Two patient groups deliver different results
AriBio's experimental Alzheimer's pill AR1001 slowed clinical decline by 31% over 52 weeks among 604 trial participants who were not taking established medicines for Alzheimer's symptoms. Across the full study of 1,535 people, however, the difference from placebo fell short of statistical significance.
Contents
- Two patient groups deliver different results
- What does 31% slower decline mean?
- Why AriBio says the trial met its goal
- Who took part in POLARIS-AD?
- Memory and daily activity tests give a mixed picture
- How the pill works, and what fluid tests found
- Similar serious event rates, more treatment withdrawals
- Why an oral treatment attracts attention
- FDA discussions and longer follow-up come next
- Key Points
The South Korean biotechnology company announced the results on November 16, 2026, after evaluating patients at approximately 230 clinical sites in 13 countries. The findings give AriBio a positive result in one of two patient groups selected for primary analysis before the results were known, alongside a weaker outcome in the broader population.
According to AriBio's results announcement, the POLARIS-AD trial met its primary efficacy objective because its statistical plan allowed success in either group. The company plans to discuss the complete findings with the U.S. Food and Drug Administration (FDA). AR1001 remains investigational and has no approval for Alzheimer's disease.
The central issue for patients and regulators is the gap between those two results. The strongest evidence of benefit came from people who were not receiving an acetylcholinesterase inhibitor, such as donepezil, or memantine when they entered the trial. Those medicines are used to manage Alzheimer's symptoms.
Sharon J. Sha, chief of Stanford University's Memory Disorders Division and the trial's global principal investigator, presented the initial findings at the Clinical Trials on Alzheimer's Disease conference, CTAD 2026. Her assessment paired encouragement about the selected patient group with a call to examine the broader results.
What does 31% slower decline mean?
POLARIS-AD measured its main outcome using the Clinical Dementia Rating-Sum of Boxes, or CDR-SB. The scale combines assessments of memory, judgment and everyday functioning. Scores range from zero to 18, with higher scores indicating greater impairment.
In the group not taking established symptom medicines, called the monotherapy population, the average score rose by 0.96 points with AR1001 and 1.40 points with placebo over 52 weeks. Patients receiving AR1001 therefore worsened by 0.44 points less. Dividing 0.44 by the placebo group's 1.40-point worsening gives the reported relative slowing of approximately 31%.
The treatment group still experienced deterioration, but at a slower average rate on this combined measure. Researchers described the difference as equivalent to approximately 3.5 months of delayed progression during the study period.
The full population produced a smaller separation. CDR-SB scores rose by 1.41 points with AR1001 and 1.52 points with placebo, a difference of 0.11 points. The corresponding p value was 0.4015, outside the trial's threshold for statistical significance.
For families, the practical promise is more time with less deterioration in abilities that support daily life. The study's main score combines several abilities, while its additional memory and daily activity assessments produced less consistent results.
Why AriBio says the trial met its goal
AriBio designed POLARIS-AD to assess two overlapping populations: all 1,535 participants and the 604 people not receiving the specified symptom medicines at entry. The latter represented approximately 39% of everyone enrolled.
The statistical plan assigned separate significance thresholds: a p value below 0.02 for the full population and below 0.03 for the monotherapy population. The monotherapy result, p=0.0266, crossed its assigned threshold.
A p value describes how compatible the observed result is with a statistical model assuming no treatment difference. Researchers set thresholds before examining outcomes to decide which findings qualify as statistically significant. When a study tests several groups, a procedure called multiplicity adjustment helps control the chance of declaring success because several comparisons were attempted.
AriBio's announcement says its primary objective required a statistically significant result in at least one population under that planned procedure. The positive monotherapy result and the unsuccessful full population result therefore both form part of the trial's outcome.
Additional analyses using different approaches to missing data and restricting the analysis to participants who followed trial procedures reportedly placed the slowing in the monotherapy population between 28% and 34%. AriBio also observed a similar pattern among patients not taking symptom medicines in its Phase 2 trial in 2021.
