Candidate Therapeutics for Long COVID & PACVS

Mechanism

100% oxygen at 1.5–2.0 atmospheres absolute (ATA) induces multiple systemic effects: hyperoxia drives angiogenesis via VEGF; neuroplasticity via BDNF and stem cell mobilisation; improves mitochondrial electron transport efficiency; reverses hypoxia-inducible factor (HIF-1α) changes in chronically hypoxic tissue; reduces microglial activation and NLRP3 inflammasome neuroinflammation.

Evidence in Long COVID

Double-blind, sham-controlled RCT (n=73). 40 daily 90-minute HBOT sessions at 2.0 ATA with 100% O₂. Primary endpoint (cognitive performance) significantly improved vs. sham. Also significantly improved: energy levels, sleep quality, and pain. Brain MRI showed improved perfusion and microstructural changes.

Limitation: Expensive; requires 40 daily sessions at specialist centres. Not studied in PACVS specifically.

Mechanism

15-nucleotide single-stranded DNA aptamer that sterically neutralises a broad spectrum of functional autoantibodies against G-protein–coupled receptors (GPCR-AAbs). Functional GPCR-AAbs (e.g. β₂AR, α₁AR, M₂ AChR, AT₁R, ETaR) were detected in all 31 Long COVID patients in one bioassay series (Wallukat et al. 2021); their prevalence in broader cohorts is uncertain. They are proposed to drive dysautonomia, POTS, tachycardia, and fatigue by constitutively activating these receptors.

Evidence

Two placebo-controlled trials in Long COVID point in different directions. (1) reCOVer (Hohberger et al. 2025; investigator-initiated phase IIa, Erlangen; EudraCT 2022-001781-35): 30 GPCR-fAAb-positive post-COVID patients with fatigue were randomised 1:1 to rovunaptabin (BC007) 1350 mg or placebo, with crossover. The primary endpoint was safety (9 vs 5 treatment-emergent adverse events, difference not significant). In secondary analyses rovunaptabin neutralised GPCR-fAAbs and improved FACIT-Fatigue, Bell score, Fatigue Severity Scale and 4 of 8 quality-of-life items; the authors call it an exploratory proof-of-concept study needing confirmation. (2) The sponsor's larger multicentre phase 2 RCT (NCT05911009, Berlin Cures; n=119; completed September 2024): Berlin Cures announced in late 2024 that topline results showed no evidence of superior efficacy over placebo. No results are posted on ClinicalTrials.gov and no peer-reviewed report was found as of this review. Earlier, a single-patient case report described improved fatigue and retinal capillary microcirculation after BC007 (Hohberger et al. 2021). GPCR-fAAbs in Long COVID were first described by Wallukat et al. (2021) using a cardiomyocyte bioassay.

Direction "Mixed": A small investigator-initiated RCT reported secondary-endpoint improvements in fatigue, but the larger sponsor phase 2 RCT was reported as showing no benefit over placebo.

Testing for GPCR-AAbs: Cell Trend GmbH (Germany) offers commercial autoantibody testing used in research contexts.

Mechanism

Synthetic vasoactive intestinal peptide (VIP) agonist. VIP receptors VPAC1 and VPAC2 are highly expressed on ACE2+ cells (alveolar type II, mast cells, enteric neurons, immune cells). VIP is a potent immunosuppressive neuropeptide: reduces IL-6, TNF-α, IL-1β; protects alveolar epithelium from apoptosis; modulates autonomic tone via parasympathetic co-transmission; down-regulates mast cell degranulation.

Evidence

Randomised placebo-controlled trial in critical COVID-19 respiratory failure (n=196, NCT04311697): the primary endpoint (alive and free of respiratory failure at day 60) was not met; a secondary analysis reported improved 60-day survival. The larger TESICO trial (n=461 in the aviptadil comparison) found no significant improvement in clinical outcomes up to day 90. Both are acute-COVID trials; no Long COVID trial results are available. Mechanism directly relevant to MCAS, autonomic dysregulation, and pulmonary manifestations of both Long COVID and PACVS. No PACVS-specific RCT yet.

Direction "Mixed": The first RCT missed its primary endpoint with a secondary survival signal; the larger TESICO RCT was null.

