GPCR Autoantibody-Driven Dysfunction
This section has moved from purely speculative to evidence-supported. Multiple studies have documented G-protein coupled receptor (GPCR) autoantibodies in ME/CFS, and treatment trials targeting these autoantibodies have shown promising results.
1 Established Evidence
1.1 Foundational Cohort Studies
The Charité Berlin group established GPCR autoantibodies as a significant finding in ME/CFS:
- Loebel et al. 2016 (Loebel et al. 2016): In 268 ME/CFS patients vs. 108 controls, 29.5% of patients had elevated antibodies against \(\geq\) 1 muscarinic (M) or \(\beta\)-adrenergic receptor. Antibodies against \(\beta_2\), M3, and M4 receptors were significantly elevated vs. controls.
- Sotzny/Freitag et al. 2021 (Freitag et al. 2021): Autoantibody levels correlated with symptom severity—fatigue, muscle pain, cognitive impairment, and GI symptoms in infection-triggered ME/CFS. First demonstration of dose-response relationship.
- Bynke et al. 2020 (Bynke et al. 2020): Swedish validation in two independent cohorts found 79–91% of ME patients had >=1 elevated antibody vs. 25% of controls. Critically: no autoantibodies detected in CSF, suggesting peripheral origin rather than intrathecal production.
- Azcue et al. 2026 (Azcue et al. 2026) (\(n=59\) ME/CFS, \(n=96\) PCC): Combined GPCR AAb profiling with comprehensive autonomic and cognitive testing. Confirmed \(\beta_2\)-adrenergic AAb elevation in ME/CFS vs PCC and HCs (\(F_{2,186}=3.15\), \(p=0.046\)). Strongest quantitative \(\beta_2\)-AAb–autonomic correlation to date (\(r=0.45\), \(p=0.001\)). Novel finding: M1, M3, M4 AAbs positively correlated with verbal/working memory. Distinct PCC vs ME/CFS profiles suggest trigger-specific immunological trajectories.
1.2 Treatment Trial Evidence
- Immunoadsorption pilot (Scheibenbogen 2018) (Scheibenbogen et al. 2018): 10 post-infectious ME/CFS patients with elevated \(\beta_2\) antibodies received 5 immunoadsorption sessions. 70% showed rapid improvement during treatment; 30% sustained improvement at 6–12 months.
- Repeat immunoadsorption (Tölle et al. 2020) (Tölle et al. 2020): 5 previously-responsive ME/CFS patients retreated ~2 years after initial IA. Modified protocol achieved 80–90% IgG/\(\beta_2\)AR-AB reduction; 4/5 showed sustained improvement at 6–12 months. Established retreatment feasibility for relapsing patients.
- Immunoadsorption cohort (Stein et al. 2024) (Stein et al. 2025): 20 post-COVID ME/CFS patients with elevated \(\beta_2\)-AR autoantibodies (interim at n=10: (Stein et al. 2023)). IgG reduced 79%, autoantibodies reduced 77%. 70% responders with \(\geq\) 10 point SF-36 Physical Function increase. Benefits sustained to 6 months. This represents the strongest observational evidence to date for autoantibody-mediated pathophysiology.
- Independent-centre immunoadsorption (Anft et al. 2025) (Anft et al. 2025): 12 post-COVID patients with elevated ANS receptor autoantibodies at Marien Hospital Herne (Ruhr-University Bochum—not Charité). Autoantibodies eliminated; pro-inflammatory cytokines (IL-4, IL-2, IL-1\(\beta\), TNF, IL-17A) reduced; soluble spike protein reduced. Neuropsychological function improved and modest grip strength gain at 30 days. However, ME/CFS symptom questionnaire scores were not significantly improved, with autoantibody rebound within 1 month. This discordance between objective biomarker improvement and subjective symptom non-response—at an independent centre—introduces genuine uncertainty about whether autoantibody removal alone is sufficient for clinical benefit, or whether the small sample (n=12) lacked power to detect a real effect.
