OSF project (full materials): https://osf.io/fgyd7
Medical disclaimer: This is a research-oriented synthesis and a proposal for testable directions — not medical advice.
TL;DR
Endometriosis isn’t “just tissue in the wrong place.” It’s a self-sustaining system:
- Ectopic lesions seed and persist.
- The immune system fails to clear them (and often becomes locally permissive).
- Lesions create a high-estrogen / progesterone-resistant microenvironment.
- Chronic inflammation + oxidative stress feed angiogenesis and survival.
- Lesions build their own blood supply and nerves.
- Over time you get fibrosis/adhesions and a pain loop that can outlive the lesions.
So the goal isn’t “one magic bullet.” The goal is to break the loop.
This post proposes a Minimal Viable Endometriosis Loop (MVEL) plus a treatment blueprint:
- Multi-vector immune programming (“vaccine-like”) to change the immune trajectory.
- Targeted clamps on estrogen/progesterone signaling, inflammatory cytokines, redox tone, angiogenesis, fibrosis, and pain sensitization.
Why endometriosis looks like uncontrolled cell growth (but isn’t cancer)
At first glance, endometriosis resembles “unwanted growth”:
- Lesions proliferate.
- They invade, recruit blood vessels, remodel tissue.
- They recur.
But the logic is closer to a wound that never resolves:
- repeated micro-bleeds → iron/ROS → inflammation → remodeling → more permissive niche.
If you’ve read my Minimal Viable Cancer Loop (MVCL) post, this is the same move: take a complex disease and compress it into a closed causal circuit.
Reference: From Paracetamol and Alzheimer’s Loops to a Minimal Viable Cancer Loop (MVCL)
https://hmwh.se/blog/2025/11/25/from-paracetamol-and-alzheimers-loops-to-a-minimal-viable-cancer-loop-mvcl/
Core biology (what the literature keeps repeating)
Across reviews and mechanistic overviews, a small set of repeating themes shows up:
1) Estrogen dependence + local estrogen production
Lesions are estrogen-driven, often with ERβ dominance and local aromatase, maintaining a pro-growth microenvironment.
2) Progesterone resistance
A large subset of patients show reduced progesterone signaling (e.g., reduced PR-B; failure to induce anti-proliferative / anti-inflammatory programs), leaving unopposed estrogenic action.
3) Immune dysfunction + chronic inflammatory peritoneal milieu
Peritoneal fluid is enriched in IL‑1β, IL‑6, IL‑8, TNF‑α, CCL2/MCP‑1, etc. Macrophages accumulate but often have impaired clearance, skewing toward M2-like wound-healing/pro-angiogenic phenotypes; NK cytotoxicity is often reduced.
4) Neuroangiogenesis
Lesions don’t just survive — they recruit blood vessels (VEGF/PDGF) and nerve fibers (NGF and other neurotrophins). That’s how a lesion becomes a pain generator.
5) Oxidative stress (iron/ROS)
Repeated micro-bleeding deposits iron → ROS → tissue damage → NF‑κB activation → more cytokines and fibrosis.
6) Fibrosis/adhesions
Deep lesions are fibrotic; TGF‑β and downstream profibrotic programs drive scarring and adhesions.
Those six bullets are enough to build a loop.
The Minimal Viable Endometriosis Loop (MVEL)
I’m defining MVEL as the smallest set of causal nodes that can keep endometriosis running once established.
The MVEL nodes (M0–M7)
M0 — Seeding event
Menstrual debris and/or endometrial-like cells reach ectopic sites (retrograde menstruation, tissue injury, metaplasia hypotheses — mechanisms may differ, but “ectopic exposure” is the common input).
M1 — Implantation & survival entry
Adhesion/MMP activity, stromal survival programs, early angiogenic signaling.
M2 — Clearance failure (the heavy node)
Macrophage phagocytosis doesn’t clean up; NK cells don’t kill effectively; immune milieu drifts toward permissive patterns.
M3 — Hormonal drive & progesterone resistance
Estrogen-rich microenvironment + progesterone resistance → growth permissive state.
M4 — Inflammation & oxidative stress amplifier
Cytokine loop (NF‑κB/MAPK) + iron/ROS from micro-bleeds.
M5 — Vascular + nerve infrastructure (neuroangiogenesis)
Lesions secure supply lines and wire themselves into pain circuits.
M6 — Fibrosis/adhesions remodeling
TGF‑β → collagen/CTGF programs → scarring, adhesions, mechanical pain.
M7 — Pain + central sensitization maintenance
Peripheral nociception + neuroimmune feedback → central sensitization; pain can persist even if lesions shrink.
The loop in one line
M0 → M1 → M2 → (M3 + M4) → M5 → M6 → M7 → back into (M4 + M2)
That’s the engine.
Why a “vaccine-like” strategy is plausible — and why it’s tricky
“Vaccine” here doesn’t necessarily mean classic sterilizing immunity. It means immune trajectory engineering.
