Your Pupil Could Test Whether Autoantibodies Impair Autonomic Function — A Hypothesis
A handheld device that measures how fast your pupil constricts and dilates in response to light costs about €3,000, is FDA-cleared, and has been validated in concussion, diabetes, multiple sclerosis, and pure autonomic failure. The measurement takes 30 seconds. You don’t have to stand up. You don’t have to exercise. You just sit in a dim room and the device records your pupil’s light reflex.
No modern automated pupillometry study has been published in ME/CFS. (No registered trials were identified in a search of ClinicalTrials.gov as of July 2026.)
Why it matters: the receptors controlling your pupil are the exact same receptors targeted by autoantibodies found in many ME/CFS patients. Your pupil could tell us whether those autoantibodies are actually doing anything.
1 The pupil is a window into the autonomic nervous system
Your autonomic nervous system controls things you don’t think about: heart rate, digestion, sweating, blood pressure, and the size of your pupils. It has two branches:
- Parasympathetic (“rest and digest”) — slows things down, conserves energy. Constricts the pupil.
- Sympathetic (“fight or flight”) — speeds things up, mobilises energy. Dilates the pupil.
The pupil’s response to a flash of light — the pupillary light reflex, or PLR — is one of the fastest, simplest, and most reproducible autonomic measurements available. A brief light stimulus hits the retina. The signal travels through the optic nerve to the Edinger-Westphal nucleus in the midbrain, then to the ciliary ganglion, and finally to the iris sphincter muscle, which constricts the pupil. The whole reflex takes about 200 milliseconds.
When the light goes off, the pupil dilates again — this time driven by the sympathetic branch, which sends a signal from the superior cervical ganglion in your neck to the alpha-1 adrenergic receptors on the iris dilator muscle.
What researchers measure with a modern pupillometer is precise: constriction amplitude (how much the pupil shrinks), constriction latency (the delay before constriction starts), dilation velocity (how fast it opens again), and baseline diameter (the resting balance between the two systems).
Each of these parameters maps to a specific receptor type. Constriction depends on M3 muscarinic receptors — these are cholinergic, parasympathetic (G protein-coupled receptors, or GPCRs). Dilation depends on alpha-1 adrenergic receptors — these are sympathetic GPCRs (Lisowski et al. 2025). This matters because both receptor types are targets of autoantibodies found in ME/CFS.
2 The autoantibody controversy
In 2016, Loebel and colleagues published a study showing that 29.5% of 268 ME/CFS patients had elevated autoantibodies against beta-2 adrenergic and M3/M4 muscarinic receptors (Loebel et al. 2016). In 2021, Szklarski and colleagues confirmed this and found that infection-triggered ME/CFS patients had significantly higher autoantibody levels than patients without an infectious trigger — and that these antibodies correlated with symptom severity (Szklarski et al. 2021). In 2026, Azcue and colleagues replicated the finding in a post-COVID and CFS cohort (Azcue et al. 2026).
But in 2025, Germain and colleagues ran the largest screen to date: 172 participants, 7,542 antibody-antigen interactions using two different high-throughput platforms (REAP and Luminex). They found zero ME/CFS-specific autoantibody signal (Germain et al. 2025). All GPCR targets — beta-1, beta-2, beta-3 adrenergic, M1 through M4 muscarinic — were negative. No autoantibody differed between patients and controls at any statistically meaningful threshold. By study size, platform diversity, and statistical rigour, Germain 2025 is the strongest single study on this question. The ELISA-positive studies (Loebel 2016, Szklarski 2021, Azcue 2026) are smaller and use a single assay platform with known specificity limitations.
This is a genuine disagreement. Two different assay platforms — ELISA (positive) and REAP/Luminex (negative) — give opposite answers on the same biological question. The disagreement cuts both ways: ELISA may detect low-specificity binding, but REAP/Luminex uses recombinantly expressed linear epitopes that can misfold, potentially missing antibodies that require native GPCR conformation (Germain et al. 2025). In 2022, Vernino and colleagues tried to replicate the ELISA findings in POTS: they found 100% of healthy controls tested positive, making the diagnostic value of the test zero (Hall et al. 2022). A parsimonious interpretation is that the ELISA detects antibodies that bind but are functionally irrelevant (low specificity); the burden of proof falls on demonstrating functional significance.
