic progestin" OR levonorgestrel OR chlormadinone OR medroxyprogesterone OR cyproterone OR desogestrel OR etonogestrel OR gestodene OR dienogest OR nomegestrol OR drospirenone) AND (myelin OR myelination). After record screening, 9 articles remained that assessed myelination or demyelination in various models of neurodegeneration, which are listed in Table 6. Screening of reference lists returned no additional relevant articles. The average number of SCYR-CLE points for these studies was 2, the average number of animals per group was 7 and the average number of dependent variables assessed was also 7. They consistently show an increase in myelination or decrease in demyelination upon progesterone administration in the absence of estrogen.
Regarding contraceptive progestins, only Nestorone, a novel progestin, which is currently being developed and not yet approved for contraceptive use, has been studied with respect to it's effects on myelination. In accordance with resuls of endogenous progesterone on myelination, two studies reported positive effects of Nestorone on myelination (El-Etr et al., 2015;El-Etr et al., 2020). In summary, both progesterone and Nestorone appear to increase myelination.

Progesterone and contraceptive progestin actions on neurotransmitter systems
Articles on progesterone actions on neurotransmitter systems were identified using the following search terms: (progesterone OR "hormonal contraceptive" OR "hormonal contraception" OR "oral contraceptive" OR "synthetic progestin" OR levonorgestrel OR chlormadinone OR medroxyprogesterone OR cyproterone OR desogestrel OR etonogestrel OR gestodene OR dienogest OR nomegestrol OR drospirenone) AND (glutamate OR GABA OR Serotonine OR Dopamine OR acetylcholine OR norepinephrine OR noradrenalin) AND brain. The search returned 1155 articles. After record screening, 160 articles focusing on neurotransmitter signalling in Table 8 Studies on the effects of progesterone-and contraceptive progestin on glutamate levels, currents or receptor binding. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol benzoate, PROG = progestin, P4 = progesterone, MPA = medroxyprogesterone acetate, NETA = norethisterone acetate, HIPPO = hippocamus, ctx. = cortex, dors. = dorsal, GS = glutamate synthetase, GLUT = glutamate, GLUT-T = glutamate transporter, NMDA-R = NMDA receptor, AMPA-R = AMPA receptor, mGLU-R = metabotropic glutamate receptor, EPSP = excitatory postsynaptic potential, LTP = long term potentiation, LTD = long term depression, bdg. = binding. Studies on contraceptive progestins are marked in italics. cognitively relevant brain areas remained, which in turn listed 8 additional relevant articles in their reference lists. These additional articles focus on older progestins, which are not currently in use in hormonal contraceptives. Results are listed in Tables 7-15, for acetylcholine (ACH), glutamate (GLUT), GABA, serotonine (5-HT) and dopamine (DOPA) and norepinephrine (NE) respectively. Neurotransmitter levels have been extensively studied in the medial preoptic area (mPOA) and other hypothalamic nuclei in relation to GnRH release and the ovulatory LH surge. Glutamate and monoamines all increase in the mPOA after estrogen-primed progesterone administration, which is in accordance with hypothalamic neurotransmitter changes along the estrous cycle (Loescher et al., 1992). Accordingly, results on hypothalamic nuclei can be found in the Supplementary Material 2, while Tables 7-15 only list results in cognitively relevant brain areas, which are the focus of the main article.

Progesterone and contraceptive progestin actions on the ACh system
Acetylcholine (ACh) has hardly been assessed with regards to progesterone actions (8 studies) and only one study assessed AcH upon contraceptive progestin administration (compare Table 7). The average number of SCYRCLE points for these studies is 1, the average number of animals per group 6 and the average number of outcome variables assessed is 9. AcH synthesis (ChAT activity) and degradation (AChE activity) both appear to increase upon combined estrogen/progesterone administration. Only one study assessed ACh levels upon progesterone administration (Muth et al., 1980) and observed reduced ACh levels, while contraceptive progestins (MPA, NETA) do not appear to alter ACh levels (Daabees et al., 1981).

