Brexanolone (SAGE-547 injection) in post-partum depression: a randomised controlled trial

Articles

Brexanolone (SAGE-547 injection) in post-partum depression: a randomised controlled trial Stephen Kanes, Helen Colquhoun, Handan Gunduz-Bruce, Shane Raines, Ryan Arnold, Amy Schacterle, James Doherty, C Neill Epperson, Kristina M Deligiannidis, Robert Riesenberg, Ethan Hoffmann, David Rubinow, Jeffrey Jonas, Steven Paul, Samantha Meltzer-Brody

Summary

Background Post-partum depression is a serious mood disorder in women that might be triggered by peripartum fluctuations in reproductive hormones. This phase 2 study investigated brexanolone (USAN; formerly SAGE-547 injection), an intravenous formulation of allopregnanolone, a positive allosteric modulator of γ-aminobutyric acid (GABAA) receptors, for the treatment of post-partum depression. Methods For this double-blind, randomised, placebo-controlled trial, we enrolled self-referred or physician-referred female inpatients (≤6 months post partum) with severe post-partum depression (Hamilton Rating Scale for Depression [HAM-D] total score ≥26) in four hospitals in the USA. Eligible women were randomly assigned (1:1), via a computer-generated randomisation program, to receive either a single, continuous intravenous dose of brexanolone or placebo for 60 h. Patients and investigators were masked to treatment assignments. The primary efficacy endpoint was the change from baseline in the 17-item HAM-D total score at 60 h, assessed in all randomised patients who started infusion of study drug or placebo and who had a completed baseline HAM-D assessment and at least one post-baseline HAM-D assessment. Patients were followed up until day 30. This trial is registered with ClinicalTrials.gov, number NCT02614547. Findings This trial was done between Dec 15, 2015 (first enrolment), and May 19, 2016 (final visit of the last enrolled patient). 21 women were randomly assigned to the brexanolone (n=10) and placebo (n=11) groups. At 60 h, mean reduction in HAM-D total score from baseline was 21·0 points (SE 2·9) in the brexanolone group compared with 8·8 points (SE 2·8) in the placebo group (difference –12·2, 95% CI –20·77 to –3·67; p=0·0075; effect size 1·2). No deaths, serious adverse events, or discontinuations because of adverse events were reported in either group. Four of ten patients in the brexanolone group had adverse events compared with eight of 11 in the placebo group. The most frequently reported adverse events in the brexanolone group were dizziness (two patients in the brexanolone group vs three patients in the placebo group) and somnolence (two vs none). Moderate treatment-emergent adverse events were reported in two patients in the brexanolone group (sinus tachycardia, n=1; somnolence, n=1) and in two patients in the placebo group (infusion site pain, n=1; tension headache, n=1); one patient in the placebo group had a severe treatment-emergent adverse event (insomnia). Interpretation In women with severe post-partum depression, infusion of brexanolone resulted in a significant and clinically meaningful reduction in HAM-D total score, compared with placebo. Our results support the rationale for targeting synaptic and extrasynaptic GABAA receptors in the development of therapies for patients with post-partum depression. A pivotal clinical programme for the investigation of brexanolone in patients with post-partum depression is in progress. Funding Sage Therapeutics, Inc.

Introduction Post-partum depression is a serious mood disorder consistently observed in an estimated 10–20% of all mothers who give birth in high-income and low-income countries worldwide.1–3 Following delivery, post-partum depression is characterised by clinically significant depressive symptoms, often co-occurring with anxiety.1–6 Severe post-partum depression is defined as a major depressive episode in the post-partum period with marked impairment in functioning in both the Inter­ national Classification of Diseases (ICD)-10 and Diagnostic and Statistical Manual of Mental Disorders (DSM)-5.7,8 Estimates of the point prevalence of severe post-partum depression generally range from 5% to 10%

Published Online June 12, 2017 http://dx.doi.org/10.1016/ S0140-6736(17)31264-3 See Online/Comment http://dx.doi.org/10.1016/ S0140-6736(17)31546-5 Sage Therapeutics Inc, Cambridge, MA, USA (S Kanes MD, H Colquhoun MD, H Gunduz-Bruce MD, R Arnold DO, A Schacterle PhD, J Doherty PhD, E Hoffmann, J Jonas MD, S Paul MD); 2b Analytics, Wallingford, PA, USA (S Raines); Department of Psychiatry, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA (Prof C N Epperson MD); University of Massachusetts Medical School, Worcester, MA, USA (K M Deligiannidis MD); Women’s Behavioral Health, Zucker Hillside Hospital, New York, NY, USA (K M Deligiannidis); Atlanta Center for Medical Research, Atlanta, GA, USA (R Riesenberg MD); and Department of Psychiatry, UNC School of Medicine, Chapel Hill, NC, USA (Prof D Rubinow MD, S Meltzer-Brody MD) Correspondence to: Dr Samantha Meltzer-Brody, Department of Psychiatry, UNC School of Medicine, Chapel Hill, NC 27514, USA samantha_meltzer-brody@ med.unc.edu

of all cases of post-partum depression, depending on the setting.1–3 Furthermore, post-partum depression is a leading cause of maternal mortality9,10 and, by affecting maternal functioning, poses serious risks to the emotional, cognitive, behavioural, and physical development of the infant and siblings.11–13 Findings from several studies implicate peripartum fluctuations in reproductive hormones (in particular, the major progesterone metabolite allopregnanolone) having pivotal pathophysio­ logical roles in post-partum depression.14–17 Allopregnanolone, a potent positive allosteric modulator of synaptic and extrasynaptic GABAA receptors,17,18 has been shown to have profound effects on