Here is where it gets clinically consequential. BC007, a DNA aptamer that neutralises GPCR autoantibodies, showed dramatic improvement in a Long COVID pilot case (Hohberger et al. 2021). An immunoadsorption trial in ME/CFS is underway (Stein et al. 2023). Some patients are getting tested for these antibodies. But nobody knows whether a positive ELISA result means the antibodies are actually impairing receptor function, or whether it means the test has a specificity problem — detecting antibodies that are present but harmless, or worse, detecting nothing real at all. Important caveat: None of these therapies are approved for ME/CFS. IVIG is off-label; immunoadsorption is trial-only; BC007 is not available as a clinical treatment. No clinical recommendation is implied by this discussion.
3 What the pupil can tell us
M3 muscarinic receptors and alpha-1 adrenergic receptors are present throughout the body — on blood vessels, on the heart, on gut smooth muscle, on immune cells, on mast cells. But in virtually every tissue, measuring receptor function is technically difficult. You’d need a biopsy, or an invasive catheter, or a specialised PET scan.
The pupil is different. The M3 and alpha-1 receptors on the iris are accessible through a transparent medium (the cornea). They are activated by a precise, repeatable stimulus (a flash of light). Their response is mechanically simple — constrict or dilate — and can be quantified at millisecond resolution with a device the size of a small flashlight.
If GPCR autoantibodies in ME/CFS are functional — meaning they bind to the receptor and impair its signalling — a pupillometer may detect the impairment, assuming (a) the antibody reaches the iris at sufficient concentration across the blood-ocular barrier, (b) the receptor isoforms expressed in iris tissue match those targeted, and (c) no compensatory mechanism (e.g., receptor upregulation) masks the effect. M3-blocking antibodies should slow constriction. Alpha-1-blocking antibodies should slow dilation.
If GPCR autoantibodies are non-functional — meaning they bind to the receptor without affecting its signalling, or they bind to something else entirely that cross-reacts with the ELISA — the pupil should be normal, regardless of what the blood test says. (The converse does not follow: abnormal PLR in a patient with antibodies could also arise from medications, brainstem pathology, deconditioning, or autonomic dysfunction independent of any autoantibody — equifinality.)
This approach approximates an in vivo bioassay, using native tissue at physiological receptor densities.
4 Cross-disease evidence
Pupillometry has been studied as an autonomic biomarker across multiple conditions with features overlapping ME/CFS. (Note: conditions listed involve different disease mechanisms — microvascular, demyelinating, traumatic — so the convergence is on method utility, not shared pathophysiology.)
In diabetic autonomic neuropathy, PLR constriction amplitude and dilation velocity are reduced, and the magnitude of reduction correlates with disease duration. Li and colleagues (2026) showed that scotopic-photopic pupil diameter difference could serve as a biomarker for long-term glycaemic management (Li et al. 2026). Thakar and colleagues (2025) confirmed the correlation between PLR parameters and retinal nerve fibre layer thickness — a structural correlate (Thakar et al. 2025).
In multiple sclerosis, Parmak Yener and colleagues (2026) demonstrated that machine learning applied to pupillographic features could discriminate patients with autonomic impairment from those without (Parmak Yener et al. 2026). De Rodez Benavent and colleagues (2019) linked retinal and pupillary autonomic abnormalities to cardiovascular autonomic dysfunction in early MS — showing that the eye reflects systemic autonomic state (Rodez Benavent et al. 2019).
In concussion, Master and colleagues (2020, JAMA Ophthalmology) published a study of 352 adolescents showing that PLR metrics distinguish concussed from healthy athletes. This was a large, rigorous validation of PLR as an objective physiologic biomarker — just in a different condition (Master et al. 2020).
In post-COVID syndrome, Smit and colleagues (2026) found altered task-evoked pupillary dynamics in 397 patients versus 129 controls. The Index of Pupillary Activity was significantly lower in post-COVID patients across all task difficulty levels (Smit et al. 2026).
In CRPS (complex regional pain syndrome), Drummond and Finch (2022) used pupil dilation to arousal stimuli as a probe of locus coeruleus function, finding that the affected side showed smaller pupils — consistent with ipsilateral noradrenergic deficit (Drummond and Finch 2022).