Progesterone and contraceptive progestin actions on the glutamatergic system
Glutamatergic signalling in response to progesterone was assessed by only 12 studies using different outcome measures and contraceptive progestin actions on glutamate levels by only one study (compare Table 8). The average number of SCYRCLE points for these studies is 1, the average number of animals per group 9 and the average number of outcome variables assessed is 10. Only one study assessed progesterone actions on glutamate levels in cognitively relevant brain areas, where they appear to be unchanged (Luine et al., 2017). In contrast, glutamate transport appears to be rather increased as indicated by an augmented Table 9 Studies on the effects of progesterone and contraceptive progestins on GABA synthesis, levels, transport and GABA-induced transmembrane currents. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol benzoate, PROG = progestin, P4 = progesterone, LYN = lynestrol, MPA = medroxyprogesterone acetate, NETA = norethisterone acetate, NG = nomegestrol, HIPPO = hippocamus, AMY = amygdala, OB = olfactory bulb, ctx. = cortex, dors. = dorsal, b. = brain, cult = culture, curr = current, transp = transporter, GAD = glutamic acid decarboxylase. Studies on contraceptive progestins are marked in italics.  transcription of glutamate transporter genes . Similarly, glutamate-evoked transmembrane currents appear to be reduced, most likely due to effects on NMDA receptors, given that AMPA receptors and metabotropic glutamate receptors appear to be unaffected (compare Table 8). Likewise, only one study assessed contraceptive progestin actions on glutamate levels in cognitively relevant brain areas (Daabees et al., 1981). While MPA did not change glutamate levels, NETA reduced glutamate levels across the whole brain. In summary, there is at least some evidence, that progesterone and some contraceptive progestins attenuates glutamate signalling.

Progesterone and contraceptive progestin actions on the GABA-ergic system
55 studies assessed progesterone or contraceptive progestin actions on the GABAergic system, including GABA synthesis via the enzyme glutamic acid decarboxylase (GAD), GABA levels, GABA currents and GABA transport (summarized in Table 9), agonistic and antagonistic GABA-A receptor binding (summarized in Table 10) and GABA-A receptor subunit expression (summarized in Table 11). The average number of SCYRCLE points for these studies was 1, the average number of animals per group was 6 and the average number of outcome variables assessed was 10.
While progesterone results regarding GABA levels per se are rare and inconsistent, activity of the GABA-producing enzyme glutamatedecarboxylase was reported to be decreased by seven studies and unchanged by three. Nevertheless, GABA-A receptor currents seem to consistently increase. Similarly, the binding of GABA-A receptor agonists has been shown to rise under progesterone, while the binding of GABA-A receptor antagonists decreases. The results of progesterone administration studies are corroborated by studies focusing on endogenous progesterone fluctuations along the estrous cycle (Bitran et al., 1991;Wilson, 1992;Finn and Gee, 1993;Martin and Williams, 1995;McCauley and Gee, 1995;Brann et al., 2002;Lovick, 2008), during pregnancy (Concas et al., 1999) or using a progesterone withdrawal paradigm (Costa et al., 1995;Smith et al., 1998;Hsu and Smith, 2003;Smith and Gong, 2005). These effects are likely attributable to the allosteric modulation of GABA-A receptors by the progesterone metabolite allopregnanolone (e.g. Hiemke et al., 1983). Several studies demonstrate that the above-mentioned effects are abolished by blocking the enzymes that convert progesterone into allopregnanolone (e.g. Frye et al., 2013), whereas the effect remains visible when blocking progesterone receptors (Maguire and Mody, 2007;Reddy et al., 2017). Furthermore, direct administration of allopregnanolone yielded similar results (Jussofie 1993a,b;Weiland and Orchinik, 1995;Uchida et al., 2002).
Several studies have also focused on the specific sub-unit composition of GABA-A receptors following progesterone administration (compare Table 11), as well as along the estrous cycle (Griffiths and Lovick, 2005;Maguire et al., 2005;Lovick, 2008;Wu et al., 2013) Table 11 Effects of progesterone and contraceptive progestins on GABA-A receptor subunit composition. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol-benzoate, EE = ethinylestradiol, PROG = progestin, P4 = progesterone, LNG = levonorgestrel, MPA = medroxyprogesterone acetate, HIPPO = hippocamus, OB = olfactory bulb, Sub = substantia. Studies on contraceptive progestins are marked in italics.  Table 12 Progesterone-and contraceptive progestin administration studies in which serotonine (5-HT) synthesis or level was the outcome variable. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol benzoate, PROG = progestogen, P4 = progesterone, NETA = norethisterone acetate, MPA = medroxyprogesterone acetate, CMA = chlormadinone acetate, HIPPO = hippocamus, VTA = ventral tegmental area, NAc = Nucleaus Accumbens, mPFC = medial prefrontal cortex. Studies on contraceptive progestins are marked in italics. and Mody, 2013). Results suggest a downregulation of α1and α4-subunit expression, while progesterone withdrawal seems to consistently upregulate α4-subunit expression (Gulinello et al., 2001;2003;Smith et al., 1998;Smith and Gong, 2005;Griffiths and Lovick, 2005;Gangisetty and Reddy, 2009;Reddy et al., 2012). These alterations in subunit compositon result in reduced responsivity of GABA-A receptors to benzodiazepines. Regarding contraceptive progestins and GABA, the majority of findings point towards an increase in GABA levels across the brain irrespective of the progestin studied. However, this is only true for older generation progestins which have been investigated here. Some studies find that the increase in GABA levels is intensified by simultaneous administration of a progestin-estrogen combination. Yet, results regarding GAD activity are mixed, raising the question whether the increase in GABA-levels is achieved by some mechanism other than upregulation of GAD levels or GAD activity. Furthermore, several studies find an upregulation of GABA-A receptors or some of their subunits, suggesting that the increase in GABA-levels is matched by an increase in the number of GABA-receptors. GABA-A receptor modulation was not observed for CMA. Other currently available contraceptive progestins have not been studied with respect to GABA-A receptor modulation.

