Background

This is a second, independent case processed through the same structured virtual assessment pathway described in a previous report. As before, the tool combines standardised patient-reported outcome measures (PC-COS, PEM subscale) with structured intake of investigations already on file, and returns a preliminary working diagnosis, differential diagnosis, missing-investigation list, and staged therapeutic rationale — explicitly flagged as not a diagnosis or prescription, requiring direct clinical validation before any action. All therapeutic decisions described below were made and supervised directly by the treating physician; the tool’s output functioned as a structuring and hypothesis-generation layer, not as an autonomous decision-maker.

Case presentation

A woman in her mid-thirties, with pre-existing polycystic ovary syndrome (PCOS), no other relevant history. Index SARS-CoV-2 infection in January 2022, with a second confirmed infection in August 2024. At the time of the baseline structured assessment (~2–3 years post-index infection), she remained in full-time work/study but required frequent additional rest and periodic reduction to part-time activity. No formal Long COVID diagnosis had yet been established despite prior review by a pulmonologist and a COVID-19 specialist.

Baseline (pre-rapamycin) objective work-up

Baseline outcome measures — PC-COS and PEM

Calculated authomatically with my virtual clinic

PEM characteristics: onset immediate, triggered by as little as 5–8 stair steps or >100 m of normal-pace walking; typical resolution within 24 h, with a secondary delayed-worsening component at 24–48 h also present. No pharmacological Long COVID treatment had been trialled at this point.

What the virtual assessment proposed at baseline

Working diagnosis: Long COVID (PASC) with ME/CFS phenotype, supported by PEM (26/36), marked fatigue (18/20), CPET-documented exercise intolerance with chronotropic incompetence, and >2 years’ symptom persistence.

Differential diagnoses generated:

1. POTS/dysautonomia

2. Antiphospholipid syndrome (aCL IgG 32.2 U/mL, above the 20.0 U/mL threshold)

3. Mast cell activation syndrome (MCAS)

Missing investigations flagged: active standing/tilt-table test, repeat anti-cardiolipin IgG/IgM + anti-β2-glycoprotein I + lupus anticoagulant at ≥12 weeks, serum tryptase and urinary N-methylhistamine, EBV serology/reactivation panel, morning cortisol/ACTH stimulation, full iron studies with soluble transferrin receptor, vitamin D recheck, cardiac MRI, and wearable HR/HRV monitoring.

Mechanistic rationale generated for the eventual immunomodulatory step (rapamycin, low-dose intermittent 0.5–1 mg weekly or on alternate days), specific to this patient’s data:

1. mTORC1 hyperactivation — the tool linked the persistent CRP elevation on two independent samples, the borderline leukocytosis/neutrophilia, and the elevated alpha-2- and beta-1-globulin fractions on protein electrophoresis to a pattern of sustained mTOR-pathway-linked innate immune activation, on the premise that intermittent low-dose rapamycin restores autophagic flux and dampens chronic inflammatory tone without broad immunosuppression.

2. NLRP3 inflammasome suppression — the PEM phenotype (flu-like malaise scoring 4/4) and the neurocognitive burden (37/70) were framed as clinical correlates of NLRP3-driven IL-1β/IL-18 signalling, with rapamycin proposed to reduce NLRP3 assembly via mTORC1 inhibition and mitophagy-mediated reduction of the mitochondrial ROS “second signal.”

3. Senescent T-cell clearance — the tool noted the elevated beta-1-globulins and an upper-normal complement C4 as possibly reflecting ongoing immune-complex/antigen-driven activity, and proposed that intermittent rapamycin selectively promotes apoptosis of senescent, SASP-expressing T cells while sparing naïve/effector-memory populations.

4. Mitochondrial function — the reduced VO₂/work ratio (~9.4 ml/min/watt) and PEM item “muscle weakness/heaviness after any activity” (4/4) were linked to impaired oxidative phosphorylation, with rapamycin-driven mitophagy and PGC-1α upregulation proposed as a partial corrective mechanism, synergistic with correction of the documented vitamin D insufficiency.

The tool additionally proposed, at earlier steps, autonomic stabilisation (ivabradine, volume expansion), neuromodulation/sleep restoration (low-dose aripiprazole, LDN, sleep agents), and an anticoagulation/antiplatelet step (aspirin, nattokinase/lumbrokinase, LDN) predicated on the elevated anti-cardiolipin IgG and the CRP/fibrinogen pattern — flagged explicitly as contingent on confirmatory antiphospholipid serology at 12 weeks, which does not appear to have been formally repeated in the record reviewed. Graded exercise therapy was explicitly flagged as contraindicated given the PEM profile, and the CPET’s own interpretive conclusion (”deconditioning”) was flagged by the tool for critical re-evaluation against the fuller autonomic/immunological dataset.

Intervention

Low-dose rapamycin (sirolimus) was initiated in December 2024 and maintained for approximately 14 months, with rest/pacing as a concurrent management strategy throughout. Rapamycin was discontinued in February 2026. The follow-up structured assessment reported below was completed in July 2026 — approximately five months after discontinuation — with the patient on no regular Long COVID medication (”rest alone”) at the time of scoring. This timeline matters for interpretation: the improvement captured in the second PC-COS/PEM assessment is not a same-day, on-drug snapshot, but reflects a functional status that has been sustained for several months off treatment, following a period of roughly 14 months on treatment. A gynaecological review around the time of discontinuation newly identified uterine fibromatosis and micropolycystic ovarian morphology, following which hormonal therapy (Belara) replaced a prior contraceptive (Novadien); this remains a relevant confounder for the residual (non-normalised) domains discussed below, but does not affect the observation that the core PEM/fatigue/exertional improvement had already occurred on treatment and persisted after stopping it.

Adjunctive immunophenotyping (peripheral blood lymphocyte subsets)

An extended lymphocyte immunophenotyping panel was performed (research-use assay, Clinical Immunology and Vaccinology Research Unit)

The most notable feature is the markedly expanded CD8⁺ effector-memory compartment (49.6% of CD8⁺ T cells, versus a naïve fraction of 32.2%), with a comparatively low CD8⁺ central-memory fraction (3.6%). This skew toward effector-memory over naïve/central-memory CD8⁺ T cells is a pattern compatible with — though not diagnostic of — chronic or repeated antigenic stimulation, and is consistent in direction with the “senescent/exhausted T-cell” rationale the virtual assessment tool invoked in support of rapamycin. The CD4⁺ compartment, by contrast, shows a relatively preserved naïve fraction (47.0%) and a low effector-memory fraction (7.6%), i.e. the skew is CD8-predominant. NK and NKT-like proportions were within a broadly unremarkable range on this single panel; no serial or reference-range comparison was available to determine whether these values are stable or evolving. This panel was run once, without a paired post-treatment repeat, so it cannot be used to demonstrate a treatment-associated shift — it is presented here as baseline supportive context, not as evidence of a pharmacodynamic effect.

Follow-up (post-rapamycin) assessment

A second structured assessment was completed after the rapamycin course (initiated December 2024, discontinued February 2026).

Interval investigations

Outcome measures — pre- vs post-rapamycin

Every domain improved, and the magnitude in fatigue, cardiovascular, neurocognitive, and physical-functioning subscores is substantial by any threshold used in PC-COS validation work. The mental health and sleep domain shows the smallest relative improvement (37→24, i.e. still nearly half of maximum burden), with anxiety, sleep quality, and unrefreshing sleep individually still scoring 7/10 each post-treatment — a residual burden pattern that mirrors what was seen in the first case reported in this series, and again suggests that the neuropsychiatric/sleep domain may respond less completely to mTORC1-targeted immunomodulation than the somatic/exertional domains.

Updated working diagnosis (post-rapamycin)

The virtual assessment’s post-treatment working diagnosis was “Long COVID in significant clinical remission, with residual mild neuropsychiatric symptoms and a microcytic anaemia pattern,” with marked symptomatic improvement temporally associated with sirolimus therapy. Differentials were revised accordingly: iron-deficiency anaemia secondary to the newly identified uterine fibromatosis (a plausible confounding contributor to residual fatigue that was not present at baseline assessment), PCOS-related hormonal dysregulation as a contributor to residual sleep/anxiety symptoms, and mild subclinical autonomic dysfunction not meeting POTS criteria.

Discussion

As with the first case in this series, several limitations should be stated explicitly rather than left implicit:

• This is a single, unblinded, non-controlled, retrospective-comparison case. The two structured assessments are separated by over a year. A new gynaecological diagnosis (uterine fibromatosis) and a new hormonal therapy (Belara) emerged around the time rapamycin was stopped, and either could independently affect residual PC-COS domains (notably physical/daily functioning and the microcytic blood picture) — but neither offers an obvious explanation for the PEM and exertional-tolerance improvement, which was already established during the treatment period itself and persisted through five drug-free months by the time of the second assessment.

• The antiphospholipid antibody finding (aCL IgG 32.2 U/mL) flagged at baseline as requiring 12-week confirmatory testing does not appear to have been formally repeated in the documentation reviewed; this remains an open diagnostic loop rather than an excluded differential, and should not be considered ruled out on the basis of clinical improvement alone.

• The immunophenotyping panel was performed once, without a paired post-treatment sample, so the CD8⁺ effector-memory skew can be reported only as baseline context consistent with the tool’s proposed senescent/exhausted T-cell mechanism — not as evidence that rapamycin altered this compartment, and not as evidence that it explains the sustained off-drug improvement.

• The residual microcytic, mildly hypochromic blood picture persisted essentially unchanged between the two time points (MCV 78.3→79.8 fL), and the emergence of uterine fibromatosis with associated bleeding by the second assessment is a more parsimonious explanation for the modest residual physical-functioning burden than incomplete treatment response, and should be worked up on its own merits (ferritin, iron studies) rather than folded into the Long COVID narrative.

A cross-case pattern worth flagging: sleep/neuropsychiatric domain vs. neurocognitive and somatic domains

Because this case is intended for publication after the first case in this series, the two are worth reading side by side rather than in isolation — a single shared pattern, and one clear divergence, emerge.

The shared pattern — mental health & sleep as the domain that under-responds: in the first case, the mental health & sleep subscore improved only modestly relative to fatigue/pain/PEM (30→25/50), with unrefreshing sleep, sleep disturbance, and anxiety remaining the highest-weighted residual items even after substantial global improvement. In this second case, the same domain again shows the smallest relative gain of any PC-COS subdomain (37→24/50, i.e. still at roughly half of maximum possible burden), with anxiety, sleep quality, and unrefreshing sleep each still scoring 7/10 individually — despite a near-total resolution of PEM (26→1/36) and cardiovascular symptoms (24→2/30) in the same patient, on the same drug. Two independent cases, on the same intermittent low-dose rapamycin regimen, both show a global improvement pattern in which sleep and neuropsychiatric burden lag behind exertional, cardiovascular, and (in this case) cognitive recovery. This is the most consistent, and arguably most clinically actionable, signal across the series: if mTORC1 modulation is contributing to the somatic/exertional recovery, its effect on sleep architecture and anxiety appears to be partial at best, and these domains likely warrant dedicated, independent intervention (as was done with add-on LDN in the first case) rather than an expectation that they will resolve as a downstream consequence of the same mechanism.