- Daratumumab pilot (Fluge et al. 2025) (Fluge et al. 2025): Anti-CD38 therapy targeting plasma cells (the antibody factories). 10 female ME/CFS patients; 60% showed marked improvement. SF-36 PF increased from 25.9 to 55.0 (\(p\)=0.002). Responders achieved near-normal function (SF-36 scores 80–95). Low baseline NK-cell count predicted non-response.
- BC007 case report (Hohberger 2021) (Hohberger et al. 2021): DNA aptamer neutralizing GPCR autoantibodies produced dramatic improvement in a Long COVID patient: fatigue normalized, brain fog resolved, retinal microcirculation improved within hours. However, the subsequent Phase II trial failed to show superiority over placebo at the population level.
The following sham-controlled RCTs and mechanistic studies have reported preliminary results or are recruiting as of May 2026:
IA-PACS-CFS (Charité Berlin) (Preßler et al. 2024): NCT05710770. Double-blind, sham-controlled RCT; 2:1 allocation; n=65 (45 IA, 20 sham). Patients hospitalised 9–12 days receiving 5 IA sessions every other day. Primary endpoint: Chalder Fatigue Scale at 60 days. Preliminary results presented at Berlin ME/CFS conference (May 2026): No statistically significant difference between treatment and placebo groups on Chalder Fatigue Scale (Rücker 2026). These are conference data, not peer-reviewed publication, and should be interpreted with caution given the poor reproducibility of conference abstracts. Criticism emerged during conference: the trial did not specifically recruit subjects with confirmed autoantibodies, potentially diluting treatment effect. Some participants showed substantial benefit in preliminary data, prompting calls for subgroup analysis focusing on antibody-positive patients. This subgroup analysis has not yet been conducted.
Charité antibody-positive case series (May 2026): Presented at Berlin conference as contrast to IA-PACS-CFS. 15 ME/CFS patients with detectable autoantibodies underwent immunoadsorption. 7 of 15 showed marked improvement in overall condition. However, the study lacked a control group, and 47% response in fatigue trials is indistinguishable from placebo effect (typical placebo response in ME/CFS trials: 40–60%). This finding should be treated as exploratory rather than evidence of efficacy (Rücker 2026).
EXTINCT (Hannover Medical School) (Hannover Medical School 2023): NCT05954325. Double-blind, sham-controlled RCT; n=63. Five IA sessions within 14 days. Primary endpoint: Chalder Fatigue Scale at 12 weeks. Completed September 2025; results not yet released as of May 2026 (Rücker 2026). First independent-institution (non-Charité) sham-controlled IA RCT.
University Medical Center Mainz trial: N=40 Long COVID patients receiving immunoadsorption. Results not yet released as of May 2026 (Rücker 2026).
Hannover COVID-associated ME trial: N=60 patients with COVID-associated ME. Results not yet released as of May 2026 (Rücker 2026).
University of Amsterdam trial: Still recruiting as of May 2026 (Rücker 2026).
IMPACT (Charité Berlin) (Charíté - Universitätsmedizin Berlin 2026): NCT07529197. Prospective observational; n=50. Five outpatient IA sessions on days 1, 2, 4, 6, and 8. Primary endpoint: SF-36 Physical Function at 8 weeks; novel secondary endpoint: memory B-cell dynamics. Recruiting since March 2026. First study to systematically measure memory B-cell changes after IA—addressing the mechanistic question of why some patients relapse (memory B-cell repopulation and re-synthesis of autoantibodies).
Together, the completed sham-controlled trials present a mixed picture: the negative IA-PACS-CFS result challenges the autoantibody hypothesis at the population level, but the methodological criticism (lack of antibody-positive preselection) and the positive antibody-positive case series suggest a subgroup-specific effect may exist. The antibody-positive responder rate in the case series was 7/15 (47%), though this lacks a control group and the sample is small. The pending results from EXTINCT, Mainz, and Hannover trials will provide critical additional data. If all sham-controlled trials are negative, the autoantibody hypothesis faces a serious challenge comparable to the rituximab-to-RituxME collapse. If positive in antibody-positive subgroups, they would validate precision-medicine approaches targeting this mechanism.
