• pediatric migraine;
  • divalproex sodium;
  • migraine prophylaxis;
  • adolescents;
  • safety


  1. Top of page
  2. Abstract
  6. Acknowledgments

Objective.— To evaluate the long-term safety and tolerability of divalproex sodium extended-release in the prophylaxis of migraine headaches in adolescents.

Background.— Divalproex sodium has been approved for migraine prophylaxis in adults. A previous double-blind, placebo-controlled study of the efficacy and safety of divalproex sodium extended-release for prevention of migraine in adolescents was followed by the present long-term extension trial, which was designed to collect additional safety and tolerability data.

Methods.— This was a 12-month, Phase 3, open-label extension of a 3-month, double-blind, placebo-controlled, multicenter study of adolescents aged 12 to 17 years with migraine headaches who had either completed the previous study or had discontinued because of lack of efficacy. Subjects from the previous trial who had experienced serious adverse events possibly or probably related to study drug were excluded. Divalproex sodium extended-release 500 mg daily was administered for 15 days then increased to 1000 mg. Study visits were conducted at days 1 and 15 and months 1, 2, 3, 6, 9, and 12. Safety assessments included adverse event collection, laboratory testing, physical and neurological examinations, vital signs, and electrocardiograms, as well as reproductive endocrine analyses for postmenarchal female subjects. Efficacy was evaluated by sequential 4-week migraine headache rates calculated from subjects' headache diaries.

Results.— A total of 112 subjects enrolled in the trial. The most common adverse events were weight gain (15%), nausea (14%), somnolence (12%), upper respiratory tract infection (11%), increased ammonia (8%), and sinusitis (8%). Five (4%) subjects experienced serious adverse events, and 15 (13%) subjects prematurely discontinued because of an adverse event. Increased ammonia levels were noted in some individuals, and the mean ammonia level for all subjects increased 19.2 µm from baseline. No other clinically significant changes were observed in laboratory values, vital signs, or electrocardiograms. Improvement in mean and median 4-week migraine headache rates occurred by the fourth month and lasted for the duration of the trial.

Conclusions.— In this long-term open-label extension study, the safety profile of divalproex sodium extended-release in adolescents with migraine was consistent with that observed in the preceding 3-month, double-blind trial and in previous adult studies. Overall, divalproex sodium extended-release was well-tolerated in adolescents aged 12 to 17 years.


adverse event


divalproex sodium extended-release




Food and Drug Administration


Medical Dictionary for Regulatory Activities


valproic acid

Among adolescents aged 12 to 17 years, an estimated 7.3% of females and 4.9% of males experience migraine headaches.1 Migraine has many adverse consequences in this age group, including missed school days, sleep disturbances, and behavioral issues,1,2 but a recent study reported that only 10.7% of migraineurs under the age of 18 currently use medication for migraine prophylaxis.3 These small numbers, coupled with a paucity of controlled studies of migraine therapies in children and adolescents,4-8 have limited our understanding of both the efficacy and safety of antimigraine agents in younger patients.

At this time, there are no Food and Drug Administration (FDA)-approved medications for the prophylaxis of migraine headaches in adolescents. Divalproex sodium (DVPX) delayed-release (DR) and extended-release (ER) tablets have been approved in the United States for prevention of migraine headaches in adults based on 3 randomized, placebo-controlled, double-blind, multicenter trials: 2 with the DR formulation9,10 and 1 with DVPX ER.11 DVPX displays a similar pharmacokinetic profile in children and adolescents 8-17 years of age compared with healthy adult controls,12 and case reports and other uncontrolled studies have suggested that DVPX may also be beneficial for the prevention of migraines in children and adolescents.13-18 These findings provided the rationale for a recently published prospective, double-blind, placebo-controlled, 12-week study of the safety and efficacy of DVPX ER for migraine prophylaxis in adolescents.19 In that study, DVPX ER failed to demonstrate efficacy in the prevention of adolescent migraine when compared with placebo.19 DVPX ER was generally well-tolerated in adolescents,19 with a safety profile similar to what has been seen previously in adults.9-11 The present report describes additional safety and efficacy data collected during a long-term open-label extension of the preceding double-blind study.