The literature already contains proof-of-concept that immune programming can alter the disease course:
- Non-specific immune stimulation (e.g., BCG pretreatment in animal models) can reduce lesion implantation in some contexts.
- Paradox: other work suggests the opposite immune state (innate tolerance) can reduce lesions, depending on timing/context.
- Xenogenic immunomodulators (e.g., RESAN in rats) report large prophylactic effects and reduced Th2 cytokine rises.
- Antigen-centric ideas exist (MUC1, lesion-associated enzymes/proteins), but specificity and fertility safety are the hard constraints.
The core design constraint
Endometriotic lesions are made of endometrial-like tissue. A strong cytotoxic response risks collateral damage to normal endometrium, implantation, and fertility. A tolerogenic strategy risks blunting surveillance too much.
So any immune strategy must be:
- Stage-aware (early seeding vs established deep disease),
- phenotype-aware (inflammatory vs tolerant/permissive niches),
- delivery-aware (regional/local where possible).
Intervention mapping (MVEL → levers)
Below is a practical “matrix” mapping nodes to intervention families. Think of it as a menu for building combinations.
Legend:
Standard = already used in practice (symptom control / hormonal suppression / surgery).
Translational = plausible near-term based on trials or adjacent fields.
Speculative = research-grade concept, needs validation.
| MVEL node | What keeps it running | What you can hit | Example lever families |
|---|---|---|---|
| M0 Seeding | ectopic exposure, tissue injury | reduce seeding frequency; change peritoneal response to debris | phenotype stratification; (speculative) prophylactic immune programming |
| M1 Implantation | adhesion/MMPs; early survival | block adhesion/invasion early | anti-adhesion strategies; MMP modulation (research) |
| M2 Clearance failure | macrophage dysfunction; NK impairment | restore clearance or reprogram phenotype | NK boosting (IL‑12/IL‑15 logic); macrophage repolarization; complement modulation (e.g., C3 axis); checkpoint-style relief (research) |
| M3 Hormonal drive | high estrogen; local aromatase; progesterone resistance | reduce estrogen tone; restore progesterone program | COCs/progestins (standard); GnRH antagonists/agonists (standard); aromatase inhibitors (selected); ER/PR pathway modulators (research) |
| M4 Inflammation/ROS | cytokines + NF‑κB + iron/ROS | break inflammatory amplification; adjust redox | cytokine targeting (e.g., IL‑8 axis in primate models); NF‑κB/MAPK pathway modulation; antioxidant/redox interventions (e.g., NAC in models) |
| M5 Neuroangiogenesis | VEGF/PDGF; NGF; nerve ingrowth | cut supply lines and wiring | anti-angiogenic strategies (research constraints); NGF/neurotrophin axis targeting (research); lesion-local delivery concepts |
| M6 Fibrosis | TGF‑β/CTGF; collagen deposition | prevent/undo scarring and adhesions | anti‑TGF‑β logic (safety constrained); anti‑fibrotic programs; post-surgical anti-adhesion + immune clamp |
| M7 Pain/central sensitization | neuroimmune feedback; central amplification | treat pain loop directly + stop upstream drivers | pain neuromodulation; anti-inflammatory clamps; central sensitization protocols; pelvic rehab (clinical) |
A treatment blueprint: “multi-vector vaccine + targeted clamps”
Here’s the minimal blueprint that respects the biology and the safety constraints.
Step 0 — Stratify (you can’t treat what you don’t segment)
Endometriosis is not one uniform disease. At minimum, stratify by:
- Lesion phenotype: superficial / endometrioma / deep infiltrating
- Immune state markers: macrophage phenotype signatures, NK function proxies, cytokine patterns
- Hormonal axis: estrogen tone, progesterone resistance markers
- Pain phenotype: peripheral-dominant vs central sensitization-dominant
Goal: pick the right immune trajectory move.
Phase 1 — “Debulk and cool” (reduce load before immune programming)
Depending on phenotype:
- Surgery for large lesions/endometriomas/deep disease (standard of care in many cases)
- Short hormonal clamp to reduce estrogen drive and inflammatory tone (standard options)
Why: fewer lesions = fewer antigens = less collateral risk when immune pathways are perturbed.
Phase 2 — Innate reprogramming (M2/M4)
This is where “vaccine-like” starts to mean something.
Two competing (and possibly stage-dependent) hypotheses exist:
H2a: The permissive niche hypothesis (attack mode)
If lesions persist due to immune tolerance/permissiveness, push toward better clearance:
- macrophage repolarization away from M2-permissive programs
- NK function restoration
- complement tuning if it’s driving dysfunction
H2b: The inflammatory amplifier hypothesis (tolerance mode)
If the peritoneum is locked in an inflammatory state that paradoxically helps lesions survive, induce a controlled innate tolerance / inflammation-resolving program.
