9.3.1. Inclusion Criteria

The investigator or other study site personnel documented in the source documents (e.g., the hospital chart) that informed consent was obtained. Laboratory tests or questionnaires that were performed as part of the screening procedures were completed prior to the first dose of study drug and recorded in the eCRF.

1. Participants aged 7 years to 17 years 11 months at time of first dose.
2. Participants must meet Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) diagnostic criteria, as assessed by a semi-structured assessment (Kiddie Schedule for Affective Disorders and Schizophrenia - Present and Lifetime (KSADS-PL)) and a separate clinical interview with a child/adolescent psychiatrist for manic or mixed episodes in bipolar I disorder.
3. Score of > 20 on the YMRS at screening and baseline.
4. The participant and legal guardian must understand the nature of the study and be able to comply with protocol requirements. The legal guardian must give written informed consent and the youth, written assent.
5. Participants with comorbid conditions [attention deficit hyperactivity disorder (ADHD), conduct disorder], except those listed in Exclusion Criterion 2, may participate.
6. If female: is premenarchal, or is incapable of pregnancy because of a hysterectomy, tubal ligation, or spousal/partner sterility. If sexually active and capable of pregnancy, has been using an acceptable method of contraception (hormonal contraceptives, intrauterine device, spermicide and barrier) for at least one month prior to study entry and agrees to continue to use one of these for the duration of the study. If sexually abstinent and capable of pregnancy, agrees to continued abstinence or to use of an acceptable method of birth control should sexual activity commence.
7. Has a negative quantitative serum ß-human chorionic gonadotrophin hormone pregnancy test at screening and a negative qualitative urine pregnancy test at baseline, if female.
8. Participants with a history of substance abuse may participate if they agree to continue to abstain from drugs during the trial and have a negative drug screen at screening or prior to baseline. Those with an initial positive drug screen during initial screening may have another screen done 1-3 weeks later while in screening, and a negative result will allow the participant to participate.
9. The participant is willing and clinically able to wash out of exclusion medications during the screening period unless previously approved by the Pediatric Off-Patent Drug Study (PODS) PI or a BPCA DCC Medical Monitor. Prior to the Day 0 visit, participants will have not used any of the following medications within the specified time frames:

   • Preceding 3 days: stimulants or melatonin
   • Preceding 1 week: all other psychotropic medications such as mood stabilizers, anticonvulsants, alpha-2 agonists, antipsychotics (other than aripiprazole) and/or antidepressants
   • Preceding 2 weeks: monoamine oxidase inhibitors, aripiprazole, fluoxetine or depot antipsychotics
   No stable participants will be asked to discontinue medications.
10. Electrocardiogram (ECG) and blood work including Complete Blood Count (CBC), electrolytes, etc showing no clinically significant abnormalities.

9.3.2. Exclusion Criteria

1. Participant who is clinically stable on current medication regimen for bipolar disorder.
2. A current or lifetime diagnosis of Schizophrenia or Schizoaffective Disorder, a Pervasive Developmental Disorder (Autism Screening Questionnaire score > 15), Anorexia Nervosa, Bulimia Nervosa, or Obsessive-Compulsive Disorder.
3. Current DSM-IV diagnosis of Substance Dependence.
4. Positive drug screen at initial screening and on retest 1-3 weeks later.
5. Participants with symptoms of mania that may be attributable to a general medical condition, or secondary to use of medications (e.g., corticosteroids).
6. Evidence of any serious, unstable neurological illness for which treatment under the auspices of this study would be contraindicated.
7. Any serious, unstable medical illness or clinically significant abnormal laboratory assessments that would adversely impact the scientific interpretability or unduly increase the risks of the protocol.
8. Current general medical condition including neurological disease, diabetes mellitus, thyroid dysfunction, or renal dysfunction that might be affected adversely by lithium, could influence the efficacy or safety of lithium, or would complicate interpretation of study results.
9. Evidence of current serious homicidal/suicidal ideation such that in the treating physician’s opinion it would not be appropriately safe for the participant to participate in this study.
10. Evidence of current active hallucinations and delusions such that in the treating physician’s opinion it would not be appropriately safe for the participant to participate in this study.
11. Concomitant prescription of over-the-counter medication or nutritional supplements (e.g., ibuprofen, naproxen, St John’s wort) that would interact with lithium or affect the participant’s physical or mental status.
12. Concomitant psychotherapy treatments provided outside the study initiated within 4 weeks prior to screening.
13. Previous adequate trial with Li⁺ (at least 4 weeks with Li⁺ serum levels between 0.8-1.2 mEq/L)
14. History of allergy to lithium or lithium intolerance
15. Psychiatric hospitalization within 1 month of screening for psychosis or serious homicidal/serious suicidal ideation
16. Clinician’s judgment that participant is not likely to be able to complete the study as an outpatient due to psychiatric reasons
17. Females who are currently pregnant or lactating
18. Sexually active females who, in the investigators’ opinion, are not using an adequate form of birth control.
19. Participants who are unable to swallow the study medication
20. Participants for whom a baseline YMRS score of < 20 is anticipated
21. Participants with an IQ less than 70 (determined using the Wechsler Abbreviated Scales of Intelligence [WASI] Vocabulary and Matrix Reasoning Subscales)

