Vistagen's social anxiety drug improved for four months without a placebo arm. Its placebo-controlled trial missed nearly three months earlier.

Updated September 22, 2026: Corrected the announcement day and clarified the study populations and four-month follow-up data.

Updated September 24, 2026: Added a third fasedienol trial, an exploratory Phase 2a repeat-dose study whose topline results Vistagen disclosed in August and whose very-severe-subgroup data it presented at a psychiatry conference this week.

Vistagen said Tuesday that patients using its social anxiety drug fasedienol as needed, in real-world settings, kept improving for four months. What the announcement does not change is what happened nearly three months earlier: in the placebo-controlled portion of the same trial, fasedienol missed its primary endpoint, and the numbers moved in placebo’s favor.

What the new data actually measured

Vistagen’s new data come from the open-label extension (OLE) of PALISADE-4, the fourth Phase 3 study in its fasedienol program. After the randomized, blinded portion of PALISADE-4 concluded, 322 participants chose to continue into the OLE, taking fasedienol as needed, up to six times a day, in ordinary anxiety-provoking situations instead of a clinic setting. Every participant knew they were receiving the drug; there was no placebo group in this portion of the study.

On the Liebowitz Social Anxiety Scale (LSAS), a clinician-administered measure, mean scores improved steadily: 20.2 points at one month, 24.6 at two, 29.1 at three, and 31.4 at four, against a baseline mean of 99.3, in the “very severe” range. The Social Phobia Inventory, a self-reported measure, showed a similar pattern, improving from 10.5 points at month one to 14.9 at month four. Discontinuation due to adverse events was low, 1.6 percent, and Vistagen reported no new safety signals after up to a year of as-needed use.

What an open-label extension cannot show

None of that establishes that fasedienol caused the improvement. An OLE has no control group, so there is nothing to separate a drug effect from the ordinary pattern of a chronic condition getting better with time, attention, and repeated clinical contact, a pattern well documented in social anxiety disorder research independent of any specific treatment. Participants also self-selected into the extension, and self-selection tends to favor people who found the drug tolerable and were already doing reasonably well.

There is a second, more specific limitation, disclosed in the release’s own footnote: Vistagen closed the OLE early, in July 2026, “for business reasons,” before every participant had the chance to reach the four-month mark. The company attributes at least part of the shrinking sample, from 298 patients at month one to 197 at month four, to that early closure, not to participants dropping out because the drug wasn’t working. The release does not specify how much of the decline reflects closure timing versus ordinary discontinuation, and it does not establish that the 197 patients assessed at month four were specifically the ones who happened to enroll earliest. What it does establish is that the month-four figures describe a smaller, unspecified subset of the original 322, not the full group followed for a complete four months.

The controlled result from nearly three months earlier

The context the announcement does not restate: in a June 30 announcement, Vistagen reported topline results from the randomized, placebo-controlled portion of this same PALISADE-4 trial, and it missed. On the primary endpoint, a single-dose public speaking challenge scored on the Subjective Units of Distress Scale, fasedienol’s least-squares mean change was -9.5 (standard error 1.7), against -11.4 (SE 1.7) for placebo, a difference of 1.9 points that narrowly favored placebo and was not statistically significant (p=0.427). Both secondary endpoints also failed to separate from placebo, and one of the two moved in the wrong direction for the drug as well.

A post-hoc analysis found a nominally significant result in a subgroup of 123 patients individually screened at a baseline LSAS of 95 or higher, a stricter, per-patient threshold than the OLE population’s reported baseline: that population’s 99.3 figure is a group average, not a floor every individual participant met, so it is not established that the OLE group as a whole meets this subgroup’s screening criterion. The subgroup showed a difference of 9.1 points favoring fasedienol (p=0.036). That subgroup was constructed after the fact, and not just by severity. Vistagen also excluded one disqualified trial site and patients showing a ceiling effect or an unusually large placebo response before running the comparison. A post-hoc analysis built on several layered exclusions is weaker evidence than a prespecified result, whatever its p-value; it can generate a hypothesis worth testing prospectively, not confirmation that the underlying effect is real.

A second consecutive miss

PALISADE-4 is not fasedienol’s first controlled setback. The compound’s original Phase 3 study, PALISADE-2, met its primary endpoint in 2023, a statistically significant result (p=0.015) in 141 patients with a lower severity bar for entry (LSAS of at least 70). Its next study, PALISADE-3, missed its primary endpoint in December 2025, in a result Vistagen’s own leadership at the time called unexpected. PALISADE-4 makes it two in a row.

A third missed trial, and a subgroup result split three ways

On September 24, Vistagen presented a poster at Psych Congress 2026 built around the PALISADE-4 very-severe subgroup result above, alongside data from a third, previously separate trial: an exploratory Phase 2a repeat-dose study (RDS, NCT06809179) testing whether a second 3.2-microgram dose of fasedienol, given ten minutes after the first, adds benefit over a single dose. Vistagen disclosed RDS’s topline results August 6. Across the trial’s 61 participants, split between placebo, a single-dose arm, and a repeat-dose arm, fasedienol again missed its primary endpoint: pooled fasedienol improved 15.0 points on the SUDS scale against 6.8 for placebo, a difference the company itself reported as not statistically significant (p=0.10), with the comparison across all three arms also not significant (p=0.2). That is a third placebo-controlled miss for fasedienol, in an exploratory Phase 2 study, not a Phase 3 trial like the two described above.

As with PALISADE-4, Vistagen highlighted a prespecified subgroup with a baseline LSAS of 95 or higher, describing the results as nominally significant. The pooled very-severe subgroup, 24 of the trial’s 61 participants, improved 16.2 points against 1.6 for placebo (p=0.04). Broken out by arm, the two active doses did not perform the same: the single-dose arm, 14 participants, reached p=0.05, at the conventional significance threshold, not clearly below it, while the repeat-dose arm, 10 participants, the one Vistagen’s chief medical officer specifically credited with showing “the potential benefit of repeat dosing,” came in at p=0.08, not statistically significant. The September 24 release describes that repeat-dose result as a “numerical improvement,” which is accurate, but it sits next to language describing the broader analysis as showing “statistically significant improvement,” a pairing that can read as stronger than what the repeat-dose arm itself supports. None of these subgroup counts, in the 10-to-24-participant range, are large enough to settle whether a second dose adds meaningful benefit.

What the company is proposing next

Vistagen says it is preparing to meet with the FDA this quarter to discuss a registrational pathway for fasedienol, built around a single future multi-dose Phase 3 trial using the LSAS as its primary endpoint, rather than the single-dose public speaking challenge design used in PALISADE-3, PALISADE-4, and RDS. The company points to PALISADE-2’s earlier result, the very-severe subgroup analyses from both PALISADE-4 and RDS, and the open-label extension data as the supporting case. The company has said its cash resources support operations into 2027, without specifying what a new trial would cost or whether existing cash covers it.

What to watch

Whether the FDA agrees that three failed placebo-controlled trials, one earlier controlled success, two small very-severe-subgroup analyses, one post hoc and one prespecified, and an uncontrolled extension add up to a viable registrational path is not yet resolved. The clearer signal will be the outcome of this quarter’s FDA meeting, and separately, whether Vistagen actually launches and funds the multi-dose trial it says it wants to run. Until a controlled trial produces a prespecified result favoring fasedienol in its total population, the open-label data announced Tuesday describes what happened to patients who knew they were on the drug, not what the drug did.