Progesterone and contraceptive progestin actions on the serotonergic system
53 studies that assessed progesterone or contraceptive progestin actions on serotonergic neurotransmission in cognitively relevant brain areas were identified. These included studies on serotonin synthesis and levels (Table 12), as well as studies on serotonin transport, degradation and receptor density (Table 13). The average number of SCYRCLE points for these studies was 1. Studies used on average 7 animals per grup and assessed 9 different outcomes.
Upon progesterone administration, several studies demonstrate an increase of serotonin synthesis and serotonin levels, along with a decrease in serotonin clearance, transport and degradation in several subcortical brain areas, including the striatum, VTA and raphe nuclei. In line with these observations, dorsal raphe neuron firing rates appear to increase during the natural endogenous progesterone elevation during pregnancy (Klink et al., 2002). Results regarding cortical areas are less conclusive, but mostly demonstrate serotonin levels to remain unchanged after progesterone administration. At the receptor level progesterone appears to reduce the number of inhibitory 5-HT1A autoreceptors, and increase the number of 5-HT2 receptors.
Likewise, serotonin levels appear to increase in response to contraceptive progestin (NETA, MPA) treatment, which is in accordance with results on progesterone administration suggesting a progesterone receptor-dependent mechanism. However, negative results have also been reported for MPA and this progestin appears to increase serotonin degradation via monoamino oxidase and decrease serotonin receptors. Regarding the mixed findings for MPA it is noteworthy, that studies on contraceptive progestins rarely differentiate between brain areas, while progesterone administration studies distinguish between a variety of cortical and subcortical areas.

Progesterone and contraceptive progestin actions on the dopaminergic system
55 of the neurotransmitter studies assessed an outcome variable related to dopaminergic signalling (Table 14). On average, the studies reached one SCYRCLE point, included 7 animals per group and assessed 7 different outcome variables. Dopamine production and dopamine levels appear to be increased in the striatum upon progesterone administration either with or without estradiol, but remain unchanged or reduced in cortical areas. Results regarding dopamine receptor expression in response to progesterone administration appear to be inconclusive.
Effects on dopamine have mostly been studied in older generation progestin, of which only MPA and LNG are currently still in use. Accordingly, results may not be applicable to newer generation progestins in modern contraceptives. Irrespective of the progestin, results show a decrease in dopamine levels, despite an increase in dopamine synthesis. These findings are probably the result of a concurrent increase in dopamine turnover and dopamine degradation, e.g., by increased MAO activity. Regarding cortical areas, these results are in agreement with studies on progesterone administration, though the fact that progestins appear to decrease dopamine levels also in the striatum ) are inconsistent with progesterone administration studies.