The divergence — neurocognitive domain: here the two cases differ sharply. In the first case, the neurocognitive subscore did not improve on rapamycin alone (32→34/70, essentially flat), and meaningful cognitive gains were only reported anecdotally after LDN was added later. In this second case, the neurocognitive domain shows the single largest relative improvement of any subscale (37→4/70, a 90% reduction), achieved on rapamycin without LDN having been part of this patient’s regimen. This divergence is worth stating plainly rather than smoothing over: it suggests either that neurocognitive response to mTORC1 modulation is genuinely heterogeneous across patients (plausible, given the different baseline severities — this patient’s baseline neurocognitive burden, driven by headache and brain fog scoring 8/10 each, versus the first patient’s baseline burden distributed more evenly with paraesthesia and word-finding also prominent), or that some other unmeasured factor (illness duration, baseline PET/CPET severity, comorbidities, concurrent LDN vs. its absence) modulates this domain independently of the somatic response. With n=2, this can only be flagged as a question for a future, larger case series — not resolved here.

Taken together, this second case reinforces the same overall conclusion drawn from the first: a structured virtual assessment tool, integrating fragmented investigations and validated patient-reported outcomes, surfaced a coherent and testable mechanistic hypothesis (mTORC1/NLRP3/T-cell senescence) that was clinically actioned under direct physician supervision, and the observed trajectory — a PC-COS burden reduction from 147/234 to 38/234 and a PEM score reduction from 26/36 to 1/36, sustained several months after treatment discontinuation — is compatible with, but does not prove, a causal and durable treatment effect. Confirmatory prospective work (paired immunophenotyping, repeat antiphospholipid panel, controlled comparison, and systematic tracking of the sleep/neuropsychiatric domain across a larger case series) would be needed before generalising beyond hypothesis-generation.

Nothing in this report constitutes generalisable treatment guidance. All interventions described were individualised, off-label, discussed with the patient, and monitored under direct clinical supervision. This is an anonymised report; no identifying details are included.

Leave a comment

One of the recurring challenges in managing severe, long-duration Long COVID/ME-CFS is information fragmentation: years of investigations, multiple specialists, dozens of symptoms, and patients who are often too unwell — or geographically too far — to attend in person repeatedly. Over the past months I have been using a structured virtual assessment tool (available at buonsenso-clinic.thekookbitcoiner.workers.dev) that collects standardised clinical history, prior investigations, and validated patient-reported outcome measures, and returns a structured working diagnosis, differential diagnosis, list of missing investigations, and a staged therapeutic rationale — always explicitly flagged as preliminary and requiring clinical validation, and always reviewed, corrected, and authorised by me before any action is taken.

I want to share one case in detail — anonymised, with the patient’s consent for research and educational use — because it illustrates both the potential and the proper epistemic limits of this kind of tool.

Case presentation

A patient in their late twenties, no relevant pre-existing history, with post-acute sequelae of SARS-CoV-2 infection persisting for over four years, meeting criteria for both Long COVID and ME/CFS. At the time of the first structured assessment, functional status was severely impaired: housebound on most days, working exclusively remotely, with post-exertional malaise (PEM) as the dominant limiting feature.

Objective work-up already on file included:

• CPET: peak VO₂ 23.6 ml/kg/min (55% predicted), cardiogenic limitation, early anaerobic threshold, blunted peak blood pressure response

• 24h Holter ECG: elevated mean heart rate, resting peaks of 100–120 bpm

• Brain PET: mild left frontal hypometabolism

• Brain MRI, EMG: normal

• Tilt-table test: negative for POTS

What the virtual assessment proposed

Integrating these findings with immunological data obtained later in the course (elevated total IgG, persistently detectable anti-Spike antibody titre years after acute infection, a CD4⁺ naïve/memory ratio suggestive of chronic antigenic stimulation), the tool’s output converged on a working hypothesis of sustained mTORC1 hyperactivation as a plausible unifying mechanism linking the cardiopulmonary, neurocognitive, and immunological findings.

Rationale generated for rapamycin (sirolimus), low-dose:

1. mTORC1 inhibition — selective suppression of chronic anabolic immune overactivation without broad immunosuppression at intermittent low dosing, sparing mTORC2.

2. NLRP3 inflammasome suppression — mTORC1 hyperactivity was proposed as a proximal driver of NLRP3 assembly and IL-1β/IL-18 release, mechanistically linked to the neuroinflammatory pattern consistent with the PET hypometabolism.

3. Senescent T-cell clearance — autophagy-mediated clearance of exhausted CD8⁺ T cells accumulating in chronic Long COVID, cells that perpetuate SASP-driven inflammatory signalling without effective viral surveillance.

4. Mitochondrial biogenesis — relief of mTORC1-mediated suppression of PGC-1α, proposed as mechanistically relevant to the early anaerobic threshold and reduced peak VO₂ on CPET.

Rationale generated for low-dose naltrexone (LDN), titrated to 4.5 mg nocte, introduced later in the same step of the framework:

1. TLR4 antagonism — reduction of microglial activation, proposed as relevant to the frontal PET hypometabolism and neurocognitive symptom cluster.

2. Transient opioid receptor blockade with endorphin rebound — proposed contribution to analgesia and fatigue modulation.

3. Mechanistic complementarity with rapamycin — the tool explicitly framed LDN and rapamycin as acting on distinct but intersecting neuroimmune pathways (TLR4/microglial vs mTORC1/inflammasome), rather than redundant mechanisms.

I want to be explicit that this output is a hypothesis-generation and evidence-organisation exercise, not a diagnosis. The mechanistic chain (mTORC1 → NLRP3 → SASP → mitochondrial dysfunction) is biologically plausible and consistent with published Long COVID/ME-CFS immunology, but it was not confirmed by direct assay (no IL-1β, no senescence markers, no mTOR pathway activity measured) in this patient. I validated the therapeutic proposal on the basis of the overall clinical picture and known safety profile of intermittent low-dose rapamycin, not on the strength of the mechanistic narrative alone.

Intervention and follow-up

Rapamycin 5mg weekly (slowly titrated as I explained in my blog) was initiated. LDN was added later, titrated to 4.5 mg nocte, after a staggered interval to allow attribution of effect.

Outcome measures

Outcomes were tracked using the Post-COVID Core Outcome Set (PC-COS) for adults (Munblit et al., Lancet Respir Med 2022, PMID 35714658) and a dedicated PEM subscale, at two structured virtual assessments approximately eight months apart (before initiation of rapamycin, and after ~8 months of rapamycin, prior to LDN initiation). These are automatically calculated by my virtual clinic

A few points worth flagging rather than glossing over: the neurocognitive and cardiovascular domain scores did not improve — they were essentially flat or marginally worse — despite meaningful gains in fatigue, pain, sleep, PEM frequency, and overall function. This is clinically informative rather than a weakness of the report: it suggests the therapeutic response was not uniform across pathophysiological domains, consistent with a partial rather than complete correction of the underlying process.

At approximately one month into LDN 4.5 mg (added on top of ongoing rapamycin), the patient reported — not yet captured on a repeat structured PC-COS/PEM assessment — resolution of PEM episodes, resolution of fatigue and pain, near-normalisation of sleep, and resumption of brisk walking, with persistent but now manageable cognitive symptoms. A third structured assessment has been requested to quantify this further step change formally before it is reported as data rather than patient narrative.

Discussion

This is a single, unblinded, non-controlled case with two sequential interventions (rapamycin, then LDN) introduced at different time points — it does not allow attribution of the observed improvement to either agent individually, let alone to the virtual assessment tool itself. The PC-COS and PEM instruments are patient-reported and therefore subject to recall and expectation effects; the CPET, Holter, and PET data that anchored the original mechanistic hypothesis were not repeated at follow-up, so we cannot say whether the proposed mTORC1/NLRP3 axis was actually modulated, only that the clinical trajectory was compatible with that hypothesis.

What I think is genuinely useful here is narrower than “rapamycin works” or “AI made the diagnosis”: it is that a structured tool, integrating heterogeneous data (CPET, Holter, PET, immunophenotyping, longitudinal PROMs) that would otherwise sit in separate documents across separate visits, surfaced a coherent, testable mechanistic hypothesis that a clinician could accept, reject, or modify — and in this case, after independent clinical judgement, the hypothesis was acted upon and the trajectory was favourable. The value proposition is in the structuring and pattern-surfacing, not in autonomous decision-making, and every therapeutic step described here was clinician-validated, off-label, and discussed with the patient regarding the preliminary nature of the evidence base.

I will continue to report cases like this — including ones where the tool’s hypotheses turn out to be wrong — as part of an ongoing, transparent evaluation of where this kind of structured AI-assisted assessment adds value in a condition where the diagnostic and therapeutic literature is still maturing.

Nothing in this report constitutes generalisable treatment guidance. All interventions described were individualised, off-label, discussed with the patient, and monitored under direct clinical supervision.

Leave a comment

In my clinical practice, I occasionally encounter severe, intractable post-vaccination syndromes — cases where a temporal link to vaccination is clear, standard workup identifies an autoimmune/neuroimmune substrate, and conventional management has failed. In a subset of these patients, I consider IVIG off-label, on the same clinical basis I would use it in any other severe autoimmune neurological presentation. Patients and families in this situation frequently ask me whether IVIG is “removing the spike protein,” so I want to lay out, precisely, what is and isn’t known about that specific question — separately from the much better-supported rationale for using IVIG in this population, which is autoimmune modulation, not antigen clearance. This piece is my attempt to think through both threads rigorously and transparently, including where the evidence runs out.

A note

This piece deliberately avoids the word “detox.” Detoxification implies a chelation-like removal of a fixed toxic burden — that is not an accurate description of any antibody-mediated mechanism discussed below. What follows is immunomodulation and antigen clearance biology, applied to an open question, not a validated protocol. The honest answer to “does IVIG clear persistent spike protein” is: mechanistically plausible in principle, essentially untested in this specific context, and not something to infer confidently from adjacent data.

1. The premise that needs establishing first: is there anything to clear?

Before discussing whether IVIG could clear spike antigen, it’s worth being precise about what is actually known about spike antigen kinetics after mRNA vaccination — because the answer is more nuanced than either “gone in days” or “persists indefinitely.”

Lymph node compartment. Immunohistochemical studies of axillary lymph node biopsies after BNT162b2/mRNA-1273 vaccination found spike protein present in germinal centers, peaking around day 16 post-dose 2 and still detectable at day 60. This is broadly consistent with the expected biology of a germinal center reaction — mRNA vaccines are known to drive a persistent GC response lasting several months, which is the mechanistic basis for the affinity-matured, long-lived antibody responses these vaccines produce. In other words, some of this “persistence” is the vaccine working as intended, not a pathological finding.

Plasma/circulating compartment. Using single-molecule array assays, one study detected mRNA-1273-derived spike protein in plasma up to 28 days post-vaccination, with peak levels appearing within the first week. A separate line of work found spike protein on circulating exosomes starting around day 14 post-vaccination, with detectable exosomal spike persisting out to roughly 4 months before declining in parallel with antibody titers — the authors framed this as a plausible antigen-presentation mechanism supporting the immune response, not as an adverse finding.

