Published on 21 September 2026
Pramipexole for anhedonic depression: what this trial demonstrates, and what it does not
Nature Medicine · 2026; 32(7): 2570-2578 · Ventorp et al.
DOI 10.1038/s41591-026-04465-9
PMID 42286321
Scientific 72
Editorial 88
The essentials
A double-blind, placebo-controlled randomized trial, conducted at a single center in Lund, tested add-on pramipexole in adults with major depressive disorder, dysthymia, or bipolar depression, and anhedonia defined not by a total score but by an item-level criterion: at least three items rated 3 or 4 on the Snaith-Hamilton Pleasure Scale (SHAPS). Eighty-five patients were randomized 1:1, 82 were analyzed in the modified intention-to-treat population, over nine weeks. The primary endpoint was met: the reduction in SHAPS score is greater under pramipexole than under placebo, with a mean difference of −4.04 points (95% CI: −6.89 to −1.18; p = 0.006; Hedges’ g 0.62). Exploratory analyses point the same way: increased light physical activity, and relative preservation of reward-related ventral striatal activation. The improvement is maintained through a six-month open-label extension. The authors conclude that anhedonia improved significantly with a favorable tolerability profile, and present pramipexole as a potential augmentation strategy. Three points frame the reading, and the last two appear only in the full text. First, in France pramipexole is authorized only for idiopathic Parkinson’s disease and restless legs syndrome, never in psychiatry. Second, the self-rated depression scale, the MADRS-S, does not separate the groups at the planned endpoint, while the anhedonia scale does. Third, tolerability is quantified in the full text, and it is markedly worse than under placebo, with sleep disturbance in seven patients in ten, nausea in six in ten, and difficulty controlling purchases reported by one patient in four at week 3.
Context
Anhedonia often outlasts treatment. The patient sleeps better, eats better, no longer cries, and still feels nothing. Serotonin reuptake inhibitors act poorly on this dimension, and part of the literature attributes this to a reward circuit that remains hypoactive. Hence the idea of targeting dopamine directly rather than serotonin. Pramipexole, a D2-family agonist with preferential affinity for D3 receptors, is the obvious candidate: an old, generic molecule with well-characterized pharmacology.
Well characterized, including for what raises concern. The summary of product characteristics, consulted on August 10, 2026 on the ANSM public medicines database, lists two indications, and only two: treatment of the signs and symptoms of idiopathic Parkinson’s disease in adults, as monotherapy or combined with levodopa, up to 3.3 mg of base per day; and symptomatic treatment of moderate to severe idiopathic restless legs syndrome, up to 0.54 mg of base per day. No psychiatric indication. Any use in depression is therefore, in France, off-label prescribing, with the information and documentation obligations that follow from it. The example here is French and dated; the underlying problem is not. Readers practicing elsewhere should check pramipexole’s authorized indications with their own national or regional medicines agency before drawing any prescribing conclusion from this trial.
The same summary of product characteristics imposes a monitoring requirement that matters here. Patients must be regularly monitored for impulse control disorders, and informed of the risk, along with those close to them: pathological gambling, increased libido, hypersexuality, compulsive buying, binge eating. The adverse-effect table lists these among the common effects in Parkinson’s disease. Somnolence is very common in Parkinson’s disease and common in restless legs syndrome, and sudden-onset sleep episodes during daily activities, sometimes without warning, are uncommon but documented. These frequencies come from Parkinson’s patients, exposed long-term and generally at higher doses than in a psychiatric trial, or from restless-legs patients exposed to far lower doses: they do not transpose directly to a depressed population. The text also calls for regular monitoring for emerging mania and delirium, and reserves dopamine agonists in patients with psychotic disorders for situations where the expected benefit outweighs the risk. Finally, discontinuation must be gradual: an abrupt stop risks a dopamine-agonist withdrawal syndrome, and abruptly stopping dopaminergic treatment can trigger a neuroleptic malignant syndrome.
Design, and why the full text changes the reading
The trial uses parallel groups: 1:1 allocation, one pramipexole arm, one placebo arm, nine weeks, with no sequence crossover. The protocol, published in 2023 in BMJ Open, describes the same design. Questions of carryover between periods and washout duration between sequences are specific to designs in which each patient receives both treatments in succession: they do not arise here, since no patient changes arm during the comparative phase.