Progesterone and contraceptive progestin actions on the noradrenergic system
22 of the neurotransmitter studies assessed an outcome variable related to neuradrenergic signalling (Table 15). The studies reached on average one SCYRCLE point, included 9 animals per group and assessed 9 different outcome variables. The majority of results suggests that norepinephrine signalling remains unchanged upon progesterone or contraceptive progestin administration. In the forebrain and midbrain both increased and decreased norepinephrine levels upon progesterone administration were observed. One study observed reduced norepinephrine levels upon levonorgestrel administration in the midbrain (Dey et al., 1991).

Discussion
In this review we systematically summarize the effects of progesterone and synthetic progestins on neurogenesis, synaptogenesis, myelination and six neurotransmitter systems. Table 16 summarizes the number of studies identified for each area and the direction of the most consistent results across those studies. Given the paucity of studies, Table 16 does not distinguish between different brain areas investigated, but lists a result if it was found in at least one brain area. The contraceptive progestins gestoden, norgestimate, drospirenone, cyproterone acetate and nomegestrol acetate are not listed in this table, since no results on their actions in the brain are available. The most consistent results include an increase in synaptogenesis and myelination, GABA signalling and dopamine release upon progesterone administration. While results on synaptogenesis and myelination appear to be mirrored by some contraceptive progestins, GABA signalling may be reduced with contraceptive progestins and dopamine release has not been studied under contraceptive progestin administration. The discussion of these findings will be organized as follows.
From human neuroimaging studies we identified the prefrontal cortex, amygdala, salience network and hippocampus as regions of interest for preclinical studies (compare section 1.3.). Following a general note on the overall confidence level of findings, we will discuss progesterone and contraceptive progestin actions on cellular and molecular parameters in these brain areas and discuss potential mechanisms of synthetic progestin effects based on comparison to progesterone results, as well as potential behavioural implications of these findings. We will also make specific suggestions for future experiments and point out the most promising future research directions.

Risk of bias
Across all domains (neurogenesis, synaptogenesis, myelination, neurotransmitter levels), the overall confidence level of studies appears low, because the majority of items on the SCYRCLE bias assessment tool were answered unclear, given that the articles assessed did not report the information relevant for bias assessment. Regarding blinding and Table 13 Progesterone-and contraceptive progestin administration studies in which serotonine (5-HT) transport, degradation or receptors were the outcome variables. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol benzoate, PROG = progestogen, P4 = progesterone, NETA = norethisterone acetate, MPA = medroxyprogesterone acetate, ENG = etonorgestrel, HIPPO = hippocamus, VTA = ventral tegmental area, NAc = Nucleaus Accumbens, mPFC = medial prefrontal cortex. Studies on contraceptive progestins are marked in italics. randomization procedures it is thus unclear whether the methodology to reduce bias was not followed or simply not reported. A major concern however is, that the number of animals per group appears low (or was not even reported!), especially in the light of the number of outcome variables assessed. On average, studies assessed as many outcome variables (different brain areas, different neurotransmitters, etc.) as animals were included per group and while the majority of studies accounted for multiple comparisons regarding the number of groups, almost no study performed a multiple comparisons correction across outcomes. Accordingly, the risk of false positive results appears to be high. Relatedly, there may be a risk for publication bias. While many studies report non-significant findings, these are usually included among significant findings in other outcome variables and no study reports only null findings. Also, the number of studies per agent and outcome variable is very low with an average of 4 studies per outcome for progesterone administration and almost no studies for the majority of contraceptive progestins currently in use (compare Table 16). On average, studies on contraceptive progestins are very old and were conducted on older progestins used in the very first contraceptives. While these studies were included to serve as hypothesis-generating comparison points for future studies on newer progestins, the results to not necessarily transfer to other progestins, given that progestins differ in their binding affinities to various receptors (see section 1.2.). Among the progestins currently in use, the highest number of findings was obtained for medroxaprogesterone acetate (MPA) and levonorgestrel (LNG), while findings on third and fourth generation progestins are almost completely lacking (see Table 16).
An important question to consider in future studies is whether animals are ovariectomized prior to contraceptive progestin administration. On the one hand, the majority of progesterone administration studies use ovariectomy to abolish endogenous hormone production and avoid interactive effects between endogenous and exogenous hormones. Accordingly, ovariectomy increases the comparability of results between contraceptive progestins and progesterone. However, in humans, hormonal contraceptives are administered to women with intact ovaries and their effects on endogenous hormone production may in fact account for some of the behavioural effects observed in hormonal contraceptive users. Accordingly, the administration of synthetic progestins to gonadally intact animals increases the relevance of findings for humans. Interestingly, among the animal studies conducted so far, MPA studies used mostly ovariectomized animals, whereas LNG studies used mostly gonadally intact animals. While the latter allows for a better comparison of findings to human neuroimaging studies, the former limits the comparability of findings across various progestins. If gonadally intact animals are studied, the cycle phase of animals at treatment start and in the vehicle treated groups appears relevant. Unfortunately, not all studies on gonadally intact animals did control for the estrous cycle phase and many human neuroimaging studies did also not control for the menstrual cycle phase of women in the comparison groups. It is thus unclear, whether the brain is particularly sensitive to contraceptive progestin actions in certain cycle phases or whether the changes elicited by contraceptive progestins are comparable in effect sizes to the changes observed in response to endogenous hormonal fluctuations across cycle phases.