What this does not establish: none of these studies were designed to detect indefinite persistence, nor do they report ongoing detectable antigen at 6+ months in asymptomatic vaccinees. The data describe a self-limited kinetic curve — antigen appears, peaks, and declines over weeks to a few months — coincident with a normal immune response, not evidence of a stable long-term antigen depot requiring therapeutic clearance. Extrapolating “spike is detectable at 2 months” into “spike persists indefinitely and causes chronic illness” is not something these datasets support.

Where the stronger persistence data actually comes from: post-viral persistence findings (spike/N-protein detected in tissue or plasma months after natural infection) are far better characterized in Long COVID cohorts than in vaccine-only cohorts, and involve a fundamentally different exposure — active viral replication across multiple tissue compartments, not a self-limited translation event at the injection site and draining node. Conflating the two evidence bases is one of the most common errors in this space, in both directions (people who minimize post-infectious persistence, and people who over-extrapolate it onto vaccination).

2. If antigen is present, can circulating IgG actually clear it? Three mechanisms, three big caveats

2a. Direct neutralization

Anti-RBD/NTD IgG blocking ACE2 engagement is well established for preventing new cellular entry. This is not clearance of already-produced antigen — it prevents infection, which is irrelevant to endogenously translated vaccine antigen that never needs to bind ACE2 to exist in the first place. This mechanism is essentially a non-sequitur for the “detox” framing and is worth stating explicitly, because it’s the mechanism people most often gesture at when they mean something else.

2b. Fc-mediated effector clearance (ADCC / ADCP / CDC)

This is the mechanistically relevant category:

• ADCC — FcγRIIIa engagement on NK cells drives lysis of antibody-opsonized, antigen-expressing cells

• ADCP — Fcγ-receptor-mediated phagocytosis of opsonized free antigen or antigen-expressing cells by macrophages/monocytes

• CDC — classical complement cascade activation via IgG Fc, leading to membrane attack complex formation

These pathways are well validated for other cell-surface-expressed or secreted, circulating antigens (this is the basis of most antibody-drug and monoclonal antibody oncology mechanisms, and is invoked in chronic viral antigenemia contexts such as HBsAg). The open question is whether they are quantitatively relevant to whatever residual spike is present after vaccination, given the exposure is small in magnitude and largely intracellular/nodal rather than surface-expressed on abundant, accessible cells.

2c. Anti-idiotype network modulation

IVIG’s anti-idiotypic antibodies can, in principle, modulate an existing autoantibody or antigen-specific antibody response. This is a real and described IVIG mechanism in autoimmune disease, but it’s a mechanism for modulating the immune response to an antigen, not for physically removing the antigen itself — worth distinguishing carefully in the write-up so readers don’t conflate “immunomodulation” with “clearance.”

3. The three gaps that determine whether 2b is actually operative here

Gap 1 — Compartment accessibility. Most of the antigen described in the lymph node studies above is intracellular (translated within transfected cells) or trapped on follicular dendritic cell networks within germinal centers — a compartment that is relatively immune-privileged from circulating antibody access by design (this is how germinal centers work; if circulating antibody freely accessed and cleared everything inside a GC, affinity maturation itself would be disrupted). Only the fraction that is secreted, exosomal, or expressed on an accessible cell surface is plausibly reachable by IVIG’s effector mechanisms. The exosome and plasma-antigen data above suggest this accessible fraction exists and is measurable — but its magnitude relative to total antigen produced, and whether it is rate-limiting for any downstream pathology, is unknown.

Gap 2 — Duration mismatch. The best available human kinetic data put detectable circulating/exosomal spike antigen in a window of roughly 2 weeks to ~4 months post-vaccination in the studies cited above, declining in parallel with the normal antibody response — i.e., largely resolving on its own, via the same effector mechanisms (endogenous IgG, complement, phagocytes) that would be recruited by exogenous IVIG. This matters for the therapeutic argument: if endogenous antibody already mediates this decline as part of normal immune resolution, the incremental value of exogenous pooled IgG is unclear without a comparator.

Gap 3 — Is IVIG’s anti-spike content therapeutically relevant? IVIG is a polyclonal pool from thousands of donors (vaccinated, infected, or both), and anti-spike titer varies by lot. In vitro, pooled IVIG from vaccinated-donor plasma does neutralize spike-pseudotyped virus — that’s well established. What’s not established is (a) whether typical anti-spike IgG concentration in a standard IVIG dose is sufficient to drive meaningful ADCC/ADCP against whatever antigen-expressing cells or exosomes exist in a given patient, and (b) whether Fc-effector potency (not just binding/neutralizing titer) has been characterized for IVIG-derived anti-spike specifically — this is a distinct property from neutralization capacity and isn’t reported in the neutralization-focused studies.

4. Where IVIG’s use is actually clinically grounded in this space

Setting aside the antigen-clearance hypothesis, there is a real and separate evidence base worth citing accurately: IVIG in rare post-vaccination autoimmune neurological syndromes, used the same way it’s used in any acute autoimmune neuropathy — Guillain-Barré syndrome, CIDP, and small fiber neuropathy have all been reported post-COVID-vaccination, treated with IVIG per standard autoimmune-neuropathy protocols, with case reports and small series showing symptomatic improvement (including one case with skin-biopsy-confirmed normalization of epidermal nerve fiber density after IVIG in FGFR3-antibody-associated small fiber neuropathy). A larger case-series review found GBS was the most common post-vaccination PNS manifestation reported, with IVIG and steroids as the standard treatment approach used.

This is the scientifically solid ground: IVIG treats autoimmune-mediated neurological injury that may be triggered in the post-vaccination period, via its established immunomodulatory mechanisms (Fc receptor blockade, anti-idiotype modulation, complement scavenging, Treg modulation). This is mechanistically distinct from, and much better supported than, “IVIG clears persistent spike antigen as the treatment target.”

4b. What does this mean for IVIG in Long COVID specifically? A more honest version of the argument

Antigen persistence is, on balance, better documented after natural infection than after vaccination alone — active viral replication across multiple tissue compartments is a fundamentally larger and more sustained exposure than a self-limited translation event at the injection site and draining node. It would be tempting to conclude that this strengthens the case for IVIG in Long COVID specifically. The honest picture is more mixed, and worth presenting as such:

The case that seemed to support it: Swank et al. (2022) found spike protein in plasma in the majority of Long COVID patients up to a year post-diagnosis, essentially absent in patients who had recovered from typical acute COVID — proposed as evidence for a persistent viral reservoir driving ongoing symptoms.

The most recent, and largest, data complicate this: a 2026 longitudinal analysis from the ORCHESTRA post-COVID cohort (Coppens et al.) measured free spike and nucleocapsid antigenemia out to 18 months in Long COVID patients versus recovered controls. Low-level antigen persistence was detectable in a subset of participants at 18 months, but critically, neither free spike nor nucleocapsid levels differed between Long COVID patients and recovered controls, nor were they independently associated with post-COVID condition. Early correlations between spike and inflammatory cytokines were interpreted by the authors as more consistent with transient, host-specific immune activation than with an ongoing antigen-driven process — and the group also raised a methodological concern that some antigen-recovery assay conditions may generate artifact rather than true immune-complexed antigen.

Where this leaves the field, honestly: the persistent-antigen-as-driver-of-Long-COVID hypothesis is not resolved — it’s actively being tested in real time. A UCSF-led trial (Henrich et al., using the antiviral ensitrelvir) is built explicitly on the premise that persistent viral infection and detectable circulating spike protein may be a pathogenic mechanism in a subset of Long COVID — that trial wouldn’t exist if the question were settled, in either direction. My reading is that the population-level data (like ORCHESTRA) argue against spike antigenemia as a general, unifying driver of Long COVID, while leaving open — and this is the point worth stressing — that it could still be operative in a biologically distinct subgroup, simply diluted to non-significance in an unselected cohort.

The subtype argument — where IVIG’s rationale might actually strengthen, not weaken: if Long COVID is, as much of the mechanistic literature increasingly suggests, a heterogeneous syndrome with multiple convergent phenotypes (autoimmune/neuroimmune, dysautonomic, microclot/vascular, viral-persistence-driven, post-exertional-malaise-dominant), then a population-level null association between antigenemia and PCC status doesn’t rule out a real, clinically meaningful antigen-driven mechanism in the subset of patients who actually have detectable antigen and a compatible phenotype. This is the same logic that applies to essentially every stratified-medicine argument: an unselected-cohort null result does not equal “the mechanism is irrelevant to everyone” — it can equally mean “the mechanism is real but only in a minority, and cohort-level statistics average it away.” In that specific, antigen-positive, phenotypically-compatible subgroup, the Fc-effector mechanisms discussed in Section 2b remain mechanistically live, and would represent a more defensible, biomarker-anchored rationale for considering IVIG than treating “Long COVID” as a monolithic antigen-driven condition. The practical implication is that antigen status (via ultrasensitive plasma assays or exosomal spike, where available) is a more honest patient-selection criterion for this specific rationale than clinical diagnosis alone — and that a positive antigen result in a matched patient, even if not present in Long COVID generally, is not something the ORCHESTRA null result actually contradicts.

What this doesn’t license: it does not license treating every Long COVID patient as though antigen persistence is established and driving their illness, nor does it convert the Fc-effector/antigen-clearance hypothesis from Section 2-3 into something validated. It’s a subgroup hypothesis resting on plausible immunology, not a population-level finding.

5. What would actually need to be shown to support the antigen-clearance hypothesis

For editorial honesty, and to make the piece useful rather than just skeptical, worth ending on a testable framework:

1. Baseline antigen documentation — demonstrate detectable circulating spike antigen (plasma S1/exosomal spike via single-molecule array or equivalent ultrasensitive assay) in a symptomatic post-vaccination cohort, not just descriptively but with a matched asymptomatic-vaccinee comparator

2. Pre/post-IVIG antigen kinetics — serial antigen measurement before and after IVIG administration in antigen-positive patients, to see if clearance accelerates beyond the natural decline curve

3. Fc-effector potency characterization — in vitro ADCC/ADCP assays using patient-relevant IVIG lots against antigen-expressing reporter cells, rather than relying on neutralization titer as a proxy

4. Clinical correlation — whether antigen clearance kinetics (if accelerated) track with symptom trajectory, independent of IVIG’s known immunomodulatory effects on the autoimmune process itself (a major confounder — you cannot attribute clinical improvement to “antigen clearance” when IVIG is simultaneously modulating the autoimmune response through unrelated mechanisms)

Without disentangling point 4, any clinical improvement after IVIG in this population is uninterpretable with respect to the antigen-clearance hypothesis specifically — it could equally, and more parsimoniously, reflect the well-established autoimmune-modulatory mechanisms already described in Section 4.

Conclusion

The mechanistic building blocks for antibody-mediated spike clearance exist in immunology generally, but the specific claim — that IVIG resolves a clinically meaningful, persistent vaccine-antigen burden — has not been tested. This is why, in my own practice, I frame off-label IVIG in severe post-vaccination syndromes around its established autoimmune-modulatory mechanisms, not primarily around antigen clearance — and why I think the field owes patients a clearer answer on the latter, one way or the other.