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

1

Articles

Research in context Evidence before this study Preclinical and clinical studies have shown that neuroactive steroids might have an important role in the pathophysiology of post-partum depression. During pregnancy in mice, expression of the γ-aminobutyric acid type A (GABAA) receptor δ subunit is downregulated as allopregnanolone concentrations increase, and at parturition, expression of the GABAA receptor δ subunit is recovered as allopregnanolone concentrations drop steeply. GABAA receptor δ-subunit-deficient mice do not adapt to the substantial changes in allopregnanolone concentration during pregnancy and parturition, and showed depression-like and anxiety-like behaviours and abnormal maternal behaviours that were reversed by administration of allopregnanolone. These findings lend support to the hypothesis that changes in neuroactive steroid concentrations during pregnancy and post partum are capable of provoking affective dysregulation. Neuroactive steroids such as allopregnanolone might function as behavioural switches, suggesting a potentially important role in treatment of reproductive and endocrine-related mood disorders such as post-partum depression. We searched PubMed (all indexed dates up to Feb 1, 2017) for clinical trials with the terms “allopregnanolone”, “neuroactive steroid”, “GABAA positive allosteric modulator”, and “postpartum depression”. This search retrieved no trials examining the neuroactive steroid or GABAA-receptor mechanism in post-partum depression. A previous report by several of this study’s investigators describes an open-label, exploratory study of brexanolone in four women with severe post-partum depression. Added value of this study To our knowledge, this study is the first randomised, double-blind, placebo-controlled trial of a therapeutic

anxiety and depression in animal models.17,19–21 Plasma allopreg­ nanolone concentrations rise in concert with progesterone throughout pregnancy, reaching the highest physiological concentrations in the third trimester.22 After childbirth, these concentrations decrease abruptly.23 Failure of GABAA receptors to adapt to these changes at parturition has been postulated to have a role in triggering post-partum depression.24,25 Alterations in concentrations or ratios of serum allopregnanolone and other neuroactive steroids have been reported in some,23,26 but not all,27 women at risk for or who develop post-partum depression. Moreover, symptoms of post-partum depression are precipitated in at-risk women by recreating hormonal fluctuations associated with pregnancy and delivery.14 For some women, onset of mood symptoms occurs in the third trimester of pregnancy, and these symptoms can substantially worsen in the immediate post-partum period.6,28,29 Whether this group of women are similar or different from women who have onset after childbirth is not known; however, women with onset in the 2

formulation of the neuroactive steroid allopregnanolone in patients with post-partum depression. The current standard of care for post-partum depression includes psychotherapy and pharmacological therapies. However, no pharmacological therapies are specifically indicated for the treatment of post-partum depression. Antidepressant medications used to treat depressive disorders outside of the perinatal period, such as selective serotonin reuptake inhibitors and tricyclic antidepressants, are commonly used in post-partum depression. However, these therapies are not directly linked with existing hypotheses regarding the causes of post-partum depression, their onset of efficacy can be delayed by several weeks or months, and their overall remission rate in post-partum depression is low. In particular, rapid onset of action is desirable in severe post-partum depression to quickly mitigate the serious, negative effects of the disorder on the mother, infant, and family. Our study suggests the potential for the development of a GABAA-positive allosteric modulator, such as the neuroactive steroid brexanolone, as a new mechanism for treatment of post-partum depression that is related to the underlying pathophysiology. Implications of all the available data Together with preclinical and clinical studies suggesting a role for neuroactive steroids and GABAA receptor regulation in the pathophysiology of post-partum depression, our findings support the rationale for further examining brexanolone in patients with post-partum depression. Several pivotal clinical trials are currently examining the efficacy and safety of brexanolone in post-partum depression.

third trimester are an important group to consider in terms of differential sensitivity to alterations in changing concentrations of neurosteroids during the peripartum period. Although the cause of post-partum depression is not entirely understood, this collective body of work supports the rationale for exploring the potential treatment of women with post-partum depression with doses of allopregnanolone that result in serum concentrations equivalent to those present during the third trimester, including women with onset of symptoms in the third trimester of pregnancy and throughout the early post-partum period. Allopregnanolone has low aqueous solubility, poor oral bioavailability, and is rapidly metabolised; however, a soluble, proprietary, β-cyclodextrin-based formulation of allopregnanolone—brexanolone (USAN; formerly SAGE-547 injection)—can be administered intravenously to produce stable physiological serum concentrations. Here, we report the results of a randomised, placebocontrolled study of brexanolone in women with severe post-partum depression.

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

Articles

Methods

Study design and participants This multicentre, randomised, double-blind, parallelgroup, placebo-controlled trial received institutional review board approval for 11 sites in the USA and recruited patients in four sites (University of Pennsylvania, University of Massachusetts, Atlanta Center for Medical Research, and the University of North Carolina, Chapel Hill; site principal investigators CNE, KMD, RR, and SM-B). Participants were enrolled by the trial site investigator and gave written informed consent. The trial was completed upon reaching the enrolment target. Additional details of the study design are provided in the appendix. Participants were accrued through both self-referral and physician referrals to clinical research and specialised psychiatric units. A summary of diagnoses of postpartum depression and time of enrolment is available in the appendix. Participants met the following complete inclusion criteria for eligibility: signed informed consent form before any study-specific procedures were performed; ambulatory female aged between 18 and 45 years of age; good physical health and no clinically significant findings as determined by the investigator on physical examination, 12-lead electro­cardiograph (ECG), or clinical laboratory tests; agreed to adhere to the study requirements; had a negative pregnancy test at screening and day 1 before the start of study drug infusion; had a major depressive episode that began no earlier than the third trimester and no later than the first 4 weeks following delivery as diagnosed by the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I);30 had a 17-item Hamilton Rating Scale for Depression (HAM-D) total score of 26 or higher at screening and day 1 (before randomisation); was within 6 months post partum at the time of enrolment; and had no detectable hepatitis B surface antigen, anti-hepatitis C virus, or HIV antibody at screening. Patients could be on stable antidepressants, but they had to be willing to delay the start of other antidepressant or anxiety medication and any new pharmacotherapy regimens, including as-needed benzodiazepine anxiolytics, until the study drug infusion and 72 h assessments were completed. Additionally, the initial inclusion and exclusion requirements included the provision that patients either must have ceased lactating at screening or if still lactating at screening, must have already fully and permanently weaned their infant(s) from breastmilk, or if still actively breastfeeding at screening, must have agreed to cease giving breastmilk to their infant(s) before receiving study drug. A protocol amendment allowed for resumption of breastfeeding after day 12 of the study. The HAM-D cutoff was chosen on the basis of previous studies of severe cases of postpartum depression (defined as perinatal major depressive disorders by both ICD-10 and DSM-5),2 studies of severe major depressive disorder defined by HAM-D score,31 and the principal investigators’ previous experiences in

an open-label trial of brexanolone in severe post-partum depression.32 Exclusion criteria included active psychosis; attempted suicide associated with an index case of post-partum depression; history of seizures, bipolar disorder, schizo­ phrenia, or schizoaffective disorder; and history of alcoholism or drug addiction (including benzodiazepines) in the 12 months before screening. Additional details about the inclusion and exclusion criteria are provided in the appendix.