9.3.3. Removal of Participants from Therapy or Assessment

Participants were removed from the study if safety or study compliance concerns occurred. Participants who withdrew or were withdrawn were not replaced.

9.4.1. Treatments Administered

The starting dose of lithium was either 600 mg/day or 900 mg/day. Participants weighing less than 30 kg received 600 mg/day as their starting dose. All other participants began lithium therapy on 900 mg/day. Dose increases of 300mg/day occurred at study visits and phone calls until any one of the following criteria was met:

1. If the participant achieved a response as measured by a rating on the Clinical Global Improvement Scale (CGI-I) of ≤ 2 and ≥ 50% decrease in the Young Mania Rating Scale (YMRS)
2. If, in the investigator’s opinion, the participant experienced uncomfortable side effects that significantly impacted functioning and were of at least moderate severity (e.g., if the participant was falling asleep in school; had a significant tremor; had daytime enuresis or protracted nausea; or did not want to take lithium anymore).
3. If the serum lithium level was >1.4 mEq/L.
4. If the dose exceeded 40 mg/kg/day (with the exception of participants weighing less than 23 kg, who could receive up to 900 mg/day)

Participants assigned to placebo were randomized in equal numbers to a maximum dose of 20, 25, 30, 35, and 40 mg/kg/day, rounded up to the next highest dose level per placebo drug supply. The placebo minimum dose was based on the participant’s baseline weight (600 mg for <30 kg vs. 900 mg for >30 kg).

9.4.2. Identity of Investigational Products

The lithium carbonate was supplied by West-Ward Pharmaceutical Corporation. Each immediate release hard gelatin capsule with opaque flesh cap/opaque flesh body contained 300 mg of lithium carbonate. Inactive ingredients consisted of D&C Yellow #10, FD&C Green #3, FD&C Red #40, FD&C Yellow #6, Gelatin and Titanium Dioxide in the capsule shell. The printing ink contained: D&C Yellow #10, FD&C Blue #1, FD&C Blue #2, FD&C Red #40, n-Butyl Alcohol, Pharmaceutical Glaze, Propylene Glycol, SDA-3A Alcohol, and Synthetic Black Iron Oxide. -+

The placebo for Lithium Carbonate 300 mg capsules was supplied by West-Ward Pharmaceutical Corporation. Each capsule contained Anhydrous Lactose, Talc and the capsule shell for the 300 mg potency contained: D&C Yellow #10, FD&C Green #3, FD&C Red #40, FD&C Yellow #6, Gelatin and Titanium Dioxide. The printing ink contained D&C Yellow #10, FD&C Blue #1, FD&C Blue #2, FD&C Red #40, n-Butyl Alcohol, Pharmaceutical Glaze, Propylene Glycol, SDA-3A Alcohol, and Synthetic Black Iron Oxide.