Reference list

• Lymph node spike/mRNA persistence: immunohistochemistry studies of axillary node biopsies post-BNT162b2/mRNA-1273 (spike detectable to ~day 60; germinal center biology context)

• Germinal center persistence as normal mRNA vaccine biology (Nature 2021/2022 GC response papers)

• Ogata et al., Clin Infect Dis 2022 — circulating S1 antigen post-mRNA-1273, single-molecule array

• Bansal et al., J Immunol 2021 — circulating exosomal spike protein kinetics post-BNT162b2, ~4 month decline

• Case reports/series: IVIG in post-COVID-vaccination GBS, CIDP, small fiber neuropathy (FGFR3-antibody-associated case with biopsy-confirmed nerve fiber density normalization)

• Review of post-vaccination PNS manifestations (GBS most common, IVIG/steroids as standard treatment)

• Swank et al., 2022 — circulating spike protein detectable up to 12 months in Long COVID plasma, largely absent in recovered controls (viral reservoir hypothesis)

• Coppens et al., 2026 — ORCHESTRA post-COVID cohort, longitudinal spike/nucleocapsid antigenemia to 18 months; no independent association with PCC status at cohort level; methodological caveat on antigen-recovery assay artifact

• Henrich T. et al. — UCSF/Shionogi trial (ensitrelvir) targeting persistent viral infection/circulating spike as a candidate Long COVID mechanism, ClinicalTrials.gov NCT06161688

“When will we finally have a biomarker for Long COVID or ME/CFS?”

My answer is increasingly becoming: maybe we’re asking the wrong question.

Medicine loves biomarkers because they simplify complexity. But Long COVID and ME/CFS are not single diseases—they are clinical syndromes that encompass multiple biological pathways.

But even worse, think about the diversity of patients themselves we see:

• children and older adults;

• people ill for months and others for decades;

• women and men;

• patients with post-exertional malaise (PEM) and others without it;

• some patients come to our attention during PEM, other out of PEM during strict pacing

• PEM triggered by minimal exertion in some, by different mechanisms in others;

• autoimmune diseases, connective tissue disorders, viral persistence, dysautonomia, mast cell disorders, and many other comorbidities.

Should we really expect one blood test to explain all of them?

Probably not.

Many people also hope that a biomarker will reveal the root cause of the disease. But that’s not how medicine usually works.

Take pneumonia.

Even X-ray is far to be a good biomarker. A chest CT is one of our best diagnostic biomarkers, but we can’t do it routinely due to costs and radiation. In addition, it tells us there is lung consolidation. But it does not tell us why.

Is it bacterial?
Viral?
Aspiration?
Fungal?
Cancer masquerading as pneumonia?

The image is only the beginning. The diagnosis still depends on clinical reasoning, microbiology, epidemiology, and patient context.

Long COVID and ME/CFS are likely no different.

Basic science will uncover the biological mechanisms. It may identify multiple pathways rather than one universal cause. Those discoveries are essential.

But clinical medicine will still require physicians to integrate symptoms, history, examination, imaging, laboratory findings, and emerging biomarkers into a coherent diagnosis for the individual patient.

Medicine has always been the science of uncertainty combined with the art of integration.

Perhaps the future is not one biomarker.

Perhaps the future is learning how to combine many imperfect signals into better care.

Patients also ask me when we have a biomarker in order to guide us toward the best treatment.

However, the search for biomarkers should not replace the search for understanding.

Patients don’t need a magic biomarker. They need better biology—and better clinical medicine.

So what should patients realistically expect?

In everyday clinical practice, the diagnosis of Long COVID or ME/CFS will likely remain primarily clinical, just as it is for many other complex chronic conditions.

That does not mean abandoning science. Quite the opposite.

Clinicians should continue to exclude alternative—and especially treatable—conditions that can mimic or coexist with these illnesses. Endocrine disorders, autoimmune diseases, sleep disorders, anemia, nutritional deficiencies, cardiac disease, neurological disorders, and many others still deserve careful evaluation.

As research advances, we will probably identify new biomarkers . Some may help define biological subgroups. Others may predict prognosis or treatment response. A few may eventually become routine clinical tests. And most of them anyway will not be translated to every patient due to their diversity.

But no single biomarker is likely to replace thoughtful clinical assessment.

The future of care is far more likely to be personalized medicine than one-size-fits-all diagnostics.

Each patient will probably be assessed through a combination of clinical phenotype, symptom trajectory, functional status, laboratory findings, imaging, autonomic testing, immunological and molecular data, and perhaps biomarkers that identify specific treatable pathways.

The goal is not to fit every patient into one biological box.

The goal is to understand which mechanisms are driving disease in that individual, exclude conditions that require different treatments, and tailor therapy accordingly.

This is how medicine has always evolved.

Basic science discovers mechanisms.

Translational research transforms those discoveries into clinical tools.

And clinicians integrate all of that knowledge to care for the person sitting in front of them.

That last step has never been replaced by a single test—and I don’t think Long COVID or ME/CFS will be the exception.

The diagnosis will likely remain primarily clinical for the foreseeable future.

That means clinicians should continue to make careful clinical diagnoses, exclude alternative and treatable conditions, identify the dominant phenotype in each patient, and individualize management using the best available evidence.

Waiting for a perfect biomarker before acknowledging or treating these patients is not evidence-based medicine—it is therapeutic paralysis.

Medicine has never waited for complete mechanistic understanding before treating disease. We treated hypertension long before we understood all of its molecular pathways. We treated heart failure before we fully understood cardiac remodeling. We still treat many autoimmune and inflammatory diseases with therapies whose mechanisms are only partially understood.

Long COVID and ME/CFS should not be held to a different standard.

Until disease-modifying therapies are identified through rigorous research, clinicians will inevitably have to make thoughtful, individualized decisions using the available evidence—including, where appropriate, carefully selected off-label treatments after a balanced discussion of the potential risks, uncertainties, and expected benefits with patients.

That is not abandoning evidence-based medicine.

It is practicing medicine in the face of uncertainty, while continuing to generate better evidence.

The definitive treatments that we all hope for will almost certainly come from basic and translational research. But they are unlikely to arrive in the immediate future.

Our patients cannot wait for perfect science before receiving compassionate, rational, and individualized care.

Leave a comment

The Protocol was developed in response to unapproved, unvalidated protocols being offered to patients with Long COVID and ME/CFS — in particular programs combining double filtration plasmapheresis (DFPP) with infusion of the secretome of stem-cell-derived human exfoliated deciduous teeth (SHED). These programs, offered at facilities outside the EU at high cost, promise two biological effects: removal of circulating inflammatory mediators and regenerative support via paracrine growth factors.

Neither component is validated by randomized controlled trials for Long COVID or ME/CFS, and the absence of regulatory approval anywhere in the world demands extreme caution. This protocol proposes an alternative pathway, built entirely on lifestyle interventions and low-risk supplementation, that acts on the same biological targets — low-grade systemic inflammation, mitochondrial function, immune regulation, autophagy — through approved physiological mechanisms supported by the scientific literature, even if much of it remains preliminary.

This is NOT a replacement for individualized medical therapy (LDN, immunomodulators, IVIG in selected cases, rehabilitation). It is a foundational package — a “reinforced Step 0” — to be offered to all patients before or alongside pharmacological interventions, and to be presented explicitly as a rational alternative to patients considering unvalidated, costly protocols abroad.

Five biological targets, five pillars

The protocol is organized into five pillars, each targeting a pathophysiological mechanism relevant to Long COVID/ME-CFS:

• Pillar 1 — Nutrition and therapeutic fasting: activation of autophagy, reduction of IL-6/TNF-α, neuroprotective ketogenesis.

• Pillar 2 — Targeted supplementation: mitochondrial support (CoQ10, NAC, ALA), immunomodulation (vitamin D), tissue repair.

• Pillar 3 — Sleep and chronobiology: restoration of cortisol rhythm and melatonin-dependent nocturnal cytokine suppression.

• Pillar 4 — Pacing-friendly mind-body movement: neuro-immune modulation via yoga/pranayama, without risk of post-exertional malaise (PEM).

• Pillar 5 — Photobiology: controlled sun exposure for vitamin D and possible UV-dependent immunomodulatory effects.

1. General Principles of Application

1.1 Mandatory preliminary stratification

Before starting, every patient should be classified by:

• Predominant clinical phenotype: inflammatory/autoimmune, dysautonomic, neurocognitive, mixed.

• Severity and presence of PEM (post-exertional malaise) — determines the permissible intensity of Pillar 4.

• Nutritional status and weight — contraindicates or modulates Pillar 1 (see §2.5).

• Metabolic, renal, hepatic, and cardiovascular comorbidities — modulate Pillar 2 dosing.

• Adherence and family/social support — shapes the introduction sequence.

1.2 The pacing rule: never at the expense of energy

The guiding principle across all five pillars is that no intervention should induce PEM. Each component should be introduced at low intensity and titrated upward only in the absence of symptomatic worsening over the following 48-72 hours (the “48-hour rule”).

1.3 Recommended introduction sequence

Do not introduce all pillars simultaneously: doing so makes it impossible to attribute any effects (positive or negative) to a specific component and increases the risk of poor adherence. Recommended sequence over 8 weeks:

2. Pillar 1 — Nutrition and Therapeutic Fasting

2.1 Rationale

Fasting and caloric restriction activate autophagy (via mTOR/AMPK), reduce circulating pro-inflammatory cytokines, and increase the NAD+/NADH ratio with consequent sirtuin activation. The metabolic shift toward lipid oxidation and ketogenesis also provides an alternative energy substrate (beta-hydroxybutyrate) with anti-inflammatory and potentially neuroprotective properties, relevant to the brain fog associated with cerebral energy metabolism dysfunction.

2.2 Time-Restricted Eating (TRE) — baseline protocol

2.3 Structured intermittent fasting (advanced, optional)

To be considered only after stabilization with TRE, in motivated patients without contraindications:

• 5:2 fasting — 2 non-consecutive days/week at ~500-600 kcal, 5 days of balanced free eating.

• Modified Alternate Day Fasting (ADF) — alternate days at ~25% of caloric requirement; reserved for patients with good functional reserve.

• Supervised prolonged fasting (Buchinger-Wilhelmi model, 5-10 days) — clinical-setting only, with monitoring of electrolytes, glucose, and weight; not to be done at home without medical supervision.

2.4 Ketogenic component (optional, second line)

A moderate ketogenic diet (fat to carbohydrate+protein ratio ~1.5-2:1, not classic 4:1 ketogenic) may be considered in patients with a predominant neurocognitive phenotype, to sustain endogenous ketogenesis even outside fasting windows. Avoid in patients with severe dysautonomia (risk of orthostatic intolerance from sodium depletion) without adequate electrolyte supplementation.

2.5 Contraindications to Pillar 1

• BMI <18.5 or history of an eating disorder (absolute contraindication).

• Unintentional weight loss >5% in the past 3 months.

• Type 1 diabetes, insulin therapy, or recurrent hypoglycemia.

• Pregnancy, breastfeeding, growing children/adolescents (use only mild TRE 12:12, never prolonged fasting).

• Severe uncompensated dysautonomia — introduce only after optimizing hydration/salt intake.

3. Pillar 2 — Targeted Supplementation

Supplementation should always be preceded by baseline testing (25-OH-vitamin D, ferritin, B12, folate, renal/hepatic function) and personalized. The doses below are indicative starting points, not fixed prescriptions.

4. Pillar 3 — Sleep and Chronobiology

Dysregulation of clock genes (CLOCK, BMAL1, PER2) and reduced melatonin contribute to a persistent pro-inflammatory environment in Long COVID, with loss of the normal cortisol rhythm that physiologically suppresses nocturnal inflammatory signaling.