Randomisation and masking Eligible patients were randomly assigned (1:1), according to a simple computer-generated randomisation schedule, with no blocking, to brexanolone or placebo. SAS was used to generate random numbers for randomisation. The randomisation schedule was produced by an independent statistician at Applied Statistics and Consulting (Spruce Pine, NC, USA), a contract research organisation entirely independent from any trial investigator. Patients, clinicians, and study teams were masked to treatment allocation. Patients in the placebo group received equivalent infusion rates, and the infusion bags for both treatments were identical in appearance. There were no differences in the characteristics of any of the infusion bags or the solutions within, including smell or colour. The pharmacist at each site who prepared the infusion bags according to the randomisation schedule, and a monitor who performed drug accountability during the study, were not masked to treatment assignment. No other study personnel were unmasked until after formal locking of the study database. The study database was locked when the final visit of the last enrolled patient was completed, data entry into the clinical database was completed, and the database for all patients was deemed clean with all queries resolved (June 22, 2016). Allocation concealment was facilitated by a secure, web-based method. Only the clinic pharmacist was given a copy of the randomisation schedule. In the event of a medical emergency, the pharmacist was to reveal actual infusion contents to the primary investigator, who was to alert the sponsor of the emergency. In all cases, if the study drug allocation for a patient had been unmasked, pertinent information (including the reason for unmasking) was to be documented in the patient’s records and on the electronic case report form. If the patient or study centre personnel were unmasked, the patient was to be terminated from the study. No such unmasking occurred during the study.

See Online for appendix

Procedures Brexanolone is a sterile solution of allopregnanolone 5 mg/mL in 250 mg/mL sulfobutylether-β-cyclodextrin (SBECD) buffered with citrate, which is diluted with sterile water for injection to render it isotonic for intravenous infusion. Each patient received a single continuous intravenous infusion of masked study drug

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

3

Articles

23 women screened

2 excluded for ineligibility*

21 enrolled and randomly assigned

10 assigned to brexanolone

11 assigned to placebo

10 received brexanolone and were followed up

11 received placebo and were followed up

10 included in safety population 10 included in efficacy population

11 included in safety population 11 included in efficacy population

for 60 h during inpatient care under the following schedule: 30 µg/kg per h (0–4 h); 60 µg/kg per h (4–24 h); 90 µg/kg per h (24–52 h); 60 µg/kg per h (52–56 h); 30 µg/kg per h (56–60 h). Infusion was initiated between 0800 h and 1000 h, apart from one patient whose infusion was initiated in the afternoon (full details provided in the appendix). Infusion rate adjustments were allowed on the basis of tolerability, side-effects, and predetermined protocol rules. The dose and infusion rate were based on the protocol of the previous open-label exploratory trial in patients with severe post-partum depression32 and on pharmacokinetic modelling. Participants remained as inpatients during the 60 h study infusion period. After dosing was complete, patients were followed up until day 30, with clinical and safety assessments done at days 7 and 30.

Outcomes Figure 1: Trial profile *One patient demonstrated continuing illicit drug use (positive screen for amphetamines, barbiturates, and opiates) and the other did not meet the symptom severity criteria (Hamilton Rating Scale for Depression score <26).

Brexanolone (n=10)

Placebo (n=11)

Characteristics Age (years)

27·4 (5·3); 27 (20–40)

28·8 (4·6); 28 (22–36)

Ethnicity Hispanic or Latino Not Hispanic or Latino

0

0

10 (100%)

11 (100%)

Black or African-American

7 (70%)

6 (55%)

White

3 (30%)

Race

Height (cm)

162·4 (7·1); 163·5 (153–175)

5 (45%) 161·7 (6·7); 162·0 (151–174)

Weight (kg)

86·7 (28·8); 76·5 (49·7–130·7)

77·0 (22·3); 73·5 (53·3–122·6)

Body-mass index (kg/m2)

32·7 (9·9); 30·5 (20·4–47·1)

29·3 (7·8); 28·2 (21·0–45·0)

Personal history Psychiatric disorder Depression (non-PPD)

6 (60%)

6 (55%)

Anxiety

2 (20%)

5 (45%)

Other

1 (10%)

2 (18%)

7 (70%)

4 (36%)

Previous PPD episodes Antidepressant medication Baseline HAM-D

3 (30%) 28·1 (27·0–30·0)

3 (27%) 28·8 (26·0–32·0)

Family history Perinatal psychiatric disorders Mother

2 (20%)

2 (18%)

Sister(s)

1 (10%)

1 (9%)

Data are n (%) or mean (SD); median (range). Age was derived from the birth date and screening date. Bodyweight and height were measured at screening. Body-mass index was programmatically calculated in the electronic case report form. Medical histories were coded according to the Medical Dictionary for Regulatory Activities (MedDRA) version 17 or later. PPD=post-partum depression. HAM-D=Hamilton Rating Scale for Depression.

Table 1: Baseline demographics and characteristics

4

The primary outcome measure was the change from baseline in HAM-D total score at the end of the treatment period (60 h). Secondary analyses included changes in HAM-D score from baseline at 2 h up to 30 days. HAM-D measurements were obtained frequently to monitor for rapid onset of improvement of symptoms and were started within a window of tolerance of 25 min before or after the designated timepoint in the first 24 h. Secondary HAM-D endpoints were the proportion of patients achieving remission (HAM-D total score ≤7), the proportion of patients achieving response (≥50% reduction in HAM-D total score), change from baseline in the Bech-6 subscore,33 which assesses the core symptoms of Montgomery-Åsberg Depression Rating Scale (MADRS) total score,34 Clinical Global Impression-Global Improve­ ment,35 major depression, and changes in the HAM-D depressed mood item score. Additional prespecified secondary and exploratory endpoints are detailed in the appendix, including the Generalised Anxiety Disorder Questionnaire,36 Edinburgh Postnatal Depression Scale,37 Patient Health Questionnaire-9,38 and Barkin Index of Maternal Function.39 To facilitate consistent scoring of the HAM-D, investigators received training on use of the scale from MedAvante (Hamilton, NJ, USA), and each patient was assessed by the same investigator throughout the study. The safety and tolerability of brexanolone were assessed by recording and summarising adverse events, clinical laboratory measurements, vital signs, and ECGs (including changes from baseline); concomitant medication was also assessed. Emergent suicidal ideation and behaviours were assessed with the Columbia-Suicide Severity Rating Scale;40 patient-reported sedation or sleepiness was assessed with the Stanford Sleepiness Scale.41

Statistical analysis On the assumption of a two-sided test at an α level of 0·10, a sample size of ten evaluable patients per group provided 80% power to detect an effect size of 1·2 between the

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

Articles

A

Placebo (n=11) Brexanalone (n=10)