9.4.3. Method of Assigning Participants to Treatment Groups

Participants enrolled into the Efficacy Phase were randomized to placebo or lithium in a 2:1 (lithium: placebo) allocation ratio. Participants who enrolled in the Long Term Efficacy Phase were treated with lithium. Participants who enrolled into the Discontinuation Phase were randomized in a 1:1 (lithium: placebo) allocation ratio. COLT1 Discontinuation Phase treatment assignments are included in Appendix 16.4.2.1. Participants who enrolled into the Restabilization Phase were treated with lithium.

Site staff were required to enter participant eligibility data for each study phase. For participants who were confirmed eligible for the Efficacy Phase and the Discontinuation Phase, treatment assignments were provided by the Data Coordinating Center’s internet data entry system (AdvantageEDC), which displayed the treatment assignment for the phase. The randomization process is further detailed in Appendix 16.1.7 (Randomization Information) and Appendix 16.1.9 (Statistical Analysis Plan).

9.4.4. Selection of Doses in the Study

Lithium doses were based on weight and increased to meet maximum tolerance; placebo doses were randomized as noted in Section 9.4.1.

9.4.5. Selection and Timing of Dose for Each Participant

The daily total dose of medication was to be divided into three doses (or 2 doses if total dose is 600 mg) throughout the day. Each dose was not to differ by more than 300 mg. Dose increases in capsule number began the day after the scheduled visit or phone call.

9.4.6. Blinding

The study personnel design during the blinded Efficacy and Discontinuation phases was based on the study design implemented by the Research Units in Pediatric Psychopharmacology (RUPP) research group (Scahill et al., 2001). There were two groups of clinicians and coordinators at each site that included a “blinded” team and an “unblinded” team. At a minimum, each team was composed of one child and adolescent psychiatrist and one site study coordinator.

To ensure the safety of the participants and to maintain the blind during the Efficacy and Discontinuation Phases, a blinded child and adolescent psychiatrist and a blinded study coordinator performed the study procedures during these randomized phases. The unblinded team assisted with randomization and monitored blood draws and the lithium blood levels of the participants.

The “unblinded” team performed study procedures during the open-label Long-Term Effectiveness Phase and the Restabilization Phase.

The Data Coordinating Center (DCC) also included a “blinded” team and an “unblinded” team, who worked with their respective site teams to provide data and protocol oversight. At a minimum, each team was composed of a protocol specialist, data manager, and statistician. The DCC blinded team provided support regarding protocol questions, procedures, and management of data unrelated to treatment information. The DCC unblinded team provided support regarding analysis, randomization, and management of the data related to treatment information, such as lithium blood levels and adverse events.

9.4.7. Prior and Concomitant Therapy

Medications that did not alter the pharmacokinetics of lithium could be prescribed for episodic or chronic use at the discretion of the treating physician. Medications that were not permitted during the study are listed in Table 9-1, and medications which were to be used with caution are listed in Table 9-2.

Table 9-1

Medications Not Permitted with Lithium.

Table 9-2

Medications to be Used with Caution.

During the study, lorazepam or hydroxyzine could be prescribed by the treating physician as a rescue medication for sleeplessness and agitation. Lorazepam was prescribed in 0.5 mg doses, not exceeding 1 mg/day, at doses at least 4 hours apart. Hydroxyzine was prescribed in 25 mg doses, not exceeding 50 mg/day, at doses at least 4 hours apart. Lorazepam and hydroxyzine were not prescribed or taken by participants within 12 hours of psychometric assessment and did not count toward the maximum of two allowable adjunctive medications.

If a participant was receiving psychotherapy at the time of study entry, therapy must have been initiated and been at a stable intensity more than 4 weeks prior to screening. No increase in intensity of sessions was allowed during the Efficacy Phase.