  1. Top of page
  2. Abstract
  6. Acknowledgments

Study Design and Subjects.— This phase 3, open-label, 12-month extension study followed subjects who either completed or prematurely discontinued because of ineffectiveness from a randomized double-blind, placebo-controlled study of DVPX ER and continued to meet the eligibility criteria from that protocol.19 Briefly, the participants were 12 to 18 years of age, in good health, were not pregnant or nursing, were practicing an effective form of birth control, had their initial migraine ≥12 months prior to entry into the double-blind study, and met International Headache Society diagnostic criteria for migraine headache.20 Exclusion criteria included a history of cluster headaches, >15 headaches of any type per month, medication noncompliance, substance abuse within the last 6 months, allergy/sensitivity to valproate, or a history of encephalopathy, hepatitis, pancreatitis, or urea cycle disorder. In addition, subjects were excluded if they needed headache medication >10 days per month, had taken valproate or an investigational agent within the last 30 days, or had not responded to 2 or more adequate trials of other migraine prophylaxis regimens or 1 trial of valproic acid (VPA) or DVPX. Subjects who had experienced a serious adverse event (AE) considered possibly or probably related to study drug during the double-blind study were not eligible for the open-label extension.

Prescription and over-the-counter medications were allowed at the investigators' discretion, but the following agents were excluded: antiepileptic drugs other than DVPX, antidepressants, aspirin-containing products, anticoagulants, chronic systemic corticosteroids, clonazepam, diazepam, erythromycin, pemoline, rifampin, tolbutamide, and zidovudine.

Study visits were conducted at days 1 and 15 and months 1, 2, 3, 6, 9, and 12. Subjects were given DVPX ER 250 mg tablets, initiated at 500 mg once daily for 15 days, then increased to 1000 mg once daily. Investigators could make dose adjustments to optimize clinical response, provided that the total daily dose remained between 250 mg and 1000 mg.

Safety.— The cumulative dataset was the primary dataset for safety. For subjects who received DVPX ER during the double-blind study and continued in the open-label extension with a gap of 7 days or less, the cumulative dataset included all safety data collected during DVPX ER treatment in both studies. For subjects who received placebo during the double-blind study, or for those with a gap of more than 7 days between studies, only safety data from the open-label study were included in the cumulative dataset.

Safety was assessed by AE collection, laboratory tests, physical and neurological examinations, vital signs, and electrocardiograms (ECGs). In addition, because of previous reports associating DVPX use with hyperandrogenism and polycystic ovary syndrome (PCOS), menstrual and reproductive endocrine parameters (testosterone and sex hormone binding globulin [SHBG]) were followed in postmenarchal female subjects who were not on hormonal contraceptives or other steroids. Testosterone levels were obtained at routine study visits and therefore did not necessarily coincide with the pre-follicular phase of the menstrual cycle.

Treatment-emergent AEs, defined as any AE that began or worsened after the first dose of DVPX ER in either study, were coded using MedDRA® (Medical Dictionary for Regulatory Activities). The mean change from baseline in laboratory and vital sign variables was calculated. For subjects treated with DVPX ER in the previous blinded study with a gap of 7 days or less between studies, the baseline was the last evaluation obtained before the first dose of DVPX ER taken in the double-blind study. For placebo-treated subjects from the previous study and for DVPX ER-treated subjects from the previous study who started the open-label extension after a gap of more than 7 days, the baseline was the last evaluation obtained prior to the first dose of DVPX ER in the open-label extension study.

Compliance, Dosing, and Exposure.— Treatment compliance was determined by pill counts and other supporting information obtained at each study visit. Study drug compliance below 70% required consultation with the subject by study site personnel. The number of subjects in the cumulative dataset with at least 6 months and 12 months of exposure to DVPX ER was reported. Trough serum total valproate concentrations were collected approximately 24 hours (±3 hours) post dose at months 1, 2, 3, 6, 9, and 12.

Efficacy.— At each visit (except the final visit at month 12), subjects received a headache and medication diary in which they recorded all headaches and any related medications taken during study participation. Migraine headaches which occurred within 48 hours of each other were considered a single migraine headache. Data from the headache and medication diary were used to calculate the 4-week migraine headache rate. Migraine headache rates and number of migraine headache days were summarized for each consecutive 4-week period throughout the duration of the long-term extension, both overall and by the treatment group assigned in the previous double-blind trial. Subjects who took at least 1 dose of open-label study drug and had at least 1 headache evaluation (intent-to-treat dataset) were included in the efficacy analyses.