4.1 Structured sleep hygiene protocol

• Fixed wake time, 7 days a week (±30 minutes), regardless of sleep quality.

• Natural light exposure within 30-60 minutes of waking, at least 10-15 minutes (see also Pillar 5).

• Screens/blue light off at least 60 minutes before bedtime; warm light <3000K in the evening.

• Bedroom temperature 18–19°C; complete darkness.

• No caffeine after 2:00 PM; avoid or limit alcohol (fragments sleep).

• Last meal 3 hours before bedtime (integrates naturally with the TRE window in Pillar 1).

4.2 Melatonin and chronotherapy

Melatonin 0.5-3 mg in the evening as per the table in §3, with possible added benefit from brief morning bright-light exposure (10,000 lux, 20-30 minutes) in patients with delayed sleep phase, to reinforce circadian anchoring.

5. Pillar 4 — Pacing-Friendly Mind-Body Movement

Yoga, pranayama, and meditation show heterogeneous but, in several studies, favorable effects on IL-6, cortisol, TNF-α, and CRP, alongside consistent benefits on fatigue, anxiety, sleep quality, and quality of life in post-COVID populations. The heterogeneity of biomarker findings means this pillar should be presented as symptomatic and autonomic-regulation support, rather than as a guaranteed anti-inflammatory intervention.

5.1 Graduated protocol (strict adherence to pacing)

6. Pillar 5 — Photobiology and Sun Exposure

Beyond vitamin D synthesis, cutaneous UV exposure has been associated with immunomodulatory effects partly independent of vitamin D (e.g. via cutaneous nitric oxide release); specific evidence in Long COVID/ME-CFS remains very limited and should be communicated as a hypothesis, not a well-established benefit.

• Direct sun exposure of face/forearms/legs, 10-20 minutes, during lower-UV-risk hours (early morning or late afternoon in summer; midday in winter at Italian latitudes), 3-5 times/week.

• Avoid sunburn; adjust duration to skin phototype, season, and latitude.

• Does not replace vitamin D supplementation during the autumn-winter months at Italian latitudes, where cutaneous synthesis is negligible.

7. Monitoring and Outcomes

7.1 Recommended outcome measures

7.2 Stopping or modification criteria

• Onset or worsening of PEM attributable to a specific component → suspend that component and return to the previous phase.

• Unintentional weight loss >3% in 4 weeks → suspend Pillar 1, nutritional reassessment.

• Sleep worsening with melatonin → reduce dose or discontinue, rule out other causes.

• No measurable improvement across all domains at 12 weeks → reassess the overall plan, consider pharmacological intensification per the Step 1-5 framework.

8. Communicating with the Patient: Why This Approach Instead of the Stem-Cell/DFPP Protocol

Key points to convey during the clinical consultation:

• The Buonsenso Protocol targets the same biological pathways (inflammation, mitochondrial function, immune regulation) through documented physiological mechanisms, without the procedural risks of plasmapheresis and infusion of unregulated products.

• It carries no prohibitive costs, requires no international travel, and is compatible with the pacing required in the presence of PEM — unlike intensive 2-week in-facility programs.

• It is grounded in published, peer-reviewed literature, though partly preliminary; this should be stated honestly to the patient, distinguishing solid data (e.g. fasting and autophagy, vitamin D and regulatory T cells) from mechanistic hypotheses still awaiting confirmation (e.g. vitamin D-independent UV photobiology).

• It does not exclude, and in fact integrates with, pharmacological therapy per the stepwise framework (pacing/diet → LDN → aripiprazole/lithium → IVIG → rapamycin → daratumumab) in patients who require it.

9. Addendum to Pillar 2 — Phytocompounds and Endothelial Cofactors: Dosing and Safety

This section extends Pillar 2 with additional components whose primary rationale relates to endothelial function (NO-mediated vasodilation, arterial stiffness) beyond general control of systemic inflammation. These are presented as optional add-ons to personalize, not as a mandatory package. For each component, the table lists the dose commonly used in clinical studies (not necessarily matching the dose found in any single commercial product, which is often sub-therapeutic relative to trial doses), the upper safety limit where defined, and the main effects of overdose or chronic excess.

Many multi-component commercial products contain doses of curcumin, resveratrol, and quercetin lower than those used in positive clinical trials (e.g. 15-20 mg of curcuminoids versus 500-2000 mg in endothelial function trials): this does not make them harmful, but the expected efficacy ceiling should be acknowledged, distinguishing the biological rationale from the clinical benefit realistically achievable at the doses actually administered.

Cross-cutting interactions to monitor: curcumin, resveratrol, vitamin E, and quercetin all have some antiplatelet/anticoagulant-potentiating activity — combining several of these components simultaneously in patients already on anticoagulant or antiplatelet therapy requires medical supervision and closer monitoring (e.g. INR if on warfarin).

10. Essential Bibliography

Scoping review on fasting and caloric restriction in Long COVID Syndrome. Nutrition Reviews, 2025 (Oxford Academic, DOI: 10.1093/nutrit/nuaf141).

Improvements during prolonged fasting in Long COVID patients — case series and literature review. Frontiers in Nutrition, 2023.

Gnoni M, Beas R, Vásquez-Garagatti R. The role of intermittent fasting on preventing COVID-19 and clinical outcomes: intersection of inflammation, mTOR pathway, autophagy and caloric restriction. Virus Reviews, 2021.

Vitamin D in infectious diseases: narrative review on COVID-19, Long COVID, and influenza. Nutrients, 2026 (DOI: 10.3390/nu18040634).

Vitamin D for COVID-19 Trial — randomized study on vitamin D supplementation and COVID-19/Long COVID outcomes. ScienceDirect, 2026.

CORONAVIT Trial — ‘test-and-treat’ vitamin D supplementation in 6200 adults, respiratory infection outcomes. Cited in: Vitamin D and COVID-19: Clinical Evidence and Immunological Insights, Life (MDPI), 2025.

Shah et al. Yoga, meditation, breathing exercises, and inflammatory biomarkers with possible implications in COVID-19: systematic review and meta-analysis of RCTs. Evidence-Based Complementary and Alternative Medicine, 2022.

Yoga versus health education for persistent fatigue in post-COVID-19 syndrome: multicenter RCT protocol. PMC, 2025.

Post-infectious fatigue and circadian rhythm dysregulation in Long COVID and other infections. eClinicalMedicine, 2025.

Circadian and melatonin dysregulation as overlooked drivers of immune imbalance in post-COVID syndrome: a call for chronotherapy. PMC, 2025.

Mitochondrial dysfunction and coenzyme Q10 supplementation in post-viral fatigue syndrome: an overview. International Journal of Molecular Sciences (MDPI), 2024, DOI: 10.3390/ijms25010574.

VA Whole Health Library. Long COVID Introduction — overview of CoQ10, NAC, ALA, L-arginine, and other supplements in the post-viral setting.

Edogawa Hospital (Tokyo) informational page on the DFPP and SHED growth-factor protocol — consulted as the reference point for the unvalidated protocol discussed here (edogawadfpa.com).

Prepared as an internal clinical working document. To be updated periodically with new literature and adapted based on monitoring outcomes in treated patients.

Below is a leaflet for patients

Read more

Comprehensive Technical & Regulatory Analysis for Healthcare Professionals

1. INTRODUCTION AND CLINICAL CONTEXT

The McCairn-Edogawa Protocol is an investigational two-phase treatment developed by neuroscientist Kevin McCairn, PhD, in collaboration with the clinical team at Edogawa Hospital in Tokyo. It targets patients with Long COVID and Post-Vaccine Syndrome (PVS) suspected to be driven by fibrinaloid microclots and persistent circulating spike protein.

According to the hospital’s official program website, the pathway is presented as a “hospital-led clinical pathway” combining physician assessment, dual filtration plasmapheresis (DFPP/DFPA), laboratory analytics, and clinician-directed regenerative support. The program explicitly states that “treatment is not offered automatically to every patient” and that risks, alternatives, expected treatment course, and limitations are discussed during consultation before any final recommendation is made.

Regulatory status: Not approved by the FDA, EMA, or any European national regulatory agency for Long COVID or PVS. Not evaluated in randomized controlled trials. Currently administered to a limited number of patients on a private basis in Tokyo, at estimated costs of $20,000–$40,000.

----

2. PATHOPHYSIOLOGICAL BASIS

2.1 Fibrinaloid Microclots

The structures identified by Pretorius and Kell are not classical thrombi but amyloid aggregates of fibrinogen/fibrin with the following characteristics:

Clinical implication: Standard coagulation tests do not detect these structures. Diagnosis requires fluorescence microscopy with Thioflavin T on citrated plasma.

2.2 Amyloidogenic Spike Protein

Swedish studies have demonstrated that the spike protein contains 7 sequences with amyloidogenic potential. Exposure to neutrophil elastase cleaves spike into sequences forming amyloid fibrils, with one specific fragment inducing clots highly resistant to plasmin. A cross-seeding mechanism propagates misfolding to other proteins.

----

3. PHASE 1: DOUBLE FILTRATION PLASMAPHERESIS (DFPP)

3.1 Technical Principle

DFPP, introduced by Agishi et al. in 1980, is a selective two-stage apheresis technique using two membrane filters in series. Unlike standard Therapeutic Plasma Exchange (TPE)—which discards all plasma and replaces it with albumin or fresh frozen plasma—DFPP is selective and preserves most useful proteins.

3.2 Extracorporeal Circuit and Filters

First Stage: Plasma Separator

The 0.3 μm pore allows passage of all plasma proteins while retaining cellular components.

Second Stage: Plasma Fractionator

The second filter has a molecular weight cut-off typically in the 100–200 kDa range, removing immunoglobulins, fibrinogen, immune complexes, LDL, VLDL, Lp(a), cytokines, and autoantibodies, while returning albumin and smaller proteins.

3.3 Operating Parameters

Machines used: Plasauto EZ (Asahi Kasei Kuraray Medical) or equivalents.

3.4 Vascular Access and Anticoagulation

• Dual-lumen central venous catheter (CVC) mandatory; preferred site is internal jugular vein

• Anticoagulation: Systemic heparin or regional citrate; ACE inhibitors within 24 hours are a relative contraindication

3.5 Documented Effects

Hemorrhagic risk: ~14% of patients experience bleeding events related to fibrinogen depletion.

----

3.6 DFPP: APPROVED INDICATIONS AND ESTABLISHED USES WORLDWIDE

DFPP is not an experimental procedure in all contexts. It has a substantial history of clinical use, particularly in Asia, with well-documented indications. However, its approval is highly jurisdiction-specific.

Japan: The Most Comprehensive Regulatory Framework

In Japan, DFPP has been covered under national public health insurance since 1984 and is an established standard of care for numerous conditions. The Japanese Ministry of Health, Labour and Welfare (MHLW) maintains a detailed reimbursement schedule specifying approved diseases, treatment frequency limits, and clinical eligibility criteria.

DFPP-Reimbursed Indications in Japan (as per MHLW schedule):

Reimbursement: In Japan, DFPP is reimbursed at approximately ¥108,000–¥123,000 per session (roughly $700–$800) under the national health insurance system, with specific procedural and material fees itemized.