Mechanism

Recombinant human IL-1 receptor antagonist (IL-1Ra). Competes at IL-1R1 to block both IL-1α and IL-1β signalling. Reduces downstream NLRP3 inflammasome effects including pyroptosis, IL-18 release, and gasdermin D pore formation. IL-1β is persistently elevated in Long COVID neuroinflammation and brain fog. Autoinflammatory hypothesis: unresolved NLRP3 activation drives chronic post-viral symptoms.

Evidence

SAVE-MORE trial (Kyriazopoulou et al. Nat Med 2021, n=594): anakinra significantly reduced mortality and ICU admission in severe COVID-19 with suPAR elevation. Multiple COVID-19 RCTs confirm IL-1 pathway relevance. No Long COVID-specific RCT yet; T2 based on IL-1β elevation in Long COVID + acute COVID RCT data.

Mechanism

Glycosaminoglycan heparinoid mixture (80% fast-moving heparin fraction + 20% dermatan sulfate). Primary action: restores and protects the endothelial glycocalyx — the polysaccharide layer lining all blood vessels — which is degraded by COVID-19-associated endothelial inflammation. Also inhibits thrombin and Factor Xa, reduces PAI-1, and suppresses NF-κB-mediated endothelial inflammation. LDL endothelial penetration is reduced.

Evidence

Randomised placebo-controlled outpatient trial (n=243 per-protocol) in high-risk patients treated within 3 days of acute COVID-19 onset: sulodexide reduced the need for hospital care (17.7% vs 29.4%) and supplemental oxygen, and lowered D-dimer and CRP; there were no between-group differences in thromboembolic events, major bleeding or mortality. This was an acute-COVID trial, not a Long COVID trial. Endothelial dysfunction is a central feature of Long COVID; no Long COVID- or PACVS-specific RCT exists.

Mechanism

SSRI with the highest sigma-1 receptor (S1R) agonist activity in its class. S1R agonism: (1) stabilises IRE1α ER stress sensor, suppressing IRE1α-driven cytokine production (IL-6, IL-8, TNF-α); (2) reduces acid sphingomyelinase (ASM) activity, impairing ceramide-mediated SARS-CoV-2 cell entry; (3) inhibits platelet serotonin uptake, relevant to the Long COVID serotonin-reduction findings (Wong et al. Cell 2023). Reduces mast cell activation signals.

Evidence

TOGETHER platform RCT (Reis et al. Lancet Glob Health 2022, n=741): 32% reduction in hospitalisation/emergency care in early COVID-19. In Long COVID, the REVIVE adaptive RCT (Reis et al. Ann Intern Med 2026; NCT06128967; n=399 adults in Brazil with fatigue 3–24 months after infection) found that fluvoxamine 100 mg twice daily for 60 days reduced Fatigue Severity Scale scores vs placebo (mean difference −0.43 at day 60 and −0.58 at day 90; posterior probability of superiority 99%) and improved quality of life, with fewer adverse events than placebo; benefit beyond 90 days and effects on other symptoms are unknown. The paper was retracted and republished by the journal in October 2026; the republished version is cited here. The previously cited "Hellmuth et al." observational SSRI study could not be verified and has been removed (citation under review).

Mechanism

Serine protease (subtilisin-like) produced by Bacillus subtilis var. natto during soybean fermentation. Directly cleaves fibrin; inactivates PAI-1 (plasminogen activator inhibitor-1), thereby prolonging tPA activity and overall fibrinolytic capacity. In vitro (Tanikawa et al. 2022): nattokinase degraded the SARS-CoV-2 spike (S) protein in a dose- and time-dependent manner in cell lysates and on the cell surface, raising interest in dissolving spike-driven amyloid fibrin microclots documented by the Pretorius group.

Evidence

T3 based on: (1) in vitro spike protein degradation (Tanikawa 2022); (2) informal use in microclot-directed Long COVID treatment protocols (no published Long COVID cohort data could be verified; citation under review); (3) cardiovascular fibrinolytic RCTs in other conditions. No Long COVID-specific RCT completed. Caution: may interact with anticoagulants; no formal safety data in LC populations.

Mechanism

Methylxanthine phosphodiesterase inhibitor. Elevates cAMP in red blood cells → improved RBC membrane flexibility and deformability, allowing passage through narrowed capillaries. Also reduces fibrinogen levels; inhibits TNF-α transcription via cAMP/NF-κB; reduces leukocyte adhesion; weak anti-platelet aggregation. Altered red-cell deformability has been reported in COVID-19, with some blood-cell changes persisting for months after hospital discharge (Kubánková et al. 2021); whether this contributes to Long COVID microvascular dysfunction is unproven.

