Aripiprazole in the Maintenance Treatment of Bipolar Disorder: A Critical Review of the Evidence and Its Dissemination into the Scientific Literature

A systematic search of the literature reveals limited evidence to support use of aripiprazole, a second-generation antipsychotic medication, in maintenance therapy of bipolar disorder, despite widespread use.

Please refer to your supplemental new drug application (sNDA), referenced above, dated January 28,2004, received January 30, 2004 of the Federal Food, Drug, and Cosmetic Act for ABILIFY (aripiprazole) Tablets.
We also acknowledge receipt of your submissions dated March 23,2004, March 24,2004, March 25,2004, July 22,2004, August 27,2004, and November 5,2004 This supplemental application provides for addition of language to product labeling relating to the maintenance of efficacy in the treatment of patients with Bipolar 1 Disorder, with a recent manic or mixed episode, who have been stabilized for at least 6 consecutive weeks on aripiprazole monotherapy.
We have completed our review of this application as amended, and it is approvable. Before this application may be approved, however, you must address the following comments and/or deficiencies: Efficacy Although we consider this application approvable, and have included draft labeling with this letter, we have concerns about the strength of the data provided in Study CN13801O. In particular, we note that, although the study is a positive study by the protocol specified analysis, this result appears to be driven by the results in Center 093 (Dr. Ignacio Rosales, Mexico City, Mexico). When this center is removed from the primary analysis, statistical significance is lost. We recognize that removing the data from this center from the analysis is, from a strictly statistical perspective, problematic; however, the dependence of the statistical significance of the overall study on the results of this small center, in which the treatment difference differs markedly from that seen in the US centers (in which there are many more patients) raises questions about the reliability of the result.
We therefore ask you to address this concern.
In addition, we note that for a number of patients who discontinued prior to completion (and who were not considered to have met relapse criteria), we do not have a detailed account of the NDA 21-436 / S-005 Page 2 reasons for their early withdrawal. We acknowledge that you have provided a brief description of the reasons for each patient's early discontinuation, but in a number of cases, these descriptions are not sufficient for us to be able to confmn that the patient did not leave the trial because of worsening disease. The absence of a detailed understanding of the reasons why patients discontinued treatment in the maintenance phase ofthis study, therefore, in addition to the concerns raised above, also raises questions about the robustness of the result reported.
For this reason, we ask you to re-examine the data to better characterize the reason that each patient (at all centers, not just Center 093) not classified as having met relapse criteria discontinued treatment. Please provide us with a sufficiently detailed description of these reasons (e.g., a brief narrative) so that we may independently examine them (for example, a patient listed as having discontinued because of insomnia might actually have been experiencing clinical worsening of his or her bipolar disorder; further, if patients had a YMRS or MADRS score at the time of early exit, we would be interested in these data, as well as previous scores for such patients). If you determine that additional patients should have been classified as having met relapse criteria, we would expect you to re-analyze the data. In any event, even if you do not re-classify any patients, we are interested to know if any patients discontinued because of evidence of worsening of their condition.

Draft Labeling
In addition to changes related to Study CN13801O, you will note that the appended draft labeling includes new WARNING language on the risk of cerebrovascular adverse events (CVAEs) in elderly patients with dementia. We believe that the draft statement represents a fair description of the data you submitted on July 30, 2003 in response to our request of January 30, 2003.
We also note changes in the labeling (package insert) addressing the manufacturer, distributor and marketer of ABILIFY. These changes were approved under supplement 8-004 on July 23, 2003, and are included here.
In addition to the changes we have indicated in the attached labeling, all other previous revisions to labeling, as reflected in the most recently approved package insert, must be included. To facilitate review of your submission, please provide a highlighted or marked-up copy that clearly shows all changes. If additional information relating to the safety or effectiveness of this drug becomes available, further revision of the labeling may be required.

Phase 4 Commitments
Prior to fmal action on this supplement, it is necessary for us to have agreement in writing concerning all pertinent Phase 4 commitments for this application. Our approval action letter for S-002 (acute manic or mixed episodes associated with Bipolar Disorder) included the following Phase 4 commitments which we believe would adequately address post-approval data needs for longer-term treatment as well: 1. Clinical Efficacy and Safety: Adult clinical studies to address efficacy and safety of aripiprazole as add-on therapy in bipolar disorder.
-You have previously agreed to submit the results of both short and longer-term studies of the efficacy and safety of aripiprazole as add-on therapy in bipolar patients currently taking mood stabilizers (e.g., lithium, valproate).
-Final report submission for the longer-term study will occur on or before September 30, 2009.
-We consider this previous commitment adequate to address Phase 4 needs for both S-002 and S-005. No additional Phase 4 commitment is necessary for S-005 if the previously agreed-upon commitment is met.
2. Pharmacologyrroxicology: Juvenile animal toxicity study/ies to support pediatric studies of aripiprazole in bipolar disorder.
-You have previously agreed to conduct and submit results of a juvenile animal study or studies to support pediatric studies of aripiprazole in bipolar disorder.
-Final report(s) submission will occur on or before June 30, 2006.
-We consider this previous commit:trent adequate to address Phase 4 needs for both S-002 and S-005. No additional Phase 4 commitment is necessary for S-005 if the previously agreed-upon commitment is met.
In addition, we request the following new Phase 4 commitment: 3. Drug-Drug Interaction: Drug interaction studies with lithium and valproate.
-We are aware that drug interaction studies are underway or recently completed examining the interaction of aripiprazole with lithium and with valproate (two separate studies).
-Please propose a date or dates for submission of the final study reports.

Chemistry, Manufacturing, and Controls
We note your request for categorical exclusion from the environmental assessment requirements, as per 21 CFR 25.15 (d) and 21 CFR 25.31(a). We have reviewed this request, and it has been found acceptable. A categorical exclusion will be approved at the time of approval of the supplemental NDA.

Promotional Materials (Draft Format)
In addition, please submit three copies of the introductory promotional materials that yo u propose to use for this product in this indication. Submit all proposed materials in draft or mockup form, not final print. Send one copy to this Division, and two copies of both the promotional materials and the package insert directly to: receptors (Kt values of 44,15,39,57,and 61 nM,respectively), and moderate affInity for the serotonin reuptake site (Kt=98 nM). Aripiprazole has no appreciable affmity for cholinergic muscarinic receptors (ICso> I 000 nM). Aripiprazole functions as a partial agonist at the dopamine P2 and the serotonin 5-HTIA receptors, and as an antagonist at serotonin 5-HT 2A receptor.
The mechanism of action of aripiprazole, as with other drugs having effIcacy in schizophrenia and bipolar disorder, is unknown. However, it has been proposed that the effIcacy of aripiprazole is mediated through a combination of partial agonist activity at D2 and 5-HTIA receptors and antagonist activity at 5-HT2A receptors. Actions at receptors other than ~, 5-HTIA, and 5-HT2A may explain some of the other clinical effects of aripiprazole, eg, the orthostatic hypotension observed with aripiprazole may be explained by its antagonist activity at adrenergic alpha 1 receptors.
Pharmacokinetics ABILIFY (aripiprazole) activity is presumably primarily due to the parent drug, aripiprazole, and to a lesser extent, to its major metabolite, dehydro-aripiprazole, which has been shown to have affmities for D2 receptors similar to the parent drug and represents 40% of the parent drug exposure in plasma. The mean elimination half-lives are about 75 hours and 94 hours for aripiprazole and dehydro-aripiprazole, respectively. Steady-state concentrations are attained within 14 days of dosing for both active moieties. Aripiprazole accumulation is predictable from single-dose pharmacokinetics. At steady state, the pharmacokinetics of aripiprazole are dose-proportional. Elimination of aripiprazole is mainly through hepatic metabolism involving two P450 isozymes, CYP2D6 aIid CYP3A4.

Tablet
Aripiprazole is well absorbed after administration of the tablet, with peak plasma concentrations occurring within 3 to 5 hours; the absolute oral bioavailability of the tablet formulation is 87%. ABILIFY can be administered with or without food. Administration of a 15-mg ABILIFY tablet with a standard high-fat meal did not significantly affect the 20f38 Cmax or AVC of aripiprazole or its active metabolite, dehydro-aripiprazole, but delayed Tmax by 3 hours for aripiprazole and 12 hours for dehydro-aripiprazole.

Oral Solution
Aripiprazole is well absorbed when administered orally as the solution. At equivalent doses, the plasma concentrations of aripiprazole from the solution were higher than that from the tablet formulation. In a relative bioavailability study comparing the pharmacokinetics of 30 mg aripiprazole as the oral solution to 30 mg aripiprazole tablets in healthy subjects, the solution to tablet ratios of geometric mean Cmax and AVC values were 122% and 114%, respectively (see DOSAGE AND ADMINISTRATION.) The single-dose pharmacokinetics of aripiprazole were linear and dose-proportional between the doses of 5 to 30 mg.

Distribution
The steady-state volume of distribution of aripiprazole following intravenous administration is high (404 L or 4.9 Ukg), indicating extensive extravascular distribution. At therapeutic concentrations, aripiprazole and its major metabolite are greater than 99% bound to serum proteins, primarily to albumin. In healthy human volunteers administered 0.5 to 30 mg/day aripiprazole for 14 days, there was dose-dependent D2 receptor occupancy indicating brain penetration of aripiprazole in humans.

Metabolism and Elimination
Aripiprazole is metabolized primarily by three biotransformation pathways: dehydrogenation, hydroxylation, and N-dealkylation. Based on in vitro studies, CYP3A4 and CYP2D6 enzymes are responsible for dehydrogenation and hydroxylation of aripiprazole, and N-dealkylation is catalyzed by CYP3A4. Aripiprazole is the predominant drug moiety in the systemic circulation. At steady state, dehYdroaripiprazole, the active metabolite, represents about 40% of aripiprazole AVC in plasma.
Approximately 8% of Caucasians lack the capacity to metabolize CYP2D6 substrates and are classified as poor metabolizers (PM), whereas he rest are extensive metabolizers (EM). PMs have about an 80% increase in aripiprazole exposure and about a 30% decrease in exposure to the active metabolite compared to EMs, resulting in about a 60% higher exposure to the total active moieties from a given dose of aripiprazole compared to EMs. Coadministration of ABILIFY with known inhibitors of CYP2D6, like 30f38 quinidine in EMs, results in a 112% increase in aripiprazole plasma exposure, and dosing adjustment is needed (see PRECAUTIONS: Drug-Drug Interactiom). The mean elimination half-lives are about 75 hours and 146 hours for aripiprazole in EMs and PMs, respectively. Aripiprazole does not inhibit or induce the CYP2D6 pathway.
Following a single oral dose of [14C]-labeled aripiprazole, approximately 25% and 55% of the administered radioactivity was recovered in the urine and feces, respectively. Less than 1% of unchanged aripiprazole was excreted in the urine and approximately 18% of the oral dose was recovered unchanged in the feces.

Special Populations
In general, no dosage adjustment for ABILIFY is required on the basis of a. patient's age, gender, race, smoking status, hepatic function, or renal function (see DOSAGE AND ADMINISTRATION: Dosage in Special Populations). The phannacokinetics of aripiprazole in special populations are described below.

Hepatic Impairment
In a single-dose study (15 mg of aripiprazole) in subjects with varying degrees of liver cirrhosis (Child-Pugh Classes A, B, and C), the AVC of aripiprazole, compared to healthy subjects, increased 31 % in mild HI, increased 8% in moderate HI, and decreased 20% in severe HI. None of these differences would require dose adjustment.

Renal Impairment
In patients with severe renal impairment (creatinine clearance <30 mL/min), Cmax of aripiprazole (given in a single dose of 15 mg) and dehydro-aripiprazole increased by 36% and 53%, respectively, but AVC was 15% lower for aripiprazole and 7% higher for dehydro-aripiprazole. Renal excretion of both unchanged aripiprazole and dehydroaripiprazole is less than 1% of the dose. No dosage adjustment is required in subjects with renal impairment.

Elderly
In formal single-dose pharmacokinetic studies (with aripiprazole given in a single dose of 15 mg), aripiprazole clearance was 20% lower in elderly (~65 years) subjects compared to younger adult subjects (18 to 64 years). There was no detectable age effect, however, 40f38 in the population phannacokineticanalysis in schizophrenia patients. Also, the phannacokinetics of aripiprazole after multiple doses in elderly patients appeared similar to that observed in young, healthy subjects. No dosage adjustment is recommended for elderly patients (see PRECAUTIONS: Geriatric Use).

Gender
Cmax and AUC of aripiprazole and its active metabolite, dehydro-aripiprazole, are 30 to 40% higher in women than in men, and correspondingly, the apparent oral clearance of aripiprazole is lower in women. These differences, however, are largely explained by differences in body weight (25%) between men and women. No dosage adjustment is recommended based on gender.

Race
Although no specific pharmacokinetic study was conducted to investigate the effects of race on the disposition of aripiprazole, population phannacokinetic evaluation revealed no evidence of clinically significant race-related differences in the pharmacokinetics of aripiprazole. No dosage adjustment is recommended based on race.

Smoking
Based on studies utilizing human liver enzymes in vitro, aripiprazole is not a substrate for CYPIA2 and also does not undergo direct glucuronidation. Smoking should, therefore, not have an effect on the pharmacokinetics of aripiprazole. Consistent with these in vitro results, population pharmacokinetic evaluation did not reveal any significant pharmacokinetic differences between smokers and nonsmokers. No dosage adjustment is recommended based on smoking status. items in the P ANSS that rates seven negative symptoms of schizophrenia (blunted affect, emotional withdrawal, poor rapport, passive apathetic withdrawal, difficulty in abstract thinking, lack of spontaneity/flow of conversation, stereotyped thinking). The Clinical Global Impression (CGI) assessment reflects the impression of a skilled observer, fully familiar with the manifestations of schizophrenia, about the overall clinical state of the patient.
In a 4-week trial (n=414) comparing two fixed doses of ABILIFY (15 or 30 mg/day) and haloperidol (10 mg/day) to placebo, both doses of ABILIFY were superior to placebo in the P ANSS total score, P ANSS positive subscale, and CGI-severity score.
In addition, the 15-mg dose was superior to placebo in the PANSS negative subscale.
In a 4-week trial (n=404) comparing two fixed doses of ABILIFY (20 or 30 mg/day) and risperidone (6 mg/day) to placebo, both doses of ABILIFY were superior to placebo in the PANSS · total score, PANSS positive subscale, P ANSS negative subscale, and CGI-severity score.
In a 6-week trial (n=420) comparing three fixed doses of ABILIFY (10, 15, or 20 mg/day) to placebo, all three doses of ABILIFY were superior to placebo in the P ANSS total score, P ANSS positive subscale, and the P ANSS negative subscale.
In a fourth study, a 4-week trial (n=103) comparing ABILIFY in a range of 5 to 30 mg/day or haloperidol 5 to 20 mg/day to placebo, haloperidol was superior to placebo, in the Brief Psychiatric Rating Scale (BPRS), a multi-item inventory of general psychopathology traditionally used to evaluate the effects of drug treatment in psychosis, and in a responder analysis based on the CGI-severity score, the primary outcomes for that trial. ABILIFY was only significantly different compared to placebo in a responder analysis based on the CGI-severity score.
Thus, the efficacy of 15-mg, 20-mg, and 30-mg daily doses was established in two studies for each dose, whereas the efficacy of the lO-mg dose was established in one study. There was no evidence in any study that the higher dose groups offered any advantage over the lowest dose group.
An examination of population subgroups did not reveal any clear evidence of differential responsiveness on the basis of age, gender, or race.
A longer-term trial enrolled 3lO inpatients or outpatients meeting DSM-IV criteria for schizophrenia who were, by history, symptomatically stable on other 80f38 ·antipsychotic medications for periods of 3 months or longer. These patients were discontinued from their antipsychotic medications and randomized to ABILIFY 15 mg or placebo for up to 26 weeks of observation for relapse. Relapse during the double-blind phase was defined as CGI-Improvement score of ~5 (minimally worse), scores ~5 (moderately severe) on the hostility or uncooperativeness items of the P ANSS, or ~O% increase in the PANSS total score. Patients receiving ABILIFY 15 mg experienced a significantly longer time to relapse over the subsequent 26 weeks compared to those receiving placebo.

Bipolar Disorder
The efficacy of ABILIFY in the treatment of acute manic episodes was established in two 3-week, placebo-controlled trials in hospitalized patients who met 'the DSM-IV criteria for Bipolar I Disorder with manic or mixed episodes (in one trial, 21% of placebo and 42% of ABILIFY-treated patients had data beyond two weeks). These trials included patients with or without psychotic features and with or without a rapid-cycling course.
The primary instrument used for assessing manic symptoms was the Young Mania Rating Scale (Y-MRS), an II-item clinician-rated scale traditionally used to assess the degree of manic symptomatology (irritability, disruptive/aggressive behavior, sleep, elevated mood, speech, increased activity, sexual interest, language/thought disorder, thought content, appearance, and insight) in a range from 0 (no manic features) to 60 (maximum score). A key secondary instrument included the Clinical Global Impression -Bipolar (CGI-BP) scale.
In the two positive, 3-week, placebo-controlled trials (n=268; n=248) which evaluated ABILIFY 15 or 30 mg/day, once daily (with a starting dose of 30 mg/day), ABILIFY was superior to placebo in the reduction of V-MRS total score and CGI-BP Severity of Illness score (mania).
A trial was conducted in patients meeting DSM-IV criteria for Bipolar I Disorder with a recent manic or mixed episode who had been stabilized on open-label ABILIFY and who had maintained a clinical response for at least 6 weeks. The first phase of this trial was an open-label stabilization period in which inpatients and outpatients were clinically stabilized and then maintained on open-label ABILIFY (15 or 30 mg/day, with a starting dose of 30 mg/day) for at least 6 consecutive weeks. One hundred sixty-one outpatients were then randomized in a double-blind fashion, to either the same dose of ABILIFY they were on at the end of the stabilization and maintenance period or placebo 90f38 and were then monitored for manic or depressive relapse. During the randomization phase, ABILIFY was superior to placebo on time to the number of combined affective relapses (manic plus depressive), the primary outcome measure for this study. The majority of these relapses were due to manic rather than depressive symptoms. There is insufficient data to know whether Abilify is effective in delaying the time to occurrence of depression in patients with Bipolar I Disorder.
An examination of population subgroups did not reveal any clear evidence of differential responsiveness on the basis of age and gender; however, there were insufficient numbers of patients in each of the ethnic groups to adequately assess intergroup differences.
The efficacy of ABILIFY in maintaining efficacy in patients with Bipolar I Disorder with a recent manic or mixed episode who had been stabilized and then maintained for at least 6 weeks, was demonstrated in a double-blind, placebo-controlled trial. Prior to entering the double-blind, randomization phase of this trial, patients were clinically stabilized and maintained their stability for 6 consecutive weeks on ABILIFY. Following this 6-week maintenance phase, patients were randomized to either placebo or ABILIFY and monitored for relapse (see CLINICAL PHARMACOLOGY: Clinical Studies). Physicians who elect to use ABILIFY for extended periods, that is, longer than 6-weeks, should periodically re-evaluate the long-term usefulness of the drug for the individual patient (see DOSAGE AND ADMINISTRATION).

CONTRAINDICATIONS
ABILIFY is contraindicated in patients with a known hypersensitivity to the product.

WARNINGS Neuroleptic Malignant Syndrome (NMS)
A potentially fatal symptom complex sometimes referred to as Neuroleptic Malignant Syndrome (NMS) has been reported in association with administration of antipsychotic drugs, including aripiprazole. Two possible cases of NMS occurred during aripiprazole treatment in the premarketing worldwide clinical database. Clinical manifestations of NMS are hyperpyrexia, muscle rigidity, altered mental status, and evidence of autonomic instability (irregular pulse or blood pressure, tachycardia, diaphoresis, and cardiac dysrhythmia). Additional signs may include elevated creatine phosphokinase, myoglobinuria (rhabdomyolysis), and acute renal failure.
The diagnostic evaluation of patients with this syndrome is complicated. In arriving at a diagnosis, it is important to exclude cases where the clinical presentation includes both serious medical illness (eg, pneumonia, systemic infection, etc) and untreated or inadequately treated extrapyramidal signs and symptoms (EPS). Other important considerations in the differential diagnosis include central anticholinergic toxicity, heat stroke, drug fever, and primary central nervous system pathology.
The management of NMS should include: 1) immediate discontinuation of antipsychotic drugs and other drugs not essential to concurrent therapy; 2) intensive symptomatic treatment and medical monitoring; and 3) treatment of any concomitant II of38 serious medical problems for which specific treatments are available. There is no general agreement about specific pharmacological treatment regimens for uncomplicated NMS.
If a patient requires antipsychotic drug treatment after recovery from NMS, the potential reintroduction of drug therapy should be carefully considered. The patient should be carefully monitored, since recurrences ofNMS have been reported.