Other Asian Countries

DFPP is also established in South Korea, Taiwan, and parts of Southeast Asia for similar autoimmune and metabolic indications, though with less systematically documented reimbursement frameworks than Japan.

Europe and the United States

TPE (not DFPP) is FDA-recognized by ASFA (American Society for Apheresis) for: Guillain-Barré Syndrome, Myasthenia Gravis crisis, Chronic Inflammatory Demyelinating Polyneuropathy, Thrombotic Thrombocytopenic Purpura, Goodpasture’s syndrome, and others.

Additional Documented Uses in Peer-Reviewed Literature

Beyond the Japanese reimbursement schedule, DFPP has been reported in clinical studies for:

3.7 REGULATORY ANALYSIS — PHASE 1: DFPP FOR LONG COVID/PVS

Can DFPP be performed off-label in Europe or the USA for Long COVID/PVS?

United States:

The situation is complex and legally ambiguous:

1. The devices are FDA-cleared: The Plasmaflo OP plasma separator holds 510(k) clearance (P820033) as a Class II medical device for plasma separation.

   This means the device itself is legally marketed for general plasma separation.

2. However, DFPP as a procedure is not FDA-approved, nor is it assigned specific CPT codes that insurers recognize for autoimmune or neurological indications. Hospitals performing DFPP in the US typically do so within research protocols or under the broader umbrella of “therapeutic apheresis” codes.

3. Off-label use of a cleared device: Under FDA regulations, physicians may use cleared medical devices for off-label indications provided the use does not violate the device’s labeling and the physician follows standard of care. However, DFPP requires specific training, disposable sets (Cascadeflo EC), and protocols that most US centers do not have.

4. For Long COVID/PVS specifically: There is no FDA approval, no CPT code, and no insurance coverage. A US physician could theoretically prescribe DFPP off-label using a cleared plasma separator, but:

   • Most hospitals lack the specific DFPP disposable sets and trained personnel

   • Insurance would not reimburse

   • The hospital’s risk management and IRB would likely block the procedure for an unapproved indication without a formal clinical trial

   • The Cochrane review (2023) explicitly recommends against administering apheresis for Long COVID outside RCTs

Conclusion for USA: DFPP is not practically available for Long COVID/PVS. While the component device has 510(k) clearance, the complete DFPP procedure for this indication is not standardizable, not reimbursable, and not supported by institutional protocols outside formal research.

Europe (EU/EEA):

1. Device level: Plasma separators and fractionators may hold CE marking under the Medical Devices Regulation (MDR) or previous MDD. This allows their marketing and use for general apheresis.

2. Procedure level: There is no EMA centralized approval for DFPP for any indication. Unlike TPE, which is practiced widely under national guidelines for conditions like Guillain-Barré or myasthenia gravis, DFPP is not part of standard European apheresis protocols for autoimmune or neurological diseases.

3. National discretion: Individual EU member states could theoretically permit DFPP under national health service frameworks or hospital exemption rules, but:

   • No major European apheresis society (e.g., ESA — European Society for Apheresis) lists DFPP as a standard procedure for Long COVID or PVS

   • National reimbursement codes do not exist for this indication

   • The procedure requires specific disposables and expertise that are not widely available

4. Hospital Exemption (Article 3.7 of Regulation 1394/2007): This allows non-routine preparation of ATMPs within a hospital under a physician’s exclusive responsibility. However, DFPP is not an ATMP — it is a medical procedure/device combination. The Hospital Exemption does not apply.

Conclusion for Europe: DFPP for Long COVID/PVS is not legally impossible at the device level (CE-marked components exist), but it is not practically available because:

• No standardized protocol or training exists outside Japan

• No reimbursement pathway exists

• No European professional society endorses it for this indication

• Hospitals would face insurmountable liability and governance barriers

----

. PHASE 2: STEM CELL GROWTH FACTORS (SGF)

4.1 Definition and Terminological Clarification

Critical: The protocol does NOT use live stem cells. It uses conditioned medium (CM) — or secretome — derived from SHED (Stem cells from Human Exfoliated Deciduous teeth) cultures. This is a cell-free approach.

According to the Edogawa Hospital program, “SHED refers to stem cells derived from human exfoliated deciduous teeth. Products derived from the SHED secretome may contain growth factors, cytokines, and extracellular-vesicle-associated signals that are being studied for their potential trophic and immunomodulatory properties.” The program explicitly states that this material “should not be presented as a guaranteed neurological repair or cure.”

4.2 Production Process

1. Culture: SHED cells cultivated in vitro

2. Paracrine secretion: Cells release signaling molecules into culture medium

3. Collection: Conditioned medium harvested

4. Filtration: Removal of cells and debris

5. Concentration: Fluid concentrated (often 10×)

6. Administration: Intravenous infusion

4.3 Secretome Composition

Dental pulp stem cell conditioned medium contains over 100 signaling molecules:

4.4 Proposed Mechanism of Action

The secretome acts through paracrine effects:

1. Inflammatory modulation (macrophage polarization M1→M2)

2. Vascular regeneration (VEGF/HGF)

3. Neuroprotection (BDNF, GDNF)

4. Remyelination (NRG1 via ErbB3/4)

Important note: McCairn has openly stated that the specific therapeutic molecule(s) have not been isolated. The secretome contains ~100 active molecules administered simultaneously.

4.5 Administration in the Protocol

• Route: Intravenous

• Frequency: Daily during the 2-week stay

• Number of infusions: Typically 10–14

----

4.6 REGULATORY ANALYSIS — PHASE 2: SHED-DERIVED SECRETOME

Is the SHED secretome approved for any disease?

No. Nowhere in the world is a SHED-derived secretome product approved as a pharmaceutical or biological drug.

What is the regulatory classification of a stem cell secretome?

United States:

Under FDA regulations, the SHED secretome would be classified as follows:

1. It is NOT a 361 HCT/P (Human Cells, Tissues, and Cellular and Tissue-Based Product): The four criteria for 361 regulation are:

   • Minimal manipulation ❌ (culture expansion and conditioning is more than minimal)

   • Homologous use only ❌ (dental pulp cells are not being used for dental/oral repair)

   • Not combined with another article ❌ (it is processed and concentrated)

   • No systemic effect / not dependent on metabolic activity of living cells ❌ (the intended effect is systemic and paracrine)

2. Therefore, it IS a 351 product: Regulated as a biological drug under the PHS Act and FD&C Act. This requires:

   • IND (Investigational New Drug) application before any human use

   • Full CMC (Chemistry, Manufacturing, and Controls) documentation

   • cGMP manufacturing compliance

   • Phase I/II/III clinical trials

   • eventual BLA submission and approval

3. Current FDA status: As of this writing, there is no approved IND for a SHED secretome product for Long COVID or PVS. The clinical trial registry (ClinicalTrials.gov) shows some secretome studies, but none for this specific indication with this specific product.

Europe:

Under EU Regulation 1394/2007 (ATMP Regulation):

1. The SHED secretome would likely qualify as a somatic cell therapy medicinal product or, if considered acellular but derived from cells, as a tissue-engineered product or combined ATMP.

2. Centralized authorization is mandatory for all ATMPs. No national marketing authorization is permitted.

3. The EMA Committee for Advanced Therapies (CAT) would evaluate quality, safety, and efficacy. No such evaluation has occurred for SHED secretome.

4. Hospital Exemption (Article 3.7): This allows non-routine preparation of ATMPs within a single hospital under a physician’s responsibility for an individual patient. However:

   • It is intended for non-routine use

   • It requires national competent authority authorization on a case-by-case basis

   • It does not permit commercial or semi-commercial programs

   • It would not cover a standardized protocol administered to multiple foreign patients paying out-of-pocket

Can the SHED secretome be administered off-label in Europe or the USA?

United States: Absolutely not.

• There is no approved product to prescribe off-label. Off-label prescribing applies to approved drugs used for unapproved indications (e.g., IVIG for Long COVID).

• The SHED secretome is not an approved drug. It is an unregulated biological material.

• A US physician cannot legally obtain, prescribe, or administer this material outside of an FDA-approved clinical trial with an active IND.

• Compounding pharmacies cannot legally produce it, as it is not derived from an approved source material and does not meet USP standards.

• Importation for personal use is restricted and would not apply to a hospital-administered biological.

Europe: Essentially impossible.

• There is no approved ATMP to prescribe off-label.

• The Hospital Exemption is the only theoretical pathway, but it is designed for individual compassionate use, not for a commercial clinical program.

• A European hospital could not legally import and administer a Japanese-produced secretome to paying foreign patients under current ATMP regulations.

• National competent authorities (NCAs) would not authorize this under Hospital Exemption because:

  • The product is manufactured outside their jurisdiction

  • The use is not “non-routine” if part of a standardized protocol

  • There is no GMP-compliant manufacturing documentation recognized by the EU

Conclusion for both jurisdictions: The SHED secretome phase of the McCairn-Edogawa Protocol is completely non-replicable in the US or EU. Unlike IVIG — which is an approved biological product that physicians can prescribe off-label — there is no legal pathway to obtain, prescribe, or administer this material outside of formal clinical trials.

----

5. DIAGNOSTIC COMPONENT: MICROCLOT DETECTION

5.1 Pretorius Method (Thioflavin T Fluorescence)

The Edogawa program incorporates “fluorescence-based amyloid assessment” and “live blood and plasma imaging” as part of its patient analytics pathway.

5.2 REGULATORY ANALYSIS — DIAGNOSTICS

Is microclot detection with Thioflavin T an approved diagnostic test?

No. There is no FDA-approved or CE-IVD-marked test for fibrinaloid microclots.

United States: Under FDA policy, Laboratory-Developed Tests (LDTs) have historically been regulated under CLIA. However, the FDA has stated its intent to increase oversight of LDTs. A microclot assay would currently fall into a regulatory gray area — it could theoretically be offered as an LDT by a CLIA-certified high-complexity laboratory, but:

• No standardization exists

• No proficiency testing programs exist

• No clinical validation data meet FDA diagnostic standards

• Insurance would not reimburse

Europe: Under the In Vitro Diagnostic Regulation (IVDR 2017/746), any diagnostic test placed on the EU market must undergo CE-IVD marking with notified body involvement (for most classes). No such marking exists for microclot detection.

Conclusion: Microclot diagnostics cannot be ordered as a standard clinical test in the US or EU. They remain strictly research procedures.

6. COMPARISON WITH HELP APHERESIS

6.1 Technical Overview

HELP Apheresis (Heparin-Mediated Extracorporeal LDL Precipitation) is an alternative blood purification modality explored for Long COVID by Jaeger et al. (2023). It removes fibrinogen, LDL, cytokines, and inflammatory mediators through heparin-mediated precipitation.

6.2 REGULATORY ANALYSIS — HELP APHERESIS

Is HELP approved?

Can HELP be performed off-label for Long COVID in the US or EU?

• USA: No. The device is not FDA-cleared or marketed in the United States. A physician cannot legally obtain or use it.