Findings in the frontal cortex and their relevance for adverse mood symptoms
Human neuroimaging studies repeatedly report changes in the volume and activation of the prefrontal cortex, though both increases and decreases in these parameters have been reported. We suggest two hypotheses explaining these opposing findings. On the one hand, decreases are more commonly observed with androgenic progestins, specifically levonorgestrel, while increases are more commonly observed with antiandrogenic progestins. On the other hand, decreases have been observed in short-term users with previous negative affect on the pill, while increases have been observed in long-term users who likely did not experience any severe mood disturbances during contraceptive treatment. Unfortunately, at the cellular level, no results regarding progesterone and contraceptive progestin actions on neurogenesis are available that might link to the gray matter changes observed in humans. However, both progesterone and MPA increase synaptogenesis in the frontal cortex (compare Table 5), while synaptogenesis in response to LNG appears to be unchanged across the whole brain (Sassoè-Pognetto et al., 2007). Accordingly, differences in neuroimaging parameters in the frontal cortex between androgenic and antiandrogenic contraceptives may at least in part result from their effects on synaptogenesis.
A likely candidate translating these findings to the molecular level is the progesterone metabolite allopregnanolone. While the androgenic progestin levonorgestrel decreases allopregnanolone in the whole brain including the frontal cortex, anti-androgenic progestins do not alter allopregnanolone content in the frontal cortex, while MPA even increases allopregnanolone in the frontal cortex (compare Table 3). Via these effects on allopregnanolone, different progestins may differentially affect GABAergic signalling and thus, the excitatory/inhibitory balance in the frontal cortex. However, while progesterone consistently increases GABA signalling across the brain (compare Table 16), including the frontal cortex, no results on progestins currently used in hormonal contraceptives are available. While older generation progestins mirror some effects of progesterone on the GABA-ergic system (compare Table 8), these results may not necessarily translate to androgenic progestins like levonorgestrel given their downregulation of allopregnanolone levels. MPA has no effect on either glutamate or GABA content in the frontal cortex, and also CMA did not change GABA-A receptor currents in the frontal cortex (compare Table 8).
In summary, a promising series of future preclinical experiments could focus on allopregnanolone content, GABA-receptor binding, synaptogenesis and neurogenesis in the frontal cortex in response to levonorgestrel on the one hand and commonly prescribed anti-androgenic progestins like chlormadinone acetate or nomegestrol acetate on the other hand. If changes in response to contraceptive progestins are observed and particularly if those changes differ between androgenic and anti-androgenic progestins, a link to preclinical behavioral markers of depression appears to be of particular relevance.
Allopregnanolone has been shown to alleviate mood lability and depressive symptoms in various hormonal transition periods (Pinna, 2020) suggesting a link between the androgenicity of the contraceptive progestin and the likelihood to develop depressive symptoms. However, given that the selective progesterone receptor modulator alleviates mood symptoms during the premenstrual period (Comasco et al., 2021), adverse mood effects during contraceptive treatment may not only be related to indirect effects on GABA-ergic signalling via alterations of allopregnanolone levels, but may also include progesterone receptor dependent mechanisms. This may for example concern alterations in the serotonergic and dopaminergic system, given their prominent link to mental health (e.g., Damsa et al., 2004). Unfortunately, results on serotonine and dopamine levels in the frontal cortex are inconsistent with respect to progesterone and MPA and no other contraceptive progestins have been studied with respect to serotonergic signalling in the frontal cortex.