Tardive Dyskinesia
A syndrome of potentially irreversible, involuntary, dyskinetic movements may develop in patients treated with antipsychotic drugs. Although the prevalence of the syndrome appears to be highest among the elderly, especially elderly women, it is impossible to rely upon prevalence estimates to predict, at the inception of antipsychotio treatment, which patients are likely to develop the syndrome. Whether antipsychotic drug products differ in their potential to cause tardive dyskinesia is unknown.
The risk of developing tardive dyskinesia and the likelihood that it will become irreversible are believed to increase as the duration of treatment and the total cumulative dose of antipsychotic drugs administered to the patient increase. However, the syndrome can develop, although much less commonly, after relatively brief treatment periods at low doses.
There is no known treatment for established cases of tardive dyskinesia, although the syndrome may remit, partially or completely, if antipsychotic treatment is withdrawn. Antipsychotic treatment, itself, however, may suppress (or partially suppress) the signs and symptoms of the syndrome and, thereby, may possibly mask the underlying process. The effect that symptomatic suppression has upon the long-term course of the syndrome is unknown.
Given these considerations, ABILIFY should be prescribed in a manner that is most likely to minimize the occurrence of tardive dyskinesia. Chronic antipsychotic treatment should generally be reserved for patients who suffer from a chronic illness that (1) is known to respond to antipsychotic drugs, and (2) for whom alternative, equally effective, but potentially less harmful treatments are not available or appropriate. In patients who do require chronic treatment, the smallest dose and the shortest duration of treatment producing a satisfactory clinical response should be sought. The need for continued treatment should be reassessed periodically.
If signs and symptoms of tardive dyskinesia appear in a patient on ABILIFY, drug discontinuation should be considered. However, some patients may require treatment with ABILIFY despite the presence of the syndrome.

Cerebrovascular Adverse Events, Including Stroke, in Elderly Patients with Dementia
In placebo-controlled clinical studies (two flexible dose and one fixed dose study) of dementia-related psychosis, there was an increased incidence of cerebrovascular adverse events (eg, stroke, transient ischemic attack), including fatalities, in aripiprazole-treated patients (mean age: 84 years; range: 78-88 years). In the fixed-dose study, there was a statistically significant dose response relationship for cerebrovascular adverse events in patients treated with aripiprazole. Aripiprazole is not approved for the treatment of patients with dementia-related psychosis. (See also PRECAUTIONS: Use in Patients with Concomitant Illness: Safety Experience in Elderly Patients with Psychosis Associated with Alzheimer's Disease.)

Hyperglycemia and Diabetes Mellitus
Hyperglycemia, in some cases extreme and associated with ketoacidosis or hyperosmolar coma or death, has been reported in patients treated with atypical antipsychotics. There have been few reports of hyperglycemia in patients treated with ABILIFY. Although fewer patients have been treated with ABILIFY, it is not known if this more limited experience is the sole reason for the paucity of such reports. Assessment of the relationship between atypical antipsychotic use and glucose abnormalities is complicated by the possibility of an increased background risk of diabetes mellitus in patients with schizophrenia and the increasing incidence of diabetes mellitus in the general population. Given these confounders, the relationship between atypical antipsychotic use and hyperglycemia-related adverse events is not completely understood. However, epidemiological studies which did not include ABILIFY suggest an increased risk of treatment-emergent hyperglycemia-related adverse events in patients treated with the atypical antipsychotics included in these studies. Because ABILIFY was not marketed at the time these studies were performed, it is not known if ABILIFY is associated with this increased risk. Precise risk estimates for hyperglycemia-related adverse events in patients treated with atypical antipsychotics are not available.
Patients with an established diagnosis of diabetes mellitus who are started on atypical antipsychotics should be monitored regularly for worsening of glucose control. 130f38 · Patients with risk factors for diabetes mellitus (eg, obesity, family history of diabetes) who are starting treatment with atypical antipsychotics should undergo fasting blood glucose testing at the beginning of treatment and periodically during treatment. Any patient treated with atypical antipsychotics should be monitored for symptoms of hyperglycemia including polydipsia, polyuria, polyphagia, and weakness. Patients who develop symptoms of hyperglycemia during treatment with atypical antipsychotics should undergo fasting blood glucose testing. In some cases, hyperglycemia has resolved when the atypical antipsychotic was discontinued; however, some patients required continuation of anti -diabetic treatment despite discontinuation of the suspect drug.
The incidence of a significant orthostatic change in blood pressure (defined as a decrease of at least 30 mmHg in systolic blood pressure when changing from a supine to standing position) for aripiprazole was not statistically different from placebo (in schizophrenia: 14% among aripiprazole-treated patients and 12% among placebo-treated patients and in bipolar mania: 3% among aripiprazole-treated patients and 2% among placebo-treated patients).
Aripiprazole should be used with caution in patients with known cardiovascular disease (history of myocardial infarction or ischemic heart disease, heart failure or conduction abnormalities), cerebrovascular disease, or conditions which would predispose patients to hypotension (dehydration, hypovolemia, and treatment with antihypertensive medications).
140f38 Seizure Seizures occurred in 0.1 % (1/926) of aripiprazole-treated patients with schizophrenia in short-tenn, placebo-controlled trials. In short-tenn, placebo-controlled clinical trials of patients with bipolar mania, 0.3% (2/597) of aripiprazole-treated patients and 0.2% (1/436) of placebo-treated patients experienced seizures. As with other antipsychotic drugs, aripiprazole should be used cautiously in patients with a history of seizures or with conditions that lower the seizure threshold, eg, Alzheimer's dementia. Conditions that lower the seizure threshold may be more prevalent in a population of 65 years or older.

Potential for Cognitive and Motor Impairment
In short-tenn, placebo-controlled trials of schizophrenia, somnolence was reported in 11 % of patients on ABILIFY compared to 8% of patients on placebo; somnolence led to discontinuation in 0.1% (11926) of patients with schizophrenia on ABILIFY in shorttenn, placebo-controlled trials. In short-tenn, placebo-controlled trials of bipolar mania, somnolence was reported in 14% of patients on ABILIFY compared to 7% of patients on placebo, but did not lead to discontinuation of any patients with bipolar mania. Despite the relatively modest increased incidence of somnolence compared to placebo, ABILIFY, like other antipsychotics, may have the potential to impair judgment, thinking, or motor skills. Patients should be cautioned about operating hazardous machinery, including automobiles, until they are reasonably certain that therapy with ABILIFY does not affect them adversely.

Body Temperature Regulation
Disruption of the body's ability to reduce core body temperature has been attributed to antipsychotic agents. Appropriate care is advised when prescribing aripiprazole for patients who will be experiencing conditions which may contribute to an elevation in core body temperature, eg, exercising strenuously, exposure to extreme heat, receiving concomitant medication with anticholinergic activity, or being subject to dehydration.

Dysphagia
Esophageal dysmotility and aspiration have been associated with antipsychotic drug use. Aspiration pneumonia is a common cause of morbidity and mortality in elderly patients, in particular those with advanced Alzheimer's dementia. Aripiprazole and other 150f38 antipsychotic drugs should be used cautiously in patients at risk for aspiration pneumonia (see PRECAUTIONS: Use in Patients with Concomitant Illness).

Suicide
The possibility of a suicide attempt is inherent in psychotic illnesses and bipolar disorder, and close supervision of high-risk patients should accompany drug therapy. Prescriptions for ABILIFY should be written for the smallest quantity consistent with good patient management in order to reduce the risk of overdose.

Use in Patients with Concomitant Illness
Safety Experience in Elderly Patients with Psychosis Associated with Alzheimer's Disease: In a flexible dose (2 to 15 mg/day), 10-week, placebo-controlled study of aripiprazole in elderly patients (mean age: 81.5 years; range: 56 to 95 years) with psychosis associated with Alzheimer's dementia, 4 of 105 patients (3.8%) who received ABILIFY died compared to no deaths among 102 patients who received placebo during or within 30 days after termination of the double-blind portion of the study. Three of the patients (age 92, 91, and 87 years) died following the discontinuation of ABILIFY in the double-blind phase of the study (causes of death were pneumonia, heart failure, and shock). The fourth patient (age 78 years) died following hip surgery while in the doubleblind portion of the study. The treatment-emergent adverse events that were reported at an incidence of ~5% and having a greater incidence than placebo in this study were accidental injury, somnolence, and bronchitis. Eight percent of the ABILIFY-treated patients reported somnolence compared to one percent of placebo patients. In a small pilot, open-label, ascending-dose, cohort study (n=30) in elderly patients with dementia, ABILIFY was associated in a dose-related fashion with somnolence.
The safety and efficacy of ABILIFY in the treatment of patients with psychosis associated with dementia have not been established. If the prescriber elects to treat such patients with ABILIFY, vigilance should be exercised, particularly for the emergence of difficulty swallowing or excessive somnolence, which could predispose to accidental injury or aspiration. (See also WARNINGS: Cerebrovascular Adverse Events, Including Stroke, in Elderly Patients with Dementia.) Clinical experience with ABILIFY in patients with certain concomitant systemic illnesses (see CLINICAL PHARMACOLOGY: Special Populations: Renal Impairment and Hepatic Impairment) is limited. 160f38 ABILIFY has not been evaluated (J' used to any appreciable extent in patients with a recent history of myocardial infarction or unstable heart disease. Patients with these diagnoses were excluded from premarketing clinical studies.

Information for Patients
Physicians are advised to discuss the following issues with patients for whom they prescribe ABILIFY:

Interference with Cognitive and Motor Performance
Because aripiprazole may have the potential to impair judgment, thinking, or motor skills, patients should be cautioned about operating hazardous machinery, including automobiles, until they are reasonably certain that aripiprazole therapy does not affect them adversely.

Pregnancy
Patients should be advised to notify their physician if they become pregnant or intend to become pregnant during therapy with ABILIFY (aripiprazole).

Nursing
Patients should be advised not to breast-feed an infant if they are taking ABILIFY.

Concomitant Medication
Patients should be advised to inform their physicians if they are taking, or plan to take, any prescription or over-the-counter drugs, since there is a potential for interactions.

Alcohol
Patients should be advised to avoid alcohol while taking ABILIFY.

Heat Exposure and Dehydration
Patients should be advised regarding appropriate care in avoiding overheating and dehydration. 170f38

Sugar Content
Patients should be advised that each mL of ABILIFY oral solution contains 400 mg of sucrose and 200 mg of fructose.

Drug-Drug Interactions
Given the primary CNS effects of aripiprazole, caution should be used when ABILIFY is taken in combination with other centrally acting drugs and alcohol. Due to its aladrenergic receptor antagonism, aripiprazole has the potential to enhance the effect of certain antihypertensive agents.

Potential for Other Drugs to Affect ABILIFY
Aripiprazole is mt a substrate of CYPlAl, CYPlA2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, or CYP2El enzymes. Aripiprazole also does not undergo direct glucuronidation. This suggests that an interaction of aripiprazole with inhibitors or inducers of these enzymes, or other factors, like smoking, is unlikely.
Both CYP3A4 and CYP2D6 are responsible for aripiprazole metabolism. Agents that induce CYP3A4 (eg, carbamazepine) could cause an increase in aripiprazole clearance and lower blood levels. Inhibitors of CYP3A4 (eg, ketoconazole) or CYP2D6 (eg, quinidine, fiuoxetine, or paroxetine) can inhibit aripiprazole elimination and cause increased blood levels.
Ketoconazole: Coadministration ofketoconazole (200 mg/day for 14 days) with a l5-mg single dose of aripiprazole increased the AVC of aripiprazole and its active metabolite by 63% and 77%, respectively. The effect of a higher ketoconazole dose (400 mg/day) has not been studied. When concomitant administration of ketoconazole with aripiprazole occurs, aripiprazole dose should be reduced to one-half of its normal dose. Other strong inhibitors of CYP3A4 (itraconazole) would be expected to have similar effects and need similar dose reductions; weaker inhibitors (erythromycin, grapefruit juice) have not been studied. When the CYP3A4 inhibitor is withdrawn from the combination therapy, aripiprazole dose should then be increased.
Quinidine: Coadministration of a lO-mg single dose of aripiprazole with quinidine (166 mg/day for l3 days), a potent inhibitor ofCYP2D6, increased the AVC of aripiprazole by 112% but decreased the AVC of its active metabolite, dehydro-180f38 aripiprazole, by 35%. Aripiprazole dose should be reduced to one-half of its nonnaI dose when concomitant administration of quinidine with aripiprazole occurs. Other significant inhibitors of CYP2D6, such as fluoxetine or paroxetine, would be expected to have similar effects and, therefore, should be accompanied by similar dose reductions. When the CYP2D6 inhibitor is withdrawn from the combination therapy, aripiprazole dose should then be increased.
Carbamazepine: Coadministration of carbamazepine (200 mg BID), a potent CYP3A4 inducer, with aripiprazole . (30 mg QD) resulted in an approximate 70% decrease in . Cmax and AVC values of both aripiprazole and its active metabolite, dehydro-aripiprazole. When carbamazepine is added to aripiprazole therapy, aripiprazole dose should be doubled. Additional dose increases should be based on clinical evaluation. When carbamazepine is withdrawn from the combination therapy, aripiprazole dose should then be reduced.
No clinically significant effect of famotidine, valproate, or lithium was seen on the phannacokinetics of aripiprazole (see CLINICAL PHARMACOLOGY: Drug-Drug Interactions).

Potential for ABILIFY to Affect Other Drugs
Aripiprazole is unlikely to cause clinically important phannacokinetic interactions with drugs metabolized by cytochrome P450 enzymes. In in vivo studies, 10-to 30-mg/day doses of aripiprazole had no significant effect on metabolism by CYP2D6 (dextromethorphan), CYP2C9 (warfarin), CYP2C19 (omeprazole, warfarin), and CYP3A4 (dextromethorphan) substrates. Additionally, aripiprazole and dehydroaripiprazole did not show potential for altering CYPIA2-mediated metabolism in vitro (see CLINICAL PHARMACOLOGY: Drug-Drug Interactions).
Alcohol: There was no significant difference between aripiprazole coadministered with ethanol and placebo coadministered with ethanol on perfonnance of gross motor skills or stimulus response in healthy subjects. As with most psychoactive medications, patients should be advised to avoid alcohol while taking ABILIFY. 190f38

Carcinogenesis
Lifetime carcinogenicity studies were conducted in ICR mice and in Sprague-Dawley (SD) and F344 rats. Aripiprazole was administered for 2 years in the diet at doses of 1,3, 10, and 30 mg/kg/day to ICR mice and 1,3, and 10 mg/kg/day to F344 rats (0.2 to 5 and 0.3 to 3 times the maximum recommended human dose [MRHD] based on mg/m 2 , respectively). In addition, SD rats were dosed orally for 2 years at 10, 20, 40, and 60 mg/kg/day (3 to 19 times the MRHD based on mg/m 2 ). Aripiprazole did not induce tumors in male mice or rats. In female mice, the incidences of pituitary gland adenomas and mammary gland adenocarcinomas ani adenoacanthomas were increased at dietary doses of 3 to 30 mg/kg/day (0.1 to 0.9 times human exposure at MRHD based on AVC and 0.5 to 5 times the MRHD based on mg/m 2 ). In female rats, the incidence of mammary gland fibroadenomas was increased at a dietary dose of 10 mg/kg/day (0.1 times human exposure at MRHD based on AVC and 3 times the MRHD based on mg/m2); and the incidences of adrenocortical carcinomas and combined adrenocortical adenomas/carcinomas were increased at an oral dose of 60 mg/kg/day (14 urnes human exposure at MRHD based on AVC and 19 times the MRHD based on mg/m 2 ).
Proliferative changes in the pituitary and mammary gland of rodents have been observed following chronic administration of other antipsychotic agents and are considered prolactin-mediated. Serum prolactin was not measured in the aripiprazole carcinogenicity studies. However, increases in serum prolactin levels were observed in female mice in a 13-week dietary study at the doses associated with mammary gland and pituitary tumors. Serum prolactin was not increased in female rats in 4-and 13-week dietary studies at the dose associated with mammary gland tumors. The relevance for human risk of the findings of prolactin-mediated endocrine tumors in rodents is unknown.

Mutagenesis
The mutagenic potential of aripiprazole was tested in the in vitro bacterial reversemutation assay, the in vitro bacterial DNA repair assay, the in vitro forward gene mutation assay in mouse lymphoma cells, the in vitro chromosomal aberration assay in Chinese hamster lung (CHL) cells, the in vivo micronucleus assay in mice, and the unscheduled DNA synthesis assay in rats. Aripiprazole and a metabolite (2,3-DCPP) 200f38 were clastogenic in the in vitro chromosomal aberration assay in CHL cells with and without metabolic activation. The metabolite, 2,3-DCPP, produced increases in numerical aberrations in the in vitro assay in CHL cells in the absence of metabolic activation. A positive response was obtained in the in vivo micronucleus assay in mice, however, the response was shown to be due to a mechanism not considered relevant to humans.

Impairment of Fertility
Female rats were treated with oral doses of 2, 6, and 20 mg/kg/day (0.6, 2, and 6 times the maximum recommended human dose [MRHD] on a mg/m 2 basis) of aripiprazole from 2 weeks prior to mating through day 7 of gestation. Estrus cycle irregularities and increased corpora lutea were seen at all doses, but no impairment of fertility was seen. Increased pre-implantation loss was seen at 6 and 20 mg/kg, and decreased fetal weight was seen at 20 mg/kg.
Male rats were treated with oral doses of 20, 40, and 60 mg/kg/day (6, 13, and 19 times the MRHD on a mg/m 2 basis) of aripiprazole from 9 weeks prior to mating through mating. Disturbances in spermatogenesis were seen a 60 mg/kg, and prostate atrophy was seen at 40 and 60 mg/kg, but no impairment of fertility was seen.

Pregnancy Pregnancy Category C
In animal studies, aripiprazole demonstrated developmental toxicity, including possible teratogenic effects in rats and rabbits.
Pregnant rats were treated with oral doses of 3, 10, and 30 mg/kg/day (1, 3, and 10 times the maximum recommended human dose [MRHD] on a mg/nl basis) of aripiprazole during the period of organogenesis. Gestation was slightly prolonged at 30 mg/kg. Treatment caused a slight delay in fetal development, as evidenced by decreased fetal weight (30 mg/kg), undescended testes (30 mg/kg), and delayed skeletal ossification (10 and 30 mg/kg). There were no adverse effects on embryofetal or pup survival. Delivered offspring had decreased bodyweights (10 and 30 mg/kg), and increased incidences of hepatodiaphragmatic nodules and diaphragmatic hernia at 30 mg/kg (the other dose groups were not examined for these findings). (A low incidence of diaphragmatic hernia was also seen in the fetuses exposed to 30 mg/kg.) Postnatally, delayed vaginal opening was seen at 10 and 30 mg/kg and impaired reproductive 210f38 performance (decreased fertility rate, corpora lutea, implants, and live fetuses, and increased post-iInplantation loss, likely mediated through effects on female offspring) was seen at 30 mg/kg. Some maternal toxicity was seen at 30 mg/kg, however, there was no evidence to suggest that these developmental effects were secondary to maternal toxicity.
Pregnant rabbits were treated with oral doses of 10, 30, and 100 mg/kg/day (2, 3, and 11 times human exposure at MRHD based on Ave and 6, 19, and 65 times the MRHD based on mg/m 2 ) of aripiprazole during the period of organogenesis. Decreased maternal food consumption and increased abortions were seen at 100 mg/kg. Treatment caused increased fetal mortality (100 mglkg), decreased fetal weight (30 and 100 mg/kg), increased incidence of skeletal abnormality (fused sternebrae at 30 and 100 mg/kg) and minor skeletal variations (100 mg/kg).
In a study in which rats were treated with oral doses of 3, 10, and 30 mg/kg/day (1, 3, and 10 times the MRHD on a mg/m 2 basis) of aripiprazole perinatally and postnatally (from day 17 of gestation through day 21 postpartum), slight maternal toxicity and slightly prolonged gestation were seen at 30 mg/kg. An increase in stillbirths, and decreases in pup weight (persisting into adulthood) and survival, were seen at this dose.
There are no adequate and well-controlled studies in pregnant women It is not known whether aripiprazole can cause fetal harm when administered to a pregnant woman or can affect reproductive capacity. Aripiprazole should be used during pregnancy only if the potential benefit outweighs the potential risk to the fetus.

Labor and Delivery
The effect of aripiprazole on labor and delivery in humans is unknown.

Nursing Mothers
Aripiprazole was excreted in milk of rats during lactation. It is not known whether aripiprazole or its metabolites are excreted in human milk. It is recommended that women receiving aripiprazole should not breast-feed.

Pediatric Use
Safety and effectiveness in pediatric and adolescent patients have not'been established.