30

Mean HAM-D total score

25 20 15 *

10

*

*

*

*

*

*

5 End of infusion 0

B

40 35 Mean MADRS total score

brexanolone and placebo groups with regard to the primary outcome variable. An effect size of 1·2 corresponds to a placebo-adjusted difference of 12 points in the change from baseline in HAM-D total score at 60 h with an assumed standard deviation of 10 points. By including two treatment groups and using a randomisation ratio of 1:1, a total of 20 evaluable patients was required. After 16 participants had completed HAM-D efficacy assessments at 60 h, a sample size re-estimation was done. For this analysis, an independent statistician from the Datatrial (Newcastle upon Tyne, UK) contract research organisation calculated the unmasked mean percentage change from baseline in HAM-D total score at 60 h for the placebo group. On the basis of this information only, the independent statistician communicated one of the following messages back to the sponsor: if placebo response was less than 35%, no adjustment to the sample size is required; or if placebo response was 35% or more, increase the sample size by five subjects per group. No personnel involved in the conduct of the study had access to the interim results generated by the independent statistician. Since the sponsor was kept masked to the data and uninformed of the interim results (ie, response rates) until final database lock, no statistical adjustment was made to the level of significance for hypothesis testing at the end of the study. No increase in sample size was needed in this trial. On the basis of the results of the sample size re-estimation, the sample size could have been increased to a maximum of 32 randomised patients. This adjustment to the sample size would have allowed for an effect size of 1·0 to be detected. The safety population included all randomised patients who started infusion of study drug or placebo. The efficacy population included the subset of the safety population who had a completed baseline HAM-D assessment and at least one post-baseline HAM-D assessment. The change from baseline in HAM-D and MADRS total score was analysed with a mixed effects model for repeated measures. This linear model included centre, treatment, baseline HAM-D total score, assessment timepoint, and timepoint-by-treatment as explanatory variables. Centre was treated as a random effect, while all other explanatory variables were treated as fixed effects. This analysis model was prespecified in the trial protocol and statistical analysis plan. Modelbased point estimates (ie, least squares means, 95% CIs, and p values) were reported for each timepoint. The primary comparison was between brexanolone and placebo at the 60 h timepoint. Other changes from baseline endpoints were analysed with similar methods. The HAM-D response and remission rates at each timepoint were analysed with Fisher’s exact test. Point estimates (ie, odds ratios [ORs]), 95% CIs, and p values are reported. Additional details of the statistical analysis plan and statistical methods are provided in the appendix. Analyses were done with SAS 9.2. This study is registered with ClinicalTrials.gov, number NCT02614547.

30 25 20 15

*

10 5 0

*

*

*

*

*

End of infusion 0

12

24

36

48

60

72

168 720 (day 7) (day 30)

Timepoint (h)

Figure 2: Mean HAM-D and MADRS total scores (A) The Hamilton Rating Scale for Depression (HAM-D) is a 17-item diagnostic questionnaire used to measure the severity of depressive episodes in patients with mood disorders. It consists of individual ratings related to the following symptoms: depressed mood (sadness, hopeless, helpless, worthless), feelings of guilt, suicide, insomnia (early, middle, late), work and activities, retardation (slowness of thought and speech; impaired ability to concentrate; decreased motor activity), agitation, anxiety (psychic and somatic), somatic symptoms (gastrointestinal and general), genital symptoms, hypochondriasis, loss of weight, and insight. Higher HAM-D scores indicate more severe depression. (B) The Montgomery-Åsberg Depression Rating Scale (MADRS) is a ten-item diagnostic questionnaire used to measure the severity of depressive episodes in patients with mood disorders. Higher MADRS scores indicate more severe depression, and each item yields a score of 0–6, producing total score ranges from 0 to 60. p values were calculated by two-sided t test. *Denotes statistical significance versus placebo, defined as p≤0·01.

Role of the funding source The funding source assisted in the study design, data collection, data analysis, data interpretation, and writing of the report. All authors had full access to all the data in the study and the corresponding author had final responsibility for the decision to submit for publication.

Results This trial was done between Dec 15, 2015 (first enrolment), and May 19, 2016 (final visit of the last enrolled patient). 23 patients were screened, of whom 21 were eligible for enrolment (figure 1). All randomised patients completed the 60 h inpatient dosing protocol and completed the 30 day follow-up period. There were little to no missing data in the study. One patient missed an assessment of

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

5

Articles

HAM-D

MADRS

Brexanolone Placebo LSmeans (n=10) (n=11) difference

p value

Brexanolone Placebo Mean (n=10) (n=11) difference

p value

2h

–6·27

–4·12

–2·2 (2·3)

0·3685

··

··

··

··

4h

–8·67

–5·21

–3·5 (2·9)

0·2476

··

··

··

··

8h

–10·57

–5·94

–4·6 (3·1)

0·1549

··

··

··

··

12 h

–12·67

–6·66

–6·0 (3·7)

0·1164

··

··

··

··

24 h

–19·37

–8·12

–11·3 (3·6)

0·0059

36 h

–20·17

–8·21

–12·0 (4·0)

0·0078

–26·36

–8·83

48 h

–21·87

–9·21

–12·7 (4·0)

0·0047

60 h

–20·97

–8·75

–12·2 (4·1)

0·0075

–27·96

–12·10

–15·9 (5·5)

0·0104

72 h

–21·07

–8·39

–12·7 (4·3)

0·0078

–28·66

–12·46

–16·2 (5·5)

0·0090

··

–17·5 (5·4)

··

–29·86

–11·46

··

0·0042 ··

–18·4 (5·3)

0·0026

7 days

–20·97

–8·06

–12·9 (3·9)

0·0038

–27·56

–11·60

–16·0 (5·4)

0·0091

30 days

–20·77

–8·84

–11·9 (4·1)

0·0095

–26·26

–11·19

–15·1 (5·2)

0·0100

Data are LSmeans (SE). The change from baseline in the Hamilton Rating Scale for Depression (HAM-D) mean total score and Montgomery-Åsberg Depression Rating Scale (MADRS) mean total score was analysed using a mixed effects model for repeated measures. MADRS was not assessed at 2, 4, 8, 12, or 36 h. p values were calculated by two-sided t test.