Findings in the amygdala and their relevance for anxiety
Human neuroimaging studies tentatively suggest reduced amygdala volume and reacitivity during contraceptive treatment, thought results cannot be linked to a specific progestin yet. In line with these observations, BDNF levels in the amygdala were reduced with desogestrel and synaptogenesis in the amygdala was reduced with levonorgestrel (Li et al., 2019). These reductions may account for the reduced amygdala responsiveness during contraceptive treatment in human neuroimaging Table 14 Studies on the effects of progesterone and contraceptive progestins on dopamine (DA) synthesis, release, turnover and receptor binding. OVX = ovariectomy, EST = estrogen, E2 = estradiol, EB = estradiol benzoate, EE = ethinylestranol, MEST = mestranol, PROG = progestogen, P4 = progesterone, LYN = lnestrol, LNG = levonorgestrel, MPA = medroxyprogesterone acetate, NETA = noresthisterone acetate, HIPPO = hippocamus, ME = median emininence, VTA = ventral tegmental area, NAc = Nucleaus Accumbens, mPFC = medial prefrontal cortex, AMY = amygdala, thal. = thalamus; osc. = oscilation, transp. = transporter. Studies on contraceptive progestins are marked in italics. studies, however a more systematic approach to delineate cellular and molecular effects of different progestins is necessary. For example, no study has assessed neurotransmitter content in the amygdala in response to contraceptive progestins, although progesterone administration studies did observe a number of changes in neurotransmitter signalling. While monoamine content of the amygdala appears to be unchanged after progesterone administration (Cone et al., 1981;Crowley, 1978), the EPSP slope in the amygdala was reduced in response to progesterone (Yang et al., 2017) and GABA-A receptor binding increased (compare Table 10), suggesting a shift from excitatory to inhibitory processes under high progesterone conditions. It is likely that the alterations in GABA-A receptor binding in the amygdala are mediated via allopregnanolone. However, allopregnanolone content in the amygdala in response to administration of different synthetic progestins has not been assessed so far. This question appears to be of particular relevance given the well-known anxiolytic properties of allopregnanolone on the one hand (Pinna, 2020), and the well-established role of the amygdala in the processing of fearful stimuli on the other hand (Choleris et al., 2008;Toufexis, 2007). Results regarding LNG effects on anxiety in animal models suggest a dosage and treatment duration dependent modulation of anxiety-like behavior with lower doses and shorter administration having an anxiolytic effect (Porcu et al., 2012;Simone et al., 2015;Picazo et al., 1998), while higher doses or longer administration increase anxiety-like behavior (Follesa et al., 2002;Graham & Milad, 2013;Simone et al., 2015;Parrish et al., 2019). Increased anxiety was also observed with the LNG derivate ENG (Maijer & Semple, 2016). The role of interactive effects with ethinylestradiol has also been discussed (Graham & Milad, 2013;Simone et al., 2015). For MPA positive (Toufexis et al., 2004), negative (Frye et al., 2010a) and null effects have been reported . Again, interactions with estradiol were observed (Pazol et al., 2004). These results support individualized approaches to hormonal contraception evaluating the precise dosage and estrogen interactions depending on the individual neuro-endocrine profiles of women.

