FDA approves Bayer’s Sunharbo for recurrent stroke and warns GLP-1 drugs over rare optic-nerve injury

STAT reports that the Food and Drug Administration approved Bayer’s Sunharbo to reduce recurrent stroke, calling it the first drug in a class meant to limit pathologic clotting without the bleeding penalty that sidelines many anticoagulants. In a separate action, the agency warned that GLP-1 receptor agonists may be associated with non-arteritic anterior ischemic optic neuropathy, or NAION, and told companies to inform clinicians and patients. The items share a news cycle. They do not share an evidence standard.

Secondary stroke is estimated to affect about one in five people after a first event, so a modest absolute reduction would matter. This roundup does not report Sunharbo’s comparator, event rates, intervals, or bleeding outcomes, and it does not disclose Bayer’s sponsorship. A mechanism thought to spare hemostasis is a hypothesis, not a clinical result. The GLP-1 review, FDA said, followed a 2024 observational study. That design can raise a safety signal. It cannot show causation.

The clearest published signal is a July 2024 JAMA Ophthalmology matched cohort from a referral clinic, not a population registry. In type 2 diabetes, cumulative NAION incidence was 8.9% with semaglutide versus 1.8% on comparator medicines, hazard ratio 4.28. In overweight or obesity it was 6.7% versus 0.8%, hazard ratio 7.64. Event counts were small, confidence intervals wide, and the clinic population was enriched for optic-nerve disease. Those figures are not community base rates. NAION is still uncommon and has no proven treatment.

What the roundup misses is the standard that should connect both items. A bleeding-sparing approval, the claim behind factor XIa-style anticoagulants, needs absolute stroke and major-bleed differences before it is a clinical fact. This account does not identify Sunharbo’s molecule. A warning for a rare, irreversible eye event needs incidence per 10,000 users, not only a specialty-clinic hazard ratio. Neither action proves benefit or harm. Observational linkage can justify a labeled warning. Only published randomized data can support Sunharbo’s safety claim.