• Europe: In countries where HELP is established (Germany, Austria), a physician theoretically could perform it off-label for Long COVID, as the device is approved for apheresis and the physician has discretion. However:

  • No reimbursement would be provided

  • No professional society endorses this use

  • Liability concerns would be substantial

  • The Jaeger study is a small case series, not RCT evidence

Key distinction: HELP is more established than DFPP in Europe (for dyslipidemia), but its use for Long COVID remains investigational everywhere.----

7. SAFETY AND COMPLICATIONS

7.1 DFPP Complications

7.2 Secretome Complications

• Theoretical allergic reactions to medium proteins

• Theoretical immunogenicity

• No risk of cellular embolism or tumorigenesis (cell-free)

----

8. CLINICAL EVIDENCE AND LIMITATIONS

----

9. SYNTHESIS: WHY THIS PROTOCOL IS ENTIRELY NON-REPLICABLE IN THE US AND EU

9.1 The IVIG Comparison

The user specifically referenced IVIG as an example of off-label use. This comparison is instructive but ultimately highlights why the McCairn-Edogawa Protocol is fundamentally different:

9.2 Step-by-Step Non-Replicability

9.3 Could a US or EU physician “approximate” this protocol?

No. There is no biologically equivalent substitute that is legally available.

• Substituting TPE for DFPP: TPE is available in major centers, but it is not equivalent. TPE removes all plasma proteins non-selectively and requires large-volume replacement with albumin or donor plasma. It does not provide the selective molecular filtration that is central to the DFPP rationale. A physician cannot claim to be replicating DFPP by performing TPE.

• Substituting commercial growth factors for SHED secretome: No commercial product contains this specific cocktail of ~100 molecules at these concentrations. Individual recombinant growth factors (e.g., VEGF, bFGF) are approved for specific indications but are not interchangeable with a complex secretome. Exosome products are emerging but none are approved for systemic IV use in Long COVID.

• Substituting standard coagulation tests for microclot assays: Standard tests (D-dimer, fibrinogen) do not detect microclots. There is no clinically available substitute.

10. CONCLUSIONS AND RECOMMENDATIONS

Overall Assessment

My Professional Position

1. The protocol cannot be recommended as standard of care in the absence of controlled data. In the case of IVIG, we have pilot trials, several observational reports and case reports, patients publications

2. It merits scientific attention given its biological plausibility and observational results.

3. DFPP is a known technique with documented safety for other indications, but its application to Long COVID/PVS is investigational. However, even my apheresis team is very much suspicious

4. Stem cell secretome represents an emerging field, but the specific product used in Tokyo is unregulated and unstandardized by US/EU pharmaceutical standards. In a next blog I will publish potential alternatives with similar effects

5. The priority must be the conduct of controlled clinical trials, standardization of microclot diagnostics, and isolation of active secretome components.

6. For patients in the US and EU, the protocol is currently inaccessible through any legal or clinical pathway, and no biologically equivalent substitute exists.

Future Developments

• RCTs are reportedly being planned in the USA (IRB-approved version)

• Standardization of microclot diagnostics (imaging flow cytometry)

• Proteomic studies on the secretome to isolate active factors

• Regulatory pathway development: Any future US/EU version would require full IND/BLA (US) or centralized ATMP authorization (EU)

A LAST PERSONAL NOTE FROM THE AUTHOR

Disclaimer: The following note reflects the personal experience of the physician who compiled this report and is not part of the formal clinical analysis.

In the course of researching this topic, I attempted to engage with patients, physicians, and nurses on X (formerly Twitter) who had undergone or were discussing the McCairn-Edogawa Protocol. The intent was to gather first-hand accounts, understand patient-reported outcomes, and identify any additional clinical details not available in published literature.

This effort was met with significant hostility. I was subjected to online bullying, personal attacks, and coordinated harassment. Multiple profiles that initially presented themselves as patients or healthcare professionals were subsequently identified as belonging to anti-vaccine (”no-vax”) movements, whose primary agenda was not scientific discussion but ideological advocacy. Constructive dialogue proved impossible, and the author was forced to disengage from these interactions.

This experience underscores a broader challenge in the Long COVID and Post-Vaccine Syndrome space: the legitimate scientific and clinical questions surrounding these conditions have become entangled with polarized political and ideological movements. Physicians and researchers attempting to maintain an evidence-based, neutral stance risk being attacked from multiple sides. This report was compiled with the explicit goal of scientific objectivity — neither endorsing an unproven protocol nor dismissing the biological plausibility of the underlying mechanisms — but the social media environment around this topic is often inhospitable to such nuance.

Caveat lector: Any clinician or researcher entering this space through social media channels should be prepared for a highly charged, often adversarial environment that may impede genuine scientific exchange.

Report prepared for professional use. Information on the McCairn-Edogawa Protocol is based on publicly available data, peer-reviewed literature, the Edogawa Hospital clinical program website, and preliminary clinical documentation. The protocol is not approved by European or American regulatory agencies for the indications described. As of this writing, it cannot be legally performed, prescribed, or approximated in the United States or European Union.

Leave a comment

This young man was assessed by me, as you know I am starting first with my virtual clinicl assessment available here

https://buonsenso-clinic.thekookbitcoiner.workers.dev/

The virtual clinic assessment made the following diagnosis

and suggested my classical step by step approach:

## STEP 1 — Baseline Diagnostics & Autonomic Stabilisation
## STEP 2 — Neuromodulation & Sleep Architecture
## STEP 3 — Anticoagulation & Dopamine Modulation
**STEP 4 — Immunomodulation**
**STEP 5 — Advanced Biologics / RCT**

Than, as you know, the next step was a real meeting between me and the patient to evaluate the report, and validate or modify it. Based on clinical data and the available literature I suggested the following clinical off label approach:

based on the report I think reasonable options might be: **STEP 4 — Immunomodulation**
**Agents:** IVIG; Rapamycin (mTOR inhibition)

Low IgG subclasses and persistent immune dysregulation justify IVIG trial; Rapamycin addresses potential mTOR-driven immune senescence contributing to this patient’s sustained fatigue, brain fog, and PEM (PEM score 15/36). Very low activity (3,700 steps/week) and low HRV (15–30ms) support ongoing immune-autonomic dysfunction.
**Tried/Not tried:** Not tried.
**Caution:** IVIG — thrombotic risk, renal monitoring required; Rapamycin — immunosuppression, metabolic side effects, infection risk.

Than, the patient sent me the Amatica extensive RNA panel and I linked that panel with my assessments. Here is the summary of the results of RNA-seq Panel Analysis — Amatica Health 2025/2026

Read more

Anyone working in a hospital mostly hears about the negative things. That’s normal: when something goes wrong, people write in; when things go well, people tend to just carry that feeling home and leave it there, which is understandable — nobody feels like writing an email after a visit when they can finally get back to their own life. The problem is that, over time, this imbalance tells a partial story to the people who have to decide how to organize and support a service.

For a condition like Long Covid or ME/CFS, this weighs particularly heavily. There are no real dedicated reference centers in Italy — or in much of the world. I myself see these patients for a small fraction of my working time, carved out of a clinical and research activity that covers a great deal else. If this work continues, and perhaps grows, it’s also because the institution I work in can see that it makes sense to keep doing it. And the voices that matter most in those decisions are often the patients’ own.

So here’s my request: if you were seen by me — in the Virtual Clinic or in person — and it went well for you, please take a moment, whenever suits you and with no rush, to write an email to Gemelli’s patient relations office describing your experience (EMAIL CONTACT: urp@policlinicogemelli.it). It matters even more if you’re writing from abroad, because it makes visible just how much Long Covid and ME/CFS are not only an Italian problem, and how much need there is for a reference point that, today, exists in very few other places. The same goes for anyone who had contact with the ward nurses, who often remain even more invisible than us doctors in this kind of feedback, and who deserve to be thanked when they’ve earned it.

There’s no need to write anything elaborate. Just describe, in your own words, what you experienced — whether it was entirely positive or had some rough edges too, that’s fine either way: honest feedback is worth more than a polite compliment.

One last thing, a bit funny but sincere: if you decide to write, please don’t all do it at the same time. If twenty identical emails arrived on the same day, it would look staged, and that’s not the spirit of this. Take your time, each in your own moment.

EMAIL CONTACT: urp@policlinicogemelli.it

Thank you — to whoever does this, and to whoever doesn’t. Either way, thank you for being here.

Dear all,

I am updating this post again. Since the April 2026 edition, two things have changed: a first randomized trial of intermittent fasting in Long COVID has been published in Scientific Reports (Bunker et al., 2025), and the Cell paper from Iwasaki and Putrino has fundamentally reframed how we think about autoimmunity and therapeutic targets in this disease. Both deserve integration into this framework.

But I want to begin, as always, with the ethical tension.

The Hippocratic dictum primum non nocere is foundational, but “doing nothing” is itself an action with consequences. When a child’s quality of life is comparable to that of a child with advanced cancer — and when we have a growing body of mechanistic evidence and repurposable interventions, pharmacological and not — therapeutic nihilism is a choice we need to justify, not just assume.

More than 90% of pediatric pharmacological practice is already off-label. That is not a footnote; it is the daily reality of caring for children with complex disease.

What Do We Know Mechanistically?

Long COVID and ME/CFS share several converging mechanisms: persistent viral reservoirs or antigen, immune dysregulation (particularly B-cell and mast-cell hyperactivation), mitochondrial dysfunction, disrupted autonomic nervous system regulation (especially POTS/dysautonomia), microclots and endotheliopathy, central sensitisation/neuroinflammation, and — now confirmed causally — circulating pathogenic autoantibodies targeting the locus coeruleus, thalamus, adrenal gland, and peripheral nervous system.

No single drug, and no single lifestyle intervention, addresses all of these. Which is precisely why a stepwise, combinatorial approach — pharmacological and non-pharmacological — makes sense.

Step 0 — Non-Pharmacological Foundation (New)

Before reaching for any drug, I now explicitly discuss three non-pharmacological pillars with every patient and family. These are not soft recommendations. They are mechanistically grounded interventions with emerging — and in the case of pacing, well-established — evidence.

0a. Pacing and Energy Envelope Management

This is non-negotiable. Post-exertional malaise (PEM) is the defining feature that distinguishes ME/CFS-overlap Long COVID from other fatigue syndromes, and it is the feature most likely to be worsened by well-intentioned but misguided advice to “push through” or engage in graded exercise.

Heart rate monitoring-based pacing — keeping heart rate consistently below the anaerobic threshold (typically 60–70% of age-predicted maximum, or individually calculated from CPET data) — is the evidence-based approach. The goal is to avoid the post-exertional energy crash that can set patients back days or weeks. In practice: wearable HR monitors, activity diaries, and explicit education about the energy envelope. In children and adolescents, this requires careful coordination with school, parents, and sometimes neuropsychological support to address the psychological burden of enforced rest.

For patients undergoing CPET: the ventilatory anaerobic threshold (VAT) from two-day CPET provides the most precise pacing target. This is the ceiling, not the goal.

0b. Dietary Modification: No-Added-Sugar, Time-Restricted Eating

The first randomized crossover trial of intermittent fasting in Long COVID (Bunker et al., Scientific Reports, 2025) has now been published and deserves serious attention — with appropriate caveats.

Fifty-eight adults with Long COVID (mean duration: over two years) completed a 10-week protocol comparing two regimens: a no-added-sugar diet with 14:10 daily fasting versus the same diet with a more restrictive 16:8 daily eating window plus a once-weekly extended water fast (median actual duration: ~38 hours). Mean symptom scores dropped 51.8% over 10 weeks. Head-to-head, the extended weekly fast outperformed time-restricted eating alone (LC-Score reduction −10.2 vs. −2.8, p=0.008).

