Vitals Knowledge Vault

GLP-1 RA NAION Safety Signal

10 min read

GLP-1 RA NAION Safety Signal

TL;DR

A 2026 meta-analysis found GLP-1 receptor agonists associated with non-arteritic ischemic optic neuropathy (NAION) — OR 1.70 (95% CI 1.23–2.36) at the class level, absolute excess risk ~0.037%, NNH ~2,700. The semaglutide-specific signal is stronger vs active comparators (HR 2.19–2.81); the FDA is in active review as of April 2026 with no Black Box or REMS mandated. Large propensity-matched cohorts find no significant or protective effect. The benefit-risk ratio remains clearly favorable for most patients.

Why it Matters for Vitals

  • Ben uses GLP-1 RAs — this is a personal safety-relevant finding
  • NAION is irreversible — early symptom recognition matters
  • Patients with prior NAION in one eye have highest absolute risk for the other eye (~20–25% bilateral within 5 years)
  • Slow GLP-1 titration may be a reasonable precaution for disc-at-risk patients (mechanistic extrapolation)
  • No wearable prodrome exists — NAION cannot be detected via HRV, sleep, or activity data
  • The NAION signal is disproportionately concentrated in semaglutide (86.5% of published GLP-1 RA NAION cases)

Key Facts

  • NAION: Second most common optic nerve blindness cause; sudden, painless, usually upon waking; vision loss is usually permanent
  • Association: OR 1.70 meta-analytic (GLP-1 class level) — real at aggregate level but context-dependent; semaglutide-specific HR range 1.51–7.25 vs active comparators
  • Absolute risk: 0.037% excess (class-level); semaglutide-specific estimates ~2.19–2.90/10,000 person-years vs SGLT2i; NNH ~2,700–4,500 in highest-signal studies
  • Population cohorts: Large propensity-matched studies find no significant risk or protective effect; active-comparator studies show weaker/null associations than broad-comparator studies
  • Regulatory: FDA in active review (April 2026) — no Black Box or REMS mandated; EMA classified NAION as “very rare” for GLP-1 RAs
  • Signal concentration: NAION signal is disproportionately semaglutide (86.5% of published GLP-1 RA NAION cases); tirzepatide shows no significant ischemic optic neuropathy signal in primary pharmacovigilance analyses
  • Mechanism: Leading hypothesis — rapid glucose correction → optic nerve hypoperfusion in disc-at-risk patients. No direct drug toxicity established.

Safety Guidance

Evidence-backed actions:

  1. Inform all GLP-1 RA patients: sudden painless vision loss upon waking → seek immediate ophthalmologic care
  2. Baseline ophthalmologic evaluation for patients with prior NAION, disc-at-risk anatomy, or optic nerve anomaly
  3. Risk-benefit counseling: NNH ~2,700–4,500 vs substantial CV/mortality benefits — benefit clearly favorable for most patients
  4. Patients with prior unilateral NAION: highest priority for explicit counseling and urgent-care protocol
  5. Semaglutide users with highest-risk profiles (disc-at-risk, prior NAION in one eye, optic disc drusen): consider formal baseline ophthalmologic evaluation before GLP-1 initiation

Reasonable extrapolation:

  • Slow titration for disc-at-risk patients (rapid-glucose-correction mechanism hypothesis)
  • Annual ophthalmologic follow-up for long-term semaglutide users with overlapping NAION risk factors (obesity, T2D, sleep apnea, disc-at-risk)

Speculative — requires clinical judgment:

  • HbA1c trajectory monitoring for rapid correction risk
  • BP optimization (avoid nocturnal hypotension)
  • Optimize treatment of obstructive sleep apnea (independent NAION risk factor)

Not actionable:

  • HRV/wearable NAION prediction — no evidence; no prodrome

Claims NOT Supported

  • ❌ “GLP-1 RAs cause NAION” — causal link not established; observational data only
  • ❌ Pharmacovigilance RORs as relative risk estimates — signal only, not causal risk estimates
  • ❌ Wearable monitoring can prevent or predict NAION — no prodromal wearable signature exists
  • ❌ “FDA has mandated a Black Box warning” — no formal regulatory action taken as of April 2026; FDA is in active review
  • ❌ “Injectable Ozempic/Wegovy is riskier than oral Rybelsus” — formulation differential not established; injectable market share is the most parsimonious explanation
  • ❌ “High-dose semaglutide carries higher NAION risk” — dose–response has not been systematically studied within the therapeutic dose range
  • ❌ “Tirzepatide/liraglutide carry the same NAION risk as semaglutide” — evidence is semaglutide-dominant (86.5% of reported cases); other agents have too few cases to assess

Evidence Quality

EvidenceGradeSource
Meta-analytic association — GLP-1 class (OR 1.70)HighPMID 41498755
Semaglutide vs SGLT2i HR 2.81 (Danish-Norwegian)ModeratePMID 40098249
Semaglutide vs antidiabetic HR 2.19 (Danish single-country)ModeratePMID 39696569
NAION OR 3.92 — 78 RCTs (JAMA Ophthalmol)ModeratePMID 40810985
Asia-Pacific meta-analysis HR 2.62; semaglutide = 86.5% of GLP-1 NAION casesModeratePMID 40962119
Military Health System OR 0.36 protectiveModeratePMID 41217382
Chou et al null association HRs 0.72–1.51 (all subgroups NS)ModeratePMID 39491755
TriNetX Taiwan delayed signal at 2–3 years HR 2.39–2.44ModeratePMID 40146102
Systematic review GRADE very low certainty (OR 2.44 CI 0.59–10.15)LowPMID 41692115
Pharmacovigilance RORs 11–68Very LowPMID 40383360, 40133108
Rapid-glucose-correction mechanismGapPMID 40055951
Dose-dependencyGapNo systematic study
Injectable vs oral formulation differentialGapPMID 40289093 (signal only)
Regulatory: FDA active review, no Black Box/REMS; EMA “very rare”SupportedPMID 41857480, 40183385

Key References

Semaglutide-Specific Evidence:

  • PMID 40098249 — Diabetes Obes Metab 2025. Danish-Norwegian cohort, n=61,377 semaglutide users vs SGLT2i. HR 2.81 pooled.
  • PMID 39696569 — J Diabetes 2024/2025. Danish single-country, n=424,152 T2D, HR 2.19.
  • PMID 40810985 — JAMA Ophthalmol 2025. 78 RCTs, 73,640 participants. NAION OR 3.92 (CI 1.02–15.02).
  • PMID 40146102 — JAMA Ophthalmol 2025. TriNetX Taiwan, 3.34M diabetic patients. Delayed signal HR 2.39–2.44 at 2–3 years.
  • PMID 40962119 — Asia Pac J Ophthalmol 2026. 10 studies pooled HR 2.62. Semaglutide = 86.5% of GLP-1 NAION cases.
  • PMID 40383360 — Am J Ophthalmol 2025. FAERS + VigiBase ROR 11–68 for ischemic optic neuropathy with semaglutide.
  • PMID 41692115 — Ophthalmology 2026. Systematic review n=4.8M, OR 2.44 (very low certainty); active-comparator subgroup OR 0.72.
  • PMID 41217382 — Mil Med 2025. Military Health System, n=1,212,775. OR 0.36 (protective).
  • PMID 39491755 — Ophthalmology 2025. Multinational cohort n=297,220, 21 countries. Null association across all subgroups.
  • PMID 40788647 — JAMA Netw Open 2025. TriNetX propensity-matched GLP-1 RA vs controls. HR 1.26 (NS).
  • PMID 40133108 — Obes Res Clin Pract 2025. FAERS, ROR 17.57 semaglutide-specific; 96 NAION cases.
  • PMID 40289093 — BMC Ophthalmol 2025. FAERS formulation analysis: injectable vs oral ocular ADE risk.

