The aromatase loop — how an endometriosis lesion makes its own estrogen

The mechanism behind that aberrant switch is specific. In endometriotic stroma, CYP19A1 is transcribed from its proximal promoter II, the same promoter used by genuinely steroidogenic tissue, and that promoter is driven by steroidogenic factor-1 (SF-1, the nuclear receptor NR5A1) operating through a cyclic-AMP and protein kinase A cascade. The lesion also overexpresses steroidogenic acute regulatory protein, StAR, which shuttles cholesterol into the mitochondrion to feed the pathway. The net result is a small, self-contained estrogen factory built inside ectopic tissue — a tissue that has co-opted the transcriptional machinery of a steroid-producing organ.

Making estradiol is only half of what makes the lesion dangerous. The other half is that it fails to get rid of it. Healthy estrogen-responsive tissue keeps a brake on local estradiol through 17-beta-hydroxysteroid dehydrogenase type 2, the enzyme that oxidizes potent estradiol into much weaker estrone. Endometriotic tissue runs low on that enzyme. So the lesion is simultaneously turning the production up and turning the inactivation down: it generates estradiol and then retains it, holding the active hormone in place rather than clearing it. Local estrogen concentration climbs for both reasons at once, in a compartment that the rest of the body's hormonal regulation barely reaches.

There is a second brake the lesion is missing, and it sits at the level of the whole reproductive cycle rather than inside the enzyme. In normal tissue, progesterone is the physiologic opposition to estrogen-driven proliferation — the luteal-phase signal that restrains growth. Endometriotic tissue tends to be relatively progesterone-resistant, so that opposing signal lands weakly even when progesterone is present. The operative problem is therefore not always an absolute excess of estrogen but a relative one: estrogen acting with too little progesterone to oppose it, in a tissue already manufacturing extra estrogen on its own account. This is the more accurate reading of the phrase estrogen dominance — a balance tilted toward proliferation rather than a single number out of range — and it is part of why the lesion proliferates under conditions that orthotopic tissue would tolerate. The lesion has both turned up its own estrogen and turned down its sensitivity to the hormone that would tell it to stop.

What turns this into a loop — rather than a one-way production line — is what that local estradiol does next. Estradiol transcriptionally induces COX-2, the enzyme encoded by PTGS2, which raises the lesion's output of prostaglandin E2, a powerful inflammatory mediator. And prostaglandin E2, signaling back through its EP2 and EP4 receptors and the cAMP they raise, is the most potent known inducer of SF-1 and aromatase in this tissue. Trace the circuit and it closes on itself: estradiol begets prostaglandin E2, prostaglandin E2 begets SF-1 and aromatase, aromatase begets more estradiol. It is a feed-forward, self-amplifying loop in which estrogen and inflammation each drive the other, accelerating until something external interrupts it. Once running, it does not need much from the rest of the body to keep going.

That last point is the clinically decisive one. The entire loop is intracrine — generated, consumed, and amplified inside the lesion — and it is therefore largely independent of ovarian estrogen output. This is precisely why the conventional logic of suppressing the ovaries is incomplete. A combined contraceptive, a progestin, or a GnRH analogue lowers the estradiol arriving from the ovary, and that helps, but it leaves the lesion's own intracrine production substantially intact. The lesion goes on making and retaining its own estrogen, and the prostaglandin arm of the loop goes on reinforcing it, beneath whatever ovarian suppression has been achieved. The partial, non-durable responses that so often follow ovary-directed therapy are not a mystery once the loop is in view; they are the expected outcome of suppressing one estrogen source while leaving a second, self-amplifying one running.

A useful feature of seeing the disease this way is that much of the loop has a measurable readout in blood, which turns the mechanism into something that can be tracked rather than merely described. Estradiol interpreted against progesterone captures the balance the lesion exploits, rather than either hormone read alone. Sex-hormone-binding globulin read against fasting insulin captures how much of that estrogen is actually free to act and whether insulin is suppressing the binding globulin. Systemic inflammatory tone shows up in markers like high-sensitivity CRP, and the estrogen-clearance machinery — methylation capacity and thyroid-dependent hepatic conjugation — has its own readouts. These are best understood as functional-optimization targets followed over time within an individual, interpreted in pairs and under clinician guidance, not as validated diagnostic cut-offs; their value is in showing the direction of change as the underlying drivers are addressed. The point for the aromatase story is simply that the loop is not an abstraction. Falling free estrogen and falling insulin are, in principle, the biochemical signature of starving it.