Participants in the Efficacy Phase could be prescribed a psychostimulant and/or melatonin (up to a maximum of 3mg at bedtime for insomnia) at the Week 4 visit. In order to begin treatment with a psychostimulant during the Efficacy Phase, the participant was required to be a “partial responder” (YMRS reduction 25-49% and CGI≤ 3 or YMRS reduction ≥ 50% and CGI-I =3) or “responder” (YMRS reduction of ≥ 50% and CGI-I 1 or 2 (with baseline as a reference point) when psychostimulant treatment was initiated. If a participant was prescribed a psychostimulant and/or melatonin during the Efficacy Phase, the participant was permitted to stay on either or both agents throughout the duration of the trial.

For participants who received active lithium during the Efficacy Phase, psychotherapy could be initiated outside of the study anytime during the first 8 weeks of the Long-Term Efficacy Phase and was required to be at stable intensity for the last 4 weeks of the Long-Term Efficacy Phase prior to enrollment into the Discontinuation Phase. For participants who received placebo during the Efficacy Phase, no increase in intensity of sessions outside of the study was allowed during the first 8 weeks of the Long-Term Efficacy Phase. Psychotherapy could be initiated outside of the study during weeks 8-16 of the Long-Term Efficacy Phase and was required to be at stable intensity for the last 4 weeks of Long-Term Efficacy Phase prior to entry into the Discontinuation Phase.

Participants could be treated with adjunctive psychotropic medication during the Long-Term Effectiveness Phase if there was a need to address symptoms for refractory psychosis; manic/mixed manic symptoms; or associated symptoms of depression, anxiety, and/or ADHD, according to the treatment management noted in Table 9-3. Psychostimulants could be initiated during the Long-Term Effectiveness Phase as noted in Table 9-3 for participants who were not “responders” in the Efficacy Phase, but who were either “partial” or “full responders” during the Long-Term Effectiveness Phase.

Table 9-3

COLT Reasonable Choices for the Management of Child and Adolescent BD.

Participants who were prescribed a psychostimulant, an antipsychotic, or an antidepressant medication during the Long-Term Effectiveness Phase were permitted to stay on those medications at the same dose throughout their participation in the Discontinuation and Restabilization Phases of this study. No dose changes to adjunctive medications or new adjunctive medications could be prescribed in the Discontinuation Phase.

During the Discontinuation Phase, initiation of psychotherapy outside of this study and any increase in intensity of sessions was not allowed.

During the Restabilization Phase, if the response to the re-institution of lithium was inadequate, adjunctive treatment as outlined in Table 9-3 was permissible. Psychotherapy outside of this study could be initiated or increased in intensity within the first 2 weeks of entry into the Restabilization Phase.

9.4.8. Treatment Compliance

Participants were asked to return lithium dosing diaries and study medication at each visit. Adherence was assessed by comparing the actual number of pills returned and the expected number of pills returned. Medication compliance was assessed by review of the lithium trough serum levels. Compliance was defined as taking ≥ 60% of medication in one week.

9.5.1. Efficacy and Safety Measurements Assessed and Flow Chart

The purpose of this study was to investigate the use of lithium for the treatment of pediatric participants with mania.

A chart of safety and psychometric assessment procedures are presented in Tables 9-5 and 9-6.

Table 9-5

Psychometric Assessment Procedures.

Table 9-6

General Assessment Procedures.

Table 9-4

Safety Assessment Procedures.

9.5.2. Appropriateness of Measurements

The efficacy or safety assessments used in this study are clinical laboratory tests and recognized as reliable and accurate. Laboratory, physical, and psychometric measurements are based on clinical study data.

9.5.3. Drug Concentration Measurements

Blood serum samples were analyzed for lithium concentrations by the site local laboratories’quantitative methods. In adults, the therapeutic range is between 0.8 and 1.2 mEq/L, with 1.5 mEq/Liter representing the lower limit of risk for intoxication (Amdisen, 1980). In the pediatric population, side effects have been reported at serum levels of 0.65 to 1.37 mEq/L (Hagino et al., 1995). Specimen collection time points are detailed in Section 9.5.1.

Lithium serum levels are summarized descriptively by visit separately for each study phase.