Geriatric Use
Of the 7951 patients treated with aripiprazole in premarketing clinical trials, 991 (12%) were ~65 years old and 789 (10%) were ~75 years old. The majority (88%) of the 991 patients were diagnosed with dementia of the Alzheimer's type.
Placebo-controlled studies of aripiprazole in schizophrenia or bipolar mania did not inchne sufficient numbers of subjects aged 65 and over to determine whether they respond differently from younger subjects. There was no effect of age on the pharmacokinetics of a single 15-mg dose of aripiprazole. Aripiprazole clearance was decreased by 20% in elderly subjects (~65 years) compared to younger adult subjects (18 to 64 years), but there was no detectable effect of age in the population pharmacokinetic analysis in schizophrenia patients.
Studies of elderly patients with psychosis associated with Alzheimer's disease have suggested that there may be a different tolerability profile in this population compared to younger patients with schizophrenia (see PRECAUTIONS: Use in Patients with Concomitant Illness). The safety and efficacy of ABILIFY in the treatment of patients with psychosis associated with Alzheimer's disease has not be~n established. If the prescriber elects to treat such patients with ABILIFY, vigilance should be exercised.

ADVERSE REACTIONS
Aripiprazole has been evaluated for safety in 7951 patients who participated in multipledose, premarketing trials in schizophrenia, bipolar mania, and dementia of the Alzheimer's type, and who had approximately 5235 patient-years of exposure. A total of 2280 aripiprazole-treated patients were treated for at least 180 days and 1558 aripiprazole-treated patients had at least 1 year of exposure.
The conditions and duration of treatment with aripiprazole included (in overlapping categories) double-blind, comparative and noncomparative open-label studies, inpatient and outpatient studies, fixed-and flexible-dose studies, and short-and longer-term exposure.
Adverse events during exposure were obtained by collecting volunteered adverse events, as well as results of physical examinations, vital signs, weights, laboratory analyses, and ECG. Adverse experiences were recorded by clinical investigators using terminology of their own choosing. In the tables and tabulations that follow, modified 230f38 COSTART dictionary terminology has been used initially to classify reported adverse events into a smaller number of standardized event categories, in order to provide a meaningful estimate of the proportion of individuals experiencing adverse events.
The stated frequencies of adverse events represent the proportion of individuals who experienced at least once, a treatment-emergent adverse event of the type listed. An event was considered treatment emergent if it occurred for the first time or worsened while receiving therapy following baseline evaluation. There was no attempt to use investigator causality assessments; ie, all reported events are included.
The prescriber should be aware that the figures in the tables and tabulations cannot be used to predict the incidence of side effects in the course of usual medical practice where patient characteristics and other factors differ from those that pr~vailed in the clinical trials. Similarly, the cited frequencies cannot be compared with figures obtained from other clinical investigations involving different treatment, uses, and investigators. The cited figures, however, do provide the prescribing physician with some basis for estimating the relative contribution of drug and nondrug factors to the adverse event incidence in the population studied.

Adverse Findings Observed in Short-Term, Placebo-Controlled Trials of Patients with Schizophrenia
The following findings are based on a pool of five placebo-controlled trials (four 4-week and one 6-week) in which aripiprazole was administered in doses ranging from 2 to 30 mg/day.

Adverse Events Associated with Discontinuation of Treatment in Short-Term, Placebo-Controlled Trials
Overall, there was no difference in the incidence of discontinuation due to adverse events between aripiprazole-treated (7%) and placebo-treated (9%) patients. The types of adverse events that led to discontinuation were similar between the aripiprazole and placebo-treated patients . .

Adverse Findings Observed in Short-Term, Placebo-Controlled Trials of Patients with Bipolar Mania
The following findings are based on a pool of 3-week, placebo-controlled, bipolar mania trials in which aripiprazole was administered at doses of 15 or 30 mg/day. 24of38

Adverse Events Associated with Discontinuation of Treatment In Short-Term, Placebo-Controlled Trials
Overall, in patients with bipolar mania, there was no difference in the incidence of discontinuation due to adverse events between aripiprazole-treated (11%) and placebotreated (9%) patients. The types of adverse events that led to discontinuation were similar between the aripiprazole and placebo-treated patients.

Commonly Observed Adverse Events in Short-Term, Placebo-Controlled Trials of Patients with Bipolar Mania
Commonly observed adverse events associated with the use of aripiprazole in patients with bipolar mania (incidence of 5% or greater and aripiprazole incidence at least twice that for placebo) are shown in Table 1. There were no adverse events in the short-term trials of schizophrenia that met these criteria. Adverse Events Occurring at an Incidence of 2% or More Among Aripiprazole-Treated Patients and Greater than Placebo in Short-Term, Placebo-Controlled Trials Table 2 enumerates the pooled incidence, rounded to the nearest percent, of treatmentemergent adverse events that occurred during acute therapy (up to 6 weeks in schizophrenia and up to 3 weeks in bipolar mania), including only those events that occurred in 2% or more of patients treated with aripiprazole (doses ~2 mg/day) and for which the incidence in patients treated with aripiprazole was greater than the incidence in patients treated with placebo in the combined dataset. 250f38 260f38 An examination of population subgroups did not reveal any clear evidence of differential adverse event incidence on the basis of age, gender, or race.

Schizophrenia
Dose response relationships for the incidence of treatment-emergent adverse events were evaluated from four trials in patients with schizophrenia comparing various fixed doses (2, 10, 15, 20, and 30 mg/day) of aripiprazole to placebo. This analysis, stratified by study, indicated that the only adverse event to have a possible dose response relationship, and then most prominent only with 30 mg, was somnolence (placebo,7.7%;15 mg,8.7%;20 mg,7.S%;30 mg,IS.3%).

Extrapyramidal Symptoms
In the short-term, placebo-controlled trials of schizophrenia, the incidence· of reported EPS for aripiprazole-treated patients was 6% vs. 6% for placebo. In the short-term, placebo-controlled trials in bipolar mania, the incidence of reported EPS-related events excluding events related to akathisia for aripiprazole-treated patients was 17% vS. 12% for placebo. In the short-term, placebo-controlled trials in bipolar mania, the incidence of akathisia-related events for aripiprazole-treated patients was IS% vs. 4% for placebo. Objectively collected data from those trials was collected on the Simpson Angus Rating Scale (for EPS), the Barnes Akathisia Scale (for akathisia) and the Assessments of Involuntary Movement Scales (for dyskinesias). In the schizophrenia trials, the objectively collected data did not show a difference between aripiprazole and placebo, with the exception of the Barnes Akathisia Scale (aripiprazole, 0.08; placebo, -O.OS). In the bipolar mania trials, the Simpson Angus Rating Scale and the Barnes Akathisia Scale showed a significant difference between aripiprazole and placebo (aripiprazole, 0.61; placebo, 0.03 and aripiprazole, 0.25; placebo, -0.06). Changes in the Assessments of Involuntary Movement Scales were similar for the aripiprazole and placebo groups.
Similarly, in a long-term (26-week), placebo-controlled trial of schizophrenia, objectively collected data on the Smpson Angus Rating Scale (for EPS), the Barnes Akathisia Scale (for akathisia), and the Assessments of Involuntary Movement Scales (for dyskinesias) did not show a difference between aripiprazole and placebo. 27of38

Laboratory Test Abnormalities
A between group comparison for 3to 6-week, placebo-controlled trials revealed no medically important differences between the aripiprazole and placebo groups in the proportions of patients experiencing potentially clinically significant changes in routine serum chemistry, hematology, or urinalysis parameters. Similarly, there were no aripiprazole/placebo differences in the incidence of dis continuations for changes in serum chemistry, hematology, or urinalysis.
In a long-term (26-week), placebo-controlled trial there were no medically important differences between the aripiprazole and placebo patients in the mean change from baseline in prolactin, fasting glucose, triglyceride, HDL, LDL, and total cholesterol measurements.

Weight Gain
In 4-to 6-week trials in schizophrenia, there was a slight difference in mean weight gain between aripiprazole and placebo patients (+0.7 kg vs. -0.05 kg, respectively), and also a difference in the proportion of patients meeting a weight gain criterion of ~7% of body weight [aripiprazole (8%) compared to placebo (3%)]. In 3-week trials in mania, the mean weight gain for aripiprazole and placebo patients was 0.0 kg vs. -0.2 kg, respectively. The proportion of patients meeting a weight gain criterion of ~7% of body weight was aripiprazole (3%) compared to placebo (2%). Table 3 provides the weight change results from a long-term (26-week), placebocontrolled study of aripiprazole, · both mean change from baseline and proportions of patients meeting a weight gain criterion of ~7% of body weight relative to baseline, categorized by BMI at baseline:  Table 4 provides the weight change results from a long-term (52-week) study of aripiprazole, both mean change from baseline and proportions of patients meeting a weight gain criterion of ;:::7% of body weight relative to baseline, categorized by BMI at baseline: Between group comparisons for a pooled analysis of placebo-controlled trials in patients with schizophrenia or bipolar mania, revealed no significant differences between aripiprazole and placebo in the proportion of patients experiencing potentially important changes in ECG parameters. Aripiprazole was associated with a median increase in heart rate of 5 beats per minute compared to a 1 beat per minute increase among placebo patients.

Additional Findings Observed in Clinical Trials
Adverse Events in Long-Term, Double-Blind, Placebo-Controlled Trials The adverse events reported in a 26-week,' double-blind trial comparing ABILIFY and placebo in patients with schizophrenia were generally consistent with those reported in the short-term, placebo-controlled trials, except for a higher incidence of tremor [9% (13/153) for ABILIFY vs. 1% (2/153) for placebo]. In this study, the majority of the cases of tremor were of mild intensity (9/13 mild and 4/13 moderate), occurred early in therapy (9/13 ~9 days), and were of limited duration (9/13 ~10 days). Tremor infrequently led to discontinuation «1%) of ABILIFY. In addition, in a long-term (52week), active-controlled study, the incidence of tremor for ABILIFY was 4% (34/859). A similar adverse event profile was observed in a long-term study in bipolar disorder. 290f38

Other Adverse Events Observed During the Premarketing Evaluation of Aripiprazole
Following is a list of modified COSTART tenns that reflect treatment-emergent adverse events as defined in the introduction to the ADVERSE REACTIONS section reported by patients treated with aripiprazole at mUltiple doses ~2 mg/day during any phase of a trial within the database of 7951 patients. All reported events are included except those already listed in Table 2, or other parts of the ADVERSE REACTIONS section, those considered in the WARNINGS or PRECAUTIONS, those event tenns which were so general as to be uninfonnative, events reported with an incidence of :::;;0.05% and which did not have a substantial probability of being acutely life-threatening, events that are otherwise common as background events, and events considered unlikely to be drug related. It is important to emphasize that, although the events reported occurred during treatment with aripiprazole, they were not necessarily caused by it.
Events are further categorized by body system and listed in order of decreasing frequency according to the following definitions: frequent adverse events are those occurring in at least 11100 patients (only those not already listed in the tabulated results from placebo-controlled trials appear in this listing); infrequent adverse events are those occurring in 11100 to 1/1000 patients; rare events are those occurring in fewer than 111000 patients.
Body as a Whole: Frequentflu syndrome, fever, chest pain, rigidity (including neck and extremity), neck pain, pelvic pain; Infrequentface edema, suicide attempt, malaise, migraine, chills, photosensitivity, tightness (including abdomen, back, extremity, head, jaw, neck, and tongue), jaw pain, bloating, enlarged abdomen, chest tightness, throat pain; Raremoniliasis, head heaviness, throat tightness, Mendelson's syndrome, heat stroke. trials did not reveal any tendency for any drug-seeking behavior, these observations were not systematic and it is not possible to predict on the basis of this limited experience the extent to which a CNS-active drug will be misused, diverted, and/or abused once marketed. Consequently, patients should be evaluated carefully for a history of drug abuse, and such patients should be observed closely for signs of ABILIFY misuse or abuse (eg, development of tolerance, increases' in dose, drug-seeking behavior).

Human Experience
In clinical studies, accidental or intentional acute overdosage of aripiprazole was identified in patients with estimated doses up to 1080 mg with no fatalities. The reported signs and symptoms observed with aripiprazole overdose included nausea, vomiting, asthenia; diarrhea, and somnolence. In the patients who were evaluated in hospital settings, there were no reported observations indicating clinically significant adverse change in vital signs, laboratory assessments, or ECG.
During postmarketing experience, the reported signs and symptoms observed in adult patients who overdosed with aripiprazole alone at doses up to 450 mg included tachycardia. In addition, reports of accidental overdose with aripiprazole (up to 195 mg) in children have been received. The potentially medically serious signs and symptoms reported include extrapyramidal symptoms and transient loss of consciousness with recovery.

Management of Overdosage
No specific information is available on the treatment of overdose with aripiprazole. An electrocardiogram should be obtained in case of overdosage and, if QTc interval prolongation is present, cardiac monitoring should be instituted. Otherwise, management of overdose should concentrate on supportive therapy, maintaining an adequate airway, oxygenation and ventilation, and management of symptoms. Close medical supervision and monitoring should continue until the patient recovers.
Charcoal: In the event of an overdose of ABILIFY, an early charcoal administration may be useful in partially preventing the absorption of aripiprazole. Administration of 50 g of activated charcoal, one hour after a single I5-mg oral dose of aripiprazole, decreased the mean AVC and Cmax of aripiprazole by 50%. 330f38 Hemodialysis: Although there is no infonnation on the effect of hemodialysis in treating an overdose with aripiprazole, hemodialysis is unlikely to be useful in overdose management since aripiprazole is highly bound to plasma proteins.

Schizophrenia Usual Dose
The recommended starting and target dose for ABILIFY is 10 or 15 mg/dayadministered on a once-a-day schedule without regard to meals. ABILIFY has been systematically evaluated and srown to be effective in a dose range of 10 to 30 mg/day, when administered as the tablet formulation, however, doses higher than 10 or 15 mg/day, the lowest doses in these trials, were not more effective than 10 or 15 mg/day. Dosage increases should not be made before 2 weeks, the time needed to achieve steady state.

Dosage in Special Populations
Dosage adjustments are not routinely indicated on the basis of age, gender, race, or renal or hepatic impairment status (see CLINICAL PHARMACOLOGY: Special Populations ).
Dosage adjustment for patients taking aripiprazole concomitantly with potential CYP3A4 inhibitors: When concomitant administration of ketoconazole with aripiprazole occurs, aripiprazole dose should be reduced to one-half of the usual dose. When the CYP3A4 inhibitor is withdrawn from the combination therapy, aripiprazole dose should then be increased.
Dosage adjustment for patients taking aripiprazole concomitantly with potential CYP2D6 inhibitors: When concomitant administration of potential CYP2D6 inhibitors such as quinidine, fluoxetine, or paroxetine with aripiprazole occurs, aripiprazole dose should be reduced at least to one-half of its normal dose. When the CYP2D6 inhibitor is withdrawn from the combination therapy, aripiprazole dose should then be increased.
Dosage adjustment for patients taking potential CYP3A4 inducers: When a potential CYP3A4 inducer such as carbamazepine is added to aripiprazole therapy, the aripiprazole dose should be doubled (to 20 or 30 mg). Additional dose increases should 34of38 be based on clinical evaluation.
When carbamazepine is withdrawn from the combination therapy, the aripiprazole dose should be reduced to 10 to 15 mg.

Maintenance Therapy
While there is no body of evidence available to answer the question of how long a patient treated with aripiprazole should remain on it, systematic evaluation of patients with schizophrenia who had been symptomatically stable on other antipsychotic medications for periods of 3 months or longer, were discontinued from those medications, and were then administered ABILIFY 15 mg/day and observed for relapse during a period of up to 26 weeks, demonstrated a benefit of such maintenance treatment (see CLINICAL PHARMACOLOGY: Clinical Studies). Patients should be periodically reassessed to determine the need for maintenance treatment.

Switching from Other Antipsychotics
There are no systematically collected data to specifically address switching patients with schizophrenia from other antipsychotics to ABILIFY or concerning concomitant administration with other antipsychotics. While immediate discontinuation of the previous antipsychotic treatment may be acceptable for some patients with schizophrenia, more gradual discontinuation may be most appropriate for others. In all cases, the period of overlapping antipsychotic administration should be minimized.

Bipolar Disorder Usual Dose
In clinical trials, the starting dose was 30 mg given once a day. A dose of30 mg/day was found to be effective when administered as the tablet formulation. Approximately 15% of patients had their dose decreased to 15 mg based on assessment of tolerability. The safety of doses above 30 mg/day has not been evaluated in clinical trials.

Dosage in Special Populations
See Dosage in Special Populations under DOSAGE AND ADMINISTRATION: Schizophrenia. 350f38

Maintenance Therapy
While there is no body of evidence available to answer the question of how long a patient treated with aripiprazole should remain on it, patients with Bipolar I Disorder who had been symptomatically stable on ABILIFY Tablets (15 mg/day or 30 mg/day with a starting dose of 30 mg/day) for at least 6 consecutive weeks and then randomized to ABILIFY Tablets (15 mg/day or 30 mg/day) or placebo and monitored for relapse, demonstrated a benefit of such maintename treatment (see CLINICAL PHARMACOLOGY: Clinical Studies).
While it is generally agreed that pharmacological treatment beyond an acute response in mania is desirable, both for maintenance of the initial response and for prevention of new manic episodes, there are no systematically obtained data to support the use of aripiprazole in such longer-term treatment (ie, beyond 6 weeks).

Oral Solution
The oral solution can be given on a mg-per-mg basis in place of the 5-, 10-, 15-, or 20-mg tablet strengths. Solution doses can be substituted for the tablet doses on a mg-per-mg basis up to 25 mg of the tablet. Patients receiving 30-mg tablets should receive 25 mg of the solution (see CLINICAL PHARMACOLOGY: Pharmacokinetics).

ANIMAL TOXICOLOGY
Aripiprazole produced retinal degeneration in albino rats in a 26-week chronic toxicity study at a dose of 60 mg/kg and in a 2-year carcinogenicity study at doses of 40 and 60 mg/kg. The 40-and 60-mg/kg doses are 13 and 19 times the maximum recommended human dose (MRHD) based on mg/nf and 7 to 14 times human exposure at MRHD based on AVe. Evaluation of the retinas of albino mice and of monkeys did not reveal evidence of retinal degeneration. Additional studies to further evaluate the mechanism have not been performed. The relevance of this finding to human risk is unknown.

Bottles of 30
NDC 59148-010-13 Blister of 100 NDC 59148-010-35 The 30-mg ABILIFY tablets are pink, round tablets, debossed on one side with "A-OIl" and "30", As the clinical team has described in detail, the single study submitted by the sponsor in support of the effectiveness of aripiprazole as maintenance treatment in patients with Bipolar I Disorder required patients to have been considered stable on aripiprazole treatment at 30 mg/day (or less if not tolerated) for at least 4 consecutive visits separated by 2 weeks (a minimum of 6 weeks of stability). At that point, patients were randomized to continued treatment on aripiprazole or placebo, for up to 26 weeks. The primary outcome was the time to relapse, appropriately defined.
The following chart displays patient flow in the randomized phase: The p-value for the drug-placebo difference (log-rank test) was 0.02.

Dr.
He performed an analysis in which he considered all patients who discontinued. the trial early (other than those classified as having relapsed) as having relapsed: the p-value for the drug-placebo contrast was 0.06. At my request, he also performed an analysis in which patients who discontinued due to an adverse event were considered to have met relapse criteria; the p-value was 0.09.
He also performed analyses of time to manic relapse (p=0.008) and time to depressive relapse (p=0.68).
The primary finding of interest, to which all clinical reviewers allude, is represented in the following charts: Clearly, based on these retrospective analyses, the study loses significance when the data from the very small Mexican center 093 is excluded.
As noted earlier, Dr. Marc Stone has reviewed additional safety data for aripiprazole.
Specifically, based on findings of increased risk of cerebrovascular adverse events with risperidone and olanzapine in patients with dementia, the division had asked sponsors of atypical antipsychotic drug products to examine their own databases for any similar potential risk. Otsuka has responded to this request with data from three controlled trials in patients with psychosis associated with Alzheimer's Disease.
Dr. Stone has reviewed these data in detail. Two of the studies (Studies 005 and 006) examined flexible doses of aripiprazole (from 2-15 mg/day; I do not know what the distribution of actual doses was in these studies) , and in one (Study 004), patients were randomized to placebo or aripiprazole 2, 5, or 10 mg/day.
In these three controlled trials, a total of 343 patients were randomized to placebo, and 595 were randomized to aripiprazole (Study 004, N=480; Study 005, N=251; Study 006, N=207). The following chart displays the comparisons between the risk of CVAEs with aripiprazole and placebo: Risk across all three randomized trials In the 2 mg group, the event was reported as a cerebrovascular accident (CVA). In the 5 mg group, one event was listed as a CVA, the other as facial paralysis. In the 10 mg group, 2 events were listed as cerebral ischemia, one was listed as a CVA, and one was listed as an intracerebral hemorrhage.
As can be inferred from the above two displays, there was no signal for CVAE risk in the flexible dose studies (in both studies together, there was one event in the aripiprazole groups and two events in the placebo-treated patients) .
As Dr. Stone notes (see his Table 8, page 11), the rate ratio for CVAEs for aripiprazole (all controlled trials) is remarkably similar to that for olanzapine and risperidone (2.43 for aripiprazole and olanzapine; 2.60 for risperidone).