Evidence

No Long COVID clinical studies were found; the previously cited post-COVID pilot data could not be verified (citation under review). A 2024 systematic review and meta-analysis of acute COVID-19 studies (Ramzi et al.) found that add-on pentoxifylline shortened hospital stay and improved some inflammatory markers, with an inconclusive effect on mortality; its application to Long COVID is speculative. Rationale also draws on clotting-protein abnormalities reported in Long COVID (Pretorius et al. 2021). No Long COVID RCT. Well-established safety profile in peripheral arterial disease. "Mixed" because specific LC RCT data is lacking despite strong mechanistic rationale.

Mechanism

Ultrasound-guided injection of long-acting local anaesthetic (bupivacaine 0.5%, 5–10 mL) at or near the stellate ganglion — the confluence of inferior cervical and first thoracic sympathetic ganglia. Transiently disrupts sympathetic efferents to the head, neck, arm, and heart. Proposed mechanism in Long COVID: resets hyperactivated sympathetic tone, normalises elevated plasma norepinephrine/epinephrine. Lipov et al. (2009) proposed that SGB causes a prolonged reduction in nerve growth factor (NGF) and brain norepinephrine, a hypothesis developed for CRPS, hot flashes and PTSD rather than Long COVID.

Evidence

Case series (Liu & Duricka 2022, n=2) in Long COVID with symptoms attributed to dysautonomia (e.g. fatigue, orthostatic intolerance, brain fog, smell/taste disturbance): both patients reported symptom reduction after SGB. A retrospective survey-based cohort (Chiang et al. 2025, n=52, Boston) reported improvement in brain fog, fatigue, dizziness and headache, with variable duration and short-lived adverse effects in about a quarter. The one randomised trial, in persistent COVID-19 parosmia (Farrell et al. 2025, JAMA Otolaryngol; NCT06253806; n=48 analysed), found SGB no better than saline (3-month response 43% vs 38%). The previously cited "Lipov et al. observational data" in post-COVID autonomic symptoms could not be verified (citation under review). Must be performed by trained anaesthesiologist under ultrasound guidance. Established use in PTSD and hot flashes supports safety profile. No RCT of SGB for broader Long COVID symptoms yet.

Direction "Mixed": Uncontrolled reports describe benefit, but the only randomised trial (for post-COVID parosmia) found no difference from placebo.

Mechanism

Removes large circulating macromolecules from blood plasma: GPCR autoantibodies; anti-phospholipid antibodies; fibrinogen; spike protein antigens; immune complexes; complement fragments; and amyloid microclot fragments. Plasma exchange (TPE) replaces removed plasma with albumin or fresh frozen plasma. Immunoadsorption (IA) selectively removes IgG (e.g. Protein-A or tryptophan adsorbers), allowing plasma return. Both methods lower circulating IgG, including GPCR autoantibodies; an effect on microclots in Long COVID has been hypothesised but not demonstrated.

Evidence

Two sham-controlled RCTs were negative. España-Cueto et al. 2025 (phase II, double-blind; NCT05445674; n=50): six sessions of therapeutic plasma exchange vs sham were equally safe, with no between-group difference in any efficacy outcome over 90 days. Stortz et al. 2026 (IAMPOCO, patient-blinded sham-controlled crossover; NCT05841498; n=40): immunoadsorption depleted GPCR autoantibodies but did not reduce symptom severity vs sham (Post-COVID-19 Functional Scale OR 1.17, p=0.77), with more adverse events after immunoadsorption (24 vs 10). Uncontrolled data are more favourable: in a Charité pre-post study of 20 post-COVID ME/CFS patients with elevated β₂-adrenergic autoantibodies, five immunoadsorption sessions raised SF-36 physical functioning by a mean 17.75 points, with 14/20 responders (Stein et al. 2025). The previously cited Jaeger case series (~100 patients) and Pretorius-group microclot data could not be verified (citation under review). Expensive, invasive, and specialist-only.

Direction "Null": Both sham-controlled trials (plasma exchange and immunoadsorption) found no benefit over sham; positive signals come only from uncontrolled cohorts.