The key is not ideology. It’s measuring which state you’re in, then moving it in the correct direction.
Phase 3 — Antigen-targeted immune move (optional, and high governance)
If (and only if) you have a defensible lesion-selective target set, you can consider:
- mRNA / DNA platform encoding lesion-enriched antigens (conceptual)
- peptide + adjuvant (classic vaccine approach)
- DC-based presentation (more controllable, more complex)
Targets discussed in the literature include:
- MUC1 (lesion-associated; but tolerance can develop)
- lesion-associated enzymes/proteins involved in implantation (examples exist in animal studies)
- in the future, patient-specific neoantigens if somatic mutation landscapes prove clinically actionable
Governance requirement: uterine safety, fertility impact, cross-reactivity risk ledger.
Phase 4 — Clamp the amplifiers (M3/M4/M5/M6)
Even a perfect immune move can be undermined if the amplifiers keep running.
This phase is combination therapy by design:
- Hormonal clamp (reduce estrogenic drive; address progesterone resistance)
- Cytokine axis targeting (e.g., IL‑8 logic; TNF/IL‑6-type axes — research-level in endo)
- Redox tone control (iron/ROS environment; antioxidant strategies are biologically coherent but need rigorous phenotype targeting)
- Anti‑angiogenic / anti‑neurotrophin concepts where safely deliverable
- Anti‑fibrotic / anti‑adhesion programs (especially peri‑surgical)
Phase 5 — Treat the pain loop as its own disease (M7)
Endometriosis pain can become a stand-alone circuit.
So pain treatment is not “cosmetic.” It’s a legitimate node of recurrence and disability.
- treat peripheral drivers (inflammation, neuroangiogenesis)
- treat central sensitization directly (clinical protocols)
The “multi-phase vaccine” concept (a concrete sketch)
If I had to express this as a single coherent program (not a shopping list), it would look like:
Vaccine program V1 (research concept)
V1a — Innate conditioning
A controlled innate reprogramming regimen (attack-mode or tolerance-mode chosen by stratification).
V1b — Antigen targeting (optional)
If safe targets exist: multi-antigen encoding + controlled delivery route.
V1c — Maintenance / recurrence prevention
A maintenance immune bias that prevents re‑seeding from turning into a self-sustaining lesion niche.
In plain language: teach the peritoneum to respond differently to menstrual debris, and then keep it in that state.
What would falsify MVEL?
A loop model is only useful if it makes crisp, testable predictions.
Here are “hard falsifiers” for the MVEL framing:
- If M2 is not impaired (normal clearance signatures, normal NK cytotoxicity), yet disease persists unchanged → M2 is not a core node.
- If you normalize M3 (hormonal axis) but the lesion niche remains stable with no meaningful regression → M3 is a modulator, not a sustaining leg.
- If you break M4 (cytokines/ROS) and neither lesions nor pain improve → inflammation/ROS is downstream noise, not an amplifier.
- If you interrupt M5 (nerve/vascular signaling) and pain remains stable → pain isn’t being driven by lesion wiring.
- If fibrosis/adhesion prevention (M6) fails despite TGF‑β/CTGF suppression → fibrosis is not the main remodeling engine.
What’s next (practical research agenda)
If you want to make this real, the early work is not “build the vaccine.” The early work is:
- Phenotype stratification protocol (immune/hormonal/pain)
- Target selection ledger (lesion specificity vs uterine safety)
- Delivery strategy (regional where possible)
- Early biomarkers that show node movement (macrophage state; NK function; cytokines; imaging proxies)
- Combination logic (which clamps are required to prevent the loop from re-forming)
All the detailed notes, PDFs, and working materials are in the OSF repository: https://osf.io/fgyd7
Sources and further reading (starter set)
- Endometriosis: A Comprehensive Research Overview (project PDF; mechanistic summary of estrogen dependence, progesterone resistance, immune dysfunction, neuroangiogenesis, oxidative stress, fibrosis) — see OSF.
- Immunomodulatory and “Vaccine” Therapies for Endometriosis (project PDF; BCG, RESAN, MUC1, innate training vs tolerance, NK boosting, complement modulation; safety constraints) — see OSF.
- Jeljeli et al., JCI (trained immunity vs tolerance logic in endometriosis models) — see PubMed.
- Maksym et al., J Clin Med (immunology and immunotherapy in endometriosis; NK sections) — see publisher.
- Li et al., Frontiers in Immunology (immunotherapy strategy review) — see publisher.
- RESAN animal model studies — see PubMed / EJO&G Reproductive Biology.
Closing
Endometriosis is a loop problem.
If you treat it as a loop problem, “vaccine-like immune trajectory engineering + targeted clamps” stops sounding like sci‑fi and starts sounding like a normal systems-engineering response to a self-sustaining biological circuit.
The details are hard. The governance constraints are real. But the shape of the problem is now clear.