COMMENTS
The sponsor has submitted a single controlled trial that, on face, provides evidence that aripiprazoJe continues to be effective in patients with Bipolar I Disorder who have been stabilized for at least 6 weeks. Two issues, however, raise concerns about how reliable these results are.
First, as noted above, the statistical significance of the study is driven by the results at a single, small center in Mexico, in which the treatment effect is considerably greater than in the combined US centers, which enrolled many more patients. Removing the data from this single center causes the study to lose significance. Although, strictly speaking, there is no formal statistical justification for removing this center's data from the analysis, the clear dependence of the study's overall statistical significance on the results of this small center raises questions about how robust and reliable these results really are. We have discussed the data from this center with Dr. Khin of the Division of Scientific Investigations (DSI), both because of the results themselves, but also because we have little experience with Mexican clinical centers. As Dr.
Podruchny notes, according to Dr. Khin, there is no obvious reason to believe that the conduct of the study at this center differed materially from that at other centers, or that this conduct deviated in important ways from that at centers in areas with which we have more experience. Nonetheless, we will ask the sponsor to address the questions raised by the extreme results at this site.
Second, as Dr. Podruchny notes, there are questions about the reasons why some patients who were not classified as having met relapse criteria discontinued treatment early. In these studies, it is critical to be able to account for the reason each patient left the trial early. Inadequate descriptions of these reasons (e.g., patient withdrew consent) raise the possibility that some of these patients may have left the study because of worsening of their clinical state, even if they did not meet formal relapse criteria (for example, the onset of insomnia, given as a reason for early discontinuation of one patient, might be the beginning of the return of manic or depressive symptoms). For these reasons, then, we will ask the sponsor to provide a more detailed account of the reasons why each patient (not just those in Center 093) who left the study early did so.
Regarding the data on cerebrovascular adverse events reviewed by Dr. Stone, there are reasons to question the propriety of combining data from all three aripiprazole controlled trials. As Dr. Stone notes, not only were the patients in Study 006 potentially different from those in the other two studies (for example, the patients in Study 006 had [some] fewer risk factors for stroke), but the study designs were importantly different (Study 004 was the only fixed dose study). Dr. Stone notes that the sponsor concludes that the absence of any signal in the two flexible dose studies "sheds doubt" on the (weak) signal seen in Study 004. Dr. Stone, on the other hand, does not agree, and suggests that "The paucity of CAEs in both the drug and placebo groups in the [005 and 006] studies ... " provides little useful data (one way or the other) on the question. He concludes, for example, that the lower risk for CVAEs in patients in Study 006 makes it "unlikely" that any meaningful differences in CVAE risk could be seen in a study of that size (the total number of patients treated with aripiprazole in these two studies was 235). He does conclude, however, that " ... almost all of the meaningful information comes from [Study 004].".
In this, I beli~ve Dr. Stone, the sponsor, and I agree. I am not sure that the patients in Study 006 are, in fact, at so much less risk for stroke than patients in the other two studies (for example, although Dr. Stone believes that baseline differences in MMSE scores suggest that patients in Study 006 were not as impaired, I am not convinced that the small differences are meaningful in this regard, nor do I believe that the patients in Study 006 are materially younger than patients in the other two studies).
Whether we would have expected to see events in Studies 005 and 006 (assuming patients in 006 are reasonably similar), it bears pointing out that there were 235 patients in these studies combined who were treated with aripiprazole; a total of 4 events were seen in the 123 patients treated with 10 mg in Study 004. Of course, as I noted above, I do not know the actual doses achieved in the flexible dose studies, and, clearly, dose could be a critical factor (the results of Study 004 clearly suggest that dose is a critical risk factor).
In any event, the dose response data from Study 004, although not confirmed in any other study, do suggest that aripiprazole may increase the risk of CVAEs in patients with Alzheimer's Disease (I do agree with Dr. Stone that the "negative" data from Studies 005 and 006 do not necessarily cast doubt on the findings in Study 004, mainly because the designs of these studies are significantly different from that of Study 004, and, for the reasons discussed by Dr. Stone, may not have been expected to yield similar results). Whether this is also true for elderly patients in general, we cannot know, given that no other elderly subjects have been so studied. Nonetheless, we will propose that a statement describing these results be placed in the Warning section of labeling, analogous to statements in the risperidone and olanzapine product labels.
For the reasons discussed above, then, I will issue the attached Approvable letter, with appended draft labeling.  . Supplement 008 to this application was submitted by Otsuka on 12/27104, and proposed specific language to describe the occurrence of strokes in elderly patients with dementia. NDA 21-713/S-003 Abilify (aripiprazole) Oral Solution, for the granting of the maintenance claim for the oral solution, was submitted by Otsuka on 2/16/05.
As noted above, in our 11/30/04 Approvable letter to NDA 21-436/S-005, we asked the sponsor to address the following issues: 1) Although the single study submitted by the sponsor to support the maintenance claim reached statistical significance on its primary effectiveness measure, the trial lost significance when the data from a single Mexican center were removed (i.e., the treatment effect seen at that center was markedly greater than that seen in the combined US centers). Although there was no obvious reason to believe that the results at that center were compromised (a DSI inspection found no major flaws), we asked the sponsor to address this issue. 2) A number of patients had discontinued the study prior to having reached an endpOint or completing the entire duration of the study. For many of these patients, the reasons for these discontinuations were unclear, and the results of the trial could have been different had some of these patients left the trial early because of worsening of their condition. For this reason, we asked the sponsor to re-evaluate these discontinuations. 3) We asked the sponsor for written confirmation of numerous Phase 4 commitments. 4) We asked the sponsor to include language in product labeling related to cerebrovascular adverse events (CVAEs; this language was submitted as supplement 008 to the NDA; see above).
The sponsor responded to the Approvable letter on 1/3/05. The response has been reviewed by Dr~ Greg Dubitsky, medical officer and Dr. Paul Andreason, Psychiatry Drugs Team Leader. The review team recommends that the application be approved once agreement has been reached with the sponsor on language for the label.
I agree that the application may be approved.
Specifically, with regard to our concern that the overall outcome was dependent upon the results at a single Mexican center, the sponsor has argued that there is no valid reason for excluding this center, and that the crude relapse rates (for all relapses, as well as for depressive and manic relapses), as well as the mean changes in the V-MRS, although not the primary outcomes, show minimal changes when the Mexican site is excluded. Although we had similar doubts about the validity of excluding this site, we believed it was worth asking the sponsor to examine the question; I am now convinced that we should accept the results of the study when analyzed as planned (that is, with the inclusion of the site).
Regarding the potential re-classification of (some) patients as having met relapse criteria, the sponsor has re-evaluated all of these patients, and believes that, in 12 of these patients, a relapse could not have been "absolutely" ruled out. When these patients were included in a re-analysis as having met relapse criteria, the results still achieve statistical significance. Further, Dr. Dubitsky has reviewed descriptions of all of these cases. In his view, only 2 of these patients could reasonably have been considered to have met relapse criteria; are-analysis including only these 2 additional patients also was significant. I agree that this answers our second question.
All other issues raised in the Approvable letter have been resolved satisfactorily.
In particular, we have reached agreement with the sponsor on final labeling. For this reason, then, I will issue the attached Approval letter, with appended agreedupon labeling.
The sponsor's efficacy claim is supported by one clinical study, eN 138010, a multicenter double-blind, randomized trial. Since the acute efficacy claim was recently reviewed and approved, it was decided by the Division that a single longer term study was sufficient for a longer term maintenance claim.
The primary clinical reviewer was Teresa Podruchny, MD and the primary statistical review was performed by Kun He, PhD of the Division of Biometrics (HFD-710).
2.0 CHEMISTRY There were no chemistry issues on this submission as aripiprazole is already a marketed product.
3.0 PHARMACOLOGY There were no animal pharmacology issues on this submission as aripiprazole is already a marketed product.

BIOPHARMACEUTICS
There were no biopharmaceutical issues on this submission given the recent review and approval ofthe acute treatment claim 5.0 CLINICAL DATA 5.1 Efficacy Data The sponsor' s proposed efficacy claim was supported by the single study 131010 in this submission. Study 138010 was based on an open-label stabilization ofa cohort of patients with DSM-IV Bipolar Disorder who presented for treatment with an acute Mixed or Manic episode. The sponsor describes study 138010 as a three phase study: 1) stabilization, 2) maintenance and 3) extension. The sponsor appears to imply that the duration ofthe claim of efficacy is 26-weeks; however, the sponsor was informed in pre-NDA discussions that the duration ofthe maintenance claim would reflect the duration of the open label period of stabilization and not the nominal duration of the double blind observation period.
The phases ofthe study progressed in the following manner. After open-label treatment with aripiprazole a patient met stabilization criteria when a Young-Mania Rating Scale (Y-MRS) Score of ~ 10 and a Montgomery-Asberg Depression Rating Scale (MADRS) Score of ~ 13 during 4 consecutive visits. The patient continued into a maintenance phase (double blind phase) after they remained stable for at least 6 weeks (4 consecutive visits separated by periods of2-weeks). Patients at this point were randomized to either continue aripiprazole or placebo. The sponsor labels this the "maintenance" phase; however, this is a double blind treatment withdrawal period. The Division defines the maintenance phase of a relapse-prevention designed trial as the 6-week period of 4 consecutively stable rating scale scores. Time to relapse was then measured over a period of up to 26-weeks. Ifa patient remained stable for the entire 26-week period then they had the option of continuing on openlabel aripiprazole ( extension phase).
A total of 633 subjects enrolled in the study. Of these 633 there were 206 who met randomization criteria. 191 were randomized, but 35 patients across several sites received randomized yet unblinded study drug. These 35 patients were disqualified so that there were 161 patients who went on to be randomized to the double-blind phase. The ITT population included 83 subjects in placebo group and 77 subjects in aripiprazole group. The primary efficacy endpoint was the time from randomization to relapse during the maintenance phase. The primary analysis was a log-rank test of time to relapse. Relapse was defined clinically as taking place when a patient discontinued from the study due to lack of efficacy, if they were hospitalized or required an addition to or increase in their allowed psychotropic medications, other than the study medication, for manic or depressive symptoms.
The log-rank test produced a p-value of 0.02 where there were 36 out of83 (43%) patients who relapsed in placebo, and 19 out of77 (25%) patients who relapsed in aripiprazole groups, respectively. Dr. Kun He noted, "One issue is whether the study is robust because center 093 in Mexico, where there were 7 in placebo and 6 in aripiprazole groups, respectively, had 5 (71 %) relapsed in placebo and 0(0%) relapsed in aripiprazole groups, respectively. The primary analysis is not significant after removing center 093."

Safety
In a safety review that was not part of this submission, the Safety Team concluded that aripiprazole labeling required the addition of the description of cerebrovascular adverse events (CV AE) in the elderly to the WARNINGS section. Judith Racoosin, MD the Safety Team Leader, provided the following draft labeling: Cerebrovascular Adverse Events, Including Stroke, in Elderly Patients with Dementia - Aripiprazole is not approved for the treatment of patients with dementia-related psychosis.

Clinical Sections of Labeling
Draft labeling is attached to this package.
6.0 WORLD LITERATURE A world literature review was provided in the safety update for the response to the approvable action letter for supplement 002. That review of the literature is adequate and supercedes the one performed in this submission.
7.0 FOREIGN REGULATORY ACTIONS I am not aware of any foreign regulatory actions regarding this claim in non-US labeling.

PSYCHOPHARMACOLOGICAL DRUGS ADVISORY COMMITTEE (PDAC) MEETING
We decided not to take this supplement to the PDAC.

DSI INSPECTIONS
The Division of Scientific Investigations (DSI) inspected three sites; one U.S. site (64) and two sites in Mexico (93 and lIS). These sites were chosen due to either sample size or impact on study significance. The DSI inspection report was written by Dr. Ni Khin. Dr Khin felt that the sites generally followed good research practices and despite some deficiencies the data were acceptable.
10.0 APPROV ABLE LETTER An approvable letter and proposed draft labeling is attached to this review package.
11.0 CONCLUSIONS AND RECOMMENDATIONS Study 13 SO lOis positive and supports the claim of 6-weeks of extended efficacy in the maintenance treatment of Bipolar I Disorder. Prior to approval the sponsor should investigate the reasons behind the unusually high rate of placebo dropout and unusually high rate of aripiprazole retention at this site.
If it can be determined that there was a bias in favor of keeping aripiprazole patients and discontinuing placebo patients in this study site, then I would consider 13 SO lOa failed study and would not recommend approving supplement 005. In order to reach final approval, I believe that the sponsor needs to address the following issues: 1. Investigate and explain the possible causes for the disproportionately high patient dropout rate in the placebo group and disproportionately high retention rate in the aripiprazole group at site 93. 2. Reach agreement on final labeling language which shall include WARNING language on the risk of cerebrovascular adverse events (CVAE) in the elderly.

This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature.
/s/

Patient population
The sponsors report the results of clinical studies that enrolled a total of 968 subjects with a diagnosis of PAD. Thirty of these subjects were in a small open label study. There were three randomized placebo-controlled trials in which 595 patients were treated with aripiprazole in the initial study phase while 343 were treated with placebo. These trials all included subsequent open-label extensions in which 269 subjects originally in the placebo groups received aripiprazole, and 625 patients originally randomized to aripiprazole continued on that therapy. Table 1 gives brief descriptions of each of these studies. The inclusion criteria of the four studies allowed only patients with diagnosis of dementia of the Alzheimer's type. Patients with diagnosis of dementia of the vascular type or mixed type were excluded. Analyses were perfonned on both the "Placebo Controlled Data Set" (data from the double-blind placebo controlled phases of CN138004, CN138005, and CN138006) and the "All Aripiprazole Alzheimer's Dementia Data Set" (data from all patients exposed to aripiprazole in all four studies, including the open label phases). Except where noted, this review will focus on the results of the placebo-controlled phases ofCN138004, CN138005, and CN138006.
Clinical Review Marc Stone, MD NDA 021-436 Definition of terms The sponsor searched and reviewed the adverse event data from all four studies for potential cerebrovascular adverse events (CAEs). The initial search included a search for specific text strings in either the investigator verbatim term reported on the adverse event case report form or the COSTART preferred terms that might indicate CAEs. The text strings searched included 'CERE', 'ISCHEMIA', 'STROKE', 'CVA', 'C.V.A.', 'TIA', and 'T.I.A.'. In addition, COST ART preferred terms reported for all four studies were reviewed for possible CAEs. All of the adverse event data for each patient reporting a possible CAE term were reviewed to determine if the event in question would be considered a potential CAE. The terms actually occurring in the reports that were considered to indicate potential CAEs were cerebrovascular accident, cerebral ischemia, intracerebral hemorrhage, intracranial hemorrhage, subarachnoid hemorrhage, mydriasis (verbatim term mentioned 'possible TIA'), facial paralysis, subdural hematoma, cerebral edema, and consciousness decreased (the verbatim term includes transient ischemic attack).
Analytic methods For all CAE incidence summaries (except time of first onset), the incidence rate of a potential CAE was calculated two ways; 1) using a patient based denominator -by Clinical Review 2 Marc Stone, MD NDA 021-436 dividing the' number of patients with at least one report of a CAE during the study or phase by the number of patients exposed to drug during the study or phase in the particular treatment ann, and 2) using a patient year denominatorby dividing the number of patients with at least one report of a CAE during the study or phase by the total number of patient years exposure in the particular treatment ann. All CAEs were considered as treatment-emergent unless the onset date proved otherwise (i.e., CAEs with an onset prior to the first day of study medication). For completed studies, CAEs with an onset more than 30 days after the last day of study medication were excluded. For ongoing studies, complete dosing infonnation may not have been available at the time of the database cut-off. All events entered into the database as of the database cut-off, regardless of the time of onset relative to the last day of dosing, were reported for the ongoing studies.
Time to first onset of a potential CAE was analyzed for the placebo-controlled studies. A log-rank test was used to compare the time to first onset of a potential CAE between aripiprazole and placebo. Patients who did not have a potential CAE were censored on the day of their last dose of double-blind medication. The paucity of potential CAEs in these studies precluded additional statistical testing adjusting for other factors; these factors were examined using tabular comparisons. The Cochran-Mantel-Haenszel correlation test was perfonned to identify any positive dose response for CAEs.

Clinical
The incidence of treatment-emergent CAEs determined from the dose groups in the CN138004 study is seen in Table 6. The incidence of treatment emergent potential CAEs increased significantly with the dose of aripiprazole (p=O.030, CMH Row Means Score test). This dose-response effect was not seen in the flexible dosing studies, CN138005 and CN138006; the number of events was too small to establish any pattern.  Aripiprazole (Abilify®) borderline statistical significance. In the patients with a low «14) MMSE, there is a higher percentage of subjects in the aripiprazole group with CAEs (1.2% vs 0 for placebo); again, statistical significance was borderline. All of the eight cases in the aripiprazole group had a prior medical history of CV Alstroke or stroke risk while there were no cases in the placebo group among those with a prior medical history of CV Alstroke or stroke risk.
Twenty-one (3.5%) aripiprazole treated patients died during or within 30 days of discontinuing the placebo-controlled phase. Six (1 .7%) placebo treated patients died within the same period. The difference in mortality rates between aripiprazole and placebo during the placebo-controlled phase was not significant (log rank test, p=0.l39).

Observations from Open-Label Studies
There were 36 subjects with CAEs in the sponsor's open label studies observing 894 patients over 648 patient-years, an incidence rate of 5.6 per 100 patient-years. All but 30 of these subjects had entered these studies after participating in placebo-controlled trials. Notably, one subject suffering a CVA during an open-label study was also taking prazosm. Clinical

Sponsors' Conclusions
The sponsors conclude that a potential signal exists for anincreased incidence of Treatment Emergent Potential Cerebrovascular Adverse Events in the population of elderly patients with psychosis associated with Alzheimer's Disease. They note that this result is a consequence of 1 of 3 placebo-controlled trials and it was not statistically different from the incidence in the placebo-controlled population. All eight CAEs in the aripiprazole group occurred in the presence of a history of CV Alstroke or other stroke risk factors; no events occurred in the absence of these factors. They request an update to the label in the Precautions Section. The sponsors also note updated all-cause mortality in this population was 3.5% for aripiprazole and 1.7% for placebo

Adverse Event Coding and Pooling of Data
The approach taken by the sponsors to identify cerebrovascular-related adverse events is acceptable. As a result of reviewing all adverse event verbatim terms using a string search, they appear to have effectively identified a number of events that were not covered by the preferred term search.
There are some potential problems with the pooling of data from the three studies. The population in CN138006 began the study as outpatients, had a smaller percentage of stroke risk factors, and was generally younger and less neurologically impaired. These differences could lead to lower risk for true events in the CN 13 8006 study because subjects were at lower risk but a greater likelihood of detecting minor or false positive events because such events may be easier to detect in subjects who are less impaired. Combining these two populations runs the risk of diluting different signals from each of them. Although the populations in the CN138005 and CN138004 studies were similar in age, neurological impairment, and percentage of risk factors, differences in dosing protocols in the two studies (flexible dosing in CNl38005 and fixed dosing in CN138004) could lead to very different adverse event profiles and other effects. In a flexible dosing regimen, subjects with similar disease severity will be titrated to dosages that produce similar pharmacologic effect while a fixed dosing regimen is more likely to show a true dose-response effect. Rather than pool the three trials together, it would be preferable to combine the observed effects of the three trials using a random effects metaanalysis technique.

Baseline Characteristics
Differences in baseline characteristics may have led to a small bias against aripiprazole; those subjects who received active drug were slightly more likely to have a history of stroke and other risk factors. As all of the subjects who experienced CAEs were women, this bias may have been offset by the lower proportion of women among those receiving active drug if female sex were a genuine risk factor.
Clinical Review Marc Stone, MD NDA 021-436 Aripiprazole (Abilify®) In the CN138004 study, where a dose-response effect was observed, increasing dosages of aripiprazole correlated with increasing baseline MMSE, male sex, and a history of CHF. The first two factors were associated with no observed increased incidence of CAEs while the last is a weak risk factor for CAEs. Taken together, the imbalance of baseline characteristics in this study probably acted to diminish the probability of observing the dose-response relationship that was seen.

Incidence of Cerebrovascular Adverse Events
Combining the effects of the three trials using meta-analysis with a random effects model, the incidence rate ratio for CAEs from aripiprazole relative to placebo is 2.43 (95% CI: 0.23-25.35, p=0.46). This is, of course, not statistically significant and is entirely consistent with what we would expect to observe if there was no real difference. The data are also completely consistent with a substantially elevated risk; CAEs were simply too uncommon in the clinical trial experience to make any conclusions based solely on these data.
The more impressive observation is the statistically significant dose-response effect observed in the CN138004 study. Because subjects in this study were randomly assigned their dosages of aripiprazole (or to placebo), drug effect should not be confounded by pharmacokinetic differences or by indication as was possible in the other two trials.