Mechanistic and Regulatory:

  • PMID 40055951 — Acta Ophthalmol 2025. Mechanistic review — rapid glucose correction hypothesis.
  • PMID 41857480 — J Chin Med Assoc 2026. EMA classified NAION as “very rare” for GLP-1 RAs.
  • PMID 40183385 — Eur J Ophthalmol 2025. EU regulatory context; case report + literature review.

GLP-1 Class Evidence:

  • PMID 41498755 — Diabetes Care 2026. Meta-analysis 15 studies, >1.5M patients. OR 1.70 class-level.

Semaglutide-Specific Deep Dive (Batch 147)

This section added from batch-147-2026-04-25 semaglutide-specific deep-dive research. The broader GLP-1 RA context above applies; this section focuses on semaglutide-specific data. The parent canonical note establishes the NAION disease context, class-level evidence, and mechanistic framework.

Semaglutide-Specific Risk Estimates

The NAION safety signal is disproportionately concentrated in semaglutide reports:

  • Asia-Pacific meta-analysis (PMID 40962119): semaglutide accounts for 86.5% of reported GLP-1 RA–associated NAION cases in the published literature.
  • FAERS/VigiBase (PMID 40383360): tirzepatide showed significant association only with diabetic retinopathy, not with ischemic optic neuropathy in primary analyses.

Strongest signal — active-comparator cohorts:

StudyPMIDHR/ORNotes
Danish-Norwegian (Simonsen et al.)40098249HR 2.81 (CI 1.67–4.75) vs SGLT2iPer-protocol HR 6.35; stronger in Norway (HR 7.25) vs Denmark (HR 2.17)
Danish single-country (Grauslund et al.)39696569HR 2.19 (CI 1.54–3.12)n=424,152 T2D; ~106,454 semaglutide-exposed; median onset 22.2 months
TriNetX Taiwan (Hsu et al.)40146102HR 2.39–2.44 at 2–3 yearsNull at 1, 3, 6, 12 months — delayed/latent effect
Asia-Pacific meta-analysis (Chen et al.)40962119HR 2.62 (CI 1.81–3.80)Time-dependent; significant after 2 years
Natividade et al. (78 RCTs)40810985OR 3.92 (CI 1.02–15.02)NAION individually adjudicated; diabetic retinopathy OR 1.04 NS
Military Health System (Lieberman et al.)41217382OR 0.36 (CI 0.25–0.51)Protective association; largest null-finding study
Chou et al. (multinational)39491755HRs 0.72–1.51 (all NS)T2D-only HR 1.51; obesity-only HR 0.72; T2D+obesity HR 1.19

Interpretation: Studies using active comparators (SGLT2i, other antidiabetic agents) with similar cardiometabolic risk profiles consistently report weaker or null associations. Studies using broad or untreated comparators consistently report elevated risk. Confounding by indication is the strongest alternative explanation.

Regulatory Status (April 2026)

  • FDA (April 2026): FDA is in active review with Novo Nordisk regarding postmarketing NAION reports. No Black Box warning, REMS, or class label change has been finalized. No formal FDA safety determination has been made.
  • EMA/JCMA 2026: Classified NAION as a “very rare” adverse effect for GLP-1 RAs — consistent with signal confirmation, not causation determination (PMID 41857480, 40183385).
  • Novo Nordisk posture: Acknowledged FDA communication; maintains confounding explains much of the observed association; causality not established.

⚠️ Do not represent that FDA has mandated a warning. As of 2026-04-25, no such mandate exists.

Dose-Dependency — Evidence Gap

Claim: Higher semaglutide doses are associated with greater NAION risk. Evidence grade: Gap — not systematically studied.

  • Wegovy (obesity) uses 0.25–2.4 mg weekly SC. Ozempic (T2D) uses 0.25–2 mg weekly SC. Rybelsus uses 3–14 mg oral daily.
  • No published study has systematically examined dose–NAION response within the semaglutide label dose range.
  • The Danish-Norwegian cohort (PMID 40098249) per-protocol analysis showed stronger association (HR 6.35) than intention-to-treat (HR 2.81), suggesting longer continuous exposure at therapeutic doses may matter — but this does not isolate dose as an independent variable.
  • This is a critical evidence gap. Do not represent that high-dose semaglutide carries higher NAION risk without explicit “evidence gap” framing.