Mechanism

Non-absorbable synthetic rifamycin derivative (<0.4% systemic absorption). Selectively targets gut microbiota without altering host commensal ecology as broadly as systemic antibiotics. Reduces small intestinal bacterial overgrowth (SIBO); decreases endotoxin (LPS) production and translocation across the intestinal barrier; reduces TLR4 signalling cascade; improves tight junction protein expression (ZO-1, occludin); increases Faecalibacterium prausnitzii — a key butyrate producer reduced in Long COVID.

Evidence

Liu Q, et al. Gut 2022: prospective cohort documenting gut microbiota disruption in Long COVID persisting for 6+ months, including reduced F. prausnitzii and Eubacterium rectale with increased opportunistic pathogens. Rifaximin RCTs in IBS without constipation (Pimentel et al. NEJM 2011, TARGET 1 and 2) and in hepatic encephalopathy establish efficacy in other gut conditions. No Long COVID clinical data for rifaximin were found; the earlier mention of emerging pilot Long COVID data could not be verified (citation under review). No Long COVID RCT published as of this review.

Mechanism

Angiotensin receptor blocker (AT₁R blocker). Among ARBs, telmisartan uniquely acts as a partial PPAR-γ agonist — providing anti-inflammatory, anti-fibrotic, and mast cell-stabilising effects beyond blood pressure reduction. AT₁R blockade shifts renin-angiotensin-aldosterone signalling toward ACE2 → angiotensin 1-7 → Mas receptor (vasoprotective, anti-inflammatory axis). SARS-CoV-2 dysregulates RAS in both acute and post-acute phases; normalisation may reduce endothelial and mast cell hyperreactivity.

Evidence

No published Long COVID studies of telmisartan were found, and no Long COVID telmisartan trial was found on ClinicalTrials.gov; the previously cited pilot data and "Lobo-Ferreira et al." could not be verified (citation under review). ARB trials in COVID-19 have addressed acute disease. No Long COVID-specific RCT as of this review. Antihypertensive with excellent long-term safety profile. "Mixed" due to absence of controlled LC trial data despite strong mechanistic case.

Mechanism

Endogenous saturated fatty acid amide (N-palmitoylethanolamine), produced on demand from membrane phosphatidylethanolamine. PPAR-α nuclear receptor agonist: suppresses NF-κB, reduces mast cell degranulation, and decreases production of TNF-α, IL-1β, and histamine. GPR55 and GPR119 partial agonism. In dorsal root ganglia and microglia, PEA reduces neuroinflammatory gene expression. Altered endogenous PEA levels in Long COVID have been proposed, but no verified source was found (citation under review).

Evidence

Post-COVID olfactory dysfunction: in a placebo-controlled multicentre RCT (Di Stadio et al. 2023, n=202, smell loss >6 months), olfactory training plus co-ultramicronised PEA with luteolin (um-PEA-LUT) gave >3-point smell-test recovery in 89.2% vs 36.8% with training plus placebo; um-PEA-LUT alone (40–41.6%) was not clearly better than training plus placebo on that endpoint. No RCT of PEA for broader Long COVID symptoms was found. Schweiger V et al. 2019: retrospective observational study (407 patients, 359 analysed) in fibromyalgia (overlapping symptomology with Long COVID) reporting reduced pain scores with add-on ultramicronised PEA, although 57.6% stopped for lack of efficacy. Use for mast-cell symptoms is mechanistic; no clinical data in MCAS-type Long COVID were verified.

Citation under review: a previously listed review ("Petrosino S, et al., J Neuroinflammation 2022") could not be found in PubMed or Crossref and has been removed.

Mechanism

Complex of six distinct fibrinolytic serine proteases extracted from Lumbricus rubellus earthworm. Acts on a broader substrate range than nattokinase: degrades fibrin, fibrinogen, and fibrin degradation products; activates both plasminogen and tissue plasminogen activator (tPA); degrades fibrin more specifically than streptokinase, potentially with less risk of systemic haemorrhagic complications. Higher molecular weight substrates mean potentially more targeted fibrinolysis.

Evidence

Post-COVID case-series use has been reported informally; no peer-reviewed post-COVID case series has been verified (citation under review). In acute ischaemic stroke, a 2025 meta-analysis of 35 RCTs (Wiyarta et al.) found adjunctive lumbrokinase improved Barthel Index and NIHSS scores without a significant increase in GI bleeding, while noting limited evidence quality. More fibrin-specific than streptokinase in mechanistic comparisons. No Long COVID RCT. Used empirically in some Long COVID microclot treatment protocols (alongside nattokinase) due to complementary substrate specificity.