Observations from Open-Label Studies
The incidence rate of CAEs observed in the open label studies was between those observed for the respective active and placebo groups in the controlled studies. None of these differences is statistically significant. Additionally, the population observed in the open label studies was subject to selection bias; almost all of the subjects entered the open-label phase after passing through a controlled trial without significant adverse expenences.
Note was made of one subject in the open label studies who suffered a CY A while taking prazosin. One postulated mechanism for an increase in CAEs with aripiprazole is hypotension resulting from the drug's adrenergic blocking effects. This effect may be magnified with the use of anti-hypertensive drugs, particularly an alpha-adrenergic blocker such as prazosin. Because hypertension itself is a risk factor for CAEs and most of the subjects experiencing CAEs had a history of hypertension and were taking antih.ypertensive drugs, it cannot be determined whether hypertension or the combination of aripiprazole with an anti-hypertensive drug was the cause of a CAE.

Sponsors' Conclusions
The sponsors correctly recognize that the slightly elevated incidence of CAEs with aripiprazole observed in the three studies combined is very weak evidence of a harmful effect from the drug. I would disagree with the implication that the absence of any increase in observed CAEs in two of the studies (CN138005 and CN138006) sheds doubt on the significance of what was observed in CN138004. The paucity ofCAEs in both the Clinical Review 10 Marc Stone, MD NDA 021-436 Aripiprazole (Abilify®) drug and placebo groups in the CN138005 and CN138006 studies means that these two studies provide little information as to whether the risk from aripiprazole is or is not elevated; almost all the meaningful information comes from CN138004. In particular, the lower risk for CAEs in the CN138006 popUlation makes it unlikely that any significant (or even suspicious) difference would be observed with the number of patients involved. The differences in dosing practices between CN 13 8004 and the other two studies could also significantly affect whether a signal would be observed. The sponsors overlook the significance of the dose-response effect observed in the fixed dose study, CN138004. Such an effect would be much less likely to be observed in the other two studies where dosages were titrated.

Comparison with Other Drugs in Its Class
A summary of the observed risks for CAEs with aripiprazole compared with olanzapine and risperidone is given in Table 8. While the observed risk ratio for aripiprazole has very wide confidence intervals and is, by itself, highly consistent with chance, these findings should be analyzed in the context of the findings for olanzapine and risperidone. All three drugs have alphal-adrenergic blocking activity, suspected to cause hypotension and diminished cerebral perfusion leading to CAEs. The observed risk ratios for all three drugs in Alzheimer's Dementia are remarkably similar. Differences in incidence rates appear to be due to differences in risk for the selected patient populations as reflected in the different incidence rates among the placebo groups. From a Bayesian perspective, the observed results for olanzapine and risperidone constitute a prior expectation for aripiprazole. The observed aripiprazole results are strongly consistent with that expectation. The posterior distribution is the same as that resulting from the combination of all of these studies using meta-analysis. The combined results for Alzheimer's patients over all trials gives a ratio (2.54) that has statistical significance with no evidence for heterogeneity (p=0.932).
The clinical trials in dementia for olanzapine and risperidone included patients with vascular or mixed dementia as well as Alzheimer's without stratifying for type of dementia. Combining all trials and including all subjects with dementia of any etiology Clinical Review 11 Marc Stone, MD NDA 021-436 gives an overall ratio of 3.24 that is statistically significant without evidence for heterogeneity (p=O.856) .

Reviewer's Conclusions
There is good evidence for an increased risk for CAEs from aripiprazole based upon three factors : • the presence of a dose-response effect • remarkably similar findings with other drugs in its class • a plausible mechanism: hypotension from adrenergic blockade One factor that may possibly mitigate this risk is the complete lack of a signal for increased risk in the flexible dosing studies. While chance is a plausible explanation for these differences, it is possible that careful titration of dosage reduces any risk, something not possible in a fixed dose study. At the same time, the dosage titration done under protocol in the clinical trials may be much more careful than what might be seen in typical medical practice.
Labeling addressing the risk of CAEs with aripiprazole should be similar to that for olanzapine and risperidone. The excess risk for CAEs associated with aripiprazole and other drugs in its class appears to be proportional to the patient's underlying risk; the greatest caution should be exerted for patients at highest risk for CAEs.

Recommendation on Regulatory Action
I recommend the Division consider an approvable .action on this supplement. While the p value is significant when comparing aripiprazole to placebo for time from randomization to relapse in the maintenance phase, it is unclear to me that this reflects efficacy of the drug. While the larger pool of data favors aripiprazole, removal of one site in Mexico (site 93) causes the study to lose significance. This site appears to have a different relapse rate than the conglomerate u.s. sites.
DSI inspection at this site revealed protocol violations, however, overall the data were deemed acceptable. As this is the only study for maintenance and given that a large number of U.S. sites were involved but alone are not powered to show significance and for other reasons listed within this review, I recommend we ask for further exploration of the data in this study with attention to the Mexican sites, more so to site 93.

Risk Management Activity
I recommend addition ofPE/DVT to the postmarketing list as well as consideration of addition of the events of hypersensitivity, hepatobiliary events, and increased creatine phosphokinase/rhabdomyolysis. A change to the label regarding the risk of cerebrovascular adverse events in elderly patients with dementia has been added as a WARNING. This is based on a recent review by Dr. Marc Stone in DNDP. Additionally, the OVERDOSAGE/Human Experience subsection currently is under review in SLR007.

Required Phase 4 Commitments
Required Phase 4 commitments were delineated in the action letter for this supplement. As commitments were made regarding adult studies to address short and longer term efficacy as add-on therapy in bipolar patients and pharmacology-toxicology studies needed to support pediatric trials with the action on supplement 002 (acute mania), no additional studies are required at this time. However, the sponsor was asked to state a date of submission of the clinical study reports for the recently completed drug interaction studies.

Other Phase 4 Requests
There are no other phase 4 requests at this time. 633 patients enrolled in the study and 196 patients were randomized. However, the efficacy maintenance dataset is comprised of 161 patients as 35 randomized patients received unblinded medication and were discontinued. The primary efficacy measure was time from randomization to relapse in the maintenance phase for aripiprazole treated versus placebo treated patients. Key secondary measures were time to manic and time to depressive relapse analyzed using a hierarchical procedure.
eN 138010 data demonstrated significance on the primary efficacy measure (p=0.02). One site in Mexico (site 93) appears to have both a low relapse rate in the aripiprazole group and a high relapse rate in the placebo group when compared to the conglomerate U.S. sites, which contain 77% of the patients in the study. The second site in Mexico (118), has low relapse rates in both the placebo and aripiprazole groups. Although the study is not powered to examine treatment by center nor for the U.S. sites to stand alone, when dropping the other large site in Mexico (118), the study does not lose significance. However, removal of site 93 causes the study to lose significance.
Admittedly, this is post-hoc analysis. DSI noted protocol violations however, as a whole the data were not felt to be globally unacceptable although a limitation of the data was that the source documents were in Spanish. It is unclear to me whether the results (site 93) represent a spectrum of the efficacy of this drug or reflect an aberrant finding at this site that is not generalizable. For reasons outlined in the body of this review, I recommended an approvable action with further exploration of the data regarding the robustness of the p-value.
The study is not fixed dose and cannot assess dose response.

Safety
The safety data to support use in the maintenance treatment of mania are derived primarily from study CN138010. Quantitative safety review with respect to EKG data, vital signs, and clinical laboratory measures was limited secondary to the design of this study. The data were reviewed for deaths, non-fatal serious adverse events, and discontinuations secondary to adverse events. Akathisia appears to remain a common adverse event in this population. CPK elevations were the cause of discontinuation of two patients in the open label phase.
Quality control review of lists of protocol violations and the Division of Scientific Investigation report indicate missed laboratories or EKGs and the sponsor submitted data from source documents at site 118 that were not included in the CRFs (email 8-12-04). The data from the 35 randomized patients who received unblinded study medication were presented separately from the randomized population. An audit of the COSTART terms was performed. Recommendations to the sponsor resulting from this audit are made in section 9.5.
Review of deaths, non-fatal serious adverse events, and discontinuations secondary to adverse events and review of adverse event terms for this trial as per the JMP,dataset did not reveal any new, previously undescribed adverse events for the bipolar population such as to preclude approval for this indication.
Non-bipolar indications: • Review of incidence data for other indications, as supplied in incidences table in the ISS of this supplement and in supplement 002 submissions, was not performed as these tables are not exposure and placebo adjusted. • Line listings of patients who died, experienced a non-fatal serious adverse event, or discontinued secondary to an adverse event in studies blinded or newly reported since September, 2002 were submitted with supplement 002, the 120 day safety update, the response to the approvable, and supplement 005. Line listings of the deaths generally do not include the cause of death. • Review of these line listings was cursory. Review of events that require further exploration is in progress but these data are not discussed in this review and will be completed as per Division leadership advisement.
Post-marketing data are discussed in section 7 of this review.

Dosing Regimen and Administration
Study CN138010 was not a fixed dose study. In this study, dosing started in the stabilization phase at 30 mg daily with reduction to 15 mg daily as tolerated or efficacious. During the maintenance phase, the initial dose of aripiprazole was the end dose at stabilization and could be adjusted as necessary for either efficacy or tolerability issues.
The mean daily dose at the endpoint of stabilization was 25.25 mg daily. For those who completed this phase and remained eligible for maintenance, the mean daily dose was similar at ] With regard to other drugs used in the treatment of bipolar disorder, the label included with this submission (1-28-04) contains information on valproate, lithium, and carbamazepine. As per this label, no dosage adjustment of aripiprazole is required when concomitantly administered with either valproate or lithium. However, if carbamazepine is added to aripiprazole treatment, it is recommended that the dose of aripiprazole be doubled and in converse, if carbamazepine therapy is withdrawn from combination therapy, the dose of aripiprazole should be reduced.

Special Populations
No additional studies in special populations were submitted with this application.
Cerebrovascular adverse events in dementia patients were reviewed by Dr. Marc Stone of the DNDP safety team. Although not labeled for this indication, additional language to the label will be added as a warning for this group.

Product Information
Aripiprazole, AbilifyTM, is an atypical antipsychotic approved in the U.S. for use in the treatment of schizophrenia and acute manic and mixed episodes associated with Bipolar Disorder. It is a partial 02 agonist acting as an agonist in an animal model of dopaminergic hypoactivity and an antagonist in animal models of dopaminergic hyperactivity. Aripiprazole also is a 5-HT 1A partial b(4) agonist and a 5-HT c ] antagonist.

Currently Available Treatment for Indications
Drugs approved for monotherapy in the maintenance treatment of Bipolar Disorder include lithium and lamotrigine. The only atypical antipsychotic currently approved is olanzapine, which received approval for two weeks of maintenance treatment, in January of 2004. The combination of olanzpine and flouxetine (SymbyaxTM) is approved for use for up to eight weeks in the treatment of depressive episodes of bipolar depression and lamotrigine (Lamictal®)

Important Issues With Pharmacologically Related Products
Neuroleptic malignant syndrome, tardive dyskinesia, and hyperglycemia and diabetes mellitus are labeled "WARNINGS" for atypical antipsychotics. Aripiprazole will soon receive a "WARNING" for cerebrovascular adverse events in the psychosis of dementia. Olanzapine and risperidone already have such language.
Other "WARNINGS" on individual atypicals include: • Clozapine-Black box warnings for agranuloctyosis, seizures, myocarditis, other adverse cardiovascular and respiratory effects (including collapse, respiratory arrest, and cardiac arrest during initial treatment). There is a required hematologic monitoring program in place for the prescribed use of this product. • Ziprasidone carries an additional warning for QT prolongation and sudden death. • Quetiapine carries an additional bolded "PRECAUTION" for cataract development seen in animal studies and recommends monitoring for cataract development. • Olanzapine carries and additional warning for a higher incidence of death in dementiarelated psychosis treated with olanzapine versus those treated with placebo, although the drug is not approved for use in this population.

Presubmission Regulatory Activity
September, 1999-BMS and Otsuka entered a co-development agreement with respect to the development of aripiprazole. This resulted in a program that allowed for additional indications beyond schizophrenia.

Other Relevant Background Information
Aripiprazole is approved for marketing for the treatment of symptoms of schizophrenia (acute

b(4)
and maintenance or acute) in multiple countries including Brazil, C. .J Puerto Rico, Australia, Peru, Korea, r : " J and Mexico. It has received approval b(4) for the treatment of acute manic and mixed episodes of bipolar disorder in C J

SIGNIFICANT FINDINGS FROM OTHER REVIEW DISCIPLINES
Statistical review through the review of the Office of Biometrics was performed by Dr. Kun He and is discussed in the efficacy section of this review (section 6.1.4).
The Division of Scientific Investigations conducted reviews of three sites. The report was authored by Dr. Ni Khin and is discussed in section 4.4 Data Quality and Integrity of this review and in the efficacy section of this review (section 6.1.4).

Review Strategy
The Clinical Study Report for CN 1380 1O;JMP files submitted with this supplement as needed, updated appendices submitted 8-27-04, the ISS for this supplement with attention focused on maintenance mania, and narratives, and case report forms were utiliied in the preparation of this document. Safety data for study CN138037 was taken from the CSR as supplied with the submission for supplement 002 on 6-23-04.

Data Quality and Integrity
There were several issues relative to data quality and integrity • 35 patients treated unblinded initially and • the findings of the DSI inspection.
35 patients were randomized into study CN138010, across several sites, and received unblinded medications. When the sponsor became aware of the problem, randomization was closed and ongoing patients discontinued. The Division discussed the circumstances of this with the sponsor. The explanations offered as to how data were handled were considered generally acceptable.
The Division of Scientific Investigations (DSl) inspected three sites; one u.s. site (64) and two sites in Mexico (93 and 118). These sites were chosen due to either sample size or impact on study significance. Dr. Ni Khin's report of these inspections notes that site 64 was classified as "minor deviations, data acceptable" (VAl). Site 93 and 118 were classified as having deviation(s) from regulations, response received and reviewed (VAl-RR).
• Site 93 screened and enrolled 18 patients and randomized thirteen patients. Records from all 18 subjects at this site were audited. Dr. Khin noted seven specific patients who either did not receive lithiumldivalproex levels pre-randomization or had these laboratory collections after randomization and eight patients who were in the open-label stabilization phase after meeting criteria for randomization. Four of these were due to the sponsor being unable to supply blinded medication. Dr. Khin's review notes there were "multiple instancesof protocol required clinical laboratory tests" that were not performed, ranging from one to fifteen per subject. Dr. Khin also note that adverse events were not reported to the sponsor on two subjects and that there were several instances when events documented in the source document did not match the CRF.
• Site 118 screened 28 subjects and enrolled 25. Temperatures for storage of the medication were outside of recommended ranges (stored at 3°C-28°C versus 15-25°C). During stabilization, four patients received lorazepam outside of protocol specifications. Safety data problems included two subjects who experienced serious adverse events that were not reported for several weeks and sixteen of seventeen EKGs reported missing in the clinical study report (CSR) for the stabilization were later recovered by the sponsor when querying the data differently.
Given the importance of the data at site 93, an internal meeting was held with Dr. Khin to discuss the inspection results and data integrity at site 93. It was my interpretation from this internal discussion that there was no obvious major problem nor the appearance of fraud such as to disqualify all data. However, it appeared the investigators may have been inexperienced in the conduct of clinical trials.
An audit of safety data was conducted by comparing CRFs, narratives, and line listing data for a sample of patients for internal consistency. With the exception of the selection of more patients at site 93, patients were randomly selected for audit.
The patients who were audited are listed in the appendix of this document. In the comparison of CRF data to narratives to line listing, most were acceptable although there were some discrepancies and two CRFs at site 93 were corrected several times (93-184, and 93-504).
• Patient 132-355: the CRF describes the serious adverse event as manic reaction and suicidal ideation. Suicidal ideation is not captured in the line listing (App. 12.1.A) and is not discussed in the narrative. • Patient 146-437: this patient was coded as discontinuation secondary to an adverse event during stabilization on the CRF. The narrative notes he was discontinued secondary to a severe manic episode with psychotic features. While both may be correct, it appears this patient may have had lack of efficacy. • Patient 64-441 is listed in the line listing as discontinuation due to an adverse event and the narrative concurs. The CRF captured this discontinuation as a withdrawal of consent. • Patient 146-459: the narrative and text were not in agreement as to the reason for discontinuation. The sponsor was asked to clarify and noted the narrative was "incomplete" .

Compliance with Good Clinical Practices
The sponsor notes that the study was conducted in accordance with Good Clinical Practice and with generally accepted standards for the protection of patient safety and welfare including the Declaration of Helsinki and amendments. Otsuka America certified that it had not used the services of any person listed as debarred as of the Date of Debarment List in connection with the application.

Financial Disclosures
The sponsor notes that C :J investigator Financial Disclosure Forms were received by November 17, 2003 and that no investigators had information to disclose. Investigators at sites in 2002. These monies included funding for a J: :l study and honoraria fees. As this was a blinded, randomized, study, it is unlikely that these payments biased the study conduct such as to disqualify the data. 29 responses had not been received as of the date of writing the original submission document.
Otsuka submitted a certification as an applicant submitting the study that due diligence had been exercised to obtain financial information from non-responders. BMS submitted certification that as the sponsor of the study, they had not entered in to any financial arrangement with the listed clinical investigators in which the compensation to the investigator could be affected by the outcome of the study as per 21 CFR 54 and that any investigator who was required to disclose did not disclose any such interests.
The ISS for the maintenance supplement included a section on safety experience in clinical pharmacology studies. In this ISS, the sponsor notes data from 129 patients were analyzed for deaths, serious adverse events (SAEs), and discontinuations due to adverse events. The sponsor reports there were no deaths or SAEs in the clinical pharmacology studies. The lists of patients who discontinued secondary to an adverse event appears identical to the one in the acute mania 120-day update with the exception of one additional patient who discontinued secondary to vomiting.

Indication
With regard to primary efficacy, the sponsor seeks the following claim, "The efficacy of ABILIFY in maintaining stability in patients with Bipolar I Disorder with a recent manic or mixed episode, was demonstrated in a double-blind, placebo-controlled, 6-month maintenance phase of a longer-term trial." 6.

General Discussion of Endpoints
The primary efficacy measure was "time to relapse" (ie, discontinuation due to lack of efficacy) during the Maintenance Phase and was evaluated on the maintenance safety sample. The log rank test was used to compare the survival distributions of the two treatment groups with estimated survival curves obtained from Kaplan-Meier estimates. Discontinuation due to lack of efficacy was defined as either hospitalization for manic or depressive symptoms or requiring an additional medication or an increase in the allowed psychotropic medications.
Key secondary endpoints were time to manic and time to depressive relapses. These analyses were performed on the efficacy sample. A hierarchical testing procedure was employed with time to manic relapse tested first after the primary analysis, then time to depressive relapse.

Study Design
This was a randomized, double-blind, multi-center, placebo-controlled trial of aripiprazole for the maintenance of stability of patients meeting DSM-IV criteria for Bipolar I disorder. Patients were in-or outpatients who were from either a recently completed 3-week acute mania study of aripiprazole or were eligible for one of the 3-week studies but declined, or patients who were not from one of these studies but with a recent (~ 3 months) manic or mixed episode requiring hospitalization and treatment. Patients who did not enter from an acute study participated in a screening period of up to 28 days before stabilization. For these patients, all antipsychotics and psychotropics outside of the prescribed protocol medications were discontinued with a minimum one day wash out for antipsychotics.
The study consisted of an open-label stabilization phase of 6-18 weeks, a blinded, randomized maintenance phase of up to 26 weeks, and a blinded extension phase of an additional 74 weeks. During open-label stabilization, visits occurred every two weeks. Eligibility for randomization required meeting both time (minimum 6 weeks) and scale criteria (YMRS and MADRS criteria of ~ 10 and ~ 13 respectively for 4 consecutive visits).
In the maintenance phase, patients assigned to aripiprazole received the same dose of drug as they were taking at the end of the stabilization phase. The dose could be adjusted for either efficacy or tolerability purposes. Patients who completed the maintenance phase without relapse were given the option to continue their current blinded study drug for an additional 74 weeks.
Complete Inclusion criteria/exclusion criteria for the study phases are included in the appendix of this document. Suicidal patients, patients requiring ECT in the previous 2 months, and patients likely to require additional prohibited medications were excluded from entering the stabilization phase.
Prohibited concomitant medications included carbamazepine, valproic acid, divalproate sodium, sodium valproate and lithium carbonate and citrate. Fluoxetine, long acting antipsychotics, other IND drugs, all other psychotropics, gingko biloba and S1. John's were generally prohibited.
Allowed concomitant medications: Lorazepam and anticholinergics for symptomatic EPS were allowed. Lorazepam was allowed in doses up to 6mg/day during screening and the first four weeks of stabilization, 3mg/day for the fifth week and 2 mg/day thereafter in stabilization. In the maintenance phase, lorazepam up to 2 mg/day during the first month, 1 mg/day during the second month, and 1 mg/day up to 4x weekly during the remaining 18 weeks was allowed. 1M flunitrazepam and midazolam were allowed in Mexico and Brazil respectively when orallorazapam was ineffective.
Anticholinergics for EPS were allowed for EPS symptom control in doses not to exceed 6mg/day equivalents of benz tropine. No doses were to be given during the day before the baseline visit and 12 hours before either efficacy or safety rating scales.