Table 2: Mean HAM-D and MADRS total scores

80

Placebo (n=11) Brexanolone (n=10)

Patients with remission of symptoms (%)

70 60 50 40 30 20 10 0

Baseline

2

4

8

12

24

36

Timepoint (h)

48

60

72

168 720 (day 7) (day 30)

Figure 3: Patients with remission of symptoms Remission was defined as a Hamilton Rating Scale for Depression (HAM-D) total score of 7 or lower. A larger proportion of patients in the brexanolone group than in the placebo group achieved HAM-D remission at each timepoint after 2 h. The difference was significant at 24 h (p=0·0561), 48 h (p=0·0897), 60 h (p=0·0449), and 72 h (p=0·0364), as well as at day 7 (p=0·0449) and day 30 (p=0·0449). p values were calculated by Z test from log transformation of the odds ratio.

HAM-D and MADRS at day 7 and had an additional, unscheduled assessment at day 17. The proportion of patients with a previous history of psychiatric disorders was similar between treatment groups, apart from history of anxiety (table 1). The proportion of patients with at least one previous episode of post-partum depression was higher in the brexanolone group than in the placebo group. Use of antidepressant medication was balanced between the groups. At the end of the 60 h infusion, mean reduction in HAM-D total score was 21·0 points (SE 2·9) in the brexanolone group, compared with 8·8 points (2·8) in the placebo group; mean difference between groups 6

–12·2 points (95% CI –20·77 to –3·67; two-tailed t test, p=0·0075). The effect size for the clinical efficacy at 60 h was 1·2. Prespecified secondary analyses showed a –11·3 point (95% CI –18·86 to –3·65) mean difference between groups at 24 h (two-tailed t test, p=0·0059), with significant improvements seen for the brexanolone group at 36, 48, 60, and 72 h, as well as days 7 and 30 (figure 2). The primary endpoint data (HAM-D) were also analysed post-hoc by a Wilcoxon Signed Rank Test, a non-parametric method, which reached the same conclusions as the MMRM method (see appendix). Change from baseline in MADRS total score showed similar results to those obtained with the HAM-D score (figure 2 and table 2). Remission from depression (HAM-D total score ≤7) was seen in seven of ten patients in the brexanolone group and in one of 11 patients in the placebo group at 60 h (OR –23·33, 95% CI –1·56 to 1152·71; Z test from log transformation of the OR, p=0·0364; figure 3). This difference was seen at 24 h (six patients in the brexanolone group vs one patient in the placebo group; OR 15·00, 95% CI 1·07–756·72; Z test, p=0·0561) and a difference was maintained until the 30-day follow-up (seven vs two; OR 10·50, 95% CI 1·01–140·57; Z test, p=0·0499). More patients demonstrated a 50% or greater reduction in HAM-D total score in the brexanolone group than in the placebo group across all timepoints. Although this finding was not significant at 60 h (70% [seven patients] in the brexanolone group vs 36% [ four patients] in the placebo group; p=0·1450), it did achieve significance at both 72 h (80% [eight patients] brexanolone vs 27% [three patients] placebo; p=0·0374) and day 7 (80% [eight patients] brexanolone vs 20% [two patients] placebo; p=0·0335; appendix). The observed improvement in symptoms of post-partum depression following brexanolone administration also extended beyond core depressive symptoms, as evidenced by the significant treatment difference observed for CGI-I response. Brexanolone was generally well tolerated. There were no deaths, serious adverse events, or discontinuations in either group. Overall, fewer patients who received brexanolone reported adverse events compared with patients who received placebo (four of ten patients in the brexanolone group vs eight of 11 in the placebo group; table 3). The most frequently reported adverse events in the brexanolone group were dizziness (two patients in the brexanolone group vs three patients in the placebo group) and somnolence (two vs none). Sedation was reported in one patient in the brexanolone group and in no patients in the placebo group. Moderate treatment-emergent adverse events in the brexanolone group were sinus tachycardia (one [10%] patient) and somnolence (one [10%] patient). One (9%) patient in the placebo group had a severe treatment-emergent adverse event of insomnia. Moderate treatment-emergent adverse events in the placebo group were infusion site pain (one [9%] patient) and tension headache (one [9%] patient). All other treatment-emergent adverse events in both groups were mild.

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

Articles

At baseline, mean Stanford Sleepiness Scale scores were similar in the brexanolone and placebo groups (2·7 vs 2·6, respectively); these scores were also similar after treatment, suggesting that there were no differences in sleepiness between groups (appendix). One patient in the brexanolone group, who was taking clonazepam (6 mg), reported sleepiness; this patient required a dose reduction of the study drug and, after dose reduction, completed participation in the study. Improvements in Columbia-Suicide Severity Rating Scale suicidal ideation items were noted in both treatment groups (appendix). Two patients in the brexanolone group reported active suicidal ideation with a specific plan and intent at baseline but not at the post-treatment assessment. No individuals experienced worsening of suicidal ideation or behaviour during treatment or follow-up.

Discussion Severe post-partum depression is a serious and disabling mood disorder that often requires admission to hospital and is a major risk factor for maternal mortality from suicide. In addition to psychotherapy, the standard of care for moderate to severe post-partum depression includes pharmacological therapies, used judiciously, particularly in nursing mothers.42 Evidence for the efficacy of antidepressants in treatment of post-partum depression is based on their empirical use in the general population and limited data from several randomised, placebo-controlled trials.42–44 Some data suggest that time to response is longer and need for polypharmacy is greater when patients with post-partum depression are treated with antidepressants, compared with nonperinatal women with major depressive disorder.13,45,46 Moreover, many women treated for post-partum depression do not achieve full remission of symptoms.47 Therefore, there remains a great need for improved pharmacological treatment options. This study is the initial placebo-controlled trial of brexanolone in a clearly defined population of women with severe post-partum depression. Our findings provide the first placebo-controlled clinical support for the role of extrasynaptic GABAA receptors in the modulation of mood and affective states in any clinical population. The large effect size (1·2) seen in this trial contrasts with that observed in studies of currently available and widely used antidepressants, including selective serotonin reuptake inhibitors (SSRIs), serotonin–noradrenaline reuptake inhibitors (SNRIs), and tricyclic antidepressants.48,49 For example, a pooled meta-analysis of placebo-controlled studies of fluoxetine showed an effect size of –0·30 in favour of fluoxetine (95% CI –0·39 to –0·21).50 A treatment modality with rapid onset of action is desirable in severe post-partum depression in view of the extensive adverse impact of the disease on the mother, infant, and family. Moreover, the 30 day maintenance of treatment effect seen in the brexanolone group in our

Brexanolone (n=10)

Placebo (n=11)

Any adverse event

4

8

Dizziness

2

3

Nausea

1

3

Abnormal dreams

0

2

Headache

0

2

Infusion site pain

0

2

Insomnia

0

2

Rash

1

1

Somnolence

2

0

Abdominal pain

0

1

Anxiety

0

1

Dizziness postural

1

0

Dry mouth

1

0 0

Hot flush

1

Infusion site extravasation

0

1

Localised oedema

0

1

Pain in extremity

0

1

Pyrexia

1

0

Sedation

1

0

Sinus tachycardia

1

0

Skin abrasion

0

1

Tension headache

0

1

Vertigo

1

0

Data are number of patients reporting treatment-emergent adverse event, defined as an adverse event with onset after the start of study drug, or any worsening of a pre-existing medical condition or adverse event with onset after the start of study drug and until the follow-up visit on day 7 (ie, approximately 4 days after the end of the infusion). Adverse events were coded according to the Medical Dictionary for Regulatory Activities version 18.0.