The decisive point lies elsewhere. The article is published open access under a CC BY license, and the full text contains nearly everything the abstract leaves out: the diagnostic breakdown, baseline scores, eligibility criteria, adverse-event frequencies, an assessment of impulse control disorders, two mood switches, a check on whether blinding held, and the result on the depression scale. Reading the abstract alone leads to writing that a piece of data is missing when it is in fact published, and correcting that was the bulk of the work this analysis required.
One nuance remains. After the nine controlled weeks, all participants could enter a six-month open-label extension. Patients initially on placebo then switch to pramipexole. This is not a randomized crossover and allows no comparison between treatments: it is an uncontrolled follow-up.
The study at a glance
| Study question (PICO) | |
|---|---|
| Population | |
| Adults aged 18 to 75 with major depressive disorder, dysthymia, or bipolar depression. The anhedonia criterion is not a total-score threshold but an item-level one: at least three SHAPS items rated 3 or 4. Ongoing psychotropic treatment for at least 4 weeks and at least one adequate antidepressant trial without remission. 85 randomized, 82 analyzed. Diagnostic breakdown: 59.8% major depressive disorder, 37.8% dysthymia, 2.4% bipolar depression, that is two patients of 82. Mean age 47.2 years (SD 13.7), 45.1% female. Single center, Lund, Sweden. Enrollment from February 16, 2023 to April 4, 2025 | |
| Intervention | |
| Oral pramipexole, once daily, preferably in the evening, weekly titration according to tolerability and efficacy, added to ongoing treatment, for 9 weeks. The trial’s methods set the maximum dose at 4.5 mg of salt per day, that is 3.15 mg of base. The mean dose actually reached by the end of the trial is not reported in the main text, which prevents situating this population’s exposure relative to Parkinson’s patients | |
| Comparator | |
| Add-on placebo, 1:1 allocation, double-blind, parallel groups | |
| Primary endpoint | |
| Change in SHAPS score from baseline to week 9 | |
| Prespecified secondary endpoint | |
| Self-rated MADRS (MADRS-S) | |
| Exploratory analyses | |
| Light physical activity measured by accelerometer, reward-related ventral striatal activation on 7-tesla functional MRI, a simplified monetary incentive delay task. 48 participants imaged, 25 under pramipexole and 23 under placebo, exclusions due to contraindications, refusals, and technical difficulties. The imaging protocol appears in the trial’s methods | |
| Design | |
| Randomized, single-center, double-blind, placebo-controlled, parallel-group trial, followed by a 6-month open-label extension. Modified intention-to-treat analysis, mixed model for repeated measures. Registered as NCT05355337 and NCT05825235, protocol published in BMJ Open in 2023 · Individually randomized trial, CEBM 1b |
On the level of evidence, an individually randomized, preregistered trial, with a protocol published before the results and a modified intention-to-treat analysis, falls into category 1b. The reservations that follow concern the scope of the result, not the soundness of the design.