9.6.1. Site Monitoring Visits

During the study, monitoring visits to the trial sites were scheduled and conducted consistent with the BPCA DCC Dynamic Monitoring Plan to ensure that all aspects of the protocol were followed. Source documents were reviewed for verification of data collected on the Electronic Case Report Forms (eCRFs). The PODS Center and participating sites guaranteed access to source documents by BPCA DCC and the IRB, and were given access when required. Site investigators and relevant site personnel were available during monitoring visits, and they set aside sufficient time for the process.

9.6.2. Clinical Data Management

An eCRF was used to record participant data in the AdvantageEDC system. The eCRF was used for recording all historical participant information and study data as specified by the protocol. The eCRF was completed by designated and trained study personnel. Testing instruments had total scores, summary scores or sub-scales scores collected in the AdvantageEDC while the individual score elements were collected on provided source documents, and were not entered into the Advantage EDC system. However, for the primary endpoint, YMRS, each individual score elements was entered into the AdvantageEDC system.

Prior to the initiation of the study, the sponsor approved of the selected clinical sites that were experts in the field and fully capable of carrying out the study. The BPCA DCC trained the investigators, sub-investigators, and their study coordinators in eCRF completion, study management tools, and regulatory compliance. The study staff at ▬ (the primary clinical site) trained other site staff on the protocol and performance of study procedures. In addition to the investigators’ meeting, the study personnel at each site were trained on the study procedures by a CRA as necessary. CRAs from The Emmes Corporation monitored each site throughout the study. At each visit, data recorded in the EDC system were compared to source documentation to ensure accuracy. Quality assurance checks were performed to ensure that the investigators were complying with the protocol and all applicable regulations. The data for all primary outcomes were 100% monitored.

In addition to the quality assurance checks incorporated into the EDC system, additional logic checks were run to check for such items, such as inconsistent study dates and outlying laboratory values. Prior to study data lock, any necessary corrections were made by site staff in the eCRF and documented via audit trail.

9.7.1. Statistical and Analytical Plans

9.7.2. Determination of Sample Size

Sample size calculations and power estimates were conducted first and foremost to determine the number of participants needed for the Efficacy Phase of the study. Secondarily, potentially detectable differences in the primary efficacy outcome for the Discontinuation Phase were estimated given assumptions about the sample size expected for that phase of the trial.

Given the single-arm, open-label design and the descriptive and exploratory statistical objectives of the Long-Term Effectiveness and Restabilization Phases, formal sample size estimates, and power calculations were not conducted for those study phases.

A blinded sample size recalculation was conducted in September 2011. In response to the BPCA Data Monitoring Committee’s (DMC) recommendation, the sample size was re-estimated in January 2012 using the observed treatment effect. Futility and inefficacy analyses were also conducted in May 2012 and July 2012 at the request of the DMC. During each interim analysis, enrollment was continued based on the DMC recommendations.

9.8.1. Interim Safety Analysis on June 2011

The first Unblinded Safety Report for the Efficacy Phase was presented to the DMC at the June 29, 2011 meeting after enrollment of 25 participants (approximately 10% of the initially planned 225 participants). Documents related to this interim analysis are provided in Appendix 16.4.1. The DMC expressed serious concerns about study accrual. However, the study was allowed to continue because the DMC found no serious safety concerns with the study.

9.8.2. Blinded Sample Size Recalculation September 2011

The protocol specified a planned interim analysis when 75% of the participants had reached the end of the efficacy phase. The purpose of this analysis was to conduct a sample size re-estimation to ensure that the study had adequate power to detect a clinically important difference without unblinding the data. Per DMC’s request the BPCA DCC used the method specified in the protocol to evaluate pooled standard deviation of the primary endpoint with the current data in addition to using the COLT1 study standard deviation and to recalculate sample size based on updated estimate of the standard deviation.

Sample size for the Efficacy Phase of the COLT2 protocol was determined based upon the number of participants required to demonstrate a statistically significant difference in the mean change from baseline to Week 8/End-of-Phase in YMRS (Summary) scores between the two treatment arms, randomized 2:1 (lithium:placebo) based upon last observation carried forward (LOCF) values.