Findings in the hippocampus and their relevance for cognition
As for the frontal cortex, human neuroimaging studies report mixed effects regarding hippocampal volumes and hormonal contraceptive use, though most studies did not control for progestin type (Pletzer et al., 2010;Pletzer et al., 2015, deBondt et al., 2013Lisofsky et al., 2016;. Non-surprisingly, the hippocampus has been studied most extensively with regards to progesterone actions on neurogenesis. However, the number of studies focusing on female samples is surprisingly low and results regarding progesterone effects on neurogenesis are mixed and suggest substantial differences between estrogenprimed progesterone actions and progesterone only actions (compare Table 4). Likewise, mixed effects on hippocampal neurogenesis were obtained for MPA, while one study suggests increased hippocampal neurogenesis after LNG administration (Li et al., 2019). In contrast, BDNF levels in the hippocampus are reduced with LNG and synaptogenesis remains unchanged, while unchanged BDNF levels and increased synaptogenesis were observed for MPA (compare Tables 4-5). These results suggest important differences between progestins, which may account for inconsistencies in human neuroimaging studies. As for other brain areas, differential modulation of allopregnanolone content by different progestins has also been observed for the hippocampus. While LNG reduces hippocampal allopregnanolone, the anti-androgenic progestins NOMAC and CMA increase allopregnanolone in the hippocampus (compare Table 3), which might in turn affect the excitatory/ inhibitory balance in the hippocampus. However, effects of contraceptive progestins on GABA and glutamate signalling in the hippocampus have not been assessed, with the exception of GABA-A receptor subunit expression. Again, differences were observed between MPA and LNG.
While MPA increased α4 subunit expression in response to MPA similar to progesterone (Pazol et al., 2009), α4 subunit expression did not change in response to LNG (Porcu et al., 2012). No results on other progestins are available. Given the central role of the hippocampus in memory formation and retrieval, as well as spatial information processing (see Opitz, 2014 for a review), the differential modulation of cellular and molecular processes in the hippocampus by androgenic and anti-androgenic progestins appears to be of high relevance. Human studies mostly point towards an improvement of some aspects of mostly emotional memory in COC users compared to non-users (Petersen et al., 2015a,b;Merz, 2017;Spalek et al., 2019;Person & Oinonen, 2020;Peragine et al., 2020;Gamsakhurdashvili et al., 2021;Gravelsins et al., 2021). However, no study controlled for the type of progestin used and effect sizes were small in all studies. Some indication for a differential modulation of hippocampusdependent cognition emerges for spatial abilites. Improved spatial abilities were observed in mixed groups of various hormonal contraceptive formulations by Beltz et al., (2015), Bernal et al. (2020) and Peragine et al. (2020), with trend effects also observed by Patel et al. (2022). Reduced spatial performance for users of anti-androgenic contraceptives was observed by Griksiene et al. (2018). It has been discussed, that androgenic COCs are more likely to elicit a spatial improvement compared to anti-androgenic COCs, and that a higher ethinyl-estradiol dose may be beneficial for spatial abilities (Beltz et al., 2015). Given that the androgenic progestin LNG appears to increase neurogenesis in the hippocampus and seems more likely to reduce inhibitory signalling given the reduced allopregnanolone content and lack of alterations in GABA-A receptor subunit-expression, this hypothesis appears to be worthwile to explore in future clinical and preclinical studies.