Quality control
| Criterion | Status |
|---|---|
| Registration and protocol | Solid |
| FindingTwo registrations on ClinicalTrials.gov, NCT05355337 for the controlled phase and NCT05825235 for the long-term follow-up, plus a protocol published in a peer-reviewed journal before the results. This is the expected standard, and it is met | |
| Design | Solid |
| FindingParallel groups, 1:1 randomization, double-blind, placebo comparator, a single prespecified primary endpoint. Questions of carryover between periods and washout between sequences do not apply to this type of design | |
| Blinding integrity | Reservation |
| FindingA blinding-integrity assessment was conducted, and it is informative. In the pramipexole arm, 43.9% of patients correctly guessed their allocation, no different from chance. In the placebo arm, 66.7% guessed correctly, with a p value at the edge of significance (p = 0.053). The blind therefore weakens on the placebo side, not the treatment side. This is not reassuring either way: on a self-rating, a patient who has realized they are receiving nothing may report less improvement, and that bias pushes in the same direction as one from a patient who knows they are being treated | |
| Analysis population | Solid |
| FindingModified intention-to-treat, 82 of the 85 randomized. The three exclusions are documented: two dropouts before week 3 with no post-baseline data, and one participant eligible at screening but not at randomization, ineligibility confirmed by external review. 80 participants completed the controlled phase. Primary analysis by mixed model for repeated measures adjusted for baseline score, with a sensitivity analysis using last observation carried forward yielding similar results | |
| Single-center nature | Reservation |
| FindingA single Swedish site. Patient selection, prescribing culture, and follow-up are homogeneous there, which helps internal validity and limits generalizability | |
| Sample size and precision | Reservation |
| Finding82 patients analyzed. The confidence interval of the primary endpoint runs from −6.89 to −1.18: the data remain compatible with a sizeable effect as much as with a very small one | |
| Funding | Solid |
| FindingSwedish Research Council, ALF government health-care research grants, Region Skåne, and several Swedish foundations. The article states this is an academic trial and that funders had no role in its design or conduct. This is the source that matters here, more than an institutional press release | |
| Authors’ conflicts of interest | Solid |
| FindingThe disclosure is public and detailed. Lindqvist reports speaker fees from H. Lundbeck AB and Janssen-Cilag AB and a collaboration with HMNC Brain Health; Asp reports speaker fees from H. Lundbeck AB. Pizzagalli, a US co-author, reports consulting fees from roughly ten industry partners, honoraria from professional and publishing organizations, research funding from public and philanthropic sources, and stock options in five neuroscience companies: Ceretype Neuromedicine, Compass Pathways, Engrail Therapeutics, Neumora Therapeutics, and Neuroscience Software. The remaining authors declare no conflicts. What matters is a negative, verifiable fact: none of these relationships involves a manufacturer of pramipexole, and none of the five equity holdings is Alto Neuroscience, the company developing ALTO-207, a fixed-dose combination of pramipexole and ondansetron. That company itself states that the trial was conducted and funded independently by Lund University | |
| Open-label extension | Reservation |
| FindingSix months with no blind and no comparator. 54 of 65 eligible participants entered it, 37 completed it, a 31% attrition rate among entrants. The maintained improvement therefore describes the trajectory of those who remained, which is enough to limit its scope: among these 37, 59.5% show at least a 50% reduction in SHAPS score and 37.8% reach remission | |
| Documentation of tolerability | Reservation |
| FindingThe abstract describes tolerability as generally good compared with placebo, with no figures. The full text provides them, and they tell a different story: sleep disturbance 72% versus 33%, nausea 60% versus 14%, dizziness 33% versus 5%, fatigue 47% versus 31%, anxiety 30% versus 7%. The claim of favorable tolerability rests on a single discontinuation for an adverse event and the absence of a serious adverse event during the controlled phase, not on the frequency of effects, which is markedly higher than under placebo | |
Results
| Endpoint | Published result |
|---|---|
| SHAPS at 9 weeks (primary) | Mean difference −4.04 (95% CI: −6.89 to −1.18); p = 0.006; Hedges’ g 0.62; 82 participants analyzed |