Table 3: Treatment-emergent adverse events

study contrasts with the antidepressant effects reported in a study of a single infusion of ketamine in patients with major depressive disorder, which were maintained in most patients in the study for less than a week.50,51 The observed improvement in symptoms of post-partum depression following brexanolone administration also extended beyond core depressive symptoms, as evidenced by the significant treatment difference observed for CGI-I response. Finally, brexanolone was well tolerated, and fewer patients receiving brexanolone had adverse events than did patients in the placebo group, and sedation did not differ between the groups, as assessed by the Stanford Sleepiness Scale. Although this multisite phase 2 study had a modest number of study participants, the study was designed to find as representative a sample as possible. To that end, we recruited study participants from a wide geographic range that included urban, suburban, and rural settings in the USA. Travel and logistic support was provided for patients who did not live near our open research sites. Thus, the study participants reflect a diverse group in terms of race, geography, and health insurance payer mix. The study sites themselves were also diverse and

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

7

Articles

included academic medical centres and clinical research organisations. Therefore, we believe our findings to be representative of women with severe post-partum depression, defined as a HAM-D score of 26 or higher. Several forms of advertising were used to ensure broad representation of eligible study participants, including recruiting directly from study sites and local physicians in the form of print media (brochures, flyers, and physician referral letters) and social media (Facebook, Google, Bing/Yahoo, and Swoop Native advertising). Although all approaches yielded potential participants, the use of social media substantially increased our ability to reach patients more broadly, reflecting the high use of social media in women of childbearing age. As with the complex causes of mood disorders in general, multiple biological and non-biological factors undoubtedly have a role in the pathophysiology of post-partum depression.14 However, we postulated that women with onset of severe depressive symptoms in the third trimester or triggered by childbirth—often associated with rapid fluctuations in gonadal steroids— are more likely to have a hormone-responsive form of post-partum depression that could respond to therapeutic doses of neuroactive steroids, such as allopregnanolone.20 The results of this trial of brexanolone support this hypothesis and are consistent with preclinical literature on the central role of neuroactive steroids and extra­ synaptic GABAA receptors in modulating affective behaviour and anxiety. Notably, this trial addresses the treatment of severe post-partum depression with brexanolone, and additional trials will be required to see if the observed effects are generalisable for varying degrees of severity of post-partum depression and if brexanolone treatment can be applied for depression. The precise pathophysiology underlying the pre­ disposition to development of post-partum depression is unknown; however, preclinical work has implicated alterations in GABAA receptors, especially extrasynaptic GABAA receptors.24 Dynamic changes in GABAA receptor expression in response to changing neuroactive steroid concentrations might be necessary for maintaining homeostasis of neurotransmission during pregnancy and the post-partum period.24 During the course of pregnancy, plasma concentrations of the endogenous neuroactive steroid allopregnanolone, a potent positive modulator of synaptic and extrasynaptic GABAA receptors, increase from less than 5 nM before pregnancy to 157 nM in the third trimester.22 Enhanced GABAergic inhibition triggered by progressively increased allopregnanolone concentrations throughout pregnancy is thought to be offset by a simultaneous alteration in GABAA receptor activity or expression, changes in activation kinetics, or a combination of both, to prevent a substantial shift in the excitatory and inhibitory balance in the brain.52 Importantly, these changes must then be quickly reversed at the time of parturition, because allopregnanolone concentrations begin to drop steeply within hours, and 8

return back to pre-pregnancy concentrations within 5 days post partum.53 Although GABAA receptor expression and tonic inhibitory currents normally return to pre-pregnancy concentrations following parturition, the inability to adapt to rapid changes in neuroactive steroid concentrations during pregnancy and parturition might contribute to the pathophysiology of post-partum depression in some women, resulting in affective symptoms upon the rapid decline of allopregnanolone concentrations post partum or possibly beginning in the third trimester of pregnancy.14 This trial also demonstrates that a study of women with post-partum depression is feasible and that complex trial designs are not necessarily required to overcome presumed placebo responses, especially with the large effect size we observed with brexanolone. Furthermore, trials in such a clearly defined and previously understudied population are crucial to develop novel treatments for post-partum depression. This trial is limited by a small sample size, although because of the large effect size and rapid response, the study was adequately powered. Notably, because of the observed response, the data were not distributed normally, but in addition to the prespecified mixed effects model for repeated measures used for analysis, a non-parametric analysis (Wilcoxon signed rank test) also showed statistical significance. This trial is also limited by the use of a very strict definition of severe post-partum depression (HAM-D ≥26), which was defined on the basis of existing studies of severe major depressive disorder assessed with HAM-D31 and our previous experience in an open-label trial of brexanolone in severe post-partum depression.32 This restriction raises two questions that will need to be addressed in future trials: first, what overall percentage of women with severe post-partum depression would ultimately respond to brexanolone; and second, how to generalise our results to the broader group of all patients with post-partum depression. Up to 20% of women giving birth will have post-partum depression (including both minor and major depression), and approximately 5–10% of these women will have severe post-partum depression (as defined by a major depressive episode).1–3 Although this population of women with severe post-partum depression is undoubtedly an important population to study, it represents only a fraction of women with post-partum depression. A phase 3 programme is currently investigating the potential use of brexanolone in post-partum depression of varying degrees of severity. Additional limitations of our trial are the 30 day follow-up period, possible respondent fatigue in HAM-D assessment from frequent administration, and the inability (because of the small study size) to stratify patients on the basis of previous history of post-partum depression. These findings replicate our recent open-label exploratory study in four women with severe post-partum depression32 and demonstrate a substantial treatment effect of brexanolone in the trial population of patients

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

Articles

with severe post-partum depression, an indication for which there are no currently approved pharmacological therapies. The rapid and marked antidepressant response associated with brexanolone administration contrasts with the 4–6 weeks needed (and low remission rates) that have been reported with SSRIs and other antidepressants in patients with post-partum depression.44 A rapid onset of antidepressant action is crucial because speed of onset is a strong determinant of the likelihood of near-term recovery, and full remission is the objective of any treatment.54 Moreover, rapid onset of action could decrease potential risks to the mother from self-harm and risks to the baby (and siblings) from lack of attention or negative interaction with the mother. The results of our study support further development of brexanolone for the treatment of post-partum depression.