Citation under review: a previously listed review ("Tao R, et al., Evid Based Complement Alternat Med 2022") could not be found in PubMed or Crossref and has been removed.

Mechanism

First-generation antihistamine with potent 5-HT2A, 5-HT2B, and 5-HT2C serotonin receptor antagonist activity (stronger serotonin antagonism than most antihistamines). Blocks serotonin signalling, including in the gut. This sits in tension with the main Long COVID serotonin finding: Wong et al. (Cell 2023) reported reduced circulating serotonin in Long COVID, and in animal models linked low peripheral serotonin to impaired vagal signalling and memory, pointing towards serotonin repletion rather than blockade. Also an H1 antihistamine (of theoretical interest for mast-cell symptoms). Appetite stimulant; used off-label for functional gastrointestinal symptoms.

Evidence

No published Long COVID or POTS data for cyproheptadine were found; the previously cited dysautonomia-clinic case series and "Rosen DA, Nature 2023" reference could not be verified (citation under review). The closest human evidence is in paediatric functional gastrointestinal disorders: a retrospective series (Madani et al. 2016, n=151) reported complete symptom improvement in 72.8%, with sleepiness (13%) and weight gain (10%) as the main adverse effects. The serotonin-reduction model (Wong et al. 2023) does not support serotonin antagonism as a Long COVID strategy. Established H1 antihistamine with decades of use; well-characterised safety profile. Causes sedation and appetite increase — relevant considerations for LC patients.

Direction "Mixed": Symptomatic GI benefit is reported outside Long COVID, but there are no Long COVID data and the leading serotonin model points the other way.

Mechanism

Carbonic anhydrase II and IV inhibitor. Reduces CSF production at choroid plexus → decreases intracranial pressure (ICP) — relevant only if Long COVID brain fog involves occult intracranial hypertension, which is an unproven hypothesis (the earlier attribution of this proposal could not be verified). Separately: sensitises CO₂ chemoreceptors, correcting the hypocapnia-mediated "air hunger" that many Long COVID patients report (despite normal O₂ saturation). Weakly diuretic. First-line treatment for idiopathic intracranial hypertension (IIH).

Evidence

No published Long COVID case series or trials of acetazolamide were found; the previously described case series and clinical practice reports could not be verified (citation under review). Use is extrapolated from established indications (IIH and altitude sickness) to hypothesised Long COVID subgroups with raised intracranial pressure or hypocapnia. Formal Long COVID RCT lacking. "Mixed" because any benefit would be expected only in such subgroups, which remain unproven; use is untested in LC.

Mechanism

Cysteine protease complex from Ananas comosus (pineapple) stem. In vitro: Akhter et al. (2021) showed that bromelain combined with acetylcysteine (BromAc) disrupted recombinant SARS-CoV-2 spike and envelope proteins and inactivated whole virus in culture in a concentration-dependent manner; neither agent alone inactivated the virus. NF-κB inhibition reduces downstream cytokine production (IL-1β, IL-6, TNF-α). Also reduces prostaglandin E2 via COX pathway modulation. Oral bioavailability is limited but some absorption confirmed by human studies.

Evidence

T5 based on in vitro spike degradation data and anti-inflammatory case reports. BromAc has been proposed as a nasal spray for early COVID-19 (Akhter et al. 2021). No Long COVID RCT. Often co-administered with nattokinase and NAC in clinical practice for putative spike protein dissolution. Anti-inflammatory RCTs exist in other conditions (sinusitis, osteoarthritis).

Mechanism

Stimulates de novo mitochondrial biogenesis via PGC-1α upregulation and CREB phosphorylation — the same pathway activated by exercise and caloric restriction. Functions as a redox cofactor in bacterial respiration and influences mammalian cells through signalling rather than direct enzymatic activity. Activates DJ-1 (PARK7), a mitochondrial neuroprotective protein mutated in early-onset Parkinson's. May upregulate nerve growth factor (NGF) synthesis. Acts as a potent antioxidant, quenching ROS in both mitochondrial and cytosolic compartments.