1.3 Methodological Controversy
Important caveats exist regarding GPCR autoantibody testing:
- POTS replication failure (2022) (Hall et al. 2022): 116 POTS patients vs. 81 controls showed no differences in ELISA-derived GPCR autoantibody concentrations. 98.3% of POTS patients and 100% of controls had \(\alpha_1\)-adrenergic receptor antibodies above threshold. The authors concluded CellTrend ELISAs “have no diagnostic value for POTS.”
- Functional vs. binding assays: The positive studies largely used CellTrend ELISAs (binding assays), while the cardiomyocyte bioassay (measuring functional antibody activity) may be more specific but is not commercially available.
- Conflict of interest: CellTrend holds a patent for \(\beta\)-adrenergic receptor antibodies in CFS diagnosis, jointly with Charité.
- Comprehensive null from orthogonal platforms (2025) (Germain et al. 2025): Germain et al. (Cornell/Hanson group) screened 172 participants across 7,542 antibody-antigen interactions using REAP (6,183 exoproteome proteins) and Luminex (1,134 autoantigens). Complete null: no \(q\)-value below 0.68. All GPCR targets (\(\beta_1\)/\(\beta_2\)/\(\beta_3\)-adrenergic, M1–M4 muscarinic) negative. Caveat: REAP displays extracellular domains individually and may miss conformational multi-loop GPCR epitopes; cohort was chronic pre-COVID ME/CFS (mean 11–14 years).
Despite methodological concerns, the treatment evidence remains notable: if autoantibody removal (immunoadsorption) and autoantibody-producing cell depletion (daratumumab) produce clinical improvement, the autoantibodies may be pathogenic regardless of assay limitations—though the BC007 Phase II placebo-controlled failure and the Germain null result together raise the possibility that the open-label treatment responses reflect non-autoantibody mechanisms of immunoadsorption and plasma cell depletion (e.g., removal of pro-inflammatory IgG glycoforms, broader immunomodulation).
Despite promising treatment signals, the GPCR autoantibody field faces unresolved methodological issues that limit the strength of current conclusions:
- The CellTrend ELISA (used in most positive studies) has been questioned by an independent replication failure in 116 POTS patients (Hall et al. 2022), where 98.3% of patients and 100% of controls exceeded the positivity threshold—suggesting the assay may lack specificity.
- CellTrend holds a joint patent with Charité for \(\beta\)-adrenergic receptor antibodies in CFS diagnosis, creating a potential conflict of interest that has not been addressed by independent validation using alternative assay platforms.
- All published treatment trials to date (immunoadsorption n=10, n=20, n=12; daratumumab n=10) are open-label without placebo control. The 60–70% response rates cannot be distinguished from placebo effect, regression to the mean, or natural fluctuation without randomised controlled trials. The BC007 case report (Hohberger et al. 2021) was similarly uncontrolled. The Anft et al. (2025) independent-centre study (Anft et al. 2025) adds a further concern: autoantibodies were successfully eliminated but ME/CFS symptom scores did not significantly improve, with rebound within one month.
- The BC007 Phase II trial—the only placebo-controlled study of autoantibody-targeted therapy—failed to show superiority over placebo, directly challenging the therapeutic hypothesis despite the positive case report.
- Two sham-controlled immunoadsorption RCTs have completed enrolment (IA-PACS-CFS, n=66 (Preßler et al. 2024); EXTINCT, n=63 (Hannover Medical School 2023)) with results expected in 2025–2026. These will provide the first definitive test of immunoadsorption efficacy.
2 Speculative Hypotheses Emerging from GPCR Research
The daratumumab success vs. rituximab failure reveals a critical insight: B cells (CD20+) are precursors; plasma cells (CD38+) are the factories. Long-lived plasma cells can survive for decades in bone marrow and gut niches, continuously secreting autoantibodies without B cell replenishment.