  1. Top of page
  2. Abstract
  6. Acknowledgments

Subjects and Exposure.— A total of 112 subjects were treated at 26 centers in the United States. Subject demographics are presented in Table 1. The most common reasons for withdrawing from the study were AEs (n = 15, 13%), lack of efficacy (n = 11, 10%), and lost to follow up (n = 10, 9%). A total of 83 subjects (74%) had at least 6 months of exposure to DVPX ER, and 53 subjects (47%) had at least 12 months of exposure to DVPX ER. The overall mean duration of DVPX ER exposure was 300 days for the cumulative dataset, with a mean modal daily dose of 741 mg (12.5 mg/kg). The mean maximum DVPX ER daily dose for the entire exposure period was 926 mg (15.4 mg/kg). Sixty-eight percent of the subjects were ≥70% compliant with taking their study drug throughout the study. The mean (SD) serum valproate concentration at the final visit was 38.9 (37.3) µg/mL (n = 108).

Table 1.—. Subject Demographics (Cumulative Dataset)
ParameterTotal (n = 112)
  • Hispanic.

  • Denotes ethnic self-identification as Hispanic, irrespective of race.

  • BMI = body mass index.

Age (years) 
 Mean ± SD14.6 (1.54)
Age distribution (years), n (%) 
 12 to 1327 (24)
 14 to 1553 (47)
 16 to 1732 (29)
Gender, n (%) 
 Female61 (54)
 Male51 (46)
Race, n (%) 
 White78 (70)
 Black30 (27)
 Asian1 (<1)
 White/black2 (2)
 Other1 (<1)
Hispanic ethnicity, n (%) 
 Yes4 (4)
 No108 (96)
Weight (kg) 
 Mean (SD)63.8 (13.33)
BMI (kg/m2) 
 Mean (SD)23.2 (4.29)
Migraine headaches within 3 months of screening of double-blind study 
 Mean (SD)17.0 (7.11)


AEs.— Adverse events reported in ≥5% of subjects are presented in Table 2. For the cumulative dataset, the most common treatment-emergent AEs (≥10% of subjects) were weight gain (15%), nausea (14%), somnolence (12%), and upper respiratory tract infection (11%). The AEs that led to premature discontinuation for 2 or more subjects included weight gain (n = 3), somnolence (n = 2), and nausea (n = 2). Five (4%) study subjects had 1 or more serious AEs, of which 4 required hospital-level care and 1 required outpatient intervention (Table 3). One subject had a history of aggression and had been discontinued from the study because of aggressive behavior. He was subsequently hospitalized after a violent outburst and diagnosed with schizophreniform disorder. This serious AE was considered to be unrelated to study medication. Another subject was hospitalized because of a peptic ulcer which resolved upon treatment and was considered probably not related to study drug. Another subject cut herself with a knife after arguing with her mother and was hospitalized for suicidal ideation, depression, and the injuries she had self-inflicted. Her valproate level was below the limit of detectability (<12.5 µg/mL), which suggested she had not been taking her study medication. She was diagnosed with new-onset bipolar disorder with borderline personality traits and was treated accordingly. None of these events were considered by the investigator to be related to study medication. One subject developed symptomatic hyperammonemia possibly related to study drug, described in more detail below, and 1 subject underwent an elective abortion that was not related to study drug.

Table 2.—. Treatment-Emergent AEs Occurring in ≥5% of Subjects (Cumulative Dataset)
AEDVPX ER (N = 112) N (%)
  • Subjects reporting the same AE ≥ 2 times were counted only once in the totals for that particular event.

  • AE = adverse event; DVPX ER = divalproex sodium extended-release.

Any AE99 (88)
 Weight gain17 (15)
 Nausea16 (14)
 Somnolence13 (12)
 Upper respiratory tract infection12 (11)
 Ammonia increased9 (8)
 Sinusitis9 (8)
 Fatigue8 (7)
 Abdominal pain7 (6)
 Influenza7 (6)
 Dizziness6 (5)
 Viral gastroenteritis6 (5)
 Tension headache6 (5)
 Vomiting6 (5)
Table 3.—. Serious Adverse Events
Gender/age (years)Preferred term(s)Reason seriousOnset (study day)SeverityRelationship
  • Adverse events that led to premature discontinuation from study.