9

Contributors All authors contributed to the design of the study and writing of the manuscript. SK, HC, and SM-B designed the study. SK and SM-B wrote the first draft. CNE, KMD, RR, and DR were investigators in the study. SR provided statistical analysis. All authors vouch for the accuracy and completeness of the data, data analyses, and the fidelity of this report to the study protocol.

15

Declaration of Interests SK, HC, HG-B, AS, JD, EH, and JJ and are employees of Sage Therapeutics, Inc, with stock or stock options, or both. SK has a patent pending (SAGE-547 for neuropsychiatric conditions). SR, CNE, and DR and are consultants for Sage Therapeutics, Inc. RA is an employee of Sage Therapeutics, Inc. CNE, KMD, and SM-B report that their institutions are receiving grants for the conduct of the clinical trial from Sage Therapeutics, Inc. CNE is a consultant for Asarina Pharma. DR will receive stock options for being on the clinical advisory board of Sage Therapeutics. Inc. SP is a founder, board member, and shareholder of Sage Therapeutics, Inc. RR declares no competing interests. Acknowledgments This study was supported by Sage Therapeutics, Inc. We thank Jeffrey R Skaar and Anna K Talaga at Boston Strategic Partners (supported by Sage Therapeutics, Inc) and Paul Miller at Sage Therapeutics, Inc, for editorial support. References 1 Gavin NI, Gaynes BN, Lohr KN, Meltzer-Brody S, Gartlehner G, Swinson T. Perinatal depression: a systematic review of prevalence and incidence. Obstet Gynecol 2005; 106: 1071–83. 2 Howard LM, Molyneaux E, Dennis CL, Rochat T, Stein A, Milgrom J. Non-psychotic mental disorders in the perinatal period. Lancet 2014; 384: 1775–88. 3 Fellmeth G, Fazel M, Plugge E. Migration and perinatal mental health in women from low- and middle-income countries: a systematic review and meta-analysis. BJOG 2017; 124: 742–52. 4 Meltzer-Brody S, Maegbaek ML, Medland SE, Miller WC, Sullivan P, Munk-Olsen T. Obstetrical, pregnancy and socio-economic predictors for new-onset severe postpartum psychiatric disorders in primiparous women. Psychol Med 2017: 47: 1427–41. 5 Kim JJ, La Porte LM, Saleh MP, et al. Suicide risk among perinatal women who report thoughts of self-harm on depression screens. Obstet Gynecol 2015; 125: 885–93. 6 Postpartum Depression: Action Towards Causes and Treatment (PACT) Consortium. Heterogeneity of postpartum depression: a latent class analysis. Lancet Psychiatry 2015; 2: 59–67. 7 WHO. International statistical classification of diseases and related health problems, 10th revision (ICD-10). Geneva: World Health Organization, 1992. 8 American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 5th edn (DSM-5). Arlington, VA: American Psychiatric Publishing, 2013.

10

11

12 13

14

16 17 18 19

20 21 22 23 24 25

26

27

28 29 30 31

Johannsen BM, Larsen JT, Laursen TM, Bergink V, Meltzer-Brody S, Munk-Olsen T. All-cause mortality in women with severe postpartum psychiatric disorders. Am J Psychiatry 2016; 173: 635–42. Khalifeh H, Hunt IM, Appleby L, Howard LM. Suicide in perinatal and non-perinatal women in contact with psychiatric services: 15 year findings from a UK national inquiry. Lancet Psychiatry 2016; 3: 233–42. Noorlander Y, Bergink V, van Den Berg MP. Perceived and observed mother-child interaction at time of hospitalization and release in postpartum depression and psychosis. Arch Womens Ment Health 2008; 11: 49–56. Field T. Postpartum depression effects on early interactions, parenting, and safety practices: a review. Infant Behav Dev 2010; 33: 1–6. Epperson CN, Jatlow PI, Czarkowski K, Anderson GM. Maternal fluoxetine treatment in the postpartum period: effects on platelet serotonin and plasma drug levels in breastfeeding mother-infant pairs. Pediatrics 2003; 112: e425. Bloch M, Schmidt PJ, Danaceau M, Murphy J, Nieman L, Rubinow DR. Effects of gonadal steroids in women with a history of postpartum depression. Am J Psychiatry 2000; 157: 924–30. Epperson CN, Gueorguieva R, Czarkowski KA, et al. Preliminary evidence of reduced occipital GABA concentrations in puerperal women: a 1H-MRS study. Psychopharmacology (Berl) 2006; 186: 425–33. Schiller CE, Meltzer-Brody S, Rubinow DR. The role of reproductive hormones in postpartum depression. CNS Spectr 2015; 20: 48–59. Paul SM, Purdy RH. Neuroactive steroids. FASEB J 1992; 6: 2311–22. Majewska MD, Harrison NL, Schwartz RD, Barker JL, Paul SM. Steroid hormone metabolites are barbiturate-like modulators of the GABA receptor. Science 1986; 232: 1004–07. Deligiannidis KM, Kroll-Desrosiers AR, Mo S, et al. Peripartum neuroactive steroid and γ-aminobutyric acid profiles in women at-risk for postpartum depression. Psychoneuroendocrinology 2016; 70: 98–107. Zorumski CF, Paul SM, Izumi Y, Covey DF, Mennerick S. Neurosteroids, stress and depression: potential therapeutic opportunities. Neurosci Biobehav Rev 2013; 37: 109–22. Belelli D, Harrison NL, Maguire J, Macdonald RL, Walker MC, Cope DW. Extrasynaptic GABAA receptors: form, pharmacology, and function. J Neurosci 2009; 29: 12757–63. Luisi S, Petraglia F, Benedetto C, et al. Serum allopregnanolone levels in pregnant women: changes during pregnancy, at delivery, and in hypertensive patients. J Clin Endocrinol Metab 2000; 85: 2429–33. Nappi RE, Petraglia F, Luisi S, Polatti F, Farina C, Genazzani AR. Serum allopregnanolone in women with postpartum “blues”. Obstet Gynecol 2001; 97: 77–80. Maguire J, Mody I. GABA(A)R plasticity during pregnancy: relevance to postpartum depression. Neuron 2008; 59: 207–13. Paoletti AM, Romagnino S, Contu R, et al. Observational study on the stability of the psychological status during normal pregnancy and increased blood levels of neuroactive steroids with GABA-A receptor agonist activity. Psychoneuroendocrinology 2006; 31: 485–92. Deligiannidis KM, Sikoglu EM, Shaffer SA, et al. GABAergic neuroactive steroids and resting-state functional connectivity in postpartum depression: a preliminary study. J Psychiatr Res 2013; 47: 816–28. Harris B, Lovett L, Smith J, Read G, Walker R, Newcombe R. Cardiff puerperal mood and hormone study. III. Postnatal depression at 5 to 6 weeks postpartum, and its hormonal correlates across the peripartum period. Br J Psychiatry 1996; 168: 739–44. Milgrom J, Gemmil AW, Bilszta JL, et al. Antenatal risk factors for postnatal depression: a large prospective study. J Affect Disord 2008; 108: 147–57. Serati M, Redaelli M, Buoli M, Altamura AC. Perinatal major depression biomarkers: a systematic review. J Affect Disord 2016; 193: 391–404. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured clinical interview for DSM-IV axis I disorders, clinician version (SCID-CV). Washington, DC: American Psychiatric Press, 1996. Zimmerman M, Martinez JH, Young D, Chelminski I, Dalrymple K. Severity classification on the Hamilton Depression Rating Scale. J Affect Disord 2013; 150: 384–88.