Hypothesis: ME/CFS is maintained by “sanctuary” plasma cells that escaped B-cell depletion:
- Initial trigger generates autoreactive B cells
- Some differentiate into long-lived plasma cells in survival niches
- These plasma cells produce GPCR autoantibodies indefinitely
- Rituximab depletes B cells but not established plasma cells—autoantibody production continues
- Daratumumab directly kills plasma cell factories, stopping production
Evidence level: Moderate. The 8–9 month delay before maximum daratumumab benefit supports this (existing autoantibodies must decay after factory elimination).
Therapeutic implication: Combining immunoadsorption (remove existing antibodies) with daratumumab (eliminate factories) might produce faster, more complete responses.
GPCR autoantibodies may exert their effects primarily through endothelial dysfunction:
- \(\beta_2\)-adrenergic receptor autoantibodies impair endothelial vasodilation
- Muscarinic receptor autoantibodies disrupt endothelial NO production
- Impaired vasodilation → tissue hypoperfusion
- Hypoperfusion → mitochondrial dysfunction
- Mitochondrial dysfunction → cellular energy crisis → symptoms
The BC007 case report supports this: retinal microcirculation improved within hours of autoantibody neutralization (Hohberger et al. 2021)—faster than any cellular recovery could explain. The vascular effect was immediate.
Evidence level: Low-Moderate. Mechanistically plausible; BC007 microcirculation data supportive; needs direct testing.
Therapeutic implication: Vascular-supportive therapies (L-citrulline, statins) might synergize with autoantibody removal.
A 2025 preprint (Hackel et al. 2025) demonstrated that GPCR autoantibodies drive monocyte dysfunction in post-COVID ME/CFS, causing elevated MIP-1\(\delta\), PDGF-BB, and TGF-\(\beta\) 3. This suggests autoantibodies don’t just block receptors—they actively drive inflammation:
- GPCR autoantibodies bind monocyte surface receptors
- Binding triggers inflammatory cytokine production
- Cytokines cause systemic inflammation and tissue damage
- Tissue damage generates more autoantigen exposure
- Cycle perpetuates autoantibody production
Evidence level: Low-Moderate (single preprint, not yet replicated).
Therapeutic implication: Monocyte-targeted therapies might complement autoantibody removal.
The daratumumab trial showed 60% marked improvement and 40% non-response. What distinguishes responders from non-responders?
Potential factors:
- Autoantibody presence: Non-responders may have different (non-GPCR) autoantibodies, or non-autoimmune ME/CFS
- NK cell status: Low baseline NK cells predicted non-response (immune dysregulation pattern)
- Illness duration: Longer illness may cause irreversible downstream damage
- Plasma cell location: Some sanctuary sites may be less accessible to daratumumab
Identifying responder biomarkers is critical for treatment personalization.
3 Corroborating Evidence from Long COVID Passive Transfer Studies
The causal link between circulating IgG and neurological symptoms has been established by passive transfer experiments from four independent research groups (2021–2026), providing the strongest evidence to date that autoantibodies are not bystanders but active drivers of post-viral symptomatology:
- Goebel et al. (2021) — fibromyalgia IgG → pain hypersensitivity and reduced locomotion in recipient mice (Goebel et al. 2021). IgG accumulated in DRG.
- Mignolet et al. (2026) — long COVID IgG → mechanical allodynia and thermal hyperalgesia without cognitive effects or neuroinflammation (Mignolet et al. 2026).
- Chen et al. (2026) — long COVID IgG → persistent mechanical hypersensitivity; pathogenic IgG persisted 2 years (Chen et al. 2026).
- Santos Guedes de Sá et al. (2026) — long COVID IgG → fatigue-like behaviour, balance/coordination loss, thermal hyperalgesia, and reduced intraepidermal nerve fibre density in recipient mice (Santos Guedes de Sa 2026). Published in Cell after peer review. >21,000 human protein array identified >70 CNS/PNS autoantigens including MED20 and USP5. IgG crossed BBB (5% at day 5). Abnormal neuronal activation in pain, fatigue, memory, and emotional regulation circuits. Three independent groups reported similar findings (including Chen et al.).