Who took part in POLARIS-AD?
POLARIS-AD, registered as NCT05531526, began in December 2022. Participants had early Alzheimer's disease, including mild cognitive impairment or mild dementia, with evidence of amyloid pathology. Amyloid is a protein that can accumulate in the brain in Alzheimer's disease.
Eligibility included ages 55 to 90, a Mini-Mental State Examination score of at least 20 and a global Clinical Dementia Rating score of 0.5 or one. Amyloid evidence came from testing cerebrospinal fluid or an amyloid positron emission tomography scan.
Participants were randomly assigned to AR1001 at 30 milligrams or placebo, taken once daily for 52 weeks. The study was double blind, meaning participants and the study staff assessing them did not know which treatment they received during that period.
The trial included sites in the United States, European Union member states, South Korea, China, Canada and the United Kingdom. Its main milestones are:
- 2021: AriBio's Phase 2 trial showed a trend toward greater benefit among patients not taking established symptom medicines.
- December 2022: The Phase 3 POLARIS-AD trial began.
- November 16, 2026: Sha presented the initial Phase 3 efficacy and safety results at CTAD 2026.
- Third quarter of 2027: The optional extension is expected to conclude.
A total of 1,349 participants, or 87.9%, completed the blinded treatment period. Of those completers, 1,284, or 95.2%, entered an optional extension lasting another 52 weeks.
Memory and daily activity tests give a mixed picture
AR1001's secondary measures generally moved in a favorable direction, with larger numerical differences in the monotherapy population. These additional tests help describe whether changes appear across different aspects of cognition and everyday functioning.
On the Alzheimer's Disease Assessment Scale-Cognitive Subscale with 13 items, known as ADAS-Cog13, the monotherapy group showed a difference of 1.16 points favoring AR1001 over placebo. Its p value of 0.0798 did not reach statistical significance. ADAS-Cog13 and the Amsterdam Instrumental Activities of Daily Living assessment also failed to show statistically significant differences in either population.
The Mini-Mental State Examination, or MMSE, gave a more positive result. Participants in the monotherapy population receiving AR1001 scored 0.7 points higher than those receiving placebo, with p=0.0232.
AriBio described the MMSE finding as nominally significant. In plain language, the individual test crossed a conventional statistical threshold; confirmation under the procedure accounting for multiple comparisons was not reported.
Sha, the trial's global principal investigator, described how the secondary findings should fit into the wider assessment:
The secondary endpoints and biomarker findings warrant further evaluation alongside the Overall population outcome to understand AR1001's therapeutic effects in early Alzheimer's disease.
How the pill works, and what fluid tests found
AR1001 is based on mirodenafil, a compound from the phosphodiesterase-5 inhibitor class, usually shortened to PDE5 inhibitor. AriBio licensed related technology from SK Chemicals in 2011 and developed it for Alzheimer's disease.
PDE5 is an enzyme that breaks down a cellular signaling molecule called cyclic guanosine monophosphate, or cGMP. By inhibiting that enzyme, AR1001 increases cGMP signaling. AriBio says the drug can cross the blood-brain barrier, the protective boundary that controls which substances move from blood into the brain.
Preclinical research has explored potential effects on brain blood flow, nerve cell protection, inflammation and chemical changes to tau, another protein associated with Alzheimer's disease. The human trial included cerebrospinal fluid testing to examine biological activity. Cerebrospinal fluid surrounds the brain and spinal cord.
At week 52, cGMP levels increased versus placebo in both the full population and the monotherapy population, with p values below 0.0001 and 0.0058, respectively. AriBio says these findings support engagement of the intended target in the central nervous system.
The tau marker p-tau217 was 5.34 picograms per milliliter lower relative to placebo in the monotherapy population, with p=0.0402, a nominally significant finding. In the full population, the difference was 3.52 picograms per milliliter, with p=0.1784, outside statistical significance. The amyloid Aβ42/40 ratio showed no difference at 52 weeks. Further biomarker analyses are continuing.
Similar serious event rates, more treatment withdrawals
Adverse events occurred in 69.0% of participants receiving AR1001 and 66.4% receiving placebo. Most were mild or moderate, and serious adverse events occurred at similar rates: 10.0% with AR1001 and 9.8% with placebo.
Diarrhea, falls and dizziness were the most common nonserious events reported with AR1001. Each occurred more often than with placebo:
- Diarrhea: 7.7% with AR1001, compared with 4.4% with placebo.
- Falls: 7.3% with AR1001, compared with 5.9% with placebo.
- Dizziness: 6.1% with AR1001, compared with 3.1% with placebo.
Adverse events led 5.2% of AR1001 recipients to stop treatment, compared with 1.7% of placebo recipients. That is a difference of 3.5 percentage points and approximately three times the placebo rate. Diarrhea prompted discontinuation in 0.7% of AR1001 recipients and vomiting in 0.4%, compared with zero for each event in the placebo group.
Fatal adverse events occurred in four people receiving AR1001, or 0.5%, and two receiving placebo, or 0.3%. The announcement did not specify whether those deaths were considered related to treatment.
Why an oral treatment attracts attention
AriBio is pursuing a medicine patients could take at home once a day. That format could reduce administration demands compared with antibody treatments such as Leqembi and Kisunla, which target amyloid and involve injections and monitoring for brain swelling or bleeding.
Existing oral medicines already treat Alzheimer's symptoms. AR1001's development goal is to slow disease progression through a different biological pathway, expanding what an oral treatment might accomplish.
Jai-Jun Choung, AriBio's chairman and chief executive officer, described the company's intended benefit for patients and caregivers:
We started developing AR1001 with the goal of creating an oral treatment that could slow Alzheimer's progression and be practical for patients and care partners to use over the long term.
The program also has a commercial partner. AriBio signed a global rights agreement with China's Fosun Pharma in May, described as worth up to $4.7 billion, followed by a strategic equity investment contract in July.
AriBio executives have pointed to possible interactions with existing symptom medicines as an explanation for the different population results. That remains a hypothesis for further investigation. The monotherapy population was defined by medication use at the start of the trial.
FDA discussions and longer follow-up come next
AriBio executives have expressed an intention to move quickly toward a U.S. approval application. Its formal announcement sets out discussions with the FDA on the complete findings as the next regulatory step, without a fixed submission date. Approval remains undecided, including how regulators will assess the positive monotherapy result alongside the full population and secondary outcomes.
The ongoing extension gives every participant AR1001 at 30 milligrams once daily. This open label phase means participants and study teams know the treatment being given. It will provide additional information about longer exposure, safety and disease progression through its expected completion in the third quarter of 2027.
James Rock, AriBio's chief clinical officer, said the company would use the complete dataset to guide development:
We will continue our review of the data and discuss the complete findings with the FDA to inform next steps.
For patients and caregivers, the immediate outcome is a promising result in a defined group, further analysis and continued follow-up. AR1001 remains an experimental medicine rather than a treatment available for routine Alzheimer's care.
Key Points
- AR1001 slowed CDR-SB worsening by 31% over 52 weeks in the 604 participants not taking specified Alzheimer's symptom medicines at entry.
- The full population of 1,535 participants did not show a statistically significant treatment difference.
- The trial's advance statistical plan allowed its primary objective to be met in either population.
- Additional cognitive and daily activity results were mixed; fluid tests showed increased cGMP signaling.
- Adverse events caused treatment withdrawal in 5.2% of AR1001 recipients versus 1.7% receiving placebo.
- AriBio plans FDA discussions, while the extension is expected to finish in the third quarter of 2027.