Male/17Schizophreniform disorderHospitalization21 days after last doseSevereNot related
Male/12Peptic ulcerHospitalization110SevereProbably not related
Female/14HyperammonemiaRequired intervention29SeverePossibly related
Female/14Depressive symptomHospitalization1 day after last doseAll severeAll not related
Intentional self-injury
Skin laceration
Suicidal ideation
Female/16Abortion induced/elective abortionRequired intervention6 days after last doseSevereNot related

Laboratory Findings.— Laboratory values were consistent with the results seen in the previous double-blind trial,19 with the majority of subjects having laboratory values within the normal range at baseline and final visits. Laboratory parameters for which ≥5% of subjects showed a shift from normal baseline to below-normal final values included red blood cell count (7% of subjects) and white blood cell count (6%). Increases from normal baseline to above-normal final values occurred in ≥5% of subjects for ammonia (34%), inorganic phosphorus (13%), prothrombin time (13%), partial thromboplastin time (12%), eosinophils (9%), neutrophils (8%), total cholesterol (8%), and triglycerides (7%).

Mean laboratory values that decreased ≥5% from baseline for the cumulative dataset included platelet count (−14.8 × 109/L, SD = 46.0, baseline = 269.4 × 109/L), basophils (−0.05%, SD = 0.30%, baseline = 0.37%), and alkaline phosphatase (−32.5 U/mL, SD = 50.2, baseline = 185.2 U/mL). Mean increases of ≥5% over baseline were observed for monocytes (0.52%, SD = 2.1, baseline = 5.5%), uric acid (30.4 µm, SD = 54.8, baseline = 269.2 µm), triglycerides (0.09 mm, SD = 0.65, baseline = 1.26 mm), and ammonia (19.2 µm, SD = 56.5, baseline = 39.3 µm).

In contrast to the double-blind study,19 the open-label extension showed an increase in average ammonia concentration between baseline and final measurements (cumulative dataset). For individual subjects, 26 (23.2%) had at least 1 ammonia level that was classified as “very high” (≥90 µm). Most of these individual increases (20/26, 76.9%) were transient, and only 1 was symptomatic. In the symptomatic case, the subject had an elevated ammonia level (>250 µm) at the month 1 visit. As the subject was experiencing fatigue, grogginess, and vomiting, her parents were advised to take her to the emergency room, where her ammonia level was 69 µm. She was treated with intravenous fluids and lactulose, and then discharged. Repeated attempts by study site personnel to obtain medical records from this emergency room visit were unsuccessful. The subject's ammonia levels on study drug declined from their maximum value to 65 µm at the month 2 visit, after which the subject was lost to follow up.

Other Safety Parameters.— Vital sign changes from baseline to final were unremarkable. Mean changes from baseline to final values for body weight and height were 4.1 kg and 2.5 cm, respectively. Mean body mass index increased 0.8 kg/m2; 3 subjects discontinued from the study because of an AE of weight gain. There were no significant abnormalities in physical examination findings or ECGs during the study.

Reproductive endocrine parameters were evaluated in postmenarchal female subjects who were not on hormonal contraceptives or other steroids. Among these subjects, average total testosterone increased by 8.05 ng/dL (SD = 27.3) from a baseline of 45.1 ng/dL (n = 41). Average bioavailable testosterone increased 0.13 ng/dL (SD = 9.93) from a baseline of 13.46 ng/dL (n = 39) and average free testosterone decreased 0.173 pg/mL (SD = 4.62) from a baseline of 6.52 pg/mL (n = 41). Mean levels of SHBG, a testosterone binding protein, increased 1.44 nm (SD = 28.8) from a baseline of 42.6 nm (n = 39).

Reproductive endocrine AEs were reported in 4 of the 41 postmenarchal female subjects who were not on hormonal contraceptives or other steroids; in 2 of the 4 subjects, these AEs were considered probably related to study drug. One subject reported irregular menses and increased acne, which led to her premature discontinuation from the study. Another subject had an increased testosterone level that was designated as an AE by the study investigator. Two additional subjects developed increased testosterone that was considered unrelated to study drug; 1 elevation occurred 44 days after discontinuation of DVPX ER, and the other was associated with pregnancy (see below).

