The last participant in POLARIS-AD swallowed their last capsule in late June, roughly three and a half years after the first patient was dosed in the United States in December 2022. That is the quiet part of drug development: no headline, no scan, just a person in a clinic somewhere in one of 13 countries finishing a year of a white pill that might be a dementia treatment or might be nothing at all. The molecule is AR1001, also called mirodenafil, and it belongs to a drug class most people associate with something else entirely. PDE5 inhibitors are the family that includes sildenafil, sold for erectile dysfunction; Alzforum's entry on the compound notes mirodenafil itself was approved in South Korea for that indication before AriBio pointed it at the brain.

What is actually registered is unglamorous and specific, and it is worth reading before anyone reads the press coverage. ClinicalTrials.gov record NCT05531526 describes a Phase 3, double-blind, randomized, placebo-controlled trial with an actual enrollment of 1,535 people aged 55 to 90, all with mild cognitive impairment or mild dementia consistent with Alzheimer's disease (NIA-AA stages 3 to 4), an MMSE of 20 or above and a global Clinical Dementia Rating of 0.5 or 1. Participants take AR1001 30 mg once daily, or placebo, for 52 weeks. The primary endpoint is the change in CDR-SB, the clinician-rated scale that sums memory, orientation, judgment, community affairs, home life and personal care, from baseline to week 52. Secondary measures include ADAS-Cog 13, the Amsterdam instrumental activities of daily living questionnaire, MMSE and a geriatric depression scale. A 52-week extension follows, with study completion estimated for December 2027.

The mechanism is the reason this trial is interesting rather than routine. Almost every Alzheimer's drug that has reached the market recently works by clearing amyloid. AR1001 does not. In the account AriBio presented at the 2026 Alzheimer's Association International Conference in London, reported by NeurologyLive in July, PDE5 inhibition raises cGMP and PKG signaling with downstream PI3K effects, which the company links to neurotrophin support, reduced tau phosphorylation and effects on blood vessels and the blood-brain barrier. That is a mechanistic rationale built largely on laboratory and animal work, not a demonstration in people, and it should be read as the hypothesis the trial is testing rather than as a result.

The safety picture so far is the only POLARIS-AD data in public, and it is preliminary and blinded. At AAIC 2026 the company reported that the 52-week double-blind phase had been completed with no unexpected safety signals, including among participants also taking anti-amyloid therapy, with falls, diarrhea, urinary tract infection, headache and dizziness among the most common events, and ocular events recorded without cases of ocular neuropathy. None of that tells you whether the drug slowed anything. AriBio said in June it would release topline numbers after data cleaning, database lock and statistical analysis; by the start of October the group was telling Korean outlets that 1,348 of the 1,535 participants had completed the 52-week main study and that topline results would be presented at the CTAD conference in Boston, which runs November 16 to 19.

The broader class carries a split record that is worth holding in mind while waiting. A 2024 analysis in the Journal of Alzheimer's Disease combined real-world patient records with experiments in neurons derived from patient stem cells and reported an association between sildenafil use and lower Alzheimer's risk; the same paper cites an earlier pharmacoepidemiology study by Desai and colleagues that found no association between starting PDE5 inhibitors and incident dementia. Observational findings like these are not evidence of cause, in either direction, and they are precisely why a randomized Phase 3 exists. Korean financial coverage has floated an FDA filing as early as 2027 if the data are positive, but that is an analyst projection attached to a result nobody has seen.

For a family living with an early diagnosis, the practical content of this story is a date, not a decision. Nothing here identifies a drug to ask for, and PDE5 inhibitors on the market are approved for other conditions entirely. If the week 52 CDR-SB number lands in November and favors AR1001, the questions that follow are about effect size, about who benefited, about what the biomarkers did and about whether regulators agree. If it does not, the trial will have answered a real question honestly, which is also worth something.