Menstrual cycle phase and endocannabinoid modulation: what the human evidence supports

About this article

This piece has three purposes, and it is worth stating them plainly rather than leaving them implied.

The first is to educate. What follows is a summary of what human research actually establishes about hormonal modulation of the endocannabinoid system, written for readers who want the study designs and the numbers rather than the headline.

The second is to demonstrate what an evidence-graded corpus can do when it is pointed at a clinical question. Every claim below was assembled from the Cannabis Knowledge Foundation's indexed corpus of more than 159,000 cannabis papers, and graded on the way through.

The third is the one that matters most, and it is the reason the article devotes as much space to absent evidence as to present evidence. Absence of evidence is only informative if somebody has genuinely looked. When a sentence below says a study has not been done, that is not a rhetorical hedge or an author's caution — it is the result of searching a corpus of that size and finding nothing there. Establishing precisely what is not known has historically been one of the harder things for a clinician to do in this field, because the literature is large, uneven, and full of preclinical work cited as though it were clinical. Knowing what we do not know is, in a field at this stage of maturity, often more useful than another restatement of what we do.

Plasma anandamide varies across the menstrual cycle in a pattern that has been replicated in independent human cohorts: a peri-ovulatory peak, a luteal trough, and a two-enzyme hormonal mechanism that accounts for both. The broader claim built on top of that finding — that the endocannabinoid system as a whole rises and falls with the cycle, and that cycle phase should therefore shape cannabinoid dosing — does not rest on evidence of the same quality. For clinicians, the distinction between those two statements is the whole clinical question.

Background: two hormones, two enzymes

Endocannabinoid tone is not fixed. It is set by the balance between synthesis and degradation, and both sides of that balance are hormone-sensitive.

Oestradiol stimulates anandamide (AEA) release and upregulates NAPE-PLD, the principal AEA-synthesising enzyme. Progesterone works in the opposite direction, upregulating fatty acid amide hydrolase (FAAH), the enzyme responsible for AEA breakdown — a relationship first characterised by Maccarrone and colleagues (2001). One hormone opens the tap; the other widens the drain. Because oestradiol peaks around ovulation and progesterone dominates the luteal phase, the predicted result is a mid-cycle rise in AEA followed by a fall.

That is roughly what is observed. It is also worth noting where the regulation is local rather than systemic. Scotchie and colleagues (2015) showed that FAAH and monoacylglycerol lipase (MAGL) protein peak in the secretory-phase endometrium of normally cycling women, with both enzymes significantly reduced in the receptive late-secretory endometrium. Circulating ligand concentrations and uterine enzyme activity are related but distinct measurements, and conflating them is a recurring source of overstatement.

The replicated finding

El-Talatini and colleagues (2010) established the plasma AEA pattern using combined cross-sectional and longitudinal cohort designs. Cui and colleagues (2017) replicated it in both arms:

AEA correlated positively with LH, FSH and oestradiol — but not with progesterone. That negative correlation is not a null result to be explained away; it is mechanistically consistent. Progesterone acts on clearance rather than release, so it would not be expected to track circulating ligand concentrations in a simple linear fashion.

The same research group extended the observation past the reproductive years. Lam and colleagues (2008) reported postmenopausal plasma AEA significantly lower than premenopausal follicular-phase values, at concentrations comparable to the luteal trough — consistent with the permanent withdrawal of the dominant synthesis driver. This is Level C evidence: a single-group, cross-sectional measurement that has not been independently replicated longitudinally across the perimenopausal transition.

Where the cycle claim overreaches

Three gaps matter, and none of them is obscure.

The first is 2-arachidonoylglycerol (2-AG). No replicated healthy-human study has quantified plasma 2-AG across follicular, ovulatory and luteal phases with anything approaching the detail available for AEA. The best indirect signal comes from endometrial gene expression showing MGLL downregulation around the proliferative-to-secretory transition (Barcena de Arellano et al., 2022) — informative, but tissue gene expression is not a circulating ligand measurement. Any statement that begins "the endocannabinoid system rises and falls across the cycle" is, on current evidence, a statement about AEA wearing a larger coat.