Estimated “true” within group sample variance was calculated using the following formula:

The original protocol sample size requirement of 225 was based on a difference of 5 points between lithium and placebo endpoints (delta of 5). The endpoint was the change in summary YMRS score from baseline to Week 8/End-of-Phase. The sample size assumed a SD of 12.0 and 83.5% power.

During Re-estimation, the sample size was recalculated (with results shown in Table 9-8) under the following alternatives:

• Protocol criteria (delta of 5 and SD of 12.0), but using a power of 80%.
• Delta of 5, the COLT1 study pooled standard deviation of the primary endpoint from Efficacy Phase of 8.8 and a power of 80%.
• Delta of 5, COLT2 study pooled standard deviation at the time of analysis (12.4) and a power of 80%.
• Original sample size calculations using a pooled standard deviation of 12.0 and power of 83.5%.

Table 9-8

Sample Size Recalculation.

The highlighted row is the COLT2 original protocol sample size requirement, based on a SD of 12.0 and 83.5% power. For a power of 80.3%, the sample size estimate was 207. Using the SD found for the COLT1 study (8.8) and a power of 80.9%, the total sample size decreased to 114. Using the current COLT2 standard deviation of 12.4, the total sample size was estimated to be 222 (for power of 80.5%). Documents related to this interim analysis are provided in Appendix 16.4.1.

The DMC made a recommendation to NICHD to conduct an interim analysis to assess the actual delta, compare it to the estimated delta, and re-estimate the sample size.

9.8.3. Response to DMC’s Recommendation to Re-estimate Sample Size Using Observed Treatment Effect on January 2012

Following the review of the re-estimated sample size calculations, the DMC recommended that the study’s sample size be re-estimated using the treatment effect and standard deviation observed in the data collected to date. The DMC proposed an interim assessment of futility using conditional power based method and an interim monitoring for efficacy be conducted.

The BPCA DCC responded with an alternative plan because sample size re-estimation based on un-blinded observed treatment effect estimates are generally not recommended in statistical literature due to multiple reasons, including:

1. High variability of treatment effect estimated early in the trial can lead to potentially misleading large or small treatment effect.
2. Re-estimation of sample size using an interim treatment effect will affect the overall Type I and Type II error rates of the clinical trial.
3. Controversy over appropriate statistical methods that should be used to adjust the Type I and II error rates.
4. Sample size should be chosen to detect a clinically meaningful treatment effect and the treatment effect estimated in the interim might not meet this criteria.

FDA guidance on adaptive designs recommends the use of group sequential designs to effectively decrease the sample size.

Given this background and continued concerns about enrollment into the COLT2 trial, the BPCA DCC recommended an interim assessment of futility using conditional power based method. Under this plan, as per DMC’s recommendation, the study would be stopped early if the conditional power to detect hypothesized treatment effect of 5 was below a pre-specified threshold. The assessment of conditional power for futility would lead to decrease in power of the study. However, the impact on power would be minimal, especially if the threshold chosen was small. In addition, following the recommendations of the FDA Guidance document on adaptive design, the BPCA DCC suggested implementing interim monitoring for efficacy. This plan would allow the study to be stopped early if the observed treatment effect was much larger than the hypothesized value. Documents related to this interim analysis are provided in Appendix 16.4.1. Amendment 6 of the Protocol, dated March 8, 2012, was adopted to reflect these changes in interim analyses.

9.8.4. Interim Safety Analysis on May 2012

The second Unblinded Safety Report on the Efficacy Phase was presented to the DMC at the May 11, 2012 meeting after the enrollment of 55 participants, with the planned safety analysis at 25% of targeted enrollment of 225. The DMC noted that there were no obvious safety concerns observed from the data, and recommended that enrollment should continue. Documents related to this interim analysis are provided in Appendix 16.4.1.

9.8.5. Interim Analysis for Futility and Efficacy on May 2012, Updated July 2012

The interim Analysis for Futility and Efficacy based on Protocol Amendment 6.0 was presented to the DMC at the May 11, 2012 meeting. The analyses were updated and presented again at the July 23, 2012 meeting, following correction of the YMRS data used for one participant.