| ReadingDemonstrated the prespecified primary endpoint is met, with a moderate effect size. This is the trial’s only primary endpoint: everything else is either a prespecified secondary endpoint or an exploratory analysis | |
| Clinical magnitude of these 4 points | Baseline scores 41.5 under pramipexole and 39.5 under placebo, on a 14-item scale rated 1 to 4, range 14 to 56 |
| ReadingCan be sized, not graded the 4.04 points represent roughly 10% of the baseline score. Neither negligible nor transformative. What is missing to settle the question is a validated minimal clinically important difference for the SHAPS, which does not exist | |
| MADRS-S at 9 weeks (secondary) | No significant between-group difference at week 9, small effect size. Significant main effect across all post-baseline measurement times (p = 0.012) |
| ReadingDiscordant this is the strongest critical point in the dossier. The anhedonia scale separates the groups at the planned endpoint, the depression scale does not. A main effect across all measurement times does not substitute for a difference at the prespecified time point. Two readings remain open: either the effect is genuinely confined to the anhedonic dimension, which is the authors’ hypothesis, or the trial is underpowered for a broader endpoint | |
| Light physical activity | Increased under pramipexole, exploratory analysis |
| ReadingSuggested a result consistent with the motivational hypothesis, but exploratory, therefore hypothesis-generating rather than demonstrative | |
| Reward-related ventral striatal activation | Significant between-group difference: p = 0.030; Hedges’ g 0.64. Reward-related activity preserved under pramipexole, decreased under placebo. 7-tesla functional MRI, 48 participants imaged |
| ReadingSuggested the result is quantified and points in the hypothesized direction, but its status remains exploratory, and the effect owes as much to the decrease observed under placebo as to any increase under treatment. This is preserved activation, not increased connectivity: the two notions are not interchangeable | |
| Impulse control disorders | Assessed with a dedicated questionnaire, administered at several time points. At week 3, difficulty controlling purchases in 25.6% of patients under pramipexole versus 4.8% under placebo, a significant difference, not found at weeks 6 and 9. No between-group difference on the gambling scale |
| ReadingSafety signal this is the most important result for a psychiatrist reader, and it is measured, not inferred. One patient in four at week 3, during titration. Its disappearance at weeks 6 and 9 should be read cautiously: it may reflect adaptation, dose stabilization, or under-reporting after the first alert | |
| Mood switch | Two patients in the pramipexole group developed mild manic symptoms: decreased need for sleep, increased energy, flight of ideas |
| ReadingSafety signal these two patients were dysthymic, not bipolar. Vigilance therefore cannot be limited to patients with known bipolar disorder; it applies to the whole treated population. In one, symptoms resolved after dose reduction | |
| Overall tolerability | Sleep disturbance 72% versus 33%, nausea 60% versus 14%, dizziness 33% versus 5%, fatigue 47% versus 31%, anxiety 30% versus 7%. One discontinuation for an adverse event. No serious adverse event during the controlled phase |
| ReadingDocumented, and worse than placebo tolerability is documented, it is simply less favorable than the abstract suggests. What is genuinely good is acceptability: patients tolerate these effects without discontinuing. With 82 patients over nine weeks, the trial in any case lacks the power to detect a rare event | |
| Open-label extension, 6 months | 54 of 65 eligible entrants, 37 completing. Anhedonia improvement maintained among remaining participants |
| ReadingUncontrolled with no blind and no comparator, this phase documents a trajectory, not an attributable effect. And 31% attrition among entrants is by itself enough to explain an improving average result, without needing to invoke anything else | |
A word on how this result will be cited. A Hedges’ g of 0.62 corresponds, by usual convention, to a moderate effect. The temptation will be to compare it with what antidepressants achieve against placebo: the comparison flatters pramipexole, since meta-analyses place those standardized differences around 0.30, higher in severe forms but never reaching 0.62. It is above all a shaky comparison, because neither the scale, the population, nor the comparator are the same: here placebo is added to a treatment already in place. This is not a spectacular effect, and it is not nothing.
