Is pindolol top-line for bipolar depression, or is this a glitch of the network meta-analytic method?
What works, and at what dose?
STUDY: Fornaro M et al, Journal of Affective Disorders 2026
STUDY TYPE: Network meta-analysis
FUNDING: Independent
Background
Earlier meta-analyses of bipolar depression pooled doses together, blurring whether a drug works at 50 mg or 400 mg. This one adds that extra layer of refinement.
The Study
- Network meta-analysis of 103 randomized controlled trials, 20,941 patients, 71 drug-dose combinations; median trial length was 8 weeks.
Results
These doses outperformed placebo, with effect sizes (SMD) and their confidence intervals in parentheses:
- Amisulpride 200 mg/day (3.36, 2.44-4.29) and 400 mg/day (3.09, 2.17-4.01)
- Ziprasidone 120 mg/day (2.05, 0.5-3.62)
- Inositol 12 g/day (1.79, 0.83-2.75)
- N-acetylcysteine (NAC) 1 g/day (1.74, 0.32-3.16)
- Lamotrigine 50 mg/day (1.53, 0.93-2.13) and 200 mg/day (1.04, 0.70-1.35)
- Valproate 2.5 g/day (1.53, 0.29-2.76)
- Olanzapine 14.4 g/day (1.42, 0.41-2.42)
- Pramipexole 3 mg/day (1.34, 0.06-2.61)
- Pindolol 7.5 mg/day (1.18, 0.09-2.25) (acute treatment during with sleep deprivation)
- Lithium 900 mg/day (1.13, 0.13-2.13)
- OFC with olanzapine 12 mg + fluoxetine 50 mg/day (0.81, 0.16-1.46) (this outperformed olanzapine 6 mg + fluoxetine 50 mg)
After restricting to low-bias trials and excluding outliers, the list narrowed to these:
- Lamotrigine 50 mg/day and 200 mg/day
- Quetiapine XR 150 mg/day and 300 mg/day
- Lumateperone 42 mg/day (but not 28 mg)
- Pindolol 7.5 mg/day (acute treatment during sleep deprivation)
Cariprazine did not make the cut, likely because its already small effect size was hindered by the inclusion of low dose trials (it didn’t work at 0.75 mg, and it also failed in bipolar II depression).
Pindolol was a surprise that needs some clarification. It was studied as an add-on to total sleep deprivation, a rarely used but likely effective behavioral intervention for depression. Sleep deprivation desensitizes the serotonin 1A autoreceptors that pindolol acts on as an antagonist. The trial was small (40), inpatient, and pindolol was used for only nine days while the patients underwent three cycles of total sleep deprivation. After this rapid therapy, they were stabilized on lithium, and 65% maintained recovery.
Supplements
NAC may work, but my read of the trials is that it takes 4-6 months to show an effect. Still, this antioxidant is worth considering with lithium, where it may reduce the mood stabilizer’s renal risks (here are lab-tested NAC brands). Inositol’s efficacy is shakey, though it is useful as an antidote to psoriasis on lithium. Missing from this analysis are coenzyme Q10 and omega-3, which have small trials in bipolar depression and are available as a combined option as Enlyte.
Personalizing the Selection
Next, they filtered the data in ways that help personalize the treatment for:
- Children: Only lurasidone 80 mg separated from placebo
- Bipolar II: Lamotrigine 200 mg and lithium 900 mg
- Anxious bipolar: Lurasidone 60 mg, valproate 2,500 mg (possibly more anxiolytic effects at max therapeutic dose), and quetiapine 300 mg
- Higher tolerability: Lurasidone 33.6 mg (unclear why lamotrigine missing here)
Lowest tolerability was found with risperidone 6 mg (mainly prolactinemia). Armodafinil had the highest dropout rates, though this was likely due to lack of efficacy as this wakefulness-promoting agent is reasonably tolerated.
Limitations
Network meta-analyses start with a faulty assumption: that all placebo groups are identical and different trials are comparable. That inserts noise into the analysis, a bit like listening to music while loud trains are rolling by. Hard to tell which song is the best.
That’s why no drug beat placebo for response or remission in this analysis, even though they beat placebo within their own trials, which is the gold-standard.
Heterogeneity was high, many comparisons relied on sparse, indirect data, and most effect sizes likely overstate real-world benefit.
Practice Implications
- For bipolar II, lamotrigine and lithium are good options, and the two have synergestic benefits.
- For anxiety, valproate and quetiapine are well-known standards, and this study adds lurasidone to the short-list of options.
- As for pindolol? Some of my colleagues have success with sleep deprivation in bipolar depression, despite the manic risk. Pindolol may accelerate that therapy, but it’s not the only option. Light therapy and lithium also have promise (as triple chronotherapy). In unipolar depression, pindolol also works as an accelerator for SSRIs, but doesn’t have lasting benefits there. I explain how to use it in Difficult to Treat Depression.
—Chris Aiken, MD
Director, Psych Partners
Editor in Chief, Carlat Psychiatry Report