Three subjects became pregnant during the study. One delivered a healthy baby and 1 underwent an elective abortion. For the third pregnancy, repeated attempts to obtain information regarding the outcome were unsuccessful.

Efficacy.— The median (mean) 4-week migraine headache rate was 1.1 (1.4) over the entire 12 months of the study and 1.0 (1.2) for the final 4-week period. The 4-week median and mean migraine headache rates decreased between the first (2.0, 2.1) and the fourth (1.0, 1.4) 28-day period; this improvement persisted for the duration of the study. The most commonly used symptomatic medications for the treatment of migraine headaches were nonsteroidal anti-inflammatory drugs and acetaminophen, which were used for an average of 33.4% and 35.3% of migraine headaches, respectively.


  1. Top of page
  2. Abstract
  6. Acknowledgments

No medications are currently FDA-approved for the prophylaxis of migraine in children or adolescents, and long-term safety and efficacy data for antimigraine agents are lacking in this population. The present study followed over 100 subjects on DVPX ER for migraine prevention for up to 1 year. The most frequently reported AEs were weight gain, nausea, somnolence, and upper respiratory tract infection, consistent with AEs reported in the prior blinded trial in adolescents19 and in studies of DVPX ER migraine prophylaxis in adults.11 The safety assessments in this study indicated no clinically significant abnormalities in vital signs, physical examinations, neurological assessments, or ECG findings. Subjects experienced an increase over baseline in average body weight and BMI, and 3 subjects left the study because of weight gain.

Five serious AEs occurred in this study, 3 of which led to hospitalization. None of these serious AEs was considered related to study drug, with the exception of 1 case of symptomatic hyperammonemia that required emergency room care, and was classified as possibly related to study drug.

In the present study, DVPX ER treatment was associated with an increase in average ammonia levels compared with baseline. However, this finding is difficult to interpret in the absence of a placebo control group. In the preceding double-blind trial, all treatment groups, including placebo, showed a numerical increase from baseline in average ammonia levels, but there was no statistically significant difference between any DVPX ER group and placebo.19 In the long-term open-label extension, 1 subject experienced symptoms related to elevated ammonia, and there were no ammonia-related premature discontinuations.

Symptomatic and asymptomatic elevations in ammonia are recognized side effects of DVPX in pediatric and adult patients21-23 and are described in the DVPX ER prescribing information.24 While hyperammonemic encephalopathy should be considered in any patient who develops unexplained lethargy and vomiting or changes in mental status on DVPX, the clinical significance of asymptomatic elevated ammonia is unclear. A 2007 meta-analysis showed that asymptomatic elevations in ammonia are remarkably common in association with VPA or DVPX use in adults, with a mean prevalence of 90.3% in prospective studies and 47.3% in cross-sectional studies.25 In addition, a recent report suggests that a high proportion of elevated ammonia false positives are observed in pediatric patients (up to 48%).26 Ammonia measurements are greatly influenced by pre-analytical factors, including sampling technique, temperature, and time elapsed before analysis.26,27 The current study utilized venous blood, which is associated with greater assay variability than arterial samples.

Recently, concerns have been raised regarding a potential relationship between the use of antiepileptic drugs and suicidality.28 A 2008 meta-analysis conducted by the FDA on 11 antiepileptic drugs used for a variety of indications (epilepsy, bipolar disorder, migraine, etc.) in children and adults found a 1.8-fold increased risk of suicidal behavior or ideation among patients receiving antiepileptic drugs compared with those receiving placebo (0.43% vs 0.22%, estimated odds ratio 1.80, 95% CI 1.24, 2.66).29 This finding prompted an FDA alert emphasizing the need for careful monitoring for signs of suicidality among patients receiving this class of medication.28 The estimated odds ratio for suicidality associated with DVPX vs placebo in this meta-analysis was 0.72 (95% CI 0.29, 1.84).29 Although the DVPX studies analyzed by the FDA included primarily adult subjects, the overall analysis for all 11 antiepileptic drugs showed no pattern in drug effect with respect to age subgroups.29 During the present study of DVPX ER in adolescents with migraine, 1 subject experienced depression with suicidal ideation and self-injury that required inpatient care. At the time of this episode, the subject's VPA level was below the limit of detectability (<12.5 µg/mL). These events were not believed to be related to the use of DVPX ER but rather to new-onset bipolar disorder with borderline personality features.