As discussed above, the interim assessment of futility would have allowed the DMC to recommend stopping the study early if the conditional power to detect hypothesized treatment effect of 5 was below a pre-specified threshold. The interim assessment of efficacy would have allowed the study to stop early if the observed treatment effect was much larger than the hypothesized value.

Table 9-9

Distribution of YMRS Total Score and YMRS Change Score by Treatment.

Table 9-10

Distribution of Difference Between Treatments in YMRS Change Score.

9.8.5.1. Futility Analysis

The interim futility analysis based on conditional power (CP) was used as a guideline for early termination of the trial. Conditional power was defined as the probability that the final study result would be statistically significant, given the observed data and a specific assumption about the pattern of the data to be observed in the remainder of the study. The conditional power was calculated based on the primary efficacy endpoint (change from baseline to Week 8/LOCF in YMRS Total Score). This interim analysis was conducted after 54 participants reached the end of the efficacy phase. Conditional power was based on a target sample size of 100 participants. The DMC requested this target sample size given the rate of enrollment at the time of the interim analyses. A decision was made to stop the study (analysis of futility) and “accept” the null hypothesis (Section 9.7.2.1) if a conditional power calculated to detect the hypothesized treatment effect (a 5-point difference in YMRS change scores) fell below the pre-specified threshold of 40%, which means that if the computed conditional power fell below this value the chance of finding a significant result at the target sample size of 100 would be less than 40%.

The conditional power for some alternative effect using a critical value of Z_α/2 for a Type I error rate of α=0.05 was computed as:

$$CP\left[Z(1)\ge Z_{\alpha /2}\mid Z(t),\theta \right]=1-\text{Φ}\left(\mid Z_{\alpha /2}-Z(t)\sqrt{t}-\theta (1-t)\mid /\sqrt{(1-t)}\right)$$

; where t = n/N_o (N_o = 100 and n is total enrolled at the interim analysis, approximately 50).

The alternative θ is defined as:

$$\theta =1/\sqrt{2}\left(\frac{{\mu }_1-{\mu }_2}{\sigma }\right)\sqrt{N_o}$$

; where μ₁ and μ₂ are the mean change from baseline in YMRS score for the placebo and lithium arms and σ is the common standard deviation. Results of the calculation for the conditional power of the study at N=100 is shown in Table 9-11.

Table 9-11

Conditional Power (Futility) Analysis at N=100.

If the remaining data collected had a change in YMRS of 5 points and SD of 12 points the conditional power would be 93.9%. As this conditional power is above the selected 40% threshold, the decision was made not to stop the study due to futility.

9.8.5.2. Efficacy Analysis

Simultaneous with the conditional power analysis, the first interim analysis for efficacy was completed. At the start of the study, the DMC planned to review results from an interim analysis for efficacy two times during the course of the study (after approximately 50 and 100 of the participants have reached the end of the Efficacy Phase). For the purpose of sequential monitoring, the primary efficacy endpoint was the change in YMRS Total Score from baseline YMRS score to Week 8/LOCF.

The following group sequential test plan for efficacy outcome describes how alpha (α) was spent across the two interim looks and a final look at the data. This interim analysis was performed based on the primary efficacy outcome and was unadjusted for any covariates. The interim looks and a final look were based on the group sequential procedure. The symmetric O’Brien-Fleming (1979) boundary was used as it is very conservative (requires a large treatment effect to signal stopping) during the early analyses. The alpha level or Type I error in a non-sequential design was assigned to one (final) analysis. The goal of a group sequential design was to control the overall Type I error rate.

Table 9-12

Efficacy Analysis.

Using the total planned sample size of 225, data from the 54 participants provided approximately 24% of the total information for the study. The test statistic of 2.25 at the interim look was below the conservative O’Brien-Fleming bound of 4.4279 and, therefore, the study did not meet threshold to declare the observed differences to be statistically significant and the study was not stopped early.