Evidence & Long COVID Rationale

Harris et al. 2013 (n=76 human crossover): 20 mg/day PQQ significantly altered mitochondrial-related urinary metabolites and reduced CRP and IL-6 vs. placebo. Stites et al. 2006 (J Nutr): PQQ-deficient weanling mice had 20–30% less liver mitochondria and lower respiratory control ratios. Plasma metabolomics in Long COVID suggest dysfunctional mitochondria-dependent fatty-acid metabolism (Guntur et al. Metabolites 2022) and in PACVS patients with fatigue and post-exertional malaise. No Long COVID RCT.

Mechanism

Ubiquinone (CoQ10) covalently linked to triphenylphosphonium (TPP⁺) cation. The permanently cationic TPP⁺ group drives ~1000-fold accumulation in the mitochondrial matrix, driven by the large negative mitochondrial membrane potential. Positioned exactly at the inner membrane surface where ETC Complexes I and III generate superoxide (O₂•⁻). Far more potent than conventional CoQ10 supplementation, which cannot efficiently reach the intramitochondrial site of ROS generation. Protects Complex I integrity and reduces mitochondrial membrane potential decay.

Evidence

Extensive preclinical data in oxidative stress, ischaemia-reperfusion, and metabolic disease models (Murphy/Smith group, MRC Cambridge). Rossman et al. 2018 Hypertension: randomised crossover trial (n=20 healthy older adults) showing improved endothelial function and lower oxidised LDL with MitoQ 20 mg/day. Broome et al. 2022 Redox Biol: RCT in 23 untrained middle-aged men showing augmented exercise-induced PGC1α mRNA and training-induced peak power gains, independent of mitochondrial content. Mitochondrial superoxide overproduction documented in Long COVID PBMCs and endothelial cells. No Long COVID RCT. Oral formulation available as dietary supplement (10–20 mg/day).

Mechanism

Natural polyamine synthesised from putrescine (also found in wheat germ, aged cheese, mushrooms). Induces macroautophagy by inhibiting EP300 (a histone acetyltransferase) — derepressing ATG5, ATG12, and other autophagy genes. Promotes mitophagy: selective clearance of damaged, ROS-generating mitochondria. Supports hypusination of eukaryotic initiation factor 5A (eIF5A), essential for translating mitochondrial proteins. Plasma spermidine levels decline with age and post-viral illness.

Evidence

Madeo F, et al. Science 2018: comprehensive review establishing spermidine's autophagy-inducing role and longevity associations. Alsaleh G, et al. 2026 (Aging Cell): double-blind placebo-controlled pilot RCT in 40 adults over 65 after a third SARS-CoV-2 vaccine dose; 6 mg/day spermidine for 13 weeks was well tolerated and enhanced spike-specific antibody and memory B-cell responses in vaccine non-responders. Impaired autophagy has been documented in Long COVID with accumulation of protein aggregates and dysfunctional organelles. No Long COVID RCT. Dietary sources include wheat germ (~24 nmol/g), aged cheese, and fermented soy products.

Mechanism

Isoquinoline alkaloid from Berberis species. AMPK activation: mild Complex I inhibition raises intracellular AMP:ATP ratio → AMPK activation → metabolic reprogramming toward catabolism, fatty acid oxidation, and mitochondrial biogenesis. NLRP3 inflammasome inhibition: reduces IL-1β, IL-18, and pyroptosis — directly relevant to post-viral autoinflammation. mTORC1 inhibition enhances autophagy. SIRT1 upregulation. In vitro antiviral activity against SARS-CoV-2 (Varghese et al. 2021, Viruses). Modulates gut microbiome: increases Akkermansia muciniphila.

Evidence

Over 100 human RCTs in metabolic syndrome, type 2 diabetes, and PCOS — demonstrating AMPK pathway effects in humans. Varghese FS, et al. Viruses 2021: SARS-CoV-2 inhibition in vitro at low micromolar concentrations, including in primary human nasal epithelial cells. NLRP3 inhibition confirmed in multiple inflammatory disease models. No Long COVID or PACVS RCT. T6 because LC-specific human data is absent despite robust mechanistic and metabolic RCT basis.

Mechanism

Taurine-conjugated ursodeoxycholic acid. Primary action: inhibits the unfolded protein response (UPR) / ER stress — specifically suppresses IRE1α/XBP1 splicing and PERK/ATF4 phosphorylation branches. This is significant because SARS-CoV-2 spike protein (S1 and full-length) activates ER stress in alveolar, endothelial, and immune cells, driving apoptosis and inflammation. TUDCA reduces ER stress-triggered apoptosis via mitochondrial membrane stabilisation (reduces cytochrome c release). Neuroprotective via PKCε and PKCα activation. Well-established in cholestatic liver disease.