This convergence — four independent labs, two disease contexts (fibromyalgia and long COVID), peer-reviewed publication in top-tier journals — establishes IgG pathogenicity with high confidence. The findings provide indirect support for the GPCR autoantibody hypothesis in ME/CFS, since the same methodological principium (pathogenic circulating IgG) has been demonstrated in related post-viral conditions. The key gap is the absence of direct passive transfer experiments using ME/CFS patient-derived IgG, rather than fibromyalgia or long COVID IgG.
Study: (synthesis across four passive transfer studies; certainty: 0.65, four independent groups, top-tier peer review, convergent behavioural + histological + proteomic evidence; but declining from 0.75 after adversarial review because: zero studies used ME/CFS IgG, Germain 2025 null is a serious counterweight, CNS findings from Santos Guedes only; ME/CFS-specific passive transfer not yet performed).
4 Competing Hypotheses for Immunoadsorption Mechanism
The Anft et al. (2025) discordance—autoantibodies eliminated but symptoms unchanged—combined with the BC007 Phase II failure and Germain 2025 comprehensive null, demands consideration of alternative mechanisms by which immunoadsorption might produce the 70% open-label response rate observed at Charité.
(Certainty: 0.40.) Immunoadsorption is non-selective—it removes all IgG, not just GPCR autoantibodies. The therapeutic effect may come from removing other pathogenic immunoglobulin fractions: pro-inflammatory IgG glycoforms (agalactosylated IgG, known to drive inflammation via Fc\(\gamma\)RIIIa in RA), circulating immune complexes that activate complement, anti-neuronal antibodies not captured by GPCR panels, anti-mitochondrial antibodies, or persistent soluble spike protein (confirmed reduced by Anft 2025 (Anft et al. 2025)). The Germain 2025 REAP/Luminex null (Germain et al. 2025) is devastating for GPCR autoantibodies specifically, but IA removes everything. The “active ingredient” may be clearance of a pathogenic IgG fraction that current assays do not measure.
Falsifiable prediction: IA responders will have more agalactosylated (pro-inflammatory) IgG pre-treatment than non-responders. Selective removal of agalactosylated IgG (using lectin affinity columns) would reproduce IA’s clinical effect without removing GPCR autoantibodies. Not yet tested (2026).
(Certainty: 0.50.) The Anft 2025 symptom non-response (Anft et al. 2025) may reflect insufficient duration of autoantibody suppression rather than mechanistic failure. Autoantibody rebound within one month means patients never achieved a sustained low-autoantibody state. Even if autoantibodies cause tissue damage (endothelial dysfunction, receptor internalization, neuroinflammation), removing the cause does not instantly reverse the consequence. Recovery requires: (1) autoantibody clearance (days—achieved); (2) receptor re-expression on cell surfaces (days to weeks); (3) endothelial repair and vascular remodelling (weeks); (4) neuroinflammatory resolution via microglial deactivation (weeks to months); (5) metabolic recovery from energy deficit (weeks to months). The Tölle 2020 retreatment data (Tölle et al. 2020)—4/5 responded to repeat IA—supports that the key may be sustained suppression via repeated courses rather than a single course.
Falsifiable prediction: In the pending RCTs, response will correlate with duration of autoantibody suppression (measured by post-IA autoantibody half-life). Patients with faster rebound will be non-responders.
(Certainty: 0.45.) Massive IgG removal (\(\sim\) 80%) is a profound perturbation that triggers homeostatic immune recalibration: compensatory B-cell activation, cytokine milieu shift (confirmed by Anft 2025: IL-4, IL-2, IL-1\(\beta\), TNF, IL-17A all reduced (Anft et al. 2025)), possible Treg expansion, and FcRn upregulation altering immune complex clearance. This would explain why BC007 failed (specific neutralisation without the broader reset), why open-label IA works (real immunomodulation, not autoantibody-specific), and why daratumumab works (plasma cell depletion also causes broad immunomodulation). A critical discriminating experiment: FcRn inhibitors like efgartigimod (FDA-approved for myasthenia gravis) produce \(\sim\) 70% IgG reduction non-invasively via accelerated IgG catabolism. If efgartigimod reproduces IA’s clinical effect in ME/CFS, the invasive procedure is unnecessary and IgG-mediated pathology—but not necessarily autoantibody-specific pathology—is confirmed.