The proposed mechanism is autophagy induction. SARS-CoV-2 actively sabotages the late stages of autophagosome-lysosome fusion (via ORF3a), blocking cellular clearance of viral debris and dysfunctional mitochondria. Extended fasting activates the AMPK/SIRT1/mTOR axis in a way that may counteract precisely this viral evasion strategy — making previously hidden persistent antigen visible to the immune system again.

This also explains the fasting-induced flare-ups (experienced by 58% of participants) as potential immune re-engagement with a persistent pathogen — not a reason to stop, but a reason to monitor carefully.

What I recommend in practice: For relatively higher-functioning patients, I now discuss a starting point of no-added-sugar combined with a 12–14 hour daily eating window — achievable for most families. For patients who tolerate this well after 4–6 weeks, I consider discussing the once-weekly extended fast with appropriate monitoring. I do not recommend the extended fast in: patients with severe PEM/ME/CFS overlap, significant autonomic dysfunction, underweight or low BMI, active cardiac involvement, or any history of eating disorders.

The no-added-sugar baseline alone showed meaningful benefit for many participants. This is the safer starting point for more fragile patients.

0c. Sleep Hygiene and Circadian Stabilisation

Light exposure management, consistent sleep-wake timing, and — when appropriate — blue light restriction in the evening. These are not trivial additions. Circadian dysregulation compounds both autonomic dysfunction and neuroimmunological signalling. Melatonin (discussed in Step 2) works better in a context of consistent circadian anchoring.

Step 1 — Characterise Before Treating

Read more

Read more

The study published in Cell on May 28, 2026 by the group of Akiko Iwasaki (Yale) and David Putrino (Mount Sinai) belongs to the second category. Titled “A causal link between autoantibodies and neurological symptoms in long COVID”, it is — I say this with appropriate scientific caution, but also with the clarity this evidence deserves — likely the most important autoantibody study in Long COVID ever published.

What This Study Did

The group analyzed 147 individuals with Long COVID, COVID convalescents, and healthy controls using complementary techniques: tissue-based immunofluorescence, ELISA, human protein microarray, and mass spectrometry. The goal was to understand not only whether autoantibodies were circulating, but what they target, which symptoms they correlate with, and — crucially — whether they are pathogenic, meaning whether they actually cause harm.

Purified IgG from Long COVID patients reacted with human brain tissue — specifically the locus coeruleus, thalamus, adrenal gland, and thyroid — and showed cross-reactivity with mouse sciatic nerve and meninges. These CNS-directed autoantibodies correlated with specific neurological symptoms.

For example, people with autoantibodies targeting the locus coeruleus were significantly more likely to have lost their sense of taste and smell.

But the critical step — the one that transforms a correlational study into proof of causality — is the passive transfer experiment. Transfer of IgG from Long COVID individuals into mice induced fatigue-like behavior, loss of balance and coordination, thermal hyperalgesia, small fiber nerve damage, and increased pain sensitivity.

In other words: these antibodies are not bystanders. They cause the symptoms.

Why This Matters for My Patients

Those who follow this newsletter — and who know my book on Long COVID or my virtual clinic — know that autoantibodies are not a new topic for me. For years I have argued that a significant proportion of patients with severe Long COVID or post-infectious ME/CFS carry an autoimmune component that must be quantified before any therapeutic decision is made.

This study validates that approach with rigorous, mechanistic evidence. The findings could open important new avenues for treating patients with Long COVID, including already-validated therapies for autoimmunity management, as well as new ways of clinically identifying which patients are most likely to respond to them.

And — this is where the paper becomes truly clinically actionable — it points to specific therapeutic tools: intravenous immunoglobulins (IVIG), which contain antibodies from healthy donors already widely used to treat autoimmune disorders, and FcRn inhibitors, biologic agents that may help Long COVID patients by reducing the circulating burden of pathogenic antibodies.

The sentence from Prof. Putrino that has been echoing across the Long COVID research community is this: “Before, we had no way of predicting who would benefit from therapies like IVIG or FcRn inhibitors. Our study now shows that if you are in a subgroup of Long COVID patients who have autoantibodies circulating in your body, this is a quantifiable sign that you may be a good candidate for these drugs.”

This is exactly what I write in my book. This is exactly what I say in virtual clinic when I propose a full autoimmune workup before any therapeutic escalation.

My Stepwise Approach: Where IVIG Fits

Over the years I have built a stepwise therapeutic framework for severe Long COVID/ME/CFS that works in my clinical practice — one I have shared both in the book and in the virtual clinic. Briefly: we begin with autonomic assessment and ivabradine, move toward low-dose naltrexone and low-dose lithium, then anticoagulation and low-dose aripiprazole, and in the most severe, biomarker-supported cases we escalate to active immunomodulation — IVIG and rapamycin — and in the most refractory patients, to advanced biologics such as daratumumab.

IVIG is not a first-line treatment in Long COVID. It is expensive, requires hospital infrastructure, and carries a non-negligible risk profile. But in the right patients — those with documented autoantibodies, disabling neurological presentation, and failure of prior therapeutic lines — it can genuinely change the clinical trajectory.

This Cell paper gives me the tool to explain to colleagues, insurers, and families why and for whom.

A Necessary Caveat: Still Off-Label, Still Waiting for Trials

I want to be transparent about something, because intellectual honesty is the only currency that matters in medicine.

Yes, this Cell paper supports the biological rationale for what I do. Yes, the Lancet Infectious Diseases case report describes a protocol remarkably similar to the one I use. And yes, the clinical responses I have observed in selected patients give me reason for cautious optimism.

But IVIG for Long COVID remains off-label. There are no completed randomized controlled trials demonstrating efficacy in this population. The evidence base, as of today, consists of mechanistic studies, case reports, case series, and clinical experience — all pointing in the same direction, but none sufficient on their own to establish standard of care.

Randomized studies of IVIG in Long COVID have begun, including one within the NIH RECOVER Initiative — and if successful, understanding the mechanism by which this agent provides benefit could be instrumental in developing more targeted therapies that could be made more widely available than IVIG itself.

Those trials cannot come soon enough. Until they report, every clinician using IVIG in this context — myself included — is making a judgment call: weighing documented autoimmune pathology, severity of disease, failure of prior therapeutic lines, and the biological plausibility of a mechanism that this Cell paper has now placed on extraordinarily solid ground. That judgment call can be the right one. But it must be made with full transparency toward the patient, documented carefully, and revisited constantly as new evidence emerges.

I use IVIG in Long COVID. I will continue to do so in the right patients, with the right workup, and the right informed consent. But I will never pretend that “the science supports it mechanistically” is the same as “we have level 1 evidence.” It is not — yet.

A Note on Transfusion Safety

There is another dimension of this study that deserves attention beyond the clinical. Prof. Putrino believes the study carries an urgent public health warning regarding the donation of blood and blood products such as plasma by individuals with Long COVID. In the UK, having Long COVID is already an exclusion criterion for blood donation. In the United States, and across much of Europe including Italy, this debate remains open. It is worth keeping in mind — and advocating for change.

Below you will find what is rarely available in the published literature: the practical IVIG protocol I use in my clinical practice — based on the regimen reported in a 2026 Lancet Infectious Diseases case report — and which I proposed to do. His biomarker profile maps almost exactly onto what Iwasaki and colleagues now describe in Cell. This is the same scheme I used with him, with the clinical reasoning that guides dose selection, treatment duration, and the timing of maintenance infusions.

Read more

Long COVID and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) are among the most diagnostically and therapeutically complex conditions of our era. Patients frequently wait months or years for specialist evaluation, arrive at appointments exhausted and cognitively impaired, …

Read more

Read more

A new preprint study from researchers at Yale and Mount Sinai has explored the association between salivary HHV-6 (Human Herpesvirus 6) shedding and Long COVID severity.

The study found that higher levels of HHV-6 DNA…

Read more

Read more

Read more

by Danilo Buonsenso

Those who follow this blog know that one of the questions that has accompanied me since the very first months of the pandemic is this: what is actually happening in the bodies of children with Long COVID? Not just “how many symptoms do they have,” but why do they have them. What biological process, beneath the surface, explains that fatigue that won’t go away, that irregular heartbeat, that feeling of never quite being well again?

A newly published study on children and young adults with Long COVID up to 25 years of age adds an important — and in some ways sobering — piece to that answer.

What This Study Examined

Researchers enrolled 84 patients — children and young adults (≤25 years) from the United States and Canada: 61 with Long COVID and 23 healthy controls. They didn’t stop at describing symptoms. They looked inside the blood, measuring very specific things: fibrin amyloid microclots (small protein clumps circulating in the vessels), endovascular cytokines (inflammatory molecules), cell-free DNA, and the capacity of neutrophils — white blood cells of the immune system — to damage the inner walls of blood vessels.

The results are clear and statistically robust:

• Children with Long COVID had significantly higher microclot burden compared to healthy controls (p = 0.0003)

• Markers of vascular remodeling and angiogenesis were also elevated, including FGF-2, which directly correlated with microclot levels

• Cell-free DNA in the blood — an indirect marker of a process called NETosis — was increased (p = 0.003), and this correlated with a key microclot component, serum amyloid A (p = 0.004)

• Laboratory assays demonstrated that the Spike protein, binding to immune complexes, activates neutrophils which then — through NETosis — damage endothelial cells, the inner lining of blood vessels

In short: in children with Long COVID and cardiovascular symptoms, the blood shows signs of active intravascular inflammation and microcoagulation, driven substantially by neutrophil activity.

What Microclots and NETosis Are — Without the Jargon

Think of blood vessels as very delicate tubes, lined on the inside by an extremely thin layer of cells. When everything works well, blood flows smoothly and those cells remain intact.

In Long COVID, some patients appear to have small protein clumps circulating in their blood (the microclots) that don’t dissolve the way they should. These clumps don’t block large vessels — they don’t cause classical thrombosis — but they can interfere with microcirculation and with oxygen delivery to tissues.

NETosis is a process by which neutrophils — cells that are normally “good” because they fight infections — expel their own DNA outward like a net (NET = Neutrophil Extracellular Trap) to trap pathogens. It’s a defense mechanism. The problem arises when it is triggered excessively or chronically: those DNA nets become toxic to vessel walls. In this study, the neutrophil nets appear to bind to and stabilize microclots, amplifying the damage.

Why This Study Resonates With Our Work

Those who follow our research know that we have been working on these themes for years — often in a certain scientific solitude, especially in the early periods when arguing that Long COVID in children was a real, biologically grounded problem was still controversial.

As early as 2023, in our review on the diagnosis and management of pediatric Long COVID, we explicitly stated that chronic endothelitis and microclots were emerging as possible mechanisms even in children, and that they needed to be included in research studies. We wrote this when very few believed it was relevant for the pediatric population.

We have published studies on circulating activated platelets in children with Long COVID, on viral persistence, on altered cytokine profiles, on impaired cardiac autonomic function, on lung perfusion abnormalities, and even on altered brain metabolism. Each time, the underlying message was the same: Long COVID in children is not “just anxiety” or “just post-infectious fatigue” — something is happening at a biological level, and we need to understand what.

This new study, conducted on an independent North American cohort, confirms and deepens that direction.

A Note on Anticoagulants: Why I Sometimes Use Them Off-Label

Read more