Critical appraisal
| Domain | Judgment |
|---|---|
| Nature of the design | Solid |
| FindingRandomized, preregistered trial, protocol published, a single primary endpoint, modified intention-to-treat analysis. Nothing to fault in the construction | |
| Population heterogeneity | Reservation |
| FindingMajor depressive disorder (59.8%), dysthymia (37.8%), and bipolar depression (2.4%) were pooled. The issue is not bipolarity, which accounts for two of 82 patients and allows no inference, but the weight of dysthymia: nearly four patients in ten have a chronic, low-intensity disorder whose anhedonia, spontaneous course, and expected treatment response differ from those of a full depressive episode. No subgroup result is reported, and the trial is not sized to produce an interpretable one. Pooling is justified if anhedonia is genuinely a transdiagnostic endophenotype, which is a working hypothesis, not an established fact | |
| Mood-switch signal | Reservation |
| FindingTwo patients under pramipexole developed mild manic symptoms, and both were dysthymic. The reflex of reserving this kind of monitoring for patients with bipolar disorder is therefore contradicted by the trial’s own data. The product’s summary of characteristics, for its part, calls for regular monitoring for emerging mania and delirium, with no diagnostic restriction | |
| Treatment-resistance status | Reservation |
| FindingEligibility required ongoing treatment for at least four weeks and at least one adequate antidepressant trial without remission. The population is therefore characterized, and it would be inaccurate to say no criterion was defined. The reservation lies elsewhere: a single documented failure falls short of the usual definition of resistance, which requires two. Media coverage describing this as treatment-resistant depression therefore uses a stronger term than what the trial actually required of its participants | |
| Self-rated primary endpoint | Reservation |
| FindingThe SHAPS is patient-completed. An imperfect blind weighs more heavily on a subjective self-rating than on an instrumental measure. Here the flaw sits in the placebo arm, where two patients in three guess their allocation: the expected bias is disappointment widening the gap, and it pushes in the same direction as the bias from the treated arm | |
| Status of mechanistic analyses | Reservation |
| FindingThe title announces target engagement, the article itself classifies these analyses as exploratory. Three statuses must stay distinct in any citation of this trial: the primary endpoint, the SHAPS change at week 9; the prespecified secondary endpoint, the MADRS-S; the exploratory analyses, physical activity and imaging. A concordant, quantified exploratory analysis strengthens plausibility, it does not demonstrate it. Confirmation would require a prespecified mechanistic endpoint and correction for multiple comparisons | |
| Time horizon | Reservation |
| FindingNine weeks under comparator. The adverse effects of greatest concern with this class, impulse control disorders in particular, often emerge after several months of exposure and at increasing doses. A single discontinuation for an adverse event over nine weeks contrasts, moreover, with the 20% reported over twelve weeks in the 2025 UK trial, a gap that likely owes as much to differing populations and titration schedules as to duration | |
| Reproducibility | Reservation |
| FindingOne center, one principal investigator, a Nordic population. But the replication question no longer stands in quite the same terms as two years ago: a UK multicenter trial published in 2025, 151 participants randomized across nine sites, showed the superiority of add-on pramipexole on a mood endpoint, with a mean difference of −3.91 on the QIDS-SR16 (95% CI −5.37 to −2.45; p < 0.0001). What remains to be replicated outside a single center is the specific effect on anhedonia and its neural substrate | |
Two symmetrical errors threaten the reading of this dossier. The first is to overstate the result: to say that a treatment for anhedonia now exists. That is not what the authors write, they speak of a potential augmentation strategy, and it is not what the design authorizes. The second is to dismiss the result on the grounds that the trial is small and single-center: the primary endpoint was single, prespecified, and it was met with a moderate effect size on a dimension our usual treatments handle poorly.
What is demonstrated: over nine weeks, in 82 patients at a single center, adding pramipexole reduces the anhedonia score more than adding placebo, without the depression scale separating the groups at the planned endpoint. What is suggested: this effect may operate through the dopaminergic reward circuit, and may come with a rebound in physical activity. What amounts to expert opinion, including our own: that anhedonia deserves to be treated as a pharmacological target in its own right rather than as one symptom among others in a depressive episode. And what is measured on the risk side, without being quantifiable beyond this sample: difficulty controlling purchases in one patient in four during titration, and two mood switches in patients who were not bipolar.
Level of evidence
PEB appraisal: moderate confidence in the existence of an effect of pramipexole on anhedonia at nine weeks in selected patients, low confidence in its true clinical magnitude, low confidence in the proposed mechanism, very low confidence in medium-term tolerability in this population. The scientific score of 72 reflects a sound design, tempered by its single-center nature, sample size, short duration, the discordance between the anhedonia and depression scales, and two safety signals that deserve more than a passing mention. The editorial score of 88 reflects the fact that residual anhedonia is an everyday problem in consultation, and that this dossier teaches as much through its results as through the gap between its abstract and its full text.
The colleague test
What an experienced colleague might say about this study in thirty seconds, in a hallway.
“First clean randomized trial that targets anhedonia rather than mood, and it’s positive, with a moderate effect size. But it’s one center, 82 patients, nine weeks, the MADRS doesn’t separate the groups at the end, and the imaging part is exploratory. We already have the multicenter one: the 2025 UK trial, positive too but on a mood endpoint, with 20% discontinuing for intolerance. And a quarter of patients losing control of their spending by week 3, that has to be said. Where I practice, the molecule is only authorized for Parkinson’s and restless legs, so the question of prescribing it doesn’t even arise yet.”
Also keep the methodological lesson: a trial’s abstract is not the trial. Here the full text is open access, and it contains the adverse-effect frequencies, a measurement of impulse control disorders, two mood switches, the baseline scores, and the depression-scale result. None of this appears in the abstract. Concluding from a silent abstract that a piece of data does not exist is a recurring error, and it is corrected simply by opening the article.
What you can do with this on Monday morning
What you can do with this on Monday morning. Nothing to prescribe, and yet several things to do. Pramipexole has no psychiatric indication in France, and this trial does not make it available for depression there, and the same caution applies wherever you practice. What changes is what you can explain, recognize, and document.
- Name anhedonia and measure it. Many patients describe their state as a treatment failure when it is in fact an identifiable residual symptom. The SHAPS takes a few minutes to administer and makes visible a dimension that mood scales capture poorly. Tracking a score over time serves better than concluding to global resistance.
- Answer a patient who has read about this in the press, with exact words. A Swedish trial, 82 patients analyzed, nine weeks, a real improvement in anhedonia relative to placebo, a moderate effect size, but no difference on the depression scale at the planned endpoint. And a molecule whose French authorization covers Parkinson’s disease and restless legs syndrome, not depression: check the equivalent authorization status wherever you practice, since it varies by country. A promising finding is not an available treatment.
- Keep in mind the figure that matters most to a psychiatrist. In this trial, at week 3, one patient in four under pramipexole reported difficulty controlling purchases, versus one in twenty under placebo. The difference then disappears, which is no reason to forget it. And the two patients who developed mild manic symptoms were dysthymic, not bipolar: monitoring for mood switch cannot be reserved for patients with known bipolar disorder.
- Know how to recognize the picture in patients you already follow in neurology. If you follow a patient with Parkinson’s disease or restless legs syndrome, monitoring for impulse control disorders is part of the product’s summary of characteristics: gambling, buying, eating, sexual behavior. These behaviors are rarely reported spontaneously. Asking family members, with the patient’s consent, changes the yield of the question.
- Systematically document somnolence in these same patients. Somnolence is a very common effect in Parkinson’s disease and a common one in restless legs syndrome, and sudden-onset sleep episodes during ordinary activities are uncommon but reported, sometimes without warning. The question of driving then arises explicitly, and the answer belongs in the chart.
- Never stop a dopamine agonist abruptly. The product’s summary of characteristics calls for gradual tapering: an abrupt stop risks a dopamine-agonist withdrawal syndrome, and abruptly stopping dopaminergic treatment can trigger a neuroleptic malignant syndrome.
- Keep the reading grid. A prespecified primary endpoint, a prespecified secondary endpoint, and an exploratory analysis are not cited the same way; effect size says something different from a p value; and an open-access article is read in full rather than in its abstract. Three habits that will serve well beyond this dossier.
Frequently asked questions
Can pramipexole be prescribed for depression in France?
The product’s summary of characteristics, consulted on the ANSM public medicines database on August 10, 2026, lists only two indications: idiopathic Parkinson’s disease in adults and moderate to severe idiopathic restless legs syndrome. There is no psychiatric indication. Use in depression would therefore be off-label, and a single, single-center, nine-week trial does not provide grounds for such a decision. Readers practicing outside France should check the equivalent authorization status in their own country: the same distinction between a promising trial result and an approved indication applies wherever pramipexole is prescribed.
Was this trial a crossover design?
No. Parallel groups, 1:1 allocation, nine weeks, with no sequence crossover, in both the published article and the 2023 protocol. The carryover and washout questions specific to crossover designs therefore do not apply here. The six-month open-label extension is not a crossover either: every participant receives pramipexole in it, with no blind and no comparator.
Does the functional MRI prove the mechanism?
No, even though the result is quantified. The imaging, performed at 7 teslas in 48 participants, shows a significant between-group difference (p = 0.030; Hedges’ g 0.64), with reward-related ventral striatal activation preserved under pramipexole and decreased under placebo. The analysis remains classified as exploratory by the authors themselves. A concordant exploratory analysis increases the plausibility of a mechanism, it does not establish it: demonstration would require a prespecified mechanistic endpoint, with correction for multiple comparisons.