Evidence

ALS: a small double-blind pilot RCT of TUDCA alone (Elia et al. 2016, n=34) reported more responders and slower ALSFRS-R decline than placebo. The phase 2 CENTAUR trial (Paganoni et al. 2020, NEJM) tested the combination sodium phenylbutyrate–taurursodiol (PB-TUDCA), not TUDCA alone, and slowed functional decline; however, the phase 3 PHOENIX trial showed no benefit and the product was voluntarily withdrawn (Ketabforoush et al. 2024). Neither trial isolates the ER stress mechanism. In cell studies, SARS-CoV-2 spike (S1) activates ER stress/UPR responses via the host protein HAX1 (Zhu et al. 2025); the previously cited "Bhatta 2021" could not be verified (citation under review). No Long COVID or PACVS RCT. TUDCA is commercially available as a supplement (250–750 mg/day). T6 because LC-specific data absent; mechanism is directly relevant to spike protein pathology.

Mechanism

NR is a bioavailable NAD⁺ precursor that bypasses the rate-limiting de novo synthesis pathway: NR → NMN → NAD⁺. Raising intracellular NAD⁺ restores activity of NAD⁺-dependent enzymes — sirtuins (SIRT1–7, mitochondrial deacetylases governing gene expression and metabolic reprogramming), PARPs (DNA damage repair), and CD38 (immune regulation). Systemic NAD⁺ depletion in Long COVID and PACVS is plausibly driven by PARP hyperactivation (DNA damage from spike/immune response), CD38 upregulation (chronic inflammation), and impaired mitochondrial ETC generating less NADH recycled to NAD⁺.

Evidence

Wu et al. 2025 (EClinicalMedicine; senior author Guzmán-Vélez; NCT04809974): 24-week double-blind trial at MGH with a placebo lead-in. 58 participants with Long COVID were randomised 2:1 to NR 2000 mg/day for 20 weeks or placebo for 10 weeks followed by NR. NAD⁺ rose 2.6- to 3.1-fold within 5–10 weeks. There were no significant between-group differences in the primary cognitive outcomes (ECog, RBANS, TMT-B) or in fatigue, sleep, anxiety or depression. Dropout was higher in the NR-first arm (32.4% at 10 weeks, 51.4% at 20 weeks vs 14.3%). A post-hoc analysis pooling all participants' first 10 weeks on NR (unadjusted for multiplicity) found within-group improvements from baseline in executive function, fatigue, sleep quality and depressive symptoms; the authors call for larger trials. A larger Norwegian 2×2 factorial RCT (NCT05703074; NR vs placebo and a mind-body programme vs usual care; n=310) was completed in June 2025; no results are posted or published as of this review. T2 classification reflects the RCT design despite the null result.

Direction "Null": The only published controlled comparison found no difference from placebo on any primary or secondary outcome. Within-group improvements were post-hoc and unadjusted, and high dropout limits interpretation. Results of the larger Norwegian RCT are awaited.

Mechanism

Photobiomodulation therapy (PBMT) uses near-infrared (NIR, 800–1100 nm) and red (630–700 nm) light to stimulate mitochondrial cytochrome c oxidase (Complex IV), the terminal electron acceptor of the ETC. Photon absorption by the copper and heme centres of Complex IV increases its catalytic activity, boosts the proton gradient, and directly raises mitochondrial ATP production. Secondary effects include reduced NO-mediated Complex IV inhibition (restoring oxygen utilisation), reduced oxidative stress, and modulation of inflammatory cytokines (IL-1β, TNF-α) via downstream NF-κB suppression.

Evidence

No completed Long COVID RCT. Mechanistic rationale (Hamblin and colleagues): cytochrome c oxidase is the primary photoacceptor. Mitochondrial dysfunction is suggested in Long COVID by plasma metabolomics (Guntur et al. Metabolites 2022), but impairment of cytochrome c oxidase specifically has not been shown. Small pilot reports in ME/CFS and post-viral fatigue have been described informally, but none were verified for this review (citation under review). T5 rating reflects the absence of controlled LC-specific data; the mechanism is directly relevant to documented mitochondrial ETC impairment in Long COVID.