Efficacy Findings
Six hundred and thirty-three patients (663) enrolled in the study. Of these, 567 entered the stabilization phase (333 from previous aripiprazole studies) and 206 completed the stabilization phase (37%). The most common reasons for discontinuation from the stabilization phase were adverse event (22%), lack of efficacy (12%), and withdrawal of consent (12%).
d During Ibc MaitUcnance Ph$c, ~otbc.known causcs" included po-silh-c drug screen, patient relocaling, and site clo-scd by SPQllsor boclIU5C prc5pccifIcd number of relapses had been auained. e Patient 13&O1~14· 7'@4 relapsed during Ibc Extension Pba. sc. aoc~rdlng to \be relapse fmm. b," discontinued from !he EXlens;icm Phase because of "Olhcr known CllUSC" according to the cnd-<lf-Slud)' foml. f During Ibc EXlcmriau Phase, Ihc primmy -othaknown causc" (5100)' closed by spon~r bCQlIISC ~;flCd numbc;r of ret.aP'lcs had been lIt1aincd) Reasons of discontinuation: Table S.8.1 of the submission (not included in the appendix of this document) listed comments for patients who discontinued secondary to "withdrew consent" or "Other known cause". Under "withdrawal of consent", some of the comments include the subject feeling conventional therapy would have greater symptom reduction (7-278), perceived adverse events (13-557), starting a new job (16-272), and "hospitalized without previous 'advice' to the investigator and withdrew consent" (93-126). "Other" also represented many reasons including the patients terminated due to unblinding, positive drug screen, and failing to meet criteria. Table S.8.l indicates that some ofthese events could have been better classified such as patient 92-136, who was noted to have a serious adverse event but was coded as "Other", patient 6-142, also coded as "Other" is listed as making suicidal threats, and patient 10-273, coded as "withdrawal of consent" is noted to have increased depression and diarrhea.

Information in
Efficacy Data: Figure 10.1, Table 10.lA, and Table 10 A (excerpted) display efficacy data and are copied from the submission below. A sponsor provided graph of the impact of censoring is included in the appendix of this document.   Defined as discontinuation due to lack of etTtcacy. F(lr Patienl~ 138010·118·214 and 1380JO·14i·604, wll{) were randomized in error UI)Qn entry into the Stabilization Phase, time from randomization to relapse is measured from the first day of dosing in Ule Maintenance Phase.       Concomitant therapy: During stabilization, the most commpn class of medication used concomitantly was the anxiolytics (51.5%). During maintenance, the most common class used concomitantly was the anxiolytics .for the placebo group (46%) and the anxiolytics and anticholinergics and for aripiprazole group (39% each).
Potential Problems with the interpretation of the data: • relapse rate and concomitant medication use Dr. He performed the primary statistical review and concluded that the primary analysis log-rank test gave a p-value oL0199 with 36/83 relapsing in the placebo group and 19/77 relapsing in the aripiprazole group. His review notes that the relapse rates for the aripiprazole group in both Mexico and Argentina are lower than in the U.S. sites and that the Mexico rate is "extremely lower" compared to Argentina and the U.S. He noted that when data from site 93 is removed, the primary analysis is not significant (log rank p = .1043) and suggested consideration of the quality of the data at this site when making final decisions. As per his review, • Infonnal analyses to look for obvious demographic differences at baseline: Dr. He perfonned several infonnal analyses at site 93 as there appeared to be an abnonnally low relapse rate in the drug group and a higher one in the placebo group. These analyses were for baseline YMRSIMADRS scores at stabilization and pre-randomization and the time of randomization versus the time of eligibility for randomization. The latter exploration was to see whether there was a difference between site 93 and the U.S. sites in tenns of when patients were randomized versus when they met criteria to be randomized and by how long. The latter analysis probably should be duplicated by the sponsor. The analyses for baseline scale scores did not yield obvious clinical differences that could be expected to differentially affect the outcome.

• Concomitant Medication and Relapse:
Nine patients from site 118 (5 placebo, 4 aripiprazole) and 3 from site 93 are listed in the appendix of prohibited or excessive concomitant medications or missing medication start or stop dates-maintenance phase (Appendix 7.3B amended). Given what appears to be a lower relapse rate in the drug group at these sites, I researched some of the patients in these lists. Of the three listed patients at site 93, two appear to have received the lorazepam in screening (as seen in the CRFs-stop date) and for the third, there is no CRF to verify.
For site 118, one could argue that as 5 patient were placebo and 4 aripiprazole, this might suggest that randomization would then basically "equal" out the effects of this type of error. As a site, taken in its entirety, dropping the entire site 118, does not render the study insignificant (log-rank p test value =.0206). One could also argue that, if some patients were maintained by the use of excessive medications, these protocol violators might represent relapse and assessment would need to be made on a case-by-case basis.
As defined by the protocol, relapse was "Patients were discontinued for lack of efficacy if they were hospitalized and/or required an addition to or increase in allowed psychotropic medication, other than study medication, for manic or depressive symptoms.". This was somewhat difficult when adverse events were not listed specifically as manic or depressive exacerbations and is second-gu~ssing the researcher who was at the site. Examples of this are: that in the 3 rd month of the maintenance phase, lorazepam can be used only for Img per day/ 4 days a week. There is no CRF for this patient but line listing of adverse events indicated "anxiety" during the time period of the lorazepam use.
• Patient 118-246 was taking aripiprazole in the maintenance phase from July 13,2001 to January 10,2002. From October 5,2001 to October 18,2001, lorazepam at 3mg daily was used for "anxiety". The MADRS was 12 at week 12, which was increased from 0 at week 10. This is a protocol violation, whether it should have been a relapse is not clear.
• Patient 118-214 (placebo) was taking 1 mg daily lorazepam for about 3 weeks in maintenance beginning at week 12 (July 23 -August 19) for insomnia. This patient relapsed September 4, 2001. The use . of lorazepam during this 3 weeks would appear to be a protocol violation.
• Patient 118-269 (placebo) is listed in appendix 7.3 B as taking lorazepam 1 mg daily for about 5 weeks in hislher 3 rd and 4 month of the study (November 12-December 21). This is a protocol violation. There is no CRF but line listing notes "insomnia" as an AE during this time. This patient relapsed on January 4.
The appendix listing 7.3B is on-face rather confusing as even though the appendix is to include violators in the maintenance phase, sometimes the medication listed was given in the stabilization phase. Additionally, I was not able to reconcile the data on patient 118-148 as per the CRF and appendix 9.5.2 (by patient listing of concomitant medications) with that in appendix 7.3B.
Other efficacy related subgroup analysis: The sponsor performed several subgroup type analyses of the data with regard to primary efficacy: episode type, gender and rapid cycler status. Dr. He's review notes that the study is not powered for subgroup analysis with respect to gender, race, and age. The following were provided by the sponsor: • Episode type-112 manic and 48 mixed. The log rank p was significant for the manic group (0.047) and not so for the mixed group (0.385). • Gender analysis-53 males and 107 females. The log rank p was significant for the females (0.065), not for the males (0.206). • Cycling status-28 rapid cycling patients and 132 non-rapid cycling. The log-rank test p value was significant for the rapid cycling group (0.033) and not the non-rapid cycling group (0.114).

24
Clinical Review Teresa A. Podruchny NDA 21436_005 Abilify (aripiprazole) 6.1.6 Efficacy Conclusions I am unsure how to interpret this study data and have some discomfort with the certain aspects of the data. Site 93 appears to have a low relapse rate in the aripiprazole group and a high one in the placebo group. Admittedly, this is found on post-hoc analysis, is difficult to interpret, and alone does not invalidate the results. Site 118 does not relapse many patients in either group (1/8 aripiprazole and 2/9 placebo) and due to this, it is perhaps not surprising that removal of this site, does not change the primary efficacy analysis.
It is possible that the data at site 93 is not aberrant and represents the spectrum of response to this drug. It is also possible that it is not a " real" finding. The DSI inspection report does not suggest disqualification of the entire site although the report notes the source documents were in Spanish and there were certain protocol violations at the site. In an internal group meeting with Dr. Khin, types of protocol violations seen at the site were discussed. At this meeting, when reviewed by types of violations, it was not clear these violations would have created a general bias against placebo and randomization would be expected to "protect" the integrity of the data.
I recommend we take an approvable action, define factors that could have affected the outcome at these sites, and ask the sponsor to demonstrate that these findings did not bias the outcome at this site. These factors might include investigator training and site monitoring (were the non-IND sites handled the same as IND sites?), demographics at baseline (type of episode, psychiatric history, in-patient or out patient status), the use and timing of concomitant anxiolytics, enrollment to randomization ratios, time in stabilization before randomization, and the number of patients who met criteria and were not randomized at that point and by how long. It might be helpful to have translation of all source documents.
It is my opinion that it is reasonable to consider asking the sponsor to re-examine the data because although the trend in the U.S. conglomerate site is in the same direction as the final efficacy result, this is the sole study to support maintenance mono therapy and 1) the study significance seems to depend on site 93. Dropping site 118, a larger site, does not make the study lose significance.
2) the patients in the randomized phase are already an enriched group (about 64% of the patients who start stabilization complete stabilization) 3) there were some quality control type violations at the sites inspected and quality control type issues involved in the initial conduct of this study in general with 35 patients receiving unblinded medications after randomization 4) the drug did not show efficacy on-face in time to relapse for depression 5) this class of drugs is associated with EPS (including akathisia), NMS, and TD. With respect to study CN138010, safety assessments included adverse event reporting, measures of vital signs, EKGs, clinical laboratory tests, physical examinations, and body weight. The summary tables and analyses from the maintenance safety sample did not include data from the 35 patients who were treated with unblinded medications. Data from these 35 patients were reviewed for deaths and serious adverse events. One additional person excluded from the safety sample was a patient who became pregnant. This patient (141-266) was randomized but did not receive medication in the randomized phase.
Review of deaths, non-fatal serious adverse events, and discontinuations secondary to adverse events and review of adverse event terms for this trial as per the JMP dataset did not reveal any new or previously undescribed serious adverse events for the bipolar population.
Non-bipolar indications: • Review of incidence data for other indications, as supplied in incidences table in the ISS of this supplement and in supplement 002 submissions, was not performed as these tables are neither exposure nor placebo adjusted. • Line listings of patients who died, experienced a non-fatal serious adverse event, or discontinued secondary to an adverse event in studies blinded or newly reported since September, 2002 were submitted with supplement 002, the 120 day safety update, the response to the approvable, and supplement 005. As a point, the line listings of the deaths generally do not include the cause of death. • Review of these listings was cursory. Review of events that require further exploration is in progress but these data are not discussed in this review and will be completed as per Division leadership advisement.

Deaths
There were two (2) deaths reported in study CN138010; one aripiprazole patient died during the stabilization phase of the study from heroin intoxication (10-47-85) and one died from a suspected pulmonary embolism 61 days after discontinuation of aripiprazole (10-134-341). The narratives of these deaths are copied from the submission in the appendix of this document.
There were no deaths in study CN 13 803 7. . Psychiatric related events were the most common of these: "reaction manic" (3.4%), "depression" (2.9%), "reaction manic depressive" (2.5%), and "thought suicidal" (2.5%). There was one case each of pancreatitis (history of chronic pancreatitis with recurrent attacks), suicide attempt, chest pain, seizure, and spontaneous abortion. Double blind maintenance: 13.3% of the placebo group and 7.8% of the aripiprazole group experienced at least one serious adverse event.
• Placebo: reaction manic (6.0% placebo, 5.2% aripiprazole) and depression (3.6% placebo, 0% aripiprazole) were the most common serious adverse events. One suicide attempt occurred in the placebo group. • Aripiprazole: reaction manic (5.2%), paralysis (1.3 % aripiprazole, 0% placebo) and alcohol intolerance (1.3% aripiprazole, 0% placebo) were the most common serious adverse events in the drug treated group. No suicide attempt occurred in the aripiprazole group in this phase. The paralysis is indicated to have occurred after an automobile accident. Extension: 29.6% of the placebo group and 7.7% of the aripiprazole group experienced a SAE.
• Placebo group: reaction manic was the #1 event at 18.5% and anxiety and depression each accounted for 3.7%. • Aripiprazole group: reaction manic (5.1 %) and reaction manic depressive (2.6%) were the only reported SAEs.
Study 138037: Three patients experienced SAEs in study CN138037. These were events of manic-depressive reaction. One patient became pregnant (day 56) and was discontinued. During the double blind maintenance phase of the study, 19.3% of the placebo-treated patients and 10.4% of the aripiprazole treated patients experienced a treatment emergent adverse event (TEAE) that led to discontinuation.
During the extension phase, 25.9% ofthe placebo group and 10.3% of the aripiprazole group experienced a TEAE leading to discontinuation:.
Study CN138037: Four of the 24 patients discontinued secondary to an adverse event. The most common adverse event leading to discontinuation was reaction manic depressive in two patients.

Other Search Strategies
The lMP file of adverse events for study CN138010 was screened for terms coded (AETXT) hepatitis, liver failure, kidney failure, renal failure, rhabdomyolysis, and jaundice. No instances were seen. The laboratoryJMP files for Study CN138010 were screened for CPK elevations. The highest value noted was about 11,500. This patient is discussed elsewhere in the safety section of this review.

Common Adverse Events
The common adverse event profile for the bipolar population was characterized in the acute mania supplement.
Data from study CN138010 are somewhat difficult to interpret. The stabilization phase has no control group, however offers some idea perhaps of introduction to the drug. The maintenance phase is confounded by withdrawal in the placebo group, differential exposure time to placebo and aripiprazole, and selection bias. The extension phase group is small and likely reflects groups who suffer from selection bias. Patients were asked about adverse events by the investigator at weekly assessments beginning at the initiation of study treatment and recorded on the CRF. (A copy of the schedule of events, as duplicated from the submission, is included in the safety appendix of this document.) 7.1.5.2 Appropriateness of adverse event categorization and preferred terms • Preferred terms (PTERM) from the JMP dataset of adverse events for trial CN138010 were compared to the adverse event text terms (AETXT). There were no obvious systematic coding issues. There were some coding terms that should be coded to more appropriate terms. For example, patient 10-64-288 PTERM is akathisia when the text term was tardive dyskinesia and patient 10-34-94. AbnormalBehavior is the PTERMcheck this patient # while the text term describes intermittent, non, purposeful lip smacking. • Preferred terms scanned (cursory review) in other QUADR.xpt files submitted with the ISS to supplement 005 (limited to events since June, 2002) did reveal several occasions where coding might alter how serious the event would be perceived or would create the need to look multiple places in an incidence table to "cover" an occurrence. Additionally, there were missing preferred terms for which text terms were present. The safety appendix of this document contains further information. It is recommended that we ask the sponsor to link these missing terms to an appropriate preferred term and optimize the translation oftext from the CRF to preferred terms.

Incidence of common adverse events
The adverse event profile of the drug was derived from the acute studies.

Additional analyses and explorations
(Copies of the study procedures/schedules are in the safety appendix of this document.) EPS: The sponsor performed scales directed at assessing treatment emergent Parkinsonism (Simpson-Angus Scale, SAS), dyskinesia (Abnormal Involuntary Movement Scale,AIMS), and akathisia (Barnes Akathisia Global Assessment Score) during stabilization, maintenance, and extension phases. SAS scores range from 10-50, AIMS total scores range from 0-28, the Barnes Akathisia scale scores range from 0 to 5 ( 5 =severe). With all scales, a negative change indicates improvement. .
Although the data generated are confounded by withdrawal from drug in the placebo group, by differential exposure times to drug in the groups, and possibly by concomitant medication use, the difference in the mean changes from baseline measures between the groups is small anyway and likely not clinically meaningful.
EPS-related adverse events: The data for these events are difficult to interpret secondary to the study design and are not discussed in detail in this part of the review secondary to this. A group (1/77) discontinued secondary to akathisia in the maintenance phase; no placebo patient did (0/83).
Suicidality: The interpretation of the change in MADRS data is also questionable as there is a differential exposure time between the placebo and aripiprazole groups in the maintenance phase. MADRS scores Were acquired at every study visit during the maintenance study. The MADRS item 10 score was used to assess treatment emergent suicidality. The sponsor notes that among patients with a baseline score of 0-2, the incidence of scores of 5-6 at any time during the study was 0% for (0/76) for the aripiprazole patients and 1.25% (1/80) for the placebo patients. This analysis appears to have been performed on the maintenance phase sample.
• Treatment emergent adverse events 1) Two events classified as "suicide attempt" occurred in the stabilization phase and 17 events (3.07%) of "thought suicidal" occurred. 2) One event of suicide attempt occurred in maintenance phase; this was a placebo patient. Two events of "thought suicidal" occurred in placebo patients and one in aripiprazole patients (1.30%) during the maintenance phase. 3) One event of "thought suicidal" occurred in the placebo group during the extension phase.
2) During maintenance, there was 1 suicide attempt in a placebo patient and one incidence of "thought suicidal". • Discontinuation secondary to suicide related events: 1) One discontinuation secondary to suicide attempt and 13 secondary to "thought suicidal" occurred in the stabilization phase. 2) One discontinuation secondary to suicide attempt occurred in a placebo patient during the maintenance phase (0 in the aripiprazole group) and two discontinuations secondary to "thought suicidal" occurred in the placebo group with 0 in the aripiprazole group.
Glucose Metabolism: The sponsor notes that no adverse events related to glucose metabolism were reported in either treatment group during the longer-term maintenance study (ISS-005).
Overdose: The sponsor searched the database for all Phase 2/3 studies to identify overdose of aripiprazole defined as >60 mg. Since the safety update of 2002, 11 patients were identified. None of these were from the bipolar mania trials.
Abuse, tolerance, and physical dependence have not been specifically studied in humans using aripiprazole. Seizure: One patient experienced an event captured as seizure-related in the stabilization phase. No patients experienced a seizure related event in the combined maintenance and extension phases. phase (10-509, 141-266), one in the extension phase (100-116), and one before treatment (91-181). One patient became pregnant during study CN138037. This patient and patient 141-266 terminated or induced abortion. The baby of patient 100-116 experienced shoulder dislocation and jaundice which reportedly resolved.
Patient 10-509, a 44 year old female, experienced spontaneous abortion on day 151. At the time of discontinuation secondary to severe depression on day 112, her pregnancy testing was negative. It appears she was treated with risperidone and buproprion with resolution of the event noted on day 138 when she returned for a follow-up visit. At this visit, she expressed that she suspected she was pregnant. This was confirmed on day 141. The patient could not recall the date of her last menstrual period. However, the narrative notes the gestational age when pregnancy was "diagnosed" was estimated at 1-4 weeks. Concomitant medication use of oral contraceptive appears to be before the pregnancy (days 38-92), if this gestational age is correct. This patient had no previous history of spontaneous abortion or stillbirth.

Less Common Adverse Events
In study CN1380l0, one case ofRaynaud's and a case of retrograde ejaculation were noted in the maintenance population. Three patients are coded as tardive dyskinesia in the JMP set. Two were placebo patients who experienced dyskinesia in the maintenance phase and one was an aripiprazole patient in the extension phase.
The appendix with this listing may be found in the submission to supplement 002 dated May 26, 2004 (p.l 09-150). Labeling'review as per Dr. Andreason.

7 Laboratory Findings
Two patients dropped out of therapy in the stabilization phase secondary to elevated CPK values; patients 99-229 (AEs include leg cramps and myalgia) and 108-348 (SAE of mania; foot wounds, patient also had increased LDH). Narratives indicate that both patients experienced elevations of CKICK-MB while on 30 mg daily of aripiprazole. Neither patient was coded as having NMS or rhabdomyolysis. An additional patient was (146-459) was originally noted to be dropped out secondary to elevated prolactin, however it was found this was incorrect as this elevation occurred after discontinuation for other adverse events.

Overview of laboratory testing in the development program
In the stabilization phase, screening EKG, urine and blood samples for routine hematology and chemistry laboratories, urinalysis, pregnancy, and drugs of abuse were not required for patients from the acute mania trials. Prolactin levels were measured during stabilization at baseline if the patient had not entered from an acute study.
During the maintenance phase and extension phases, samples were collected for routine laboratory analysis at scheduled intervals (a schedule of events may be found in the appendix of this document). Prolactin levels were measured at randomization and throughout the doubleblind phases.
Quality Control issues: It appears from the amended appendices of protocol deviations, there were a fairly large number of patients who did not get lithiumlvalproic acid levels or had other missing labs at stabilization. About 25 patients had some clinical lab tests missing on or before the maintenance phase start date and five women did not receive pregnancy testing on or before the maintenance phase dose start date. Dr. Khin's inspection report noted that some laboratory measures were missing on many patients at site 93.

7.2 Selection of studies and analyses for drug-control comparisons of laboratory values
Study CN138010 was the only study submitted to support the indication for maintenance use in bipolar patients.

Analyses focused on measures of central tendency
Median change from baseline data were provided for the maintenance phase and combined for the maintenance and extension phases. These data were not reviewed as interpretation is problematic for the reasons discussed previously LFTs: The ISS notes that there were four patients with treatment emergent abnormal hepatic laboratory measures. One patient had both an elevated AST and AL T during stabilization, two patients had a single transaminase elevation, and a fourth had an elevated bilirubin at the visit prior to the last visit. It was noted that none of these patients had simultaneous elevations of transaminase and bilirubin.