Is a four-point change on the SHAPS a lot?
It is roughly 10% of the baseline score. The SHAPS comprises 14 items rated 1 to 4, a range of 14 to 56, and patients entered the trial with mean scores of 41.5 and 39.5. The second benchmark is the effect size, a Hedges’ g of 0.62, which conventionally corresponds to a moderate effect. What is missing to settle the question is a validated minimal clinically important difference for this scale, which does not exist. The confidence interval of the difference, −6.89 to −1.18, remains wide.
Were impulse control disorders observed in this trial?
Yes, and they were measured with a dedicated instrument, not only collected spontaneously. At week 3, 25.6% of patients under pramipexole reported difficulty controlling purchases, versus 4.8% under placebo, a significant difference. It is no longer found at weeks 6 and 9, and no difference appears on the gambling scale. This is the safety result most directly useful in consultation. It should be added that 82 patients followed for nine weeks cannot detect a rare adverse effect, and that these disorders often emerge after several months of exposure.
Did mood switches occur?
Yes. Two patients in the pramipexole group developed mild manic symptoms: decreased need for sleep, increased energy, flight of ideas. The clinical point is that both were dysthymic, not bipolar. In one, symptoms resolved after dose reduction. Vigilance therefore cannot be limited to patients with known bipolar disorder, especially since the trial counted only two such patients among 82. The product’s summary of characteristics, for its part, calls for regular monitoring for emerging mania and delirium in patients treated with pramipexole.
Was the trial funded by a pharmaceutical company, and do the authors have conflicts of interest?
Funding came from the Swedish Research Council, ALF government health-care research grants, Region Skåne, and several Swedish foundations, and the article states that funders had no role in the trial’s design or conduct. The conflict-of-interest disclosure is public: Lindqvist reports speaker fees from H. Lundbeck AB and Janssen-Cilag AB and a collaboration with HMNC Brain Health; Asp reports speaker fees from H. Lundbeck AB; and Pizzagalli, a US co-author, reports consulting fees from roughly ten industry partners, honoraria from professional and publishing organizations, research funding from public and philanthropic sources, and stock options in five neuroscience companies. None of these relationships involves a manufacturer of pramipexole. None of the five equity holdings is Alto Neuroscience, which is developing a fixed-dose combination of pramipexole and ondansetron and has commented on this publication: that company states itself that the trial was conducted and funded independently by Lund University.
Annotated bibliography
Ventorp F, Asp M, Olsson S, Lindahl J, Möller S, Svensson M, Ängeby F, Pravdinske A, Tjernberg J, Schrey S, Ståhl D, Magnusson V, Eliasson E, Agelii-Weber A, Stålhammar E, van Westen D, Deierborg T, Månsson KNT, Pizzagalli DA, Hammar Å, Tornberg ÅB, Björkstrand J, Lindqvist D (2026). Efficacy and target engagement of dopamine agonist pramipexole for anhedonic depression: a randomized placebo-controlled trial. Nature Medicine, 32(7), 2570-2578. Epub 2026 Jun 12. DOI 10.1038/s41591-026-04465-9 · PMID 42286321. Source study analyzed here, read in full text, open access under a CC BY license. Single-center, parallel-group trial, published online June 12, 2026. Registered as NCT05355337 and NCT05825235. The deposited manuscript (Research Square rs-7887410 v1, PRIME-PRAXOL trial) reports the same results. Conflict-of-interest disclosure, confirmed on PubMed: Lindqvist reports speaker fees from H. Lundbeck AB and Janssen-Cilag AB and a collaboration with HMNC Brain Health; Asp reports speaker fees from H. Lundbeck AB; Pizzagalli reports consulting fees from about ten companies (Arrowhead Pharmaceuticals, Boehringer Ingelheim, Circular Genomics, Compass Pathways, Engrail Therapeutics, Karla Therapeutics, Neumora Therapeutics, Neurocrine Biosciences, Neuroscience Software), honoraria from the American Psychological Association, the Psychonomic Society, Springer, and Alkermes, research funding from the Bird Foundation, the Brain and Behavior Research Foundation, the Dana Foundation, Millennium Pharmaceuticals, NIMH, and Wellcome Leap, and stock options in five companies (Ceretype Neuromedicine, Compass Pathways, Engrail Therapeutics, Neumora Therapeutics, Neuroscience Software); the remaining authors declare no conflicts; none of these relationships concerns a manufacturer of pramipexole, and none of the five equity holdings is Alto Neuroscience.
