Lamotrigine efficacy, safety, and tolerability for women of childbearing age with bipolar I disorder: Meta-analysis from four randomized, placebo-controlled maintenance studies

This meta-analysis investigated the efficacy, safety, and tolerability of lamotrigine versus placebo in preventing relapse and recurrence of mood episodes in women of childbearing age with bipolar I disorder. Following up to 16 weeks' open-label lamotrigine treatment, responders were randomized to double-blind treatment, including lamotrigine 100-400 mg/day or placebo, in four trials of up to 76 weeks. Women aged 18-45 years who received ≥ 1 dose of study treatment and had ≥ 1 efficacy assessment in the double-blind phase were pooled for efficacy analysis. The primary outcome was median time to intervention for any mood episode (TIME). Of 717 eligible women in the open-label phase, 287 responded and were randomized to lamotrigine (n = 153) or placebo (n = 134). The randomized group had a mean (SD) of 2.0(2.02) manic and 2.5(2.02) depressive episodes in the 3 years before screening. Median TIME was 323 days with lamotrigine and 127 days with placebo (HR 0.69; 95% CI 0.49, 0.96; p = 0.030). Lamotrigine delayed time to intervention for any depressive episode (HR 0.59; 95% CI 0.39, 0.90; p = 0.014) with no treatment difference for manic episodes (HR 0.91; 95% CI 0.52, 1.58; p = 0.732). 2/717 (< 1%) participants experienced serious rash-related adverse events (AEs) during the open-label phase, and 52/717 (7%) had non-serious rash-related events leading to study withdrawal. Incidence of AEs and AEs leading to withdrawal were similar between lamotrigine and placebo groups. Lamotrigine delayed relapse and recurrence of mood episodes, largely by preventing depressive episodes, and was well tolerated in women of childbearing age.


Introduction
Pharmacological management of bipolar disorder, centered around long-term maintenance treatment ( Grunze et al., 2013 ), is clinically challenging in women of childbearing age (WOCBA).Appropriate management of WOCBA should ensure a prolonged euthymic phase while avoiding potential complications related to the teratogenic effects of some drugs used in the treatment of bipolar disorder ( Epstein et al., 2015 ;Grunze et al., 2013 ;Thomson and Sharma, 2018 ).Women who discontinue pharmacological treatment during pregnancy are at a much greater risk of experiencing relapse or recurrence of symptoms than those who continue during pregnancy ( Viguera et al., 2007 ;Wesseloo et al., 2016 ).With up to 50% of pregnancies in women with bipolar disorder being unplanned ( Goodwin et al., 2016 ;Marengo et al., 2015 ;Zengin Eroglu and Lus, 2020 ), pharmacological treatment that can be used with fewer concerns during pregnancy may be advantageous for WOCBA with bipolar disorder.
Most available maintenance treatments for WOCBA with bipolar I disorder have notable adverse effects.Antidepressant treatment for bipolar depression is complicated by the potential risk of mood switching, particularly with older generation antidepressants and in bipolar I disorder ( Baldessarini et al., 2020 ).Because of this risk and the paucity of supporting evidence, long-term use of antidepressants as maintenance therapy is controversial ( Grunze et al., 2013 ;Pacchiarotti et al., 2013 ).Carbamazepine and valproate carry potential teratogenic risks that restrict their use during pregnancy ( Goodwin et al., 2016 ;Grunze et al., 2013 ).Valproate has particularly high teratogenicity, with 10% of children born to mothers receiving valproate having a major congenital malformation and 30-40% having a neurodevelopmental disorder ( Bromley et al., 2010 ;Cummings et al., 2011 ).Because of these teratogenic risks, valproate is contraindicated during pregnancy and is unsuitable for use by patients with bipolar disorder for whom pregnancy is a possibility unless an appropriate contraceptive plan is in place ( European Medicines Agency, 2018 ;Goodwin et al., 2016 ;Grunze et al., 2013 ;Royal College of Psychiatrists, 2018 ).Despite other treatments, including lamotrigine, having more favorable safety profiles for use during pregnancy ( Hernandez-Diaz et al., 2012 ;Tomson et al., 2018 ), national surveys suggest that in some parts of the world, valproate continues to be prescribed more commonly than might be expected for this population ( Kan et al., 2022 ;Ng et al., 2021 ;Paton et al., 2018 ;Tachibana et al., 2020 ).Antipsychotics, lithium, and valproate are associated with metabolic dysfunction and weight gain ( Chang et al., 2010 ;Grootens et al., 2018 ;Schneider et al., 2020 ;Vancampfort et al., 2013 ), reducing their long-term tolerability in these patients.
Lamotrigine is indicated as a maintenance treatment to delay the time to occurrence of mood episodes, particularly depressive episodes, in adults with bipolar I disorder ( Food and Drug Administration, 2015 ).The safety and efficacy of lamotrigine as a long-term maintenance therapy for bipolar I disorder were established in Phase 3 clinical trials and corroborated by systematic review ( Besag et al., 2021 ;Bowden et al., 2003 ;Calabrese et al., 2003 ).Among people with bipolar disorder, women are more likely than men to have a predominant depressive polarity ( Altshuler et al., 2010 ;Nivoli et al., 2011 ), suggesting that prevention of depressive episodes with lamotrigine represents a relevant benefit for WOCBA with bipolar I disorder.Further potential advantages of lamotrigine over other treatments for bipolar disorder include absence of the weight gain associated with antipsychotics and valproate ( Grootens et al., 2018 ;Schneider et al., 2020 ), and a lower risk of ovulatory dysfunction, hyperandrogenism, and polycystic ovarian syndrome among women treated with lamotrigine compared with valproate ( Morrell et al., 2008 ;Sidhu et al., 2018 ).
Although robust evidence supports the benefits of lamotrigine therapy for adults with bipolar I disorder, data for WOCBA specifically are limited.Because of its efficacy for prevention of depressive episodes, and absence of contraindications in pregnancy ( Hernandez-Diaz et al., 2012 ;Tomson et al., 2018 ), lamotrigine may represent an attractive maintenance treatment option for this population.This meta-analysis of four international, randomized, placebocontrolled trials, including individual patient data for more than 700 women, is the first study to our knowledge, investigating the efficacy, safety, and tolerability of lamotrigine maintenance therapy in WOCBA with bipolar I disorder.

Study design
Four Phase 3, multicenter, randomized, double-blind, placebocontrolled trials investigating the efficacy and safety of lamotrigine as a maintenance therapy for bipolar I disorder were pooled for meta-analysis.These were two international trials ( Bowden et al., 2003 ;Calabrese et al., 2003 ), a trial in China ( Zhang et al., 2022 ), and a trial in Japan ( Koyama et al., 2011 ) ( Table 1 ).Patients enrolled in these trials received lamotrigine monotherapy (titrated to 200 mg/day) or adjunct therapy in an open-label (OL) phase of 8- 16 weeks in the international trials and the trial in Japan, or 6-16 weeks in the trial in China.Participants who responded to lamotrigine therapy (Clinical Global Impressions -Severity scale [CGI-S] score ≤3, maintained for ≥ 4 weeks) were randomized to receive double-blind (DB) maintenance treatment.In the international pivotal trials, participants were randomized to lamotrigine (Calabrese et al.: 50,200, or 400 mg/day; Bowden et al.: 100-400 mg/day, target dosage 200 mg/day), lithium, or placebo for 76 weeks.In the trial conducted in China, participants were randomized to lamotrigine (200 mg/day) or placebo for 36 weeks, with the same treatment arms assessed over 26 weeks in the trial conducted in Japan.
Concomitant psychotropic medications were permitted during the OL phase but discontinued with appropriate washout periods before randomization (1 week for oral antipsychotics, selective serotonin reuptake inhibitors, tricyclic antidepressants, lithium [3 weeks in the trial in China], and benzodiazepines; 2 weeks for anticonvulsants, monoamine oxidase inhibitors, and reversible monoamine oxidase inhibitors).During the OL and DB phases, the short-term, intermittent use of low-dose psychotropic medication was permitted as required for control of agitation, irritability, insomnia, and hostile behavior.Drugs permitted for this use included chloral hydrate, lorazepam, temazepam, oxazepam, or midazolam (Calabrese et al. only) in the international trials; chloral hydrate, lorazepam, oxazepam, clonazepam, or estazolam in the trial in China; and triazolam, zopiclone, or brotizolam in the trial in Japan.

Patient inclusion
All studies enrolled patients with bipolar I disorder.Calabrese et al. enrolled patients with a current/recent depressive episode while Bowden et al. enrolled patients with a current/recent manic or hypomanic episode; Koyama et al. enrolled Japanese patients with a current depressive or manic episode and Zhang et al. enrolled Chinese patients with a current/recent depressive, manic, hypomanic or mixed episode ( Table 1 ).Participants were 18 years of age or older, except for Japanese participants who were 20 years of age or older.WOCBA were required to follow an appropriate contraceptive plan to avoid pregnancy.Medications containing estrogen could not be used because they increase lamotrigine clearance ( Food and Drug Administration, 2015 ).
Across the four trials, patients were excluded if they experienced rapid cycling or the number of manic, hypomanic, mixed or depressive episodes exceeded four (trial in China), six (international trials) or seven (trial in Japan) in the year before enrollment.Other exclusions were a Diagnostic and Statistical Manual of Mental Disorders, 4th Edition diagnosis of, or treatment within the year before enrollment for, panic disorder, obsessive compulsive disorder, social phobia, or bulimia nervosa; history of or current epilepsy; clinically significant cardiac, renal, hepatic, neoplastic or cerebrovascular disease; actively suicidal or a score of ≥ 3 on item 3 (suicidality) of the Hamilton Rating Scale for Depression.
In all trials, participants who experienced rash of any severity in either the OL or DB phase had treatment discontinued unless the rash was clearly not treatment related.

Outcomes
The primary efficacy outcome measure for the international trials and the trial in China was the time from entry into the DB phase to the first intervention for relapse or recurrence of a mood episode (TIME).For the trial in Japan, time to withdrawal from study (TWS), defined as time from randomization to the point at which the participant was withdrawn from the DB phase for any reason, was the primary efficacy outcome measure and TIME was a secondary outcome.All trials included time to intervention for relapse and/or recurrence of a depressive mood episode and time to intervention for relapse and/or recurrence of a manic/hypomanic/mixed mood episode as secondary efficacy outcomes.
During the DB phase, clinical assessments were made at weekly intervals for the first month, biweekly intervals for the second month, and then at monthly intervals for the remaining duration, except in the trial in Japan, where biweekly intervals were maintained from Week 4 to Week 26.

Meta-analysis
Individual patient data were available from the four trials and were pooled for analysis.All trials were assessed for risk of bias using the Cochrane risk-of-bias 1 tool ( Higgins et al., 2017 ) (Supplementary Table 1).All risks of biases were deemed low except attrition bias, for which the risk was unclear.Data for WOCBA, defined as female participants aged 18-45 years at screening were selected from the total population.
The primary endpoint was intervention for relapse or recurrence of any mood episode, defined as prescription of any additional pharmacotherapy or electroconvulsive therapy determined by the investigator to be necessary for treatment of a relapse and/or recurrence of a depressive, manic, hypomanic, or mixed episode.The associated outcome measure was TIME.TIME was selected as the primary efficacy outcome measure to account for the mixed enrollment criteria of the source studies (current/recent depressive episode [Calabrese et al.] and current/recent manic or hypomanic episode [Bowden et al.]).In a sensitivity analysis (TIME-Survival in Study [SIS]), the premature discontinuation of a participant for any reason before reaching the primary endpoint was considered an event related to bipolar disorder, and these participants were assumed to have reached TIME.
The secondary outcome measures were the time elapsed from randomization to the first intervention for the treatment of depression and predominantly depressed mixed episodes (TIDep) and the time elapsed from randomization to the first intervention for the treatment of mania, hypomania, mixed, and predominantly manic/hypomanic mixed episodes (TIMan).
The efficacy population was defined as participants within the WOCBA category who took at least one dose of DB treatment and had at least one post-baseline efficacy assessment during the DB phase.Participants who were randomized to the lithium arms of the international trials were excluded, as were participants in the Calabrese et al. trial who were randomized to lamotrigine 50 mg/day, which was assumed to be subtherapeutic, and in which recruitment was halted.
Safety and tolerability were assessed by evaluating adverse events (AEs) and reasons for withdrawal in the OL and DB phases.The safety population in the OL phase was defined as all participants within the WOCBA category who took at least one dose of lamotrigine in the OL phase.In the DB phase the safety population was defined as all participants within the WOCBA category who were assigned to receive either lamotrigine or placebo and received at least one dose of study treatment in the DB phase.Rash was an AE of particular interest, the incidence of which was determined as patients who experienced events with any of the following preferred terms: drug eruption, rash, dermatitis allergic, dermatitis, urticaria, blister, rash morbilliform, rash papular, rash erythematous, or rash pruritic.

Statistics
A random-effects model was employed to account for heterogeneity in the component studies.Kaplan-Meier survival curves were generated for time-to-event data and estimates of median TIME for the lamotrigine and placebo groups were derived with 95% confi- dence intervals (CIs).The difference in TIME between the lamotrigine and placebo groups was assessed using Cox proportional hazards regression analysis with treatment as the variable of interest.Because the covariate of importance, CGI-S, was uniform ( ≤ 3) across participants, no covariates were included in the analysis.Hazard ratios (HRs) were calculated with 95% CIs.
Additional sensitivity analyses were performed to investigate potential confounders to the primary analysis.To analyze the effect of differing study durations on TIME, HRs were calculated incorporating censoring at 26 weeks (all four studies), 36 weeks (international trials and the trial in China), or 76 weeks (international trials only).To investigate the effect of polarity of mood episode at entry to the study, HRs were calculated for TIME, TIDep, and TIMan among subgroups of participants whose baseline/most recent mood episode at enrollment was depressive or hypomanic/manic.

Handling missing data
This study incorporated an estimand framework and identified discontinuation from the study for any reason as an intercurrent event that could affect the outcome.A hypothetical strategy was employed, wherein the intercurrent event did not occur and the patient hypothetically continued the trial.These patients were modeled by censoring at the time of discontinuation and missing data were not imputed.Participants who completed the study without meeting the primary endpoint were censored at the time of study completion.For the sensitivity analysis of overall survival (TIME-SIS), study discontinuations for any reason other than study closure were imputed as primary endpoints.Reasons for discontinuation in TIME and TIME-SIS are reported in Supplementary Table 2. Secondary efficacy outcomes TIDep and TIMan were analyzed similarly to TIME, with additional censoring of data for patients who received intervention for a mood episode differing from the specified endpoint.Baseline demographic and clinical characteristics, and AEs during treatment were reported using descriptive statistics.

Ethics
Ethics Committee/Institutional Review Board approval was not required for this meta-analysis, as the analysis used anonymized patient data reported previously.All original trials included in this meta-analysis underwent Ethics Committee/Institutional Review Board reviews before patient enrollment.

Participant disposition and demographics
Participant disposition through the OL and DB phases of the study is shown in Fig. 1 .Across the four trials, 1940 people with bipolar I disorder were enrolled, 717 (37%) of whom were WOCBA.
370 WOCBA completed the OL phase and 346 withdrew from the study, the most common reason for which was AEs ( n = 100).Of the 370 WOCBA who completed the OL phase, 83 were removed from the analysis because they were randomized to receive lithium or lamotrigine 50 mg/day.Of the remaining 287 participants, 153 were randomized to receive lamotrigine and 134 were randomized to receive placebo.Of the 153 WOCBA who received lamotrigine, 57 (37%) completed the trial without meeting the primary endpoint and 96 (63%) withdrew.Of the 134 WOCBA who received placebo, 29 (22%) completed the trial without meeting the primary endpoint and 105 (78%) withdrew.
Demographic and clinical characteristics are summarized in Table 2 .The mean (standard deviation [SD]) age of participants at entry to the OL phase was 33.2 (7.38) years and mean (SD) age of onset was 21.2 (8.30) years for depressive episodes and 23.8 (8.16) years for manic/mixed episodes.

Time to intervention for a mood episode (primary endpoint analysis)
In the lamotrigine group, 65/152 (43%) participants had an intervention for a mood episode, compared with 73/133 (55%) in the placebo group.Median TIME was 323 days for the lamotrigine group; the lower limit of the 95% CI was 158 days, and the upper limit could not be estimated from the data because there were too few events after the median timepoint.For the placebo group, median TIME (95% CI) was 127 days (88, 232) ( Fig. 2A ).The HR for TIME significantly favored lamotrigine compared with placebo (HR 0.69; 95% CI 0.49, 0.96; p = 0.030).Further, in the TIME-SIS sensitivity analysis, 96/152 (63%) participants in the lamotrigine group, and 106/133 (80%) in the placebo group had an intervention for a mood episode or discontinued the study.Median (95% CI) TIME-SIS was 131 days (90, 188) for the lamotrigine group and 92 days (62, 127) for the placebo group ( Fig. 2B ) (HR 0.69; 95% CI 0.52, 0.91; p = 0.008).In further sensitivity analyses, the treatment effect for TIME was consistent when accounting for varying study duration (Supplementary Table 3) or the polarity of a participant's baseline/most recent mood episode at enrollment (Supplementary Table 4).

Time to intervention for a depressive episode or manic episode (secondary analysis)
In the lamotrigine group, 38/152 (25%) participants had a depressive episode requiring intervention, compared with  0.91 (0.52, 1.58); p = 0.732.† Mania, hypomania, mixed, or predominantly manic/hypomanic mixed episode.TIDep, time to first intervention for relapse and/or recurrence of a depressive episode; TIMan, time to first intervention for a manic/hypomanic/mixed episode.50/133 (38%) participants in the placebo group.Median TIDep could not be estimated for the lamotrigine group and was 269.0 days for the placebo group; the lower limit of the 95% CI was 137 days, and the upper limit could not be estimated from the data.The HR for TIDep was 0.59 (95% CI 0.39, 0.90; p = 0.014) ( Fig. 3A ).Individual-level data for TIDep are presented in Supplementary Fig. 1.A manic/hypomanic/mixed mood episode requiring intervention was experienced by 27/152 (18%) participants in the lamotrigine group, and 23/133 (17%) in the placebo group.Median TIMan was not reached for either treatment group.The HR was 0.91 (95% CI 0.52, 1.58; p = 0.732) ( Fig. 3B ).The treatment effect for TIDep and TIMan was consistent irrespective of baseline/most recent mood episode at enrollment (Supplementary Table 4).
§ Reported by > 1 participant in the open-label phase, or the lamotrigine or placebo group of the double-blind phase.The following serious adverse events were each experienced by one participant in the open-label phase: affective disorder, bipolar I disorder, bipolar disorder, confusional state, hypomania, intentional self-injury, major depression, mood altered, psychotic disorder, somatic symptom disorder, ankle fracture, face injury, skin laceration, chest pain, appendicitis, pneumonia, urinary tract infection, leukopenia, seizure, vaginal hemorrhage, and deep vein thrombosis; double-blind phase (lamotrigine): fetal death, syncope, and heavy menstrual bleeding; double-blind phase (placebo): depression suicidal, impulsive behavior, suicidal behavior, ectopic pregnancy, diarrhea, pancreatitis acute, cholecystitis, and hypersensitivity.
AEs led to the permanent discontinuation of the study drug or withdrawal from the study for 127/717 (18%) participants in the OL phase.In line with the trial protocols, which required participants who experienced rash to discontinue treatment unless it was clearly not treatment related, rash-related AEs were the events that most frequently led to permanent discontinuation of the study drug or withdrawal from the study, with 52/717 (7%) participants reporting nonserious rash-related AEs leading to withdrawal in the OL phase.
AEs led to the permanent discontinuation of the study drug or withdrawal from the DB phase for 12/152 (8%) lamotrigine-treated participants and 18/133 (14%) placebotreated participants.
The AEs that most frequently led to study withdrawal in the DB phase were psychiatric events (lamotrigine n = 5/152 [3%]; placebo, n = 9/133 [7%]).Non-serious rash-related AEs led to the permanent discontinuation of the study drug or withdrawal for 2/152 [1%] participants in the lamotrigine group and 3/133 [2%] in the placebo group during the DB phase.

Discussion
This meta-analysis, using patient-level data from four randomized, placebo-controlled trials (of up to 76 weeks' duration), showed that maintenance treatment with lamotrigine delayed intervention for relapse and recurrence of mood episodes compared with placebo among WOCBA with bipolar I disorder who responded to induction treatment of up to 16 weeks.Lamotrigine's therapeutic effect could be characterized by a delay in relapse and recurrence of depressive episodes, which was the predominant symptom for this patient population before enrollment in the studies.Lamotrigine was well tolerated during the DB phase, with 7% of participants discontinuing because of AEs compared with 12% of patients in the placebo group, and no new safety signals emerged in this population.
Median TIME was extended from 127 days with placebo to 323 days with lamotrigine, with a 31% reduction in hazard rate.Although data for participants who discontinued for reasons other than intervention for a mood episode were censored from this analysis, the robustness of these findings was supported by a sensitivity analysis of overall survival in which censored observations or premature discontinuation were imputed as an event defined by the primary endpoint, for which the median time to intervention was extended from 92 days with placebo to 131 days with lamotrigine.
The efficacy of lamotrigine for delaying depressive episodes but not manic mood episodes in this analysis of WOCBA with bipolar I disorder is consistent with observations from trials in the broader bipolar I disorder population and systematic literature reviews, for which lamotrigine's utility is most evident in preventing recurrence and relapse of depression ( Besag et al., 2021 ;Bowden et al., 2003 ;Calabrese et al., 2003 ;Kishi et al., 2021 ).This has important clinical implications for the treatment of WOCBA with bipolar disorder, who are more likely than men to have a predominant depressive polarity, and this should be a consideration when selecting therapy ( Altshuler et al., 2010 ;Nivoli et al., 2011 ;Popovic et al., 2014 ).Although predominant polarity may be difficult to determine, polarity at onset may be used as a proxy, particularly for patients with depressive symptoms at onset ( Yoldi-Negrete et al., 2021 ), and lamotrigine may also benefit patients with undetermined polarity ( Fico et al., 2022 ).No predominant polarity was identified in this study, but depressive episodes were reported with greater average frequency in the 3 years before screening and had earlier onset than manic episodes, supporting lamotrigine's benefit for patients with undetermined polarity or a depressive onset.Findings from a sensitivity analysis in this study support lamotrigine's use for the prevention of mood episodes, particularly depressive episodes, regardless of the polarity of the patient's most recent episode.
Rash is a known adverse drug reaction associated with lamotrigine ( Grunze et al., 2013 ) that was carefully monitored in the original trials.In line with the labeled recommendation, participants who experienced rash were required to discontinue treatment unless the rash was clearly not treatment related.During the OL phase, 54 rash-related AEs led to participants permanently discontinuing lamotrigine or withdrawing from the study.However, most of these cases were mild or moderate, suggesting that careful titration of lamotrigine dose according to prescribing information minimized the risk of serious events.Only one serious event each of rash and drug eruption was reported, which suggests that WOCBA are no more likely than the broader bipolar I disorder population to experience serious rash with lamotrigine treatment.Few cases of rash were reported with maintenance treatment during the DB phase, with a similar incidence between treatment groups, so although clinicians should continue to monitor for rash during maintenance treatment, the risk can be mitigated by appropriate initial dose titration complying with product label recommendations.
Some mood stabilizers, including valproate, and antipsychotics used in the treatment of bipolar disorder are known to cause weight gain ( Grootens et al., 2018 ;Schneider et al., 2020 ) and many drug combinations used in clinical practice are associated with an increased risk of type 2 diabetes ( Nestsiarovich et al., 2020 ).In contrast, lamotrigine is not associated with clinically meaningful weight gain, and this observation was consistent for WOCBA in this study, where few participants reported weight increase as an AE with lamotrigine.Weight gain is not only of clinical importance to a patient's health but also contributes to how well a treatment plan is tolerated.In the treatment of bipolar depression, weight gain is a major patient-reported adverse effect leading to medication discontinuation ( Rosenblat et al., 2019 ).The absence of a clinically meaningful effect on weight with lamotrigine may be considered an important factor in improving adherence to long-term maintenance treatment for this population.
Accounting for the high incidence of unplanned pregnancies among WOCBA with bipolar disorder ( Marengo et al., 2015 ;Zengin Eroglu and Lus, 2020 ), treatment decisions should always consider the possibility of pregnancy in this patient group, and WOCBA should be advised of the ter-atogenic risks associated with some medications and the need for careful family planning ( Goodwin et al., 2016 ;Yatham et al., 2018 ).Despite valproate being unsuitable for use by women or girls of childbearing age under most circumstances ( Goodwin et al., 2016 ;Grunze et al., 2013 ;Royal College of Psychiatrists, 2018 ;Yatham et al., 2018 ), drug utilization data suggest that it is still used in this population more commonly than might be desired ( Kan et al., 2022 ;Ng et al., 2021 ;Tachibana et al., 2020 ).These prescribing practices may be compounded by inadequate communication from healthcare providers regarding the teratogenicity of valproate and the importance of taking appropriate precautions to avoid pregnancy when women are prescribed the medication ( Paton et al., 2018 ).Although the current study does not provide data supporting the use of lamotrigine during pregnancy, in contrast to valproate, lamotrigine is recognized to have one of the safer pharmacological profiles for use during pregnancy of anti-epileptic drugs used in the management of bipolar disorder ( Hernandez-Diaz et al., 2012 ;Pariente et al., 2017 ;Tomson et al., 2018 ).Postpartum, the presence of a drug in breast milk is another consideration for patients who wish to continue pharmacological treatment while breastfeeding.Serious adverse effects and altered cognitive development in infants are not associated with the use of lamotrigine during breastfeeding ( Pacchiarotti et al., 2016 ).However, because infants may be exposed to lamotrigine concentrations up to 50% of that received by the mother, ( Ohman et al., 2000 ), it is advisable to monitor breastfed infants for potential effects including sedation and rash.
Interactions between lamotrigine and hormonal contraceptives should also be considered when prescribing lamotrigine to WOCBA with bipolar disorder.Estrogen-containing oral contraceptives are recognized to increase lamotrigine clearance, and lamotrigine may reduce plasma levorgestrel levels, although not to an extent that indicates a loss of contraceptive efficacy ( Food and Drug Administration, 2015 ;Sidhu et al., 2006 ).Investigation of the progestin component of several combined oral contraceptives showed that lamotrigine serum concentrations were reduced by drospirenone and levonorgestrel but not gestodene, with no evidence that lamotrigine usage altered the serum concentrations of any of these progestins ( Rauchenzauner et al., 2020 ).
Differences between the trials included in this metaanalysis, such as location, inclusion criteria, dropout rates, and duration of follow-up in the DB phase may limit assumptions of homogeneity; however, the trials were similar in most aspects and availability of individual patient data facilitated a robust pooled analysis.Despite differences between trials for maintenance phase durations, for all the studies, most mood events occurred within the first 26 weeks of randomized treatment and a sensitivity analysis confirmed that differing study duration did not appear to impact the outcomes.Therefore, there were no adjustments for the unequal study durations.The trials included in this meta-analysis were not prospectively designed to investigate the effect of lamotrigine for WOCBA, and individually were not powered to evaluate this subgroup.Limiting this meta-analysis to four trials for which participant-level data were available is a potential source of bias, but the trials selected represent a diverse range of regions with two in-ternational and two Asian trials.As the data from the trials were pooled post hoc, the interpretation of these analyses should be treated with caution.
This study is the first meta-analysis to investigate lamotrigine as maintenance treatment in WOCBA with bipolar I disorder, and it indicates that lamotrigine is efficacious for delaying mood episodes requiring intervention, predominantly depressive episodes, compared with placebo and is well tolerated.Safety data for this group of patients is important and reassuring with respect to lamotrigine's known adverse effect of rash, and long-term tolerability is benefited by the absence of weight gain with lamotrigine in comparison with other treatments for bipolar disorder.Maintenance treatment decisions for WOCBA should balance the risks of untreated illness with potential risks associated with pharmacological treatment during the perinatal period and avoid the use of treatments that are contraindicated in populations where pregnancy is a possibility.

Role of the funding source
This study was funded by GSK (GSK study 217193 ).Authors of this paper who are GSK employees were involved in the study design; the collection, analysis and interpretation of data; writing the report; and the decision to submit the paper for publication.

Fig. 1
Fig. 1 Patient disposition.† Excluded because participants were randomized to lamotrigine 50 mg (recruitment halted) or lithium.‡ Koyama et al.: 100 mg/day at investigator discretion; Bowden et al.: target dose of 200 mg/day within dosage range 100-400 mg/day.§ Includes protocol violation, sponsor discontinuation, lost to follow-up, withdrew consent, and "other" categories from the international trials.Includes protocol deviation, study termination, lost to follow-up, withdrew consent, and investigator discretion categories from Zhang et al. and Koyama et al.WOCBA, women of childbearing age.

Table 1
Summary of primary studies included in meta-analysis.Participants randomized to 50 mg/day dosage of LTG were excluded from this analysis.§ Lithium group was excluded from this analysis since this treatment group was not included in the trials in Japan and China. ‡

Table 2
Demographic and clinical characteristics of women of childbearing age.

Table 3
Most common ( ≥ 5%) † adverse events, adverse events of special interest, and serious adverse events reported by more than one patient.