The use of anticonvulsants (including DVPX) has been implicated in the development of reproductive complications in female patients, including teratogenicity,24 hyperandrogenism, and PCOS.30,31 In the current study, among 41 postmenarchal female subjects not on hormonal contraceptives or other steroids, average testosterone levels showed numerical changes over 1 year of follow up, with the direction and degree of change varying depending on the particular testosterone fraction being measured. Overall, the lack of a placebo control group and the variability of testosterone levels obtained randomly during the menstrual cycle limit interpretation of these changes. Four of 41 subjects developed reproductive endocrine AEs, which in 1 case (irregular menses and increased acne) led to premature discontinuation from the study. Of the 3 remaining subjects, 2 had AEs that were considered by the study investigator to be unrelated to DVPX ER treatment.

Although use of adequate birth control was required for study participation, 3 pregnancies occurred during the 1-year trial. For the one live birth for which information is available, there was no evidence of an adverse outcome in the infant.

In terms of efficacy, the 4-week migraine headache rate improved between the first and the fourth 28-day period, with this effect persisting to the end of the 12-month study. However, this finding must be interpreted in the context of the open-label, nonplacebo-controlled conditions of the trial. Similar improvements were seen in the preceding 3-month, double-blind, placebo-controlled study, but there was no statistically significant difference between DVPX ER treatment groups and placebo.19

In this long-term, open-label extension study of DVPX ER in adolescents with migraine, the safety profile of DVPX ER was similar to that seen in the preceding 3-month, blinded, placebo-controlled study19 and in previous DVPX adult migraine trials.9-11 AEs and laboratory changes associated with DVPX ER use in adolescents (eg, nausea/vomiting, sedation, weight gain, elevated ammonia, decreased platelets) were consistent with those observed previously in adults. Overall, DVPX ER was generally well-tolerated in male and female migraine patients aged 12-18 years.


  1. Top of page
  2. Abstract
  6. Acknowledgments

Acknowledgments: The participation of the following investigators is gratefully acknowledged:

Ricardo Ayala, MD, AMO Corporation, Tallahassee, FL; James F. Bale, MD, Pediatric Neurology, Salt Lake City, UT; Barney Blue, DO, Eminence Research, LLC, Oklahoma City, OK; Jan Brandes, MD, Nashville Neuroscience Group, PC, Nashville, TN; Roger Cady, MD, Clinvest/Headache Care Center, Springfield, MO; Mary K. Dominski, MD, Dean Foundation Medical Research, Madison, WI; Arthur H. Elkind, MD, Elkind Headache Center, Mt. Vernon, NY; Jerome Goldstein, MD, San Francisco Clinical Research Center, San Francisco, CA; David Harris, MD, Granger Medical Clinic, West Valley City, UT; Thomas C. Klein, MD, Heartland Research Associates, Wichita, KS; Toan Lam, MD, Alpine Medical Group, Salt Lake City, UT; Steve Linder, MD, Dallas Pediatric Neurology Associates, Dallas, TX; Ninan T. Mathew, MD, Houston Headache Clinic, Houston, TX; Bhagwan Moorjani, MD, Children's National Medical Center, Washington, DC; Eric Pearlman, MD, PhD, Savannah Neurology, PC, Savannah, GA; Keith Reisinger, MD, Primary Physicians Research, Pittsburgh, PA; Terrance Riske, MD, Hayden Lake Family Physicians, Hayden Lake, ID; Ned Rupp, MD, National Allergy, Asthma, & Urticaria Centers of Charleston, PA, Charleston, SC; Joel R. Saper, MD, Michigan Head Pain & Neurologic Institute, Ann Arbor, MI; Jerry J. Tomasovic, MD, Texas Association of Pediatric Neurology, San Antonio, TX; Daniel Wynn, MD, Consultants in Neurology, Northbrook, IL; Donald Younkin, MD, Children's Hospital of Philadelphia, Philadelphia, PA.

Special thanks to Lindsey Bensman, Senior Clinical Research Associate, for ensuring the careful and timely completion of this study, and to Muriel Cunningham for her medical writing and editing expertise.


  1. Top of page
  2. Abstract
  6. Acknowledgments