Prolactin:
Stabilization: Although the CRS noted one discontinuation secondary to elevated prolactin, upon further information, this was not correct as it appears the elevated prolactin level was after discontinuation.

Vital Signs
Quantitative interpretation of these data are limited secondary to similar issues as discussed in section 7.1.5 Common Adverse Events.

Overview of vital signs testing in the development program
Supine and standing systolic and diastolic blood pressures and radial artery pulses were measured at scheduled visits after the patient was supine for five minutes. Upon standing, the measurement was taken after two minutes. Vital signs scheduled on simultaneous visits as blood draws were measure before the blood draw. (A schedule of events is included in the safety appendix of this document.) Quality Control: Additionally, Dr. Khin' review noted stabilization phase EKGs on 17 patients at site 118 were noted as missing in the original study report, however, upon inspection, the sponsor had identified all except one. The updated appendices noting violations have adjusted for these EKGs (8-27-04 submission).

Selection of studies and analyses for overall drug-control comparisons
Study CN138010 is the only maintenance study with placebo-controlled data

Standard analyses and explorations of vital signs data
Review of the analyses of central tendency was not performed. Outlier data review was limited. The interpretation of these data given this study design is problematic. This review has focused on serious adverse events and drop-outs secondary to vital sign related events.

Marked outliers and dropoutsjor vital sign abnormalities
During stabilization three patients treated with aripiprazole dropped out of therapy secondary to a vital sign related adverse event; hypotension (10-359), syncope (71-59-amphetamine use also had reportedly seizure) and tachycardia (64-605). One aripiprazole patient discontinued therapy during the maintenance phase for hypertension (73-574).

Additional analyses and explorations
QTc: QT data were described in the review for the indication of acute mania. Due to the difficulty in interpretation of these data secondary to the study design, limited discussion of QT is included in this review other than section 7.1.9.3.2 below.
Orthostatic Blood Pressure Measures: An orthostatic blood pressure measure was defined as any systolic blood pressure decrease ~ 30 mm Hg supine to standing. Interpretation of this type of infonnation is also limited. In the combined extension and maintenance phases, the difference between the treatment groups in patients experiencing orthostatic blood pressure measures was about 7%.
The overall incidences of orthostatic-related adverse events during the maintenance phase were about the same in the two groups. Syncope was seen in one placebo patient and no aripiprazole patient.
Body weight: Body weight and waist circumference were recorded at scheduled visits (see appendix for copy of schedules) on the same scale for a given patient and in a standardized manner. Waist circumference was measured at the level of the umbilicus.
Most of the data from CN138010 relative to this will not be discussed in detail as the interpretation is limited. The mean change from baseline in patient weight using endpoint LOCF in the stabilization safety sample was only computed on patients who discontinued because weight was not measured in patients who continued. The mean change in 308 patients was 0.16 kg.
7.1.9 Electrocardiograms (EeGs) 7.1.9.1 Overview of ECG testing in the development program, including brief review of preclinical results Twelve-lead EKGs were acquired during study CN138010. The schedule of procedures may be found in the safety appendix of this document.
7.1.9.3 Standard analyses and explorations ofECG data 7. J. 9.3. J Analyses focused on measures of central tendency These data were not reviewed due to problems with interpretation as previously discussed throughout this review.

Analyses focused on outliers or shiftsfrom normal to abnormal
The following infonnation is provided, however, the interpretation of this is limited. During the stabilization phase, two patients had potentially clinically significant QT interval changes( ~450

34
Clinical Review Teresa A. Podrucbny NDA 21436_005 Abilify (aripiprazple) msec and 10% increase from baseline) when corrected with Bazett's formula and none did when corrected with the DNDP formula (QTcN=QT/RR). The sponsor notes that no aripiprazole treated patient had PCS QTc changes during the Maintenance or Extension phases and no patient discontinued secondary to a QTc abnormality.

Marked outliers and dropoutsfor ECG abnormalities
One patient discontinued for an EKG related event (109-67) that was an SAE. This patient was hospitalizaed secondary to bigeminy on day 56 of the study and medication was discontinued. The event was reported as resolved on day 58. The CRF notes there was no previous cardiac history. The patient was noted to have palpitations on a visit 19 days earlier.
There was symmetrical T-wave inversion in one175 aripiprazole patient (36-399) in the . maintenance phase and 0177 placebo patient. There was no adverse event listed in the line listing that would correlate with this EKG abnormality. No additional studies of abuse were submitted with this submission.

Human Reproduction and Pregnancy Data
Pregnancy issues in longer term studies of mania were discussed above in section 7.1.5.6. A complete listing of patients who have become pregnant while on aripiprazole treatment is included in the appendix of this document.

Assessment of Effect on Growth
Not assessed in this supplement.

Overdose Experience
No overdoses of study medication occurred in study CN13801O.

Postnlarketing Experience
The postmarketing information with supplement 002 encompasses or supercedes that of this submission. There were four cases of DVTIPE events in the reporting period from July 17,2003 -January 16,2004. One was of DVT, two were PE, and one case with PE and DVT. Although there are confounders in these cases such as history of smoking, obesity, or the use of other medications such as clozapine, venlafaxine, or risperidone, causality cannot be totally ruled out. One case of pUlmonary embolism occurred in a 27 year old bipolar patient treated for one month.[.J

b(5) C ~
The sponsor also notes that cumulative reports of hypersensitivity, hepatobiliary events, increased creatine phosphokinase/rhabdomyolysis, and syncope "suggest a possible causal b(5) relationship". C ::J • Hepatobiliary events included 17 cases of AST/ALT elevation-lO were classified as serious. Values for the peak ALT were reported in 14 cases, the highest was 684. Values for the peak ALT was reported in 8 cases and was 374. The sponsor notes that in the majority of the 17 cases although confounded, a causal role for aripiprazole could not be totally ruled out and that in six cases there was positive dechallenge. A hepatitis reported as drug induced occurred in a 34 year old patient who was taking concomitant medications and had taken a months worth of multivitamins at once one week prior to the event. However, the synopsis notes her AST/ALT were normal two months before and returned to normal after discontinuation of aripiprazole. • One case synopsis of increased bilirubin (12379806)  The mean age at randomization was similar between groups. About 60% of the aripiprazole patients in the maintenance phase were female and about 70% of the placebo patients. Most patients in both groups were White (67% aripiprazole, 62% placebo). Hispanic/Latino comprised the next largest group representing 26% of the aripiprazole patients and 20% of the placebo patients.
With respect to psychiatric history, 78% (458/633) ofthe enrolled patients were not rapid cycling patients. Most patients enrolled were coded as current episode manic (61 %). In the randomized population, about 17% were rapid cycling patients and most were coded as current episode manic (78% placebo and 62% aripiprazole).

Postmarketing experience
The postmarketing information with supplement 002 encompasses or supercedes that of this submission.
There were four cases of DVT/PE events in the reporting period from July 17,2003 -January 16,2004. One was ofDVT, two were PE, and one case with PE and DVT. Although there are confounders in these cases such as history of smoking, obesity, or the use of other medications such as clozapine, venlafaxine, or risperidone, causality cannot be totally ruled out. One case of lulmonary embolism occurred in a 27 year old bipolar pa~nt treated for one month-[l b(5) The sponsor also notes that cumulative reports of hypersensitivity, hepatobiliary events, increased creatine phosphokinase/rhabdomyolysis, and syncope "suggest a possible causal  clinical labs. At one ofthe sites in Mexico, 17 EKGs were uncovered at an audit by the sponsor. These EKGs were reported as missing in the stabilization phase in the Clinical Study Report and therefore probably are not captured in summary tables. Serious EKG or laboratory-related events should have been captured as serious adverse events elsewhere, so this may not be a practical safety issue but it does raise issues as to how sites were monitored and how quality control was conducted during compilation of the study report.
The coding of text from the CRFs to the COSTART term for study CN138010 appears generally to be adequate and is discussed in section 7.1.5.2. Recommendations to the sponsor regarding the non-bipolar indications and coding are included in sections 7.1.5.2 and 9.5.
• With regard to non-bipolar indications, the incidence tables of adverse events by indications for all aripiprazole treated patients cannot be meaningfully interpreted and were not reviewed as tables are not exposure and placebo adjusted. • Line listings of patients who died, experienced a non-fatal serious adverse event, or discontinued secondary to an adverse event in studies blinded or newly reported since September, 2002 were submitted with supplement 002, the 120 day safety update, the response to the approvable, and supplement 005. Line listings of the deaths generally do not include the cause of death.

Additional Submissions, Including Safety Update
There were no safety updates for supplement 005. Data from other submissions such as line listings as discussed in section 7.1 were utilized as were safety updates from submissions to S002.

Summary of Selected Drug-Related Adverse Events, Important Limitations of Data, and Conclusions
Drug related adverse events were captured in the acute studies. Limitations of the data are discussed above.

Dosing Regimen and Administration
Study CN138010 was not a fixed dose study. At the end of stabilization, the mean daily dose was about 25mg with a range from approximately 9-34 mg. Of patients who completed the stabilization phase and remained eligible for the maintenance phase (161), the mean dose was about the same at 24 mg and the range was about l3 to 30 mg daily. 63% of these patients were on 30 mg daily at endpoint and 37% were on 15mg daily.

Drug-Drug Interactions
There is no new information provided with the supplement. This is an update to a previous article published in 2000. MEDLINE was searched (1999-2003 using terms for various drugs such as the atypical antipsychotics, including aripiprazole, with the terms hypomania and mania. 34 new cases of mood switch were noted. The authors conclude that more than half of the new cases are "highly suggestive" of a causal link. The majority of these cases did not have a diagnosis of bipolar disorder and many were schizophrenia or schizophreniform disorder. The authors note that no reported cases were with clozapine. Although none were with aripiprazole, the authors note that the lack of reporting with aripiprazole and sertindole may reflect worldwide drug use.

Recommendation on Regulatory Action
I recommend the Division consider ali approvable action on this supplement. While the p value is significant when comparing aripiprazole to placebo for time from randomization to relapse in the maintenance phase, it is unclear to me that this reflects efficacy of the drug. While the larger pool of data favors aripiprazole, removal of one site in Mexico (site 93) causes the study to lose significance. This site appears to have a different relapse rate than the conglomerate u.s. sites.
DSI inspection at this site revealed protocol violations, however, overall the data were deemed acceptable. As this is the only study for maintenance and given that a large number of U.S. sites were involved but alone are not powered to show significance and for other reasons listed within this review, I recommend we ask for further exploration of the data in this study with attention to the Mexican sites, more so to site 93.

Required Phase 4 Commitments
Required Phase 4 commitments were delineated in the action letter for this supplement. As commitments were made regarding adult studies to address short and longer term efficacy as add-on therapy in bipolar patients and pharmacology-toxicology studies needed to support pediatric trials with the action on supplement 002 (acute mania), no additional studies are required at this time. However, the sponsor was asked to state a date of submission of the clinical study reports for the recently completed drug interaction studies.

Comments to Applicant
It is recommended that the sponsor re-examine the JMP databases for ISSQADRl, ISSQADR2, and ISSQADR3 for text terms that are missing preferred terms. Additionally, there appear to be terms such as "Abnormal Lab" or "Abnormal EeG" when more optimal terms for these instances would have noted that these were clinically relevant or at high levels (for example, patient 4-83-271 and patient 4-68-582). It is recommended that preferred terms be modified to more accurately reflect the potential seriousness of the text terms.

INCLUSION AND EXCLUSION CRITERIA: INCLUSION:
Inclusion Criteria into the Stabilization phase: 1) Patients with DSM-IV diagnosis of Bipolar I Disorder who had experienced at least 2 previous manic or mixed episodes including the most recent episode. 2) Patients who experienced a recent manic or mixed episode requiring hospitalization and treatment with medications that began no more than 3 months before entry into the stabilization phase 3) Patients who were eligible for an acute mania study but declined. 4) Men and women 18 and older. Women of child bearing potential must have had a negative pregnancy test within 72 hours of starting study medication, were to use an acceptable form of contraception, and could not be pregnant or lactating. 5) Patients were able to give informed consent or had an acceptable legal representative ot give consent prior to initiation of any protocol procedures 6) Patients were able to comprehend and satisfactorily comply with the protocol. Inclusion Maintenance: 7) Patients who continued to meet criteria I, 3, and 5. 8)· Patients who were in the stabilization phase for at least six weeks. 9) Stable as per YMRS ~ 10 and MADRS ~ 13 during 4 consecutive visits. Inclusion Extension: 10) Patients who completed 26 weeks of maintenance. II ) Women as in criteria 4 above. 12) Patients deemed suitable for participation in a long-term trial, for example, regarding compliance.
Th" boby "'". hom ,,;t1o a dL<k"'''.lIwd ,*,uldCf "')(\ jmlndi""ll",! ',,«lived within 2·jl",w·,. ~ N:uralh.",. fur l\lc!lC ""lients nr" J>1c.cro(~d in tho Cli"i»:.l SIIl<i'1 Rcp,lIt for Ihnt <1ud,'. This is a representation of an electronic record that was signed electronically and this page is the manifestation of the. electronic signature. Aripiprazole is an atypical antipsychotic agent that is currently approved for the acute and maintenance treatment of schizophrenia and for the treatment of acute manic or mixed episodes associated with bipolar disorder. This supplement (S-005) was submitted on J.-28-04 to gain approval for maintenance treatment in bipolar disorder. Supporting clinical data was derived from a single trial (CN138-010) which consisted of an open-label aripiprazole stabilization phase followed by randomization of stabilized patient to continued aripiprazole treatment or placebo for a 26 week maintenance period; patients who remained stable during the 26 week period could receive open-label aripiprazole during an extension phase.

/s/
The clinical data was reviewed by Dr. Teresa Podruchny and an approvable letter for S-005 was sent on 11-30-04. The sponsor was informed that the following issues would need to be addressed before this supplement could be approved: 1) The efficacy results of study CN138-010 appear to be driven by a single center (93) in Mexico City.
The treatment effect at this center differs markedly from the effect observed at u.S. centers.
This finding raises questions about the reliability of the efficacy results. The sponsor was requested to address this concern.
2) A number of patients dropped out of study CN138-010 during the maintenance treatment phase but were not counted as relapses.
The available information on these patients 1 was insufficient to assure us that these patients did not prematurely discontinue study participation due to worsening bipolar disorder. The sponsor was requested to carefully reexamine the data for these patients and provide us with a sufficiently detailed description of the reasons for dropout so that we could independently verify that the patients were not relapsing at the time of discontinuation. If any patients were reclassified as relapses, we asked the sponsor to reanalyze the data after reclassification.
3) The draft labeling attached to the approvable letter included a new section under WARNINGS that described the risk of cerebrovascular adverse events (CVAE's) in elderly patients with dementia who received aripiprazole. Additionally, we asked that all previous revisions to labeling, as reflected in the most recently approved package insert, be included. We requested that their revised labeling proposal clearly indicate all changes.

4) We requested written agreement regarding all pertinent Phase 4 commitments:
a) adult clinical studies to address the clinical efficacy and safety of aripiprazole as add-on therapy in bipolar disorder. b) juvenile animal toxicity studies to support pediatric studies of aripiprazole in bipolar disorder. c) drug interaction studies with lithium and valproate.
The first two commitments were agreed to as part of the approval action of S-002 (for the treatment of acute manic and mixed episodes associated with bipolar disorder) .
This submission contains adequate responses to all issues above. These responses are summarized below.

A. Center 93
Center 93 had the second largest number of randomized patients (N=13 or 8% of the total sample) and had results strongl y favoring drug over placebo (0% (0 / 6) of aripiprazole patients relapsed compared to 71% (5/7) of placebo patients). Based on crude relapse rates, the results at this center were markedly superior to those in the overall study (28% for aripiprazole and 51% for placebo).
In the primary analysis of time to relapse using Kaplan-Meier methodology, the log-rank p-values based on all sites was 0.020; with center 93 excluded, the p-value becomes non-significant: 0.104. BMS asserts that this loss of significance with the exclusion of center 93 is, in part, due to loss of statistical power. The sponsor also points out that there was no a priori rationale to exclude the data from center 93 in the efficacy analysis.
The FDA Division of Scientific Investigations (DSI) inspected center 93 and found the data from this site ·to be acceptable despite a number of deficiencies (see 11-3-04 review by Dr. Ni Khin). Additionally, BMS states that their internal monitoring reports were consistent with the DSI conclusions. Due to the small numbers of patients enrolled at most of the sites in study CN138-010, statistical testing for an interaction between treatment and center was not feasible. 1 The sponsor also cites the crude relapse rates for all relapse, for depressive relapse, and for manic relapse as well as the mean changes in the Y-MRS from randomization to endpoint with all sites versus with center 93 excluded to demonstrate minimal impact from excluding center 93.
Since the primary and key secondary variables were time to event and not crude relapse rates or mean change from baseline, this data has less relevance.

Reviewer's Comments
In the absence of evidence of a study characteristic which would have produced a biased result at center 93, this site should be retained in the primary analysis. This is consistent with the intent-to-treat principle. Also, in my opinion, to embark on a policy of discarding a site solely on the basis of an outlying result invites a number of difficult questions, such as how to quantitatively define an outlying result which merits exclusion of the site and whether we would allow the exclusion of a very poorly performing site which would render an otherwise negative study positive.
Furthermore, it is not entirely clear that quantitatively defining an outlying result that merits exclusion is a reasonable approach because it begs the question of whether a deviant finding was part of the natural variation in the responsivity of the illness to the drug versus the result of a peculiarity in the patient sample or study conduct. This question would be extremely difficult to answer in many cases. Admittedly, my approach does not preclude the possibility of a source of bias that has been undetected. The inspection done by DSI and monitoring performed by BMS may not have been sufficiently sensitive to reveal the source of the remarkable result at center 93. Nonetheless, without a clear reason to do so, I do not feel that the exclusion of data from center 93 is warranted.

B. Potential Misclassification of Relapse
A total of 38 patients in study CN138-010 dropped out during the maintenance phase for reasons other than relapse, which was defined a priori as either 1) hospitalization or 2) the addition or increase in allowed psychotropic medication for manic or depressive symptoms.
Individual patient data for all 38 patients was provided by the sponsor. This information consisted of a narrative summary; Y-MRS, MADRS, and CGI scores during the maintenance phase; and the End of Study form completed by the investigator.
The sponsor reexamined the information for these patients and concluded that in 12 cases worsening of bipolar illness could not be absolutely ruled out as a reason for dropout. If these 12 patients were reclassified as relapses, aripiprazole remained superior to placebo in the survival analysis of time to any relapse (p=O.033) and time to manic relapse (p=O.003). The comparison for time to depressive relapse remained non-significant.
Reviewer's Comments In the above reanalysis, the sponsor appeared to have used very broad criteria for relapse.
To verify that the results would not be changed by using stricter criteria, I personally examined the patient data for all 38 patients to identify any where relapse could be reasonably inferred. My examination revealed two patients who, in my opinion, appear to have experienced a relapse prior to discontinuation from the study: • Patient 118-438 (placebo) and  In both cases, the relapse appeared to be a manic episode. Based on submitted data for these patients, I assumed that the times to relapse were days 128 and 16 of the maintenance phase, respectively. The Statistical Reviewer, Dr. Kun He, then reanalyzed the data assuming that these two patient experienced a manic relapse on the above days. In the survival analyses for any relapse and for manic relapse, aripiprazole was superior to placebo (p-values of 0.0216 and 0.0104, respectively).2 Thus, the reclassification of these two cases did not change the conclusions of the original analysis. c.

