The Best Prevention in Bipolar Disorder

Art: Ellsworth Kelly, Spectrum V, 1969

What works, what is tolerable, and at what dose

STUDY: Fornaro, J Affect Disord. 2026 Apr 13:121798.

STUDY TYPE: Systematic review and network meta-analysis

FUNDING: Independent (but most trials industry funded)

Background

Most bipolar guidelines tell you which drugs to use — but not which doses. This network meta-analysis set out to fix that, comparing agents head-to-head across doses and age groups in the largest synthesis to date. They focused on the maintenance (ie, preventative) phase.

The Study

Forty-four randomized controlled trials enrolling 10,867 adults and children with bipolar I or II disorder were included, with follow-up ranging from 6 to 24+ months. Twenty-three treatment combinations — mood stabilizers, antipsychotics, and combination regimens — were compared against each other and placebo for relapse prevention and tolerability.

In the primary sensitivity analysis (low-risk-of-bias trials only), 15 treatments beat placebo on any-mood-relapse prevention. Here they are with top-ranking first:

1. Asenapine 20 mg/day
2. Aripiprazole 20 mg/day
3. Quetiapine 300 mg/day
4. Lithium 800 mg/day (though in reality is dosed by serum level, target 0.6-0.8 mEq/L)
5. Lithium 600 mg/day
6. Quetiapine 600 mg/day
7. Lurasidone 80 mg/day (with full meal)
8. Carbamazepine 400 mg/day (you can check serum levels, but the target range isn’t well established)
9. Valproate 71-125 μg/mL (serum level) [see Footnote]
10. Olanzapine 20 mg/day
11. Long-acting aripiprazole 400mg/4 weeks
12. Long-acting risperidone 25mg/2 weeks
13. Aripiprazole 30mg/day combined with lamotrigine 200mg/day
14. Carbamazepine 650 mg/day
15. Quetiapine 550mg/day

However, the confidence intervals for most of the top 5 treatments overlapped, so don’t take the ranking too seriously:

They measured tolerability as “discontinuation due to any side effect.” Here, carbamazepine ranked #1…

1. Carbamazepine 650mg/day
2. Asenapine 20mg/day
3. Long-acting risperidone 25mg/2 weeks
4. Lamotrigine 200mg/day (logRR=-1.62; 95%C.I. =-2.34;-0.90)

What? Risperidone is more tolerable than lamotrigine? Again the confidence intervals overlap a lot.

Quetiapine 550 mg/day emerged as the most protective overall for both depressive and manic relaps. For children, lithium 1400 mg/day and lamotrigine 200 mg/day ranked top [see Footnote]. I agree with that for lithium, but note that lamotrigine failed in its main child trial and is not FDA approved there.

45% of the trials enrolled a mix of bipolar I and II, but the researchers did not parse out response by subtype.

A Few Problems

Network metaanalysis is notoriously unreliable because it assumes the placebo effect was similar across all trials. This paper further blurs reality by combining bipolar I and II together. It’s a skewed snapshot, so I never apply network results without checking them against the full body of research. And that tells us….

• Bipolar II. Lamotrigine alone is often adequate for bipolar II, as we don’t need to prevent mania there. This analysis relied on a large trial of lamotrigine as monotherapy in bipolar I, where it failed, and it should not be used as monotherapy there.
• Other subtypes. Classic bipolar (pure manias or hypomanias followed by depression, with long periods of recovery) responds better to lithium, while other bipolar types responds better to anticonvulsants and antipsychotics. That might matter more than the ranking here. Comorbidities also matter — consider quetiapine for anxiety disorders, valproate for alcohol use disorder.
• Functional outcomes. This analysis looked at relapses, but lithium has better long-term functional outcomes than antipsychotics, and in some trials better prevention, which is why ISBD ranks it #1 for new onset bipolar disorder. Had they looked at hospitalization rates, suicide, or overall mortality, lithium also would have come up on top.
• Antipsychotic heavy. Meta-analyses usually rely on placebo-controlled trials, which can favor industry-sponsored meds like antipsychotics.

Practice Implications

With all those caveats, what can we take away from this analysis?

1. The long-term superiority of lithium and quetiapine is supported by an analysis they late Terrence Ketter and David Osser’s guidelines. I agree, and find quetiapine tops for anxiety, depression, and insomnia; lithium tops for classic bipolar.
2. Asenapine rose to the top for both efficacy and tolerability. Although it barely has any data in acute bipolar depression, it did prevent depression in long-term trials, so its dual efficacy is worth considering. And it is reasonably well tolerated (though sedating).
3. Carbamazepine is under-utilized, but offers a tolerability advantage that patient’s appreciate. It is daunting to dose, but I’ve simplified that process in Prescribing Psychotropics.
4. Lamotrigine is favored in bipolar II, and as an add-on treatment it has a good balance of efficacy/tolerability in bipolar I.
5. The dosing targets here are reasonable for bipolar I. Carbamazepine’s is surprisingly low, but in line with the trials.
6. The study reassures us about what works long-term, but personalizing these options is more important than the ranking.