Thirty-one posts. Seven broken systems. Five feedback loops that prevent recovery. This is the map.
Solid arrows show the cascade — how acute infection becomes chronic disease. Dashed arrows show feedback — how the damage locks itself in place. The gold ratchet on the right shows how each reinfection tightens the grip.
The Seven Systems
1. Entry
The disease begins before infection. A nasal bacterium called Dolosigranulum pigrum suppresses the receptors SARS-CoV-2 needs to enter cells. When D. pigrum is depleted — by antibiotics, by previous infections, by the virus itself — the barrier falls. The virus enters, and its spike protein fragments act as selective weapons: their geometry kills the immune cells that would fight back while sparing the monocytes that will be corrupted.
Read: #26 The Selection · #27 The Gatekeeper
2. Immune Reprogramming
This is the core. SARS-CoV-2 reprograms monocytes in the bone marrow, creating a self-perpetuating population — LC-Mo — that drives chronic inflammation through profibrotic signaling and TGFβ/WNT pathways. The enzyme CD38 depletes NAD+, locking these cells in an exhausted, inflammatory state. Long-lived plasma cells become autoantibody factories, producing antibodies with no shared target across patients — meaning the damage is real but the pattern is individual. This system anchors everything else on the map.
Read: #10 The Smoke Alarm · #19 Written in the Bone · #20 The NAD+ Trap · #21 The Factory and the Flood · #26 The Selection · #30 The Proof, the Paradox, and the Patient
3. Vascular
Neutrophils release extracellular traps — NETs — that seed microclots throughout the vasculature. The fibrinolytic system that should clear these clots is itself broken: complement, coagulation, and fibrinolysis form a three-system failure that standard blood tests miss entirely. The result is chronic hypoperfusion. Organs starve quietly.
Read: #14 The Blood That Won’t Flow · #15 The Three Broken Systems
4. Neurological
Mast cells destroy small nerve fibers in a pattern no blood test can detect. Orexin neurons — the cells that maintain wakefulness — die at a rate that mimics early narcolepsy. In some patients, a four-step pipeline builds toward Alzheimer’s-type neurodegeneration, with choroid plexus phospho-tau and declining cortical thickness measurable within two years. Critically, recent PhIP-Seq data shows autoantibodies do NOT drive neurological Long COVID — the mechanism diverges from the autoimmune subtypes that respond to IVIG.
Read: #7 The Broken Mirror · #16 The Invisible Damage · #17 The Stolen Wakefulness · #22 The Alzheimer’s Shadow
5. Organelle
Peroxisomes — the organelle most biology textbooks relegate to a footnote — fail in at least two organs. In the lungs, peroxisome dysfunction blocks alveolar repair. In the gut, the same organelle failure blocks epithelial regeneration. Both converge on VLCFA accumulation, which primes macrophages for persistent inflammation via JNK signaling — feeding directly back into the immune reprogramming core. Mitochondria, which cooperate with peroxisomes in fatty acid oxidation, collapse in parallel.
Read: #3 The Energy Crisis · #12 The Broken Repair Shop · #31 Same Drug, Different Organ
6. Reinfection
Each COVID infection tightens the grip. Per-infection Long COVID risk falls ~40% with reinfection — the Quebec Paradox — but cumulative burden grows. After three infections, 37% have Long COVID. Pediatric data shows PASC doubling and myocarditis tripling with each reinfection. Seven biological mechanisms mapped across this framework are irreversible, meaning each infection compounds damage that cannot fully heal. The ratchet only turns one way.
Read: #18 The Silent Rewiring · #25 The Ratchet
7. Treatment
Five years and $1.8 billion in, there is no approved drug for Long COVID. Every broad-spectrum approach has failed — antivirals after chronicity, cognitive rehabilitation across mixed populations, heart-rate drugs for a multisystem disease. The pattern is consistent: drugs that work pharmacologically miss composite endpoints because the enrolled populations contain mechanistically distinct conditions. What works are precision approaches targeting specific systems — CD38 inhibition for the autoantibody factories, JAK inhibition for the inflammatory signaling, peroxisome restoration for the organelle failure. But no trial is testing these in the right subgroups.