This is also exactly where the picture turns from explanatory to actionable, because there is a clean clinical proof-of-principle that targeting the loop itself works. Aromatase inhibitors — letrozole and anastrozole — are non-steroidal competitive inhibitors of CYP19A1, the terminal enzyme in the chain. They do not merely reduce ovarian estrogen; they collapse aromatization wherever it occurs, including inside the lesion, which breaks the estrogen-to-COX-2-to-PGE2-to-aromatase circuit at its source. In trials, aromatase inhibition was associated with reduced implant size and reduced pain. That result is the cleanest available demonstration of the central claim: targeting lesional steroidogenesis, rather than ovarian output alone, produces measurable benefit. It is the reason aromatase inhibition functions as the proof of concept for the whole idea that endometriosis can be approached as lesional biology and not only as anatomy.

It is worth being precise about what the lesion is and is not independent of, because the autonomy is relative rather than absolute. The intralesional factory is the reason ovarian suppression alone disappoints, but it is not the only source of estrogenic drive in the body, and the two interact. Adipose tissue is itself the dominant site of extragonadal aromatization — fat converts circulating androgens into estrogen outside the ovary entirely — so systemic estrogenic tone is set partly by body composition, not just by the gonad. At the same time, the bioavailable fraction of whatever estrogen is present is governed by sex-hormone-binding globulin, and SHBG is suppressed by insulin; hyperinsulinemia therefore lowers SHBG and raises the free, active estrogen the lesion is exposed to. The picture that results is layered: the lesion makes its own estradiol intracrinely, the body's fat makes more peripherally, and high insulin frees up more of the total to act. The intralesional loop is the engine, but it runs in an estrogenic environment that other parts of the metabolism can stoke or starve.

That layering also explains why the aromatase loop is best understood as one circuit inside a larger, reinforcing network rather than the whole disease. The same local estradiol that feeds the prostaglandin loop sits alongside chronic peritoneal inflammation, alternatively activated macrophages that supply fibrosis and a blood supply, NF-kB integrating the inflammatory signals, and the insulin/IGF-1 axis amplifying proliferation underneath all of it. These circuits are not independent of the estrogen loop — estradiol induces the COX-2 that drives the prostaglandin arm, and the prostaglandin arm feeds back to inflammation more broadly — but they are partly redundant, which is why even a clean blockade of one node tends to produce a partial rather than complete response. The aromatase loop is the most directly druggable and best-validated of these circuits, and that is exactly why it makes such a clean teaching case: it shows, in isolation, the general principle that the lesion sustains itself from the inside.

There is an important physiological caveat in how aromatase inhibitors are actually used, and it follows directly from the same hormonal wiring. In a premenopausal woman, sharply lowering circulating estradiol removes the negative feedback on the pituitary, which responds with a reflex rise in follicle-stimulating hormone that stimulates the ovary toward folliculogenesis — partly working against the intended estrogen suppression and producing ovarian cysts. For that reason, aromatase inhibitors are co-prescribed in premenopausal women with an ovarian suppressant — a progestin, a combined contraceptive, or a GnRH analogue — to hold that reflex in check. This is not a hedge against the mechanism; it is a recognition that the loop sits inside a larger feedback system, and that interrupting one node means accounting for how the system compensates.

The implication of the aromatase loop runs deeper than any single drug. It reframes what the lesion is. An endometriotic implant is not passive misplaced tissue waiting on hormonal weather from elsewhere in the body; it is a locally autonomous, estrogen-producing, self-inflaming micro-tissue that manufactures the very signal it depends on and shields that signal from inactivation. Once that is the operating model, the long-standing puzzle of incomplete endocrine response resolves into something straightforward — you cannot fully starve a tissue of a hormone it makes for itself by turning down a supply it has learned to do without. And the same logic points forward: the meaningful question becomes not only how much estrogen the ovary is sending, but how much the lesion is generating in place, and what it would take to switch that local factory off.