Table 16
Summary of progesterone and contraceptive progestin actions on cellular and molecular brain parameters. Arrows indicate the direction of results, numbers indicate the number of consistent studies relative to the number of total studies available. Only domains for which more than one progesterone study is available are listed. The most consistent results are highlighted in bold and dark colors, less consistent results are hightlighted in lighter colors. Inconsistent results or singular findings are not highlighted. Green: increase, red: decrease, yellow: no change. Areas for which no findings are available so far are displayed in black. P4 = progesterone, E = estrogen, MPA = medroxyprogesterone acetate, LNG = levonorgestrel, DSG = desogestrel (the column also includes results on etonorgestrel), CMA = chlormadinone acetate, NET = nestorone, BDNF = brain derived neurotrophic factor, ACh = acetylcholine, GLUT = glutamate, 5-HT = serotonine, DA = dopamine, NE = norepinephrine.
2014). Two studies demonstrate no alterations of cognition after MPA administration (Frye et al., 2010b;Frye et al., 2013), while only one study argues for an improvement of cognition after MPA administration (Chisholm & Juraska, 2012b). Apart from that, only two animal studies assessing cognition after progestin administration were identified, both administering a combination of EE and LNG (Simone et al., 2015;Prakapenka et al., 2018). Both studies suggest that effects on performance depended on the interactive effects of EE and LNG. In the study by Simone et al. (2015) LNG in combination with lower doses of EE impaired novel object recognition, while LNG in combination with higher doses of EE improved novel object recognition. In the study by Prakapenka et al. (2018), EE and LNG alone enhanced spatial working memory, while their combination impaired spatial working memory. In summary, converging evidence of animal and human studies suggests that the cognitive effects of COC are dependent on the EE dose anddepending on the cognitive domainmay differ between androgenic and anti-androgenic progestins. These results are in accordance with estrogen-primed progestin actions on hippocampal neurogenesis, synaptogenesis, BDNF and allopregnanolone content. Based on progesterone administration studies a promising avenue with regards to hippocampal volume changes appears to be neuronal survival, given that neuroprotective effects of progesterone have been extensively described (Guennoun, 2020). However, there is a paucity of studies in female samples and effects of contraceptive progestins on neuronal survival in the hippocampus have not been assessed (compare Table 4).

Findings in other brain areas
Apart from brain areas responsive to contraceptive treatment in human neuroimaging studies, preclinical studies frequently focused on the striatum and midbrain with regards to progesterone and contraceptive progestine actions. Particularly, monoamine content appears to be increased in the striatum after progesterone administration, while levonorgestrel and medroxyprogesterone acetate were related to reduced dopamine levels in the striatum. Accordingly, this may be one brain area, where effects of endogenous progesterone and synthetic progestins diverge irrespective of progestin type. Only few neuroimaging studies focused on the basal ganglia as a target for contraceptive progestin actions, demonstrating for example and increased volume with longer contraceptive treatment duration (Pletzer et al., 2019a,b). However, fronto-striatal circuits emerge as particularly responsive to progesterone in neuroimaging studies of the menstrual cycle (Pletzer et al., 2018;Pletzer et al., 2019a,b;Hidalgo-Lopez & Pletzer, 2019;Hidalgo-Lopez et al., 2020;Hidalgo-Lopez & Pletzer, 2021), particularly during executive functions. Based on these findings, future clinical and preclinical studies on contraceptive progestin actions should include the striatum among their areas of interest. For example, like the prefrontal cortex and amygdala, the striatum has been confirmed as one of the sites for extra-hippocampal adult neurogenesis (Ernst et al., 2014). However, even though striatal volumes appear to increase in response to progesterone (Pletzer et al., 2018), it has not been assessed whether progesterone or synthetic progestin administration increases neurogenesis in the striatum. Likewise, studies on striatal synapse formation in response to progesterone or synthetic progestins are lacking. Given that the majority of striatal neurons is GABAergic and progesterone and synthetic progestin dependent modulation of GABA-signalling has been discussed with regards to the prefrontal cortex (compare section 4.1), it appears worthwhile to extend future approaches to fronto-striatal circuits.

Conclusions
In conclusion, human neuroimaging studies have identified the prefrontal cortex, amygdala and hippocampus as areas of interest for animal studies on neurogenesis, synaptogenesis and neurotransmitter signalling. However, there is a massive lack of animal studies assessing the effects of contraceptive progestins on these cellular and molecular brain parameters, especially with regards to the progestins currently in use (compare Table 16). Regarding neurogenesis, progesterone effects are also understudied and poorly understood. Accordingly, no hypotheses regarding contraceptive progestin administration can be drawn from results on progesterone administration. Regarding synaptogenesis, serotonergic and GABAergic signalling parallels between progesterone and older contraceptive progestins arise, suggesting actions via nuclear progesterone receptors. However, regarding allopregnanolone and BDNF levels divergent results were observed depending on the androgenicity of progestins, suggesting that indirect actions of contraceptive progestins may also arise from their metabolization to and actions on neuroactive steroids. From this overview a clearcut series of experiments arises to address the current challenge for individualized contraceptive approaches is to identify predictors of tolerability for each progestin in combination with various estrogen doses.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.