Browning M, Cowen PJ, Galal U, et al. (2025). Pramipexole augmentation for the acute phase of treatment-resistant, unipolar depression: a placebo-controlled, double-blind, randomised trial in the UK. Lancet Psychiatry, 12(8), 579-589. DOI 10.1016/S2215-0366(25)00194-4. UK multicenter trial, 151 participants randomized across nine sites, twelve weeks. Mean difference on the QIDS-SR16 of −3.91 (95% CI −5.37 to −2.45; p < 0.0001) favoring add-on pramipexole. Discontinuation for adverse events: 20% under pramipexole versus 5% under placebo. An essential reference here for two reasons: the multicenter replication of the benefit already exists, on a mood endpoint rather than an anhedonia endpoint, and this trial supplies an order of magnitude for the discontinuation rate that the Lund trial, with a single discontinuation, cannot estimate.
Lindahl J, Asp M, Ståhl D, et al. (2023). Add-on pramipexole for anhedonic depression: study protocol for a randomised controlled trial and open-label follow-up in Lund, Sweden. BMJ Open, 13(11), e076900. DOI 10.1136/bmjopen-2023-076900. Protocol published before the results. Confirms the parallel-group design, the planned sample of 80 patients, and a population combining unipolar depression, bipolar depression, and dysthymia. The final eligibility criteria, the item-level anhedonia criterion, and the maximum dose of 4.5 mg of salt per day, that is 3.15 mg of base, are taken here from the published trial’s methods, not from the protocol.
Agence nationale de sécurité du médicament et des produits de santé. Summary of product characteristics, PRAMIPEXOLE BIOGARAN 0.18 mg, tablet. base-donnees-publique.medicaments.gouv.fr, consulted August 10, 2026. Regulatory reference for this article: indications limited to idiopathic Parkinson’s disease and moderate to severe idiopathic restless legs syndrome, maximum doses, monitoring for impulse control disorders, somnolence and sudden-onset sleep episodes, monitoring for mania and delirium, gradual discontinuation. A second summary of product characteristics, for PRAMIPEXOLE ZENTIVA 0.18 mg, was consulted the same day and agrees on indications and warnings. Readers practicing outside France should consult the equivalent regulatory source in their own country, since pramipexole’s authorized indications vary by jurisdiction.
Lund University. Parkinson’s medication shows promise in treating treatment-resistant depression. lunduniversity.lu.se, consulted August 10, 2026. Institutional press release, a secondary source, cited only to document how the trial was presented to the public. Its title describes the population as treatment-resistant, whereas eligibility required only a single adequate antidepressant trial without remission, short of the usual definition of resistance. Neither the funding, nor the academic nature of the trial, nor the 7-tesla imaging protocol are drawn from this release: they appear in the article itself.
Alto Neuroscience. Alto Neuroscience highlights Nature Medicine publication demonstrating significant effects of pramipexole on anhedonia. investors.altoneuroscience.com, consulted August 10, 2026. Corporate press release, cited only to document the existence of a commercial interest around ALTO-207, a fixed-dose combination of pramipexole and ondansetron. The company states there that the trial was conducted and funded independently by investigators at Lund University. To be read as financial communication, not as a scientific source.
Editorial collections
Tags
Verified on August 10, 2026 against the full text of the publication and its supplementary material where available. This analysis underwent an independent double reading. The English version was checked for conformity on September 21, 2026, against the figures of the French version and against the source. How we verify what we publish
This analysis is intended for healthcare professionals. It does not constitute a prescribing recommendation and does not replace individual clinical judgment.
Analysis from Psychiatry Evidence Base, evidence-based psychiatry, explained with rigor.