On August 4, 2023, the US Food and Drug Administration (FDA) granted marketing authorization for zuranolone (Zurzuvae) for postpartum depression. Prior to this recent approval, patients with postpartum depression had only 1 FDA-approved option: brexanolone, an intravenous medication requiring infusion across multiple days, with a cost of $34 000 for a single course of treatment.¹ Zuranolone thus fills an important niche as an oral option. However, there are several features of the design and conduct of the drug’s pivotal trials that undermine confidence in this agent, which should give physicians pause in prescribing it. Here, we explore the evidence base of the FDA approval of zuranolone.

Postpartum depression is a subset of major depressive disorder that meets the diagnostic criteria of major depressive disorder but occurs with onset during pregnancy or within 4 weeks of delivery. Current guidelines advise the mainstays of major depressive disorder therapy for individuals with postpartum depression. For instance, first-line treatment is psychotherapy. The American College of Obstetricians and Gynecologists (ACOG) recommends a selective serotonin reuptake inhibitor (SSRI) as first-line pharmacotherapy for women with postpartum depression.² For women with no previous exposure to SSRI or serotonin and norepinephrine reuptake inhibitor treatment, sertraline and escitalopram are considered reasonable first-line agents.

Zuranolone is a neuroactive steroid that is a positive modulator of the GABA_AR (γ-aminobutyric acid “A” receptor). The drug works by increasing the inhibitory tone of the central nervous system. Zuranolone exerts its effect via the same receptor as benzodiazepines and barbiturates. Although it is similar to brexanolone in action, zuranolone is an oral formulation.

The drug’s mechanism of action underpins its adverse effect profile (somnolence, dizziness, sedation) and may explain the black box warning stating that patients should not drive a motor vehicle or engage in potentially hazardous behavior within 12 hours of taking this medication. The similarity to other GABA-ergic compounds may also explain why the drug demonstrated a dose-dependent misuse potential comparable with that of alprazolam (Xanax)³ and why human participants have been shown to exhibit symptoms of withdrawal after cessation of zuranolone in clinical trials. This is specifically stated in the FDA’s package leaflet, which states that “Adverse reactions reported upon discontinuation of zuranolone in healthy subjects who received 50 mg of zuranolone for 5 to 7 days (on the 7th day subjects received 50 mg or 100 mg) included: insomnia, palpitations, decreased appetite, nightmare, nausea, hyperhidrosis, and paranoia. These adverse reactions indicate a potential for physical dependence with zuranolone.”³

There are significant safety concerns regarding zuranolone’s use in both pregnant and breastfeeding individuals. Trial participants were required to abstain from breastfeeding from the first dose administered through 7 days following the last dose of the 14-day treatment. They were also required to be receiving effective contraception. These precautions seem appropriate, given that preclinical animal models demonstrated neurotoxic effects on the developing brain. According to the FDA package insert, “Oral administration of zuranolone to rats during pregnancy and lactation resulted in developmental toxicity in the offspring, including, perinatal mortality, at maternal exposures similar to that in humans at the [maximum recommended human dose].”³

There are 2 trials that support the use of this drug. The first study, ROBIN, examined the effect of 30-mg capsules of zuranolone, once per day (evening with meal), vs placebo in patients with severe postpartum depression.⁴ Severity was determined with inclusion criteria that required a baseline Hamilton Depression Rating Scale (HAMD)-17 score of 26 or higher. Thus, the trial is restricted to individuals with severe postpartum depression. The mean baseline score was just above 28 in both groups. Although the drug achieved statistical significance, there was a marked placebo effect. The primary end point (change from baseline in HAMD-17 score at day 15) improved in the placebo group by 13.6 points, three-quarters of the 17.8-point improvement seen with zuranolone.

The second registration study, SKYLARK, had an almost identical trial design, testing a higher dose of zuranolone (50 mg). Participants similarly had severe depressive episodes, with mean baseline HAMD-17 scores higher than 28. A similar result was shown, with 74% of the improvement at day 15 achieved with placebo vs zuranolone.⁵

There are at least 3 concerns with the trials used to support this approval. First, the use of placebo control is inappropriate. Multiple guidelines endorse treatment of severe postpartum depression with SSRIs and psychotherapy. Yet, in ROBIN and SKYLARK, just 20% and 15% of women, respectively, were using antidepressants at baseline. Patients were required to delay the start or alteration of any psychotropic treatment regimens until after day 15 in ROBIN and until after day 45 in SKYLARK. Each participant should have been prescribed an SSRI and psychotherapy based on the current standard of care. The median length of time to response expected with these interventions does not justify withholding treatment from the control group. We are unable to assess whether the improvement in the treatment group would have been observed vs standard of care.

Second, these trials assessed drug efficacy in individuals with severe postpartum depression. Meanwhile, the FDA’s drug approval does not qualify its indication for individuals with postpartum depression of this severity, instead granting a blanket approval for milder forms of postpartum depression—an indication that lacks evidence.

Third, the placebo effect and/or regression to the mean is a large effect. In fact, roughly 75% of the effect of the product occurred with placebo in both trials. Patients with postpartum depression in this study tended to get better, regardless of treatment. The authors suggest this could be due to the high number of study visits, although another potential contribution is regression to the mean. To enter either study, patients needed a HAMD-17 score of 26 or greater. Any participant needed to achieve this both on initial screening and day 0. Thus, a person whose score improved between these points, eg, from day 28 to 24, would be excluded from the study. Data are not provided on the reason for screening failures. We do know that 44% of patients in ROBIN who were enrolled and screened were excluded prior to randomization and 60% were excluded in SKYLARK. This raises the possibility that a substantial number of eligible individuals were improving without any intervention prior to day 0.

Zuranolone failed to gain approval for major depressive disorder. Two trials failed to achieve the primary end point and the third trial achieved the primary end point statistically, but the magnitude of improvement vs placebo was an underwhelming 1.7-point reduction in HAMD-17 score at day 15.⁶ Sage representatives have stated that they planned to price the 14-day course less than $10 000 only if it was also approved for major depressive disorder. Analysts estimate the price will be $10 000 to $30 000 for the 2-week course.¹ Whether the cost of this medication will impact accessibility remains to be seen.

The FDA saw fit to approve zuranolone based on modest efficacy data in trials that compared it with substandard medical care. This medication has significant limitations, including misuse potential, impaired psychomotor functioning, lack of compatibility with breastfeeding, and anticipated exorbitant cost. We suggest that the FDA require control groups that reflect standards of care to better inform and equip physicians with medications that might improve the health of patients, and we caution the current use of zuranolone.