"For seven decades, we've really been at this altar of serotonin, norepinephrine and dopamine." That is Roger McIntyre, reportedly professor of psychiatry and pharmacology at the University of Toronto per the JCP podcast, and the metaphor is doing exactly what he wants it to do: it turns the monoamine hypothesis into a cult, seventy years of misplaced worship, and positions what comes next as apostasy with data.
Here is the sentence that matters, from the same episode: GLP-1s "have the potential to transform the health span and the lifespan of people living with mental illnesses by targeting on-label considerations today and potentially targeting the underlying pathophysiology of the brain-based disorder tomorrow."
Read that again and notice the load-bearing word. Tomorrow.
Everything McIntyre can defend sits in the clause about today—the on-label metabolic stuff. Everything that would constitute a paradigm shift sits in the clause about tomorrow, hedged with "potentially," and closed out with "so stay tuned." The rhetoric is running well ahead of the evidence calendar, and it is worth slowing down to see by exactly how much.
The rhetoric is running well ahead of the evidence calendar, and it is worth slowing down to see by exactly how much.
Start with what the field's own reviews actually say, because the fine print is more honest than the headlines.
The Oxford systematic review reportedly by De Giorgi and colleagues sifted, depending on the source, 278 (per Nature Mental Health) or 280 (per BJPsych Open) preclinical studies and 96 clinical ones. That ratio alone tells the story: the mechanism is where the action is, not the patients.
In substance use disorders, the signal replicates—preclinical work plus emerging clinical data, mediated, plausibly, through dopaminergic modulation of reward. This is the strongest card in the deck, and it is a real one.
In neurocognitive disorders, there's a defensible case for reduced dementia risk and cognitive protection across various cohorts. Also real, also mechanistically coherent.
Then the case thins fast. In psychotic disorders, the review concludes GLP-1 receptor agonists "primarily mitigate antipsychotic-induced metabolic side effects, with minimal evidence for direct effects on psychosis itself."
Sit with that. When you read that GLP-1s "help" in schizophrenia, what the evidence supports is that they treat the weight gain and diabetes the antipsychotics caused. That is a genuine clinical good. It is not a new theory of psychosis.
And mood—the disorder the whole reconceptualization is being sold on? The verdict is "inconclusive," with "some studies reporting antidepressant properties while others show no clear benefit."
Inconclusive is not a paradigm. Inconclusive is a to-do list.
The tell is that McIntyre knows this. In his own paper for Psychiatry and Clinical Psychopharmacology, the title asks whether incretin agonists are psychiatric drugs or a "creative superlative," and the abstract answers plainly: whether they "meet the evidentiary threshold to be classified as psychiatric drugs remains to be determined by ongoing phase III trials."
That is the same author who invoked the altar of serotonin. The podcast promises a paradigm; the peer-reviewed abstract promises a phase III readout. Both are McIntyre. Only one is under oath.
Now watch how a real, measurable signal gets rhetorically inflated, because this is the pattern that should make you cautious every time.
A Maastricht group ran a clean study: randomized, double-blind, placebo-controlled crossover, 18 older overweight men, ketone monoester versus taste-matched placebo. They pushed β-hydroxybutyrate to a peak of 2.9 mmol/L—reportedly genuine nutritional ketosis. This is exactly the kind of metabolic-brain experiment the new paradigm rests on.
The result the story wants: reaction time improved. It did—reportedly by 9 milliseconds. The confidence interval ran from -18 to -1 milliseconds. Statistically real, p equals 0.029, and about the width of an eyeblink's margin.
The results the story tends to skip: no effect on memory, no effect on executive function. And the ketones lowered whole-brain cerebral blood flow by 4.7 mL per 100g per minute, and dropped cerebral perfusion velocity too.
So the honest one-line summary is: acute ketosis shaved 9 milliseconds off psychomotor speed while reducing blood flow to the brain. That is a fascinating, ambiguous physiological finding. It is not "fuel your neurons and think faster," and the distance between those two sentences is where metabolic psychiatry keeps losing its discipline.
Nine milliseconds is the whole genre in miniature: a defensible measurement wearing a headline it can't support.
None of this means the mechanistic story is wrong. On the bench, it is genuinely strong—GLP-1 receptors are expressed across the brain, and the proposed effects on neurogenesis, neuroplasticity, and neuroinflammation are biologically plausible and heavily studied.
The problem is the specific inference being marketed: that because metabolism shapes the brain, GLP-1s are antidepressants. That inference requires a study design that mostly does not exist yet.
Here is the design that would settle it. Most GLP-1 trials showing mood benefit run in people with diabetes or obesity, where the drug reliably drops weight and glucose. If your depression lifts after you've lost weight, slept better, and stopped feeling metabolically wrecked, was that a direct central nervous system effect—or the predictable psychological dividend of being physically healthier?
To separate those, you need matched trials: participants who lose the same weight and achieve the same glycemic control through different means, with mood measured against both. Absent that control, "GLP-1s treat depression" and "feeling less sick makes you feel less depressed" produce identical data. The MDPI review concedes exactly this, listing "disentangling central from peripheral effects" as an open challenge, and noting that large trials in primary psychiatric populations are simply lacking.
This is the same missing-control problem that sinks so much metabolic-lifestyle research: the confound is baked into who gets studied.
So here is the distinction that actually governs what a clinician should do this week, and it is sharp.
Prescribing a GLP-1 to a patient on clozapine to combat drug-induced weight gain and diabetes is well-justified today—McIntyre says so in plain language, and the evidence backs him. Given the excess premature mortality in persons living with psychiatric disorders reported in McIntyre's paper, this may reportedly confer meaningful mortality reduction. This is not speculative. It is arguably underused.
Prescribing a GLP-1 as an antidepressant, on the theory that it targets the metabolic root of the mood disorder, is a hypothesis wearing a paradigm's clothes.
The two get blurred precisely because they involve the same molecule and the same population, and the blur is where the overclaim lives. "GLP-1s improve outcomes in psychiatric patients" is true and important. "GLP-1s are antidepressants that replace the monoamine model" is neither established nor, yet, testable at scale.
The field has even named its own restraint. The Oxford authors, watching enthusiasm build "in both the scientific community and in the media," urge approaching adoption in psychiatry "cautiously." The Nature Mental Health group calls it a "nascent" field of "key questions." Nascent is the right word.
A paradigm shift is a claim about the past being wrong. What metabolic psychiatry has right now is a claim about the future being promising—which is a different, humbler, and entirely legitimate thing to say.
The honest version of McIntyre's pitch is already sitting in his abstract: use these drugs now for the cardiometabolic mortality that is killing patients, and wait for the phase III trials before deciding whether they are psychiatric drugs at all.
That version doesn't need an altar. It just needs the trials to read out. Until they do, the fine print is the finding.