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3

9

Articles

32 Kanes SJ, Colquhoun H, Doherty J, et al. Open-label, proof-ofconcept study of brexanolone in the treatment of severe postpartum depression. Hum Psychopharmacol Clin Exp 2017; 32: e2576. 33 Bech P, Allerup P, Gram LF, et al. The Hamilton depression scale. Evaluation of objectivity using logistic models. Acta Psychiatr Scand 1981; 63: 290–99. 34 Montgomery SA, Asberg M. A new depression scale designed to be sensitive to change. Br J Psychiatry 1979; 134: 382–89. 35 Guy W. ECDEU assessment manual for psychopharmacology: Rockville, MD: US Department of Health, Education, and Welfare, Public Health Service, Alcohol, Drug Abuse, and Mental Health Administration, National Institute of Mental Health, Psychopharmacology Research Branch, Division of Extramural Research Programs, 1976. 36 Spitzer RL, Kroenke K, Williams JB, Lowe B. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med 2006; 166: 1092–97. 37 Cox JL, Holden JM, Sagovsky R. Detection of postnatal depression. Development of the 10-item Edinburgh Postnatal Depression Scale. Br J Psychiatry 1987; 150: 782–86. 38 Spitzer RL, Kroenke K, Williams JB. Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study. Primary Care Evaluation of Mental Disorders. Patient Health Questionnaire. JAMA 1999; 282: 1737–44. 39 Barkin JL, Wisner KL, Wisniewski SR. The psychometric properties of the Barkin Index of Maternal Functioning. J Obstet Gynecol Neonatal Nurs 2014; 43: 792–802. 40 Posner KBG, Stanley B, Brent DA, et al. The Columbia-Suicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults. Am J Psychiatry 2011; 168: 1266–77. 41 Hoddes E, Zarcone V, Smythe H, Phillips R, Dement WC. Quantification of sleepiness: a new approach. Psychophysiology 1973; 10: 431–36. 42 Yonkers KA, Wisner KL, Stewart DE, et al. The management of depression during pregnancy: a report from the American Psychiatric Association and the American College of Obstetricians and Gynecologists. Obstet Gynecol 2009; 114: 703–13. 43 Hantsoo L, Ward-O’Brien D, Czarkowski KA, Gueorguieva R, Price LH, Epperson CN. A randomized, placebo-controlled, double-blind trial of sertraline for postpartum depression. Psychopharmacology (Berl) 2014; 231: 939–48.

10

44 De Crescenzo F, Perelli F, Armando M, Vicari S. Selective serotonin reuptake inhibitors (SSRIs) for post-partum depression (PPD): a systematic review of randomized clinical trials. J Affect Disord 2014; 152–54: 39–44. 45 Hendrick V, Altshuler L, Strouse T, Grosser S. Postpartum and nonpostpartum depression: differences in presentation and response to pharmacologic treatment. Depress Anxiety 2000; 11: 66–72. 46 Epperson CN, Anderson GM, McDougle CJ. Sertraline and breast-feeding. N Engl J Med 1997; 336: 1189–90. 47 Cox EQ, Sowa NA, Meltzer-Brody SE, Gaynes BN. The perinatal depression treatment cascade: baby steps toward improving outcomes. J Clin Psychiatry 2016; 77: 1189–200. 48 Fournier JC, DeRubeis RJ, Hollon SD, et al. Antidepressant drug effects and depression severity: a patient-level meta-analysis. JAMA 2010; 303: 47–53. 49 Bech P, Cialdella P, Haugh MC, et al. Meta-analysis of randomised controlled trials of fluoxetine v. placebo and tricyclic antidepressants in the short-term treatment of major depression. Br J Psychiatry 2000; 176: 421–28. 50 Murrough JW, Iosifescu DV, Chang LC, et al. Antidepressant efficacy of ketamine in treatment-resistant major depression: a two-site randomized controlled trial. Am J Psychiatry 2013; 170: 1134–42. 51 Zarate C, Singh JB, Carlson PJ, et al. A randomized trial of an N-methyl-D-aspartate antagonist in treatment-resistant major depression. Arch Gen Psychiatry 2006; 63: 856–64. 52 Concas A, Mostallino MC, Porcu P, et al. Role of brain allopregnanolone in the plasticity of γ-aminobutyric acid type A receptor in rat brain during pregnancy and after delivery. Proc Natl Acad Sci USA 1998; 95: 13 284–89. 53 Klak J, Hill M, Pařízek A, et al. Pregnanolone isomers, pregnenolone and their polar conjugates around parturition. Physiol Res 2003; 52: 211–21. 54 van Calker D, Zobel I, Dykierek P, et al. Time course of response to antidepressants: predictive value of early improvement and effect of additional psychotherapy. J Affect Disord 2009; 114: 243–53.

www.thelancet.com Published online June 12, 2017 http://dx.doi.org/10.1016/S0140-6736(17)31264-3