Every major psilocybin depression trial this desk has covered excludes a current or past history of bipolar disorder as a matter of protocol, stated directly in trial documents. Compass’s own trial protocols make that exclusion explicit. A trial now actively recruiting at the University of Texas Health Science Center in Houston enrolls exactly that excluded diagnosis, and layers a second dimension on top: participants are selected specifically for elevated markers of suicide risk, then dosed after coming off their mood stabilizers. It does not enroll patients in active suicidal crisis, the trial excludes that population too, consistent with standard safety practice in this field. What it does enroll is a diagnosis most major programs have never tested at all, in patients carrying real risk factors well short of that crisis threshold.

What the trial is actually testing

NCT06706232 is a non-randomized, open-label study assessing the safety and acceptability of up to two sequential 25 milligram doses of psilocybin, paired with mindfulness-informed psychotherapeutic support, in patients with Bipolar II depression and elevated suicide risk. It has been recruiting since July 2025, with primary completion estimated for August 2027. Eligibility is built around validated suicide-risk markers rather than depression severity alone: participants must meet elevated cutoffs on the Interpersonal Needs Questionnaire, specifically perceived burdensomeness and thwarted belongingness, layered on top of mild to moderate depression by MADRS criteria. The trial explicitly excludes active suicidal ideation with intent or a plan, and a medically significant suicide attempt within the past six months, the same acute-crisis threshold major programs use to keep the highest-risk patients out entirely. And the protocol requires participants to taper off mood stabilizers and other relevant medications starting five weeks before dosing, remaining off them for at least two weeks prior. This is not a trial dosing patients in active suicidal crisis. It is a trial enrolling the one diagnosis every major program excludes by name, in patients carrying real suicide-risk markers short of that crisis threshold, without the medication that would normally provide a mood-stabilizing floor underneath them.

Why bipolar disorder and suicide risk are excluded in the first place

The concern behind both exclusions is well established in psychiatry independent of psychedelics. Any drug capable of producing rapid, significant mood elevation carries a theoretical risk of triggering hypomania or mania in a bipolar patient, and any intervention with unpredictable acute psychological effects carries a real, non-theoretical safety burden when tested in people already experiencing active suicidal ideation. Those are reasonable, standard reasons for a large, pivotal registration trial to exclude both populations, and Compass’s protocols make that judgment explicitly rather than by omission.

What the existing evidence in this exact population actually shows

The evidence base here is thin, and more mixed than a simple “historically excluded, therefore unknown” framing suggests. The first dedicated study, led by Scott Aaronson at Sheppard Pratt and funded by Compass, published in JAMA Psychiatry in December 2023, is described in its own announcement as believed to be the first psilocybin trial ever conducted specifically in bipolar II depression. It was small, fifteen participants, a single 25 milligram dose, and the results were clean: a mean MADRS reduction of 24 points at three weeks, eleven of fifteen participants in remission, and critically, no increase in suicidality scores and no manic symptoms observed.

A second, more recent open-label pilot, posted as a preprint in mid-2025, used a larger dose-escalation design, fourteen participants receiving 10 milligrams followed by 25 milligrams if symptoms persisted. Depression scores again improved significantly and durably, holding through 90 days. But the safety picture here is less uniformly clean: three of the fourteen participants, roughly a fifth, experienced notable psychiatric adverse events specifically including suicidal ideation and hypomania, both of which resolved with clinical support. The study’s own authors describe the overall adverse event profile as broadly comparable to psilocybin trials in other populations, a fair characterization, but “comparable to other populations” is a different and more modest claim than “no signal of the specific risk this population was excluded to avoid,” which is what the 2023 study alone might have suggested.

Why the Houston trial’s specific design matters

Two design choices raise the stakes on top of an already thin evidence base. First, this trial selects specifically for elevated suicide-risk markers rather than depression severity alone, meaning the population entering the dosing session carries real, validated risk factors even though the most acute crisis presentations are excluded by protocol. Second, the mandatory mood-stabilizer washout removes a layer of pharmacological protection that, in ordinary clinical practice, exists specifically to blunt the kind of mood destabilization bipolar patients are vulnerable to. Whether that washout meaningfully changes the risk profile relative to the prior two studies, both of which appear to have enrolled patients regardless of their baseline medication status, is exactly the kind of question this trial’s own safety data will need to answer, not something the existing literature has tested directly.

The caveats

Neither prior study was randomized, placebo-controlled, or powered to detect anything beyond a preliminary safety signal, and combined they represent fewer than thirty total patients ever dosed with psilocybin specifically for bipolar II depression. The hypomania and suicidal ideation events in the 2025 pilot resolved with support and did not represent uncontrolled harm, which is a materially different and better outcome than an adverse event that does not resolve. And this trial is explicitly a safety and acceptability study, not an efficacy trial, its stated purpose is determining whether this approach is feasible at all in this population, an appropriately cautious first step rather than a claim that the excluded-population question is already answered.

The frame

The honest way to read this trial is not as reckless expansion into forbidden territory, and not as reassuring proof the field’s caution was overblown. It is the first study designed specifically to test, in patients carrying the formally excluded diagnosis and real, validated suicide-risk markers, whether the concerns that built bipolar disorder into every major protocol’s exclusion list actually materialize under real dosing conditions, including the specific condition, medication-free bipolar patients with elevated risk markers, that prior studies did not isolate. The existing data offers a real but limited reassurance, not a clean bill of safety: one small trial found nothing, a larger one found a real if resolved signal in about a fifth of participants. A trial built to answer this question directly, in exactly the diagnosis everyone else has excluded, is going to matter for how the rest of the field eventually decides whether that exclusion should stay standard practice or start narrowing. Given how carefully this desk has learned to read a topline number, this is one where the actual safety data, not the design’s stated ambition, will be worth waiting for.