The second is receptor availability. No human imaging study compares CB1 availability across cycle phases, or before and after menopause. Claims that CB1 density shifts with the cycle in women are extrapolated from rodent work, including ovariectomy models.

The third is the clinical question itself. No trial in any indication has tested whether menstrual cycle phase modifies therapeutic response to cannabis-based medicinal products.

Does cycle phase change cannabinoid response?

The single controlled human test is instructive, and its result is not the one the mechanism predicts. Pabon and de Wit compared oral THC at 7.5 mg and 15 mg in women tested in the early versus late follicular phase. Across most outcome measures, responses were similar; the only detectable differences were that ratings of "wanting more" and anxiety emerged slightly earlier in the early follicular phase. The authors concluded that the oestradiol difference between those two phases does not strongly influence acute THC response.

Two caveats cut in opposite directions. That comparison brackets a relatively modest oestradiol difference and does not test the luteal phase, where the AEA trough actually sits — so the question is not closed. But the preclinical-derived assumption that higher oestrogen reliably means greater THC sensitivity is not confirmed either, and it should stop being presented as though it were.

Use behaviour may be more cycle-sensitive than pharmacodynamics. A 32-day daily-diary study in normally cycling women found premenstrual cannabis quantity exceeded both follicular and ovulatory use, while earlier work by Griffin and colleagues found no systematic phase relationship in women without severe premenstrual dysphoria. A premenstrual increase in use is plausible in subgroups where pain, stress or coping-motivated use are prominent. It is not a general rule, and it is worth distinguishing from evidence of benefit.

Where cycle biology does appear to matter

The most mechanistically coherent case is primary dysmenorrhoea. The luteal AEA trough coincides with peak endometrial prostaglandin production and maximal pain. CB1 and CB2 are expressed in human myometrium, and both AEA and THC produce CB1-mediated relaxation of oxytocin-stimulated myometrial strips (Dennedy et al., 2004) — though the only human contractility data come from pregnant tissue obtained at caesarean section, which makes extension to the dysmenorrhoeic uterus inferential. In rodent uterus, CB1 agonism selectively inhibits spontaneous prostaglandin-mediated contractions without blunting responses to exogenous prostaglandin E2 (Pagano et al., 2017), suggesting upstream modulation rather than general smooth-muscle relaxation. MAGL is also the proximal source of the arachidonic acid from which uterine prostaglandins are synthesised. The chain is coherent. It has not been demonstrated in the condition itself, and the small cannabidiol studies published to date are preliminary rather than practice-changing.

Endometriosis is more interesting and less tidy. In the secretory phase, circulating AEA, 2-AG and OEA are elevated while CB1 expression in endometrial stromal cells is reduced (Sanchez et al., 2016); CB2 is reduced in lesion tissue (Lingegowda et al., 2021). Elevated ligands alongside reduced receptor expression is not what a simple endocannabinoid deficiency model predicts. The most direct link to the disease's endocrine phenotype is that progesterone normally upregulates endometrial CB1 in the secretory phase, and that regulation is disrupted in women with endometriosis (Resuehr et al., 2012) — Level B human tissue evidence, and a plausible tie to progesterone resistance. Importantly, the endocannabinoid system appears dysregulated differently in primary dysmenorrhoea, endometriosis and adenomyosis, so findings should not be transferred between them.

Implications

The defensible position is narrower than the popular one, and still useful. Cycle phase is a source of measurable biological variability, not a dosing variable with an established algorithm behind it. For clinicians, that argues for asking premenopausal patients whether symptoms — and any perceived benefit or adverse effects — vary around ovulation or the premenstrual week, and for recording the answer rather than acting on an assumption. It also argues for treating the phase of sampling as a real methodological variable in any study measuring endocannabinoids in women. Two experiments would move the field quickly: a properly powered plasma 2-AG series across all four phases, and a trial stratifying therapeutic response by cycle phase in a single indication. Neither has been done.