Falsifiable prediction: IgG-depleting interventions that do not remove autoantibodies specifically (efgartigimod, rozanolixizumab) should produce similar clinical effects to IA in ME/CFS. Not yet tested (2026).
5 Undocumented Biological Phenomena
Based on the GPCR autoantibody literature, several biological phenomena have never been directly examined:
- Bone marrow plasma cell populations: Do ME/CFS patients have expanded long-lived plasma cells producing GPCR autoantibodies? No bone marrow studies have examined this.
- Gut-associated plasma cells: The gut wall contains plasma cell niches. Do these contribute to autoantibody production in ME/CFS?
- Autoantibody epitope specificity: Which specific receptor epitopes do ME/CFS autoantibodies target? Epitope mapping might predict functional effects.
- Functional vs. binding antibody correlation: How well do ELISA-detected antibodies correlate with functional bioassay results in the same patients?
- Autoantibody fluctuation with symptoms: Do autoantibody titers change during PEM episodes or remissions?
- GPCR receptor internalization: Do autoantibodies cause receptor downregulation through chronic stimulation?
6 Evidence Assessment Summary
| Finding | Evidence Level | Notes |
|---|---|---|
| GPCR autoantibodies elevated in ME/CFS | Low-Moderate | CellTrend ELISA cohorts positive; Germain 2025 REAP/Luminex null |
| Symptom correlation with titers | Low-Moderate | Sotzny 2021; Swedish replication weak (\(r < 0.3\)) |
| Immunoadsorption efficacy (Charité) | Moderate | Lancet 2024 (n=20); no placebo control; single centre |
| Immunoadsorption efficacy (independent) | Low | Anft 2025 (n=12): autoantibodies cleared but symptoms not improved (Anft et al. 2025) |
| Immunoadsorption RCTs | Pending | IA-PACS-CFS (n=66) and EXTINCT (n=63) completed; results awaited |
| Daratumumab efficacy | Moderate | 60% response; open-label |
| BC007 efficacy | Low | Case reports positive; Phase II failed |
| Peripheral (not CNS) origin | Moderate | No CSF autoantibodies (Bynke 2020) |
| CellTrend assay specificity | Controversial | POTS study questions diagnostic value |
| Broad autoantibody screen (REAP/Luminex) | Strong null | Germain 2025: 7,542 interactions; no signal |
Overall assessment: The GPCR autoantibody hypothesis is at a critical inflection point. The detection evidence has weakened substantially: the most comprehensive screen to date (Germain 2025 (Germain et al. 2025), two orthogonal platforms, 7,542 interactions) found no autoantibody signal, and the only placebo-controlled treatment trial (BC007 Phase II) failed. The Anft 2025 (Anft et al. 2025) independent-centre study adds a further complication: autoantibodies were successfully eliminated but symptoms did not improve—raising the possibility that even the open-label treatment responses from Charité reflect non-autoantibody mechanisms of immunoadsorption (bystander IgG clearance, immune reset, or placebo effects of invasive procedures; see Bystander Clearance: IA Works by Removing Non-Autoantibody IgG, Immune Reset: IA as Immunomodulation Independent of Autoantibodies). However, four pivotal events are imminent: (1) IA-PACS-CFS sham-controlled RCT results (n=66 (Preßler et al. 2024)), (2) EXTINCT sham-controlled RCT results (n=63, independent centre (Hannover Medical School 2023)), (3) RESETME daratumumab RCT, and (4) IMPACT memory B-cell dynamics study (Charíté - Universitätsmedizin Berlin 2026). If all RCTs are negative, the autoantibody hypothesis takes a blow comparable to the rituximab-to-RituxME collapse—the second promising open-label immune intervention to fail under placebo control. If positive, they would constitute the most important ME/CFS therapeutic breakthrough to date and would force re-examination of why detection assays fail while removal therapies succeed.