Based on the interim analyses, the goal was to randomize 100 participants. However, the contract for the study ended in April 2013 and the period of performance could not be extended.

The study was closed to enrollment in December 2012, with 81 participants randomized, in order to meet the timeline for the administrative closure of the study in April 2013.

9.8.6. Final Safety Analysis on September 2013

The final Unblinded Safety Report on the Efficacy Phase was presented to the DMC after enrollment was closed at 81 participants. Documents related to this interim analysis are provided in Appendix 16.4.1.

9.9.1. Changes in Conduct of Study

There were six amendments implemented to the original protocol (dated April 8, 2009). The original protocol and the six amended protocols are provided in Appendix 16.1.1. No participants were screened or enrolled into the study under of the original protocol.

The six amendments to the protocol are summarized in this section.

9.9.2. Changes in Planned Analyses

The following changes were made to the analyses after finalization of the statistical analysis plan.

1)

The definition of Major Protocol Deviation was modified to include received incorrect randomized study medication. Review indicated that the major protocol deviation definition was too stringent and a second definition of major protocol deviations was also analyzed. This new definition revised the compliance criteria to missing >40% in two consecutive weeks and dose criteria to exceed 50 mg/kg/day (instead of 40 mg/kg/day). Other deviations remained the same.

2)

In the analyses of YMRS response status and YMRS improvement status during the Long-Term Effectiveness Phase, Discontinuation Phase and Restabilization Phase, baseline values were re-defined to be the pre-treatment day 1 Efficacy Phase value (as defined in item 7 below).

3)

Analyses by Efficacy Phase randomized treatment group were added to the Long-Term Effectiveness Phase. All tables are re-labeled with “A” for by age strata and “B” for by Efficacy Phase randomized treatment group.

4)

Tertiary Endpoint of Child Mania Rating Scale – Parent Version was not collected during the Long-Term Effectiveness, Discontinuation and Restabilization Phases and will not be included in the results.

5)

An additional sensitivity analysis of primary endpoint by treatment received was added to the Discontinuation Phase.

6)

CGI-Severity and CGI-Improvement for the Long-Term Effectiveness Phase, Discontinuation Phase and Restabilization Phases was defined as follows:

• CGI-Severity (Mania, Depression and Overall Illness):

  • Success: Baseline = 1 or 2 and less or same severity level at LOCF
    Baseline = 3 and less severe at LOCF
    Baseline = 4, 5, 6 and LOCF=1, 2, 3 and at least a 2 point decrease
  • Failure: Otherwise
• CGI-Improvement (Mania, Depression and Overall Illness):

  • Success: Baseline CGI-S = 1 or 2 and LOCF CGI-I = 1,2,3,4
    Baseline CGI-S = 3 and LOCF CGI-I = 1,2,3
    Baseline CGI-S = 4, 5, 6 and LOCF CGI-I = 1,2
  • Failure: Otherwise

P-value for the Discontinuation Phase will be calculated using Fisher’s Exact Test.

7)

Response, remission and improvement status were added to the Discontinuation Phase. Definition of response for Long-Term Effectiveness Phase, Discontinuation Phase and Restabilization Phase was revised to use pre-treatment Efficacy Phase day 1 as the baseline summary YMRS value and the CGI-I (overall illness) at LOCF was modified to the CGI-S (overall illness) at LOCF. See below:

• Response: YMRS reduction ≥ 50% from Efficacy Phase baseline and CGI-S (Overall illness) = 1 or 2 at LOCF
• Partial response: YMRS reduction 25-49% % from Efficacy Phase baseline and CGI-S (Overall illness) ≤ 3 at LOCF OR
  YMRS reduction ≥ 50% and CGI-S (Overall illness) = 3 at LOCF
• Non-response: YMRS reduction < 25% % from Efficacy Phase baseline or CGI-S (Overall illness) ≥ 4 at LOCF

8)

Due to low accrual, exploratory analyses for all phases were not performed.

9)

The adjusted and unadjusted effect size (cohen’s d) were added for the Efficacy Phase primary efficacy endpoint (Lipsey and Wilson, 2001).