Formulation Differential (Injectable vs Oral) — Evidence Gap

Claim: Injectable semaglutide (Ozempic/Wegovy) carries higher NAION risk than oral (Rybelsus). Evidence grade: Gap — signal-generating data insufficient for conclusion.

  • FAERS formulation analysis (PMID 40289093): subcutaneous injection had higher frequency of reports concerning ischemic optic neuropathy, retinal detachment, retinal tear, and retinal hemorrhage compared to oral tablets. Median time to onset: injectables 7 days vs oral 3.5 days.
  • However, injectable market share dramatically exceeds oral; observed differences may reflect channeling bias (injectable users are more likely to be higher-dose, higher-weight patients with more comorbidities) rather than a true formulation-specific risk.
  • Most NAION case reports are for Ozempic/Wegovy. Whether this reflects pharmacology or market share is unknown.
  • No published study has compared injectable vs oral semaglutide NAION incidence in the same patient population.
  • Do not claim oral Rybelsus is safer without explicit evidence-gap framing.

Algorithm Hooks (Evidence-Boundary Labeled)

Human signoff required before any Vitals feature uses these claims.

What Vitals CAN do:

  • Safety flag (strong, unconditional): Any Vitals user on semaglutide or other GLP-1 RA who records sudden, painless vision loss — particularly upon waking — must receive an immediate prompt to discontinue GLP-1 therapy and seek urgent ophthalmologic evaluation. This flag must be non-dismissible without acknowledgment.
  • Baseline risk nudge: Vitals users initiating semaglutide with pre-existing risk factors (prior NAION in one eye, disc-at-risk, optic disc drusen, T2D + obesity + sleep apnea) should receive a prompt recommending baseline ophthalmologic examination before starting therapy.
  • OSA referral trigger: Vitals sleep staging consistently flagging moderate-to-severe obstructive sleep apnea (AHI >15) in a semaglutide user should prompt referral for polysomnography and consider baseline ophthalmology, given OSA is an independent NAION risk factor.

What Vitals CANNOT do:

  • No Vitals biometric (HRV, RHR, SpO2, sleep staging, activity, EDA, skin temperature) can detect, predict, or monitor NAION risk. NAION is a structural microvascular ischemic event with no known prodromal wearable signature.
  • Do not surface any personalized risk score, probability estimate, or trajectory projection for NAION.
  • Do not advise patients to stop semaglutide based on the NAION signal alone — the cardiovascular and mortality benefits substantially outweigh the rare risk for most patients.

Implementation flag (suggested):

⚠️ GLP-1 Vision Safety: If you experience sudden, painless vision loss — especially upon waking — stop your GLP-1 medication immediately and go to urgent eye care. This cannot wait. For questions, contact your prescriber. Vitals cannot assess your individual risk level.

# Vitals safety flag — GLP-1 + sudden vision loss (non-negotiable)
def glp1_naion_vision_flag(user, event):
    if user.on_glp1 and event.type == "sudden_painless_vision_loss":
        return {
            "flag": True,
            "urgency": "immediate",
            "dismissible": False,
            "message": "Stop GLP-1 medication now. Seek urgent ophthalmology immediately.",
        }
    return {"flag": False}
 
# Vitals risk nudge — GLP-1 + disc-at-risk anatomy
def glp1_naion_baseline_nudge(user):
    risk_factors = [
        user.has_prior_naion_one_eye,
        user.known_small_cup_to_disc_ratio,
        user.has_optic_disc_drusen,
        user.T2D and user.BMI > 30,
        user.has_osa,
    ]
    score = sum(risk_factors)
    if score >= 2 and user.on_semaglutide:
        return {
            "flag": True,
            "urgency": "routine",
            "message": "Baseline eye exam recommended before starting GLP-1 therapy.",
        }
    return {"flag": False}

Graph View

Backlinks