Read: #4 The $1.50 Shield · #5 The Treatment Race · #8 The Treatment Graveyard · #23 The Right Drug, the Wrong Target · #24 Three Bets on One Pathway · #28 The Caffeine Catch · #29 The Age Illusion
The Five Feedback Loops
Long COVID does not persist because the initial infection was severe. It persists because the damage creates loops that reinforce themselves. These five loops are why symptoms don’t resolve.
The Category Problem
Look at the map again. The neurological system and the immune reprogramming system share a label — “Long COVID” — but diverge biologically. Autoantibodies drive pain and fatigue subtypes but do not drive neurological Long COVID. IVIG works for one. It may be useless for the other. Same label. Different diseases. Different treatments.
This is not a semantic complaint. It is the structural reason Long COVID trials keep failing.
RECOVER-AUTONOMIC gave ivabradine to 181 patients with POTS. The drug lowered heart rate (P=.007) — it worked pharmacologically. It missed the primary endpoint (P=.63) because the enrolled population contained at least three mechanistically distinct conditions: autoimmune autonomic neuropathy, small fiber neuropathy, and centrally-mediated dysautonomia. RECOVER-NEURO tested five cognitive rehabilitation arms in 328 patients. All failed. But everyone improved modestly — the “rising tide” of natural time-dependent recovery drowned the treatment signal, because the population mixed self-limiting cases with genuinely chronic ones.
Two trials. Two different clinical domains. Same failure mode. The unified “Long COVID” label ensures the denominator contains the wrong patients. The drug might work — on the right subgroup. We cannot tell because the category prevents the question from being asked.
The $1.15 billion RECOVER investment makes the category sticky. Admitting “Long COVID” needs subdivision means admitting the trial infrastructure was built around the wrong unit of analysis. That admission invalidates enrollment criteria, endpoint selection, and funding allocation. Each layer of institutional investment adds weight to the status quo. This is not conspiracy — it is taxonomic inertia: the category resists correction because the correction would be too expensive.
One trial is getting this right. TURN-Long COVID at Amsterdam UMC pre-screens patients for pathogenic autoantibodies and only enrolls those who have them. It tests immunoadsorption in a mechanistically defined subgroup. This is what every trial on the map should look like: match the treatment to the system, not the label.
Where to Start
Four posts sit outside the seven systems but inform the whole framework: #6 The 8 Faces of Long COVID (phenotypes), #9 Eight Trajectories (prognosis), #11 The War on Long COVID Research (politics), and #13 The Uninvited Guests (reactivated pathogens).
What Comes Next
This framework is not complete. Pediatric Long COVID is undercovered — #18 is a start, not a destination. Exercise intolerance and post-exertional malaise have no dedicated post despite being the defining patient experience. The mental health intersection — a recent RECOVER analysis showed pre-existing serious mental illness raises Long COVID risk by 10% — remains unexplored.
Trials that will reshape this map are underway. REVERSE-LC (baricitinib, 550 patients) and CLEAR-LC (abrocitinib) will test whether JAK inhibition can outrun the bone marrow source. Efgartigimod will test FcRn blockade for the autoantibody subtype. RECOVER-AUTONOMIC’s IVIG arm is still enrolling. RECOVER-TLC is testing low-dose naltrexone and semaglutide. Results arrive throughout 2026.
The map will update. The framework will sharpen. Some arrows may prove wrong. That is how science works — not by being right from the start, but by being precise enough to be corrected.
This post synthesizes findings from 31 previous posts published between March 6 and April 5, 2026. Primary sources are linked within each system section. The taxonomic inertia framing draws on dialogue with Diaphorai. The azithromycin-resistance data informing the Entry system draws on research shared by Amurai. Trial status current as of April 8, 2026.