Across all these conditions, pupillometry detects autonomic dysfunction Rizzuto et al. (2025). The pattern varies by disease, but the method works. A notable caution: Egg and colleagues (2002) found that the one PLR parameter most intuitively related to autonomic instability — pupillary hippus, the spontaneous oscillation of pupil size — was inversely correlated with fatigue in MS (r = -0.52). More pupil oscillation meant less fatigue (Egg et al. 2002). This null should temper expectations: not every PLR parameter will track symptoms, and the direction of effect may not be what intuition predicts.
The only direct CFS pupillometry study was published in 1997 by Sendrowski and colleagues. It used pharmacological methodology (phenylephrine mydriasis) rather than modern automated pupillometry, had fewer than 20 participants, and tested a specific hypothesis about sympathetic denervation hypersensitivity (Sendrowski, Buker, and Gee 1997). It was never replicated.
A 29-year gap between the only ME/CFS study and a technology that is now routine in neurology, ophthalmology, and sports medicine — that is the evidence gap.
5 What would the study look like?
A first-pass study would require nothing that doesn’t already exist:
| Component | Status |
|---|---|
| Device | NeurOptics PLR-3000 or equivalent — FDA-cleared, €2–5K, used daily in concussion clinics |
| Participants | 60 GPCR-autoantibody-positive ME/CFS + 60 negative + 60 healthy controls — achievable at a site with an existing ME/CFS cohort and validated autoantibody assay (screening ~200+ patients to identify 60 positives at ~29.5% seropositivity rate) |
| Autoantibody testing | Already established at Charité (CellTrend ELISA) and through published REAP/Luminex protocols (Germain 2025) |
| Protocol | Single session, 30 seconds per measurement, supine, no exercise required. Medication confounds: beta-blockers, anticholinergics, SSRIs, and antihistamines all affect pupil function. These will be documented and analysed as covariates, but with 180 participants and multiple subgroups, statistical control is limited. Medication washout is not proposed due to ethical and practical barriers |
| Analysis | Compare constriction velocity and dilation velocity between groups. Control for age, medications, and ambient light |
| Duration | Data collection: 30 seconds per measurement, ~30–45 minutes per participant including setup, consent, and autoantibody blood draw. Total: feasible within 6–12 months at a single site |
| Cost | Device €2–5K (~1–5% of total; personnel, recruitment, and antibody testing dominate) |
The primary analysis tests two predictions: 1. ME/CFS patients with elevated GPCR autoantibodies will show slower PLR constriction velocity (if M3 antibodies are functional) and/or slower dilation velocity (if alpha-1 antibodies are functional) compared to antibody-negative patients and controls. 2. PLR parameters will correlate with autonomic symptom severity (COMPASS-31 score) and orthostatic intolerance measures.
The secondary analysis tests whether autoantibody subtype — M3-dominant versus alpha-1-dominant — predicts which PLR parameter is affected. If this double dissociation holds, pupillometry could serve as a receptor-specific bioassay, telling clinicians which autonomic pathway is impaired without a blood draw.
6 Three scenarios: what each would mean
6.1 Scenario A: Positive — PLR is abnormal in autoantibody-positive patients
The pupil test reveals functional receptor impairment in patients with positive autoantibody titres — but not in patients with negative titres and not in healthy controls. The dissociation between M3 antibodies and constriction (versus alpha-1 antibodies and dilation) is clean.
What changes for patients: A positive pupillometry finding would be the first evidence that GPCR autoantibodies are pathogenic in ME/CFS — not just a laboratory finding. It would mean the autoantibodies are impairing autonomic receptor function at physiological concentrations in native tissue. This directly strengthens the rationale for immunomodulatory therapies (IVIG, immunoadsorption, BC007). A patient could be told: “Your antibodies are impairing your receptors. Therapies that remove or neutralise them have a mechanistic target we can actually measure.”
Pupillometry could also serve as a treatment-response monitor. If a therapy works, pupil function should improve. A 30-second measurement before and after treatment replaces subjective symptom reports with an objective physiological endpoint.
6.2 Scenario B: Null — PLR is normal regardless of autoantibody status
The pupil responds normally in all ME/CFS groups — antibody-positive, antibody-negative, no difference from controls. The autoantibodies detected by ELISA are either non-functional, non-pathogenic, or not reliably detected.