Labeling
The sponsor included proposed labeling in this submission. The following revisions (indicated by strikethrough font or underlining) to the approvable labeling are proposed by the sponsor: • INDICATIONS/Bipolar Mania, third paragraph: The efficacy of ABILIFY in maintaining efficacy in patients with Bipolar I Disorder with a recent manic or mixed episode who had been stabilizedC question of how long a patient treated with aripiprazole should remain on it, patients with Bipolar I Disorder who had been symptomatically stable on ABILIFY (15mg/day or 2 The results for depressive and mixed relapse remained unchanged since these two patients were assumed to have had manic relapses. These results were communicated to me by Dr . He in an Email on 1-18-05. 30mg/day with a starting dose of 30mg/day) for at least 6 E ~ weeks and then randomized to ABILIFY (15mg/day or 30mg/day) or placebo ~ ~ for relapse demonstrated a benefit of such maintenance treatmentL ~ it is generally agreed that pharmacological treatment beyond an acute response in mania b(4) is de~irable, both for maintenance of the initial response and for prevention of new manic episodes, there are no systematically obtained data to support the use of aripiprazole in such longer-term treatmentC J L J I have no strong objection to the above changes proposed by the sponsor. However, there are three additional revisions that are warranted: 1) The survival analysis of any relapse and of manic relapse, the latter being one of two key secondary variables, showed a statistically significant advantage of aripiprazole over placebo. However, the survival analysis of depressive relapse, the other key secondary variable, did not demonstrate statistical superiority of aripiprazole over placebo. Without an active control, it is impossible to know whether this represents lack of effectiveness of aripiprazole in delaying depressive relapse or an inability to detect such an effect due to other factors in this trial. Nonetheless, it is important for prescribers to be aware that such an effect for depressive relapse has not been demonstrated. This information regarding the key secondary variables should be added to both the Clinical Trials section of CLINICAL PHARMACOLOGY and INDICATIONS.
2) Final language for the labeling of information regarding cerebrovascular adverse events in elderly patients with dementia under WARNINGS was successfully negotiated with the sponsor on 1-19-05. 3 This language is as follows:

Cerebrovascular Adverse Events, Including Stroke, in Elderly Patients with Dementia
In placebo-controlled clinical studies (2 flexible dose and 1 fixed dose study) of dementia-related psychosis, there was an increased incidence of cerebrovascular adverse events (e.g., stroke, transient ischemic attack), including fatalities, in aripiprazole-treated patients (mean age: 84 years; range: 78-88 years).
In the fixed dose study, there was a statistically significant dose response relationship for cerebrovascular adverse events in patients treated with aripiprazole. Aripiprazole is not approved for the treatment of patients with dementia-,related psychosis. (See also PRECAUTIONS: Use in Patients with Concomitant Illness: Safety Experience in Elderly Patients with Psychosis Associated with Alzheimer's Disease.) 3) The section under ADVERSE REACTIONS entitled Additional Findings Observed in Clinical Trials/Adverse Events in a Long-Term, Double-Blind, Placebo-Controlled Trial should clarify that the cited figures were derived from the longterm study in schizophrenia and that similar findings were observed in the long-term study in bipolar disorder C' 3

D. Phase 4 Commitments
In the cover letter of this submission, BMS commits to providing reports of the two lithium/valproate interaction studies by 6-30-05.
Regarding the adult studies of aripiprazole as add-on therapy and the juvenile animal toxicity studies to support pediatric studies in bipolar disorder, the sponsor has reaffirmed their commitment to complete these trials and submit final study reports on or before 9-30-09 and 6-30-06, respectively.

III. Conclusions and Recommendations
The sponsor has provided reasonable responses to our concerns about the robustness of the efficacy results from study CN138-010. Additionally, both the sponsor and I have proposed some minor revisions to labeling which are described above . Lastly, the sponsor has agreed to the three Phase 4 commitments delineated above.
From a clinical standpoint, once final revisions to labeling have been agreed upon, this supplement may be approved.
1 I concur with Dr Dubitsky's view that site 93 should be included in the ITT analysis as the basis for approval of this submission. 2 I concur with Dr. Dubitsky' assessment of the efficacy analysis and re-analysis that explores the results of the study based on some deviations in patients' dropout status. 3 The sponsor revised draft labeling differs slightly yet significantly from the Division's previously proposed labeling in the Approvable Action letter of November 30,2004. This draft labeling implies a maintenance claim of ,/ -weeks. I do not agree with this language.

b(4)
4 The sponsor has agreed to the Divisions proposed phase IV requirements.
CONCLUSIONS/RECOMMENDATIONS I agree with Dr Dubitsky that the sponsor has adequately addressed our questions about site 93 and addressed our concerns about proper accounting of patients who dropped out due to relapse versus other reasons. I also note that the sponsor has committed to the Divisions proposed phase IV studies. Negotiating mutually agreeable labeling remains the only outstanding task to perform before approval ofthis supplement.

LABELING
There are two outstanding labeling topics: the length of the implied maintenance claim, and whether or not differential treatment effects on depression and mania should be mentioned.
• Length of Maintenance Claim-The sponsor has submitted draft labeling that refers to the C. 1 [ :I: The Division has adopted the policy that the length oftime that patients are stable in the open-label treatment phase shall be used as the description of the length ofa maintenance claim that is supported by a relapse-prevention designed study. The double blind "observation" period of the study tests whether treatment with aripiprazole that was continued to that point is still helpful at that point.
• Differential Effects on Treatment of Mania and Depression-Dr Dubitsky suggests that we include labeling that states that Abilify was not effective in preventing depressive relapse and the study was positive solely on the basis of preventing manic relapse. I am not sure that this distinction needs to be made; however, there is some precedence for doing this. Lamotrigine labeling for the maintenance treatment ofbipolar disorder mentions a sub-analysis that states that the strongest treatment effect was seen for preventing relapse of the depressive state. In the case ofLamotrigine, there were two pooled studies from which this conclusion was drawn, and there was no acute treatment effect. In one study, the patients entered with an acutely depressive index episode and in the other patients entered with an acutely manic index episode.
In this case with Abilify, there is an approved acute treatment with only one maintenance study that was positive. The requirement for only one positive study was approved in advance because Abilify was approved for acute treatment of bipolar disorder. In that study patients entered after having an initial manic episode. I hesitate to draw a conclusion of a differential effect on depression from just this one study of initially manic patients especially when we are only granting an additional6-weeks efficacy. Nonetheless, I do not strongly object either.
The study was originally designed to observe for both manic and depressive relapses in the primary efficacy analysis. The sponsor was able to pass this overall test and therefore may claim 6-weeks of maintenance efficacy. Had the study only been positive for maintenance of non-mania, but had failed the overall test, we would not have approved it for this indication.

APPEARS THIS WAY
ON ORIGINAL This is a representation of an electronic record that was signed elec~ronica"y and this page is the manifestation of the electronic signature. .:l c: :l This newly provided analysis was part of a response to draft labeling that was proposed by the -b(4)

/s/
Division. Item #5 ofthis e-mail from the company stated, "ECGs: During the fmallabeling discussions for the acute mania approval, Dr. Katz requested that we submit a pooled analysis of the ECG data from the short term schizophrenia and bipolar disorder studies in conjunction with this application. We are sending the pooled ECG results in the attached document to support the change to this section." The results ofthe pooled analyses may be found in the appendix of this memo.
The currently approved Abilify labeling states, Between group comparisons for pooled, placebo-controlled trials in patients with schizophrenia, revealed no significant differences between aripiprazole and placebo in the proportion of patients experiencing potentially important changes in ECG parameters; in fact, within the dose range of 10 to 30 mg/day, aripiprazole tended to slightly shorten the QTc interval. Aripiprazole was associated with a median increase in heart rate of 4 beats per minute compared to a 1 beat per minute increase among placebo patients.

Review of Pooled ECG Analysis
The analysis ofthe pooled ECG data does not give data on the dose range of from 1O-30-mg( J C ~~ ~ new analysis also shows a mean increase in the heart rate of 5 beats per minute instead (BPM) of 4 BPM. Pooled placebo patients continued to show a mean increase in heart rate of I BPM in the placebo group.

Page 1
The mean changes in QTcE and QTcN were roughly equal. The Abilify QTcB showed a lesser decrease over the treatment period than placebo. This is likely due to the greater mean heart rate with Abilify treatment over placebo. Given this higher mean heart rate with Abilify treatment, the QTcB will show a falsely increased duration and is theref()re not an appropriate correction method for raw QT. Therefore, QTcN and QTcE are more appropriate correction methods over QTcB.
Outlier analyses ofQTcN and QTcE also. showed a roughly equal proportion of patients meeting outlier criteria. QTcB showed a higher proportion of patients meeting outlier criteria for the aripiprazole group; however, due to the higher heart rate with aripiprazole the QTcB is not an appropriate correction method.

CONCLUSIONS/RECOMMENDATIONS
Given this new analysis I propose the following labeling changes to the current ECG section of labeling: Between-group comparisons for a pooled analysis of placebo-controlled trials in patients with schizophrenia or bipolar mania, revealed no significant differences between aripiprazole and placebo in the proportion of patients experiencing potentially important changes in ECG parameters. Aripiprazole was associated with a median increase in heart rate of 5 beats per minute compared to a 1 beat per minute increase among placebo patients      a Includes all patients with both a baseline and an endpoint measurement.

ON ORIGINAL
b Includes all patients with an on-study measurement.
c Includes all patients with both a baseline and an on-study measurement.

APPEARS THIS WAY
ON ORIGINAL  Includes all patients with an on-study measurement.
Includes all patients with both a baseline and an on-study measurement. 17. ADDITIONAL COMMENTS: The applicant has not identified any changes to the CMC portion of this application.

CONCLUSIONS & RECOMMENDATIONS:
The applicant has provided adequate information to support this change. From a CMC perspective, it is recommended that this supplement be APPROVED. Evaluation: Adequate Based on 21 CFR 25.31(b), a categorical exclusion should be granted as the expected introduction ofthe substance at the point of entry into the aquatic environment will be below 1 ppb.

Appears Thts Way On Original
This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature.

3-dichlorophenyl)-I-piperazinylJ butoxyJ-3, 4dihydro-2 (1 H)-quinolino, ne
Formulation: Tablet (5 mg,  The submitted study demonstrates that aripiprazole is a partial D2 agonist in pituitary cells in vitro. The authors transduced retrovirally the short or the long form of human dopamine D2 receptor gene into rat pituitary cell line (GH4Cl) and examined the effect of aripiprazole on prolactin release and cAMP accumulation in either D2L or D2S receptor expressing GH4Cl cells. Aripiprazole inhibited forskolin-stimulated prolactin release in both D2Lor D2S receptors, however the maximal inhibition of prolactin release was less than that of dopamine. In addition, aripiprazole antagonized the suppression attained by dopamine in both cells. The maximal inhibition of prolactin release and cAMP level by aripiprazole were greater for the D2S-than for D2L-receptor expressing cells. Saturation binding analysis showed that the maximal binding capacity was approximately 4-fold higher at the D2S-than at D2L-receptor expressing cells, while affmity was similar at these cells. The results indicate that "aripiprazole acts as a partial agonist at both D2S and D2L receptors expressed on rat pituitary cells with high affmity, and that its agonist-antagonist properties may depend upon the amount of D2 binding capacity on the cells." Study 015812: Validation of the assay method for OPC-14597 in 1% lactic acid solution by high performance liquid chromatography. This study validated a modified HLPC method for analyzing the purity of OPC-14597 (dissolved in 1 % aqueous solution in lactic acid) with regard to specificity, linearity, accuracy, reproducibility, precision, and stability during the assay period.

Conclusions and Recommendations
The conclusion is that the primary analysis for the time from randomization to relapse during the maintenance phase is significant comparing aripiprazole and placebo in evaluating subjects with Bipolar I Disorder but one should consider whether the quality of operations in center 093 is high, which was suggested to be inspected by DSI, when making final decision.

Brief Overview of Clinical Studies
This was a randomized, double-blind, multicenter, placebo-controlled trial in USA, Mexico, and Argentina, evaluating the use of aripiprazole in the maintenance of stability of patients with Bipolar I Disorder. There were 3 phases in this study: a Stabilization Phase, a Maintenance Phase, and an Extension Phase. A total of633 subjects enrolled in the study, and resulting 161 randomized to . maintenance phase. ITT included 83 subjects in placebo group and 77 subjects in aripiprazole group. The primary efficacy endpoint was the time from randomization to relapse during the maintenance phase. The primary analysis is log-rank test.

Statistical Issues and Findings
The primary analysis is log-rank test which gives p-value .0199 where there were 36 outof83 (43%) relapsed in placebo, and 19 out of77 (25%) relapsed in aripiprazole groups, respectively.
One issue is whether the study is robust because center 093 in Mexico, where there were 7 in placebo and 6 in aripiprazole groups, respectively, had 5 (71%) relapsed in placebo and 0 (0%) relapsed in aripiprazole groups, respectively. The primary analysis is not significant after removing center 093. One should consider whether the quality of operations in center 093 is high, which was suggested to be inspected by DSI, when making final decision.

Appears This Way
On Original The current submission, NDA 21436 S005 was to support aripiprazole in treating subjects with Bipolar I Disorder. The study was a randomized, double-blind, multicenter, placebo-controlled trial in USA, Mexico, and Argentina, evaluating the use of aripiprazole in the maintenance of stability of patients with Bipolar I Disorder. There were 3 phases in this study: a Stabilization Phase, a Maintenance Phase, and an Extension Phase. Stabilization phase, maintenance phase, and extension phase. A total of633 enrolled in the study, and resulting 161 randomized to maintenance phase.

Evaluation of Efficacy
Texts, tables, and graphs in Sections 3.1.1 -3.1.7 are mainly adapted from the Applicant's Study Report.

Objective
The primary objective of this study was to compare the maintenance of stability of aripiprazole versus placebo as measured by the time to relapse (Le., discontinuation due to lack of efficacy) during the Maintenance Phase. Patients were discontinued from the study due to lack of efficacy if they were hospitalized for manic or depressive symptoms or required an addition to or increase in their allowed psychotropic medications.

Study Design
This was a randomized, double-blind, multicenter, placebo-controlled trial evaluating the use of aripiprazole in the maintenance of stability of patients with Bipolar I Disorder. The patient sample was diagnosed with Bipolar I Disorder, according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria, and had recently experienced a manic or mixed episode.

of 21
There were 2 routes of entry into this study. Patients who had recently completed a 3-week acute mania study of aripiprazole (CN138007, CN138009, CN138062, CN138074, or CN138077) were eligible to enter this study. Also, patients who had recently experienced (:s 3 months) a manic or mixed episode requiring hospitalization and treatment, but who had not participated in a 3-week aripiprazole study were eligible to enter this study. Patients who were eligible to participate in a 3-week aripiprazole acute mania study, but declined participation, were considered for this maintenance of stability study. Patients entered the study as inpatients or as outpatients.
There were 3 phases in this study: a Stabilization Phase, a Maintenance Phase, and an Extension Phase.
Stabilization Phase: During this phase patients received open-label aripiprazole treatment with a starting dose of30 mg/day. The dose could be decreased to 15 mg/day at any time, if necessary for tolerability. The Stabilization Phase was from 6 to 18 weeks in duration, with visits every 2 weeks. Patients continued in the Stabilization Phase until symptoms of their Bipolar Disorder were stable. Stability was defined by a Young-Mania Rating Scale (Y -MRS) Score of:s 10 and a Montgomery-Asberg Depression Rating Scale (MADRS) Score of:S 13 during 4 consecutive visits over a minimum of 6 weeks.
Patients entered the Maintenance Phase only after meeting stabilization criteria for 4 consecutive weeks and after remaining in the Stabilization Phase for a minimum of 6 weeks. Patients who entered the study and did not roll over directly from an acute mania study participated in a screening period of up to 28 days before entering the Stabilization Phase. There was a minimum 1 day wash-out period for antipsychotics. All antipsychotic treatment and psychotropic medications outside ofthose prescribed by this protocol were discontinued during the screening phase.
Maintenance Phase: Patients meeting stabilization criteria during the Stabilization Phase were randomized to either aripiprazole or placebo. Patients assigned to aripiprazole started the Maintenance Phase at the same dose they were taking at the end of the Stabilization Phase. The dose ofaripiprazole was 15 mg/day or 30 mg/day and could be changed at any time during the study, as necessary based on therapeutic effect and tolerability. Patients continued in the Maintenance Phase ofthe study for up to 26 weeks (6 months).
Extension Phase: Patients who completed 26 weeks of the Maintenance Phase without a relapse had the option to continue on their current double-blind study drug treatment in the Extension Phase for an additional 74 weeks (17 months).

Efficacy Measures
The primary efficacy outcome measure was the time to relapse (as defined by discontinuation due to lack of efficacy) from randomization in the 26-week Maintenance Phase. Patients were NDA 21-436 6 of 21 discontinued from the study due to lack of efficacy if they were hospitalized and/or required an addition to or increase in their allowed psychotropic medications, other than study medication, for manic or depressive symptoms.
Secondary efficacy measures included the time to manic relapse and the time to depressive relapse during the Maintenance Phase.
Relapses were classified into 3 categories: manic type, depressive type, or mixed type. A relapse was classified as a manic or depressive type if the patient was hospitalized for manic or depressive symptoms or required an addition to or increase in allowed psychotropic medication, other than study medication, for manic or depressive symptoms, as indicated on the relapse CRF page. A relapse was classified as a mixed type if the patient required intervention for both manic and depressive symptoms, as indicated on the relapse CRF page. The numbers and percentages of relapses falling into each of the 3 categories are presented by treatment group.

Statistical Analysis Plan
For time-to-event analyses, such as time to relapse, the log-rank test was used to compare the survival distributions of the 2 treatment groups. The estimated survival curves for each treatment group were obtained from the Kaplan-Meier estimates. Analysis of the primary efficacy measure will be performed using the Maintenance Safety Sample, which comprises all patients in the Randomized Sample who take at least one dose of study medication during the double-blind treatment phase, as identified on the dosing record. Other efficacy analyses will be performed using the Maintenance Efficacy Sample, which comprises all patients who are in the Maintenance Safety Sample Phase and have at least one post-randomization efficacy evaluation.
Sample size calculation: it was expected that the 6-month placebo relapse rate would be 45% and the aripiprazole relapse rate would be 20%. A total of 45 events would be required to yield 87% power to detect a 25% difference in the percentage of patients relapsing between placebo and the aripiprazole treatment groups, assuming these relapse rates, a dropout rate for reasons other than relapse of 18%, and a 2-sided test at the 0.05 level. These assumptions were based on results from 3 previous studies. Based on these assumptions, it was expected that 152 patients would have to be randomized to obtain 150 evaluable patients (75 per treatment group) to yield 45 events (number of patients who relapsed). The hazard ratio for these relapse rates and sample size was 2.7.

Protocol Amendments and Deviations
Protocol Amendment: There were 6 amendments and 4 administrative letters during the study. Amendments 4,5, and 6 affected the analysis ofthe study.
Amendment 4 added an Extension Phase so that patients may have continued on double-blind therapy upon completion of26 weeks ofthe Maintenance Phase. In addition, data handling for those  7 of 21 patients who inadvertently received unblended Maintenance Phase study medication was addressed. The amendment clarified that these patients were to be replaced and only safety data were to be analyzed.
Amendment 5 modified the criteria for closing and completing the study. The rationale of the amendment was to incorporate new information on maintenance treatment that became available after the initiation of the study. The original power calculations, which assumed relapse rates of 17% for aripiprazole and 47% for placebo, were based on a 30% difference in the expected relapse rate between aripiprazole and placebo; however, new information from maintenance treatment studies in bipolar patients indicated that the differences in the relapse rates between an active treatment and placebo might be less than 30%. Thus, new expected relapse rates for aripiprazole and placebo were calculated and were based on the assumption of a 25% expected difference in relapse rates between the 2 treatments, a clinically meaningful difference. It was then assumed that the placebo relapse rate would be approximately 45% and the aripiprazole relapse rate would be approximately 20%. Based on these new sample size calculations, the number of patients needed for relapse changed from 36 to 45.
Amendment 6 added 2 key secondary efficacy analyses to the study: time to manic relapse and time to depressive relapse. These were to be analyzed using a hierarchical testing procedure.
Protocol Deviations: On December 7, 2000 it was discovered that blinded supplies for the Maintenance Phase ofthe study were labeled in error with product information. Due to this labeling error, randomization was closed, and the 35 patients who had been randomized into the Maintenance Phase of the study at that time were discontinued from the study. Randomization was resumed on January 1, 2001, after appropriately repackaged supplies were available. The 35 patients who received unblinded study medication were not included in the analyses of efficacy or the Maintenance Safety Sample, but were analyzed separately. The safety data for these patients are presented in supplemental tables.

Study Population
A total of 633 patients were enrolled in the study, and 567 entered the Stabilization Phase, where 361 (64%) discontinued from this Phase, and 206 (37%) completed. Of , the 206 patients who, completed the Stabilization Phase, 161 were randomized to double-blind treatment in the Maintenance Phase. An additional 35 patients were randomized to the double-blind Maintenance Phase, but are not included in the Randomized Sample because of a labeling error, as described in Protocol deviation section. Ninety-four (58%) of the 161 patients discontinued from the Maintenance Phase of the study: 55 (66%) placebo-treated patients and 39 (50%) aripiprazoletreated patients. The most common reason for discontinuing from therapy in both treatment groups was because oflack of efficacy (43% placebo; 24% aripiprazole).
The disposition of all patients enrolled in the study is presented by treatment and study phase in  During the Stabilization Phase, "other known causes" included such things as SCreen failure, positive drug screen, did not meet inclusion criteria, and site closed by sponsor because prespecified number of relapses had been attained. In addition, I patient discontinued because of an SAE (thought suicidal) and was included in this category.
Forty-six patients completed the Stabilization Phase: 35 patients were randomized to the double-blind Maintenance Phase but were discontinued because of a labeling error; 11 patients discontinued because of other reasons (eg, Y -MRS or lV1ADRS criteria not met for randomization, reason not stated) and were not randomized to the double-blind Maintenance Phase; and 1 patient (Patient 138010-141-266) did not complete the Stabilization Phase but \-vas randomized in error to double-blind treatment.
During the Maintenance Phase, "other known causes" included positive drug screen, patient relocating, and site closed by sponsor because prespecified number of relapses had been attained.
Patