Definium Therapeutics announced positive topline results from Panorama, its second Phase 3 trial of DT120 ODT in generalized anxiety disorder, on September 14. The 100-microgram arm beat placebo on the primary endpoint, a Hamilton Anxiety Rating Scale (HAM-A) reduction at Week 12, by a placebo-adjusted 5.1 points (p<0.0001). That is DT120’s third consecutive positive Phase 3 readout, following Emerge in major depressive disorder in June and Voyage, the first GAD trial, in August. It is also, on the standardized measure investors and regulators use to compare trials, a smaller effect than Voyage produced. Panorama’s Cohen’s d is 0.64. Voyage’s was 0.81.
The raw benefit held. The standardized effect size didn’t
Read the two trials side by side and the gap is specific, not vague. At Week 12, Voyage’s placebo-adjusted HAM-A difference was 5.4 points; Panorama’s is 5.1, a decline of about 6 percent. Voyage’s standardized effect size was 0.81; Panorama’s is 0.64, a decline of about 21 percent. A raw point difference falling by single digits while its standardized effect size falls by a fifth is not a contradiction. It is what happens when the variability of individual patient responses rises between trials, since Cohen’s d divides the mean difference by that variability. Definium’s release does not report the standard deviations underlying either figure, so whether variance rose, as opposed to some other factor moving the two measures apart, cannot be established from the disclosed numbers. What the disclosed numbers do not support is the simpler interpretation that Panorama showed a large collapse in placebo-adjusted symptom benefit: the primary, placebo-adjusted endpoint difference was nearly identical across the two trials. Its magnitude relative to the observed variability was not.
Other endpoints move in the same direction, some further. Voyage’s Week 1 HAM-A advantage was 7.7 points; Panorama’s is 5.3, a 31 percent decline, a materially bigger drop than the primary endpoint’s 6 percent. The response-rate advantage (HAM-A improvement of 50 percent or more) fell from 27 percentage points in Voyage to 18 in Panorama, and the share reaching mild-or-better symptoms fell from a 28-point advantage to 19. Remission moved the other way, ticking up from a 10-point advantage to 12. CGI-S improvement at Day 2 was identical in both trials, 0.8 points ahead of placebo. That is not a uniform pattern of decline. It is a mixed one, and the responder secondaries that cleared p<0.0001 in Voyage are reported for Panorama only at p<0.05 and p<0.01, a small but material difference in the strength of the separations the two releases report.
The arm built to answer the field’s open question wasn’t built to answer it
Panorama’s design innovation was a 50-microgram arm, randomized alongside the 100-microgram and placebo groups at a 2:1:2 ratio. Definium is explicit about this arm’s purpose: it was included to “help mitigate functional unblinding,” the problem of GAD patients correctly guessing their dose assignment because a psychoactive drug’s own perceptual effects give it away, a concern the FDA’s own psychedelic clinical trial guidance flagged directly. It is the exact design feature this desk singled out in its Voyage coverage as the detail worth watching once Panorama’s data arrived.
The data arrived, and Definium’s release states plainly that this arm “was not designed or powered for statistical comparisons.” The 50-microgram group showed a placebo-adjusted HAM-A benefit of its own, roughly 29 percent smaller than the 100-microgram arm’s at Week 12. A non-zero benefit at half the target dose is compatible with dose-dependent pharmacology. It does not, however, separate pharmacological dose response from expectancy or functional unblinding, particularly when perceptual effects remained common in the 50-microgram arm: the trial’s own adverse-event table shows illusion, the perceptual effect most likely to tip a patient off, in 61 percent of the 50-microgram group and 68 percent of the 100-microgram group, against an overall treatment-emergent adverse event rate of 96.1 percent and 94.8 percent respectively, compared with 62.9 percent on placebo. Both readings remain available because the comparison the field most wanted this arm to resolve was never statistically powered to resolve it.
What the rest of the release supports
Away from that open question, Panorama’s safety data reads as unremarkable in the way sponsors want a fourth dosing cohort to read: TEAEs mostly mild to moderate, concentrated on dosing day, no new signal, no suicidality signal, and discontinuation rates within a point of each other across all three arms (10.4 percent, 11.5 percent, 10.3 percent). Across more than 1,000 treatment sessions in the Phase 3 program through September 10, 97 percent of participants met the program’s structured end-of-session monitoring criteria by hour eight, a concrete operational data point for a drug Definium wants delivered in outpatient settings instead of extended overnight observation. Panorama’s 40-week open-label extension, in which participants may receive up to four additional 100-microgram doses based on symptom severity, carries the same limitation this desk raised about Voyage’s identically structured extension: retreatment triggered by a clinician’s judgment of symptom severity, not prespecified on a blinded schedule, is a meaningful signal about durability but not the same evidentiary weight as a design built to isolate it from the start.
Definium’s own characterization of the results, in a company statement from chief executive Rob Barrow describing “unprecedented efficacy” across “four complementary studies,” is the company’s assessment of its own program, not an independently verified claim. What is independently verifiable from the release is narrower and still consequential: a third positive Phase 3 trial for the same compound, a primary endpoint that held its raw magnitude while its standardized effect size fell by a fifth, and the specific methodological question the trial was designed to help answer left open by the same release that reports it. Definium says a pre-NDA meeting is scheduled for the fourth quarter of 2026, with an NDA filing anticipated in the first half of 2027. Fuller trial data will be needed to determine why the standardized effect moved more than the raw placebo-adjusted difference, and whether that difference matters to the overall consistency of the efficacy package.