What changes for patients: The Germain 2025 null result is functionally confirmed. The search for GPCR autoantibody-depletion therapies should be de-prioritised. Patients with positive ELISA results but no functional receptor impairment can avoid pursuing treatments that have no mechanistic rationale.
What changes for researchers: The GPCR autoantibody chapter in ME/CFS would close or narrow sharply — provided the blood-ocular barrier is not blocking antibody access. An antibody that is functionally relevant but cannot cross the barrier would produce a false-negative PLR while still being pathogenic elsewhere. This can be addressed by measuring antibody concentration in tear fluid or aqueous humour in a subset of participants. If barrier penetration is confirmed and PLR is still normal, the GPCR hypothesis in ME/CFS would narrow sharply. Resources shift to non-GPCR antibody targets (ANT, thyroid, ganglionic AChR), which are a separate question not addressed by this study. The pupillometry method, having failed to detect any functional effect, would not be pursued as a biomarker in ME/CFS — though it remains useful in the overlapping conditions where it is already validated.
Important: A null pupillometry result does not disprove all autoantibody involvement in ME/CFS — only the specific GPCR hypothesis that was tested.
6.3 Scenario C: Mixed — PLR abnormal only in a subset, or effects too small for clinical use
A subset of antibody-positive patients shows PLR abnormalities, but the effect is small, or the overlap between groups is too large for individual diagnosis. PLR may track autonomic dysfunction but can’t distinguish pathogenic from non-pathogenic antibodies at the individual patient level.
Interpretation boundary: Scenario C is distinguishable from a true positive only if a pre-specified effect-size threshold is met (e.g., Cohen’s d ≥ 0.5 for constriction velocity difference between antibody-positive and negative groups). Effects below this threshold should be treated as a null result, not as “partial support.” Without this boundary, the hypothesis is unfalsifiable (lakens2013calculating?).
What changes for patients: Ambiguity persists. The PLR effect may be real but too small to guide individual clinical decisions. Larger cohorts or confound-standardised protocols would be needed. The GPCR autoantibody question remains open, but this would favour the view that if these antibodies are functional, their effect is modest or restricted to a narrow subset.
What does not happen under this scenario: no clinical test, no treatment monitor, no receptor-specific bioassay, no individual-level diagnostic stratification.
7 Why this matters now
There is a specific quality to this evidence gap that is unusual. Most biomarker gaps in ME/CFS are expensive to close. You need mass spectrometry for metabolomics. You need sequencing for genomics. You need specialised MRI for neuroimaging.
Pupillometry needs a €3,000 device, a dim room, and 30 seconds per patient. (The device cost is ~1–5% of total study cost; researcher time, participant recruitment, and autoantibody testing dominate.)
The technology is FDA-cleared. Normative data exist. Cross-disease validation has been done. The mechanistic rationale — GPCR autoantibodies targeting the exact receptors controlling the pupil — is specific and falsifiable.
What’s missing is a study — specifically, someone with an existing ME/CFS cohort, a GPCR autoantibody dataset, and a ~€3,000 equipment budget to spend one year answering a question that the field has been arguing about for a decade.
Note: This article draws on the systematic literature review and evidence synthesis documented in the pupillometry-autonomic-biomarker integration plan and the formal integration into the ME/CFS paper (Chapter Biomarker Research, Biomarker Research; Chapter 8, Neurological Dysfunction; Chapter 14d, Cross-Disease Connections) completed on 2026-07-09.
8 Clinical bottom line
This article describes a research hypothesis, not a clinical tool. No clinical action is warranted at this time. Specifically: - Do not order pupillometry for ME/CFS patients — it is not validated for this population and there are no interpretation norms. - Do not order GPCR autoantibody testing for clinical decision-making — the assay controversy (ELISA vs REAP/Luminex) means the result’s meaning is unclear. - Do not pursue immunomodulatory therapy (IVIG, immunoadsorption, BC007) based on a positive autoantibody test — none are approved for ME/CFS and mechanistic rationale is unconfirmed. - If a patient asks about this topic, the honest answer is: “We don’t know yet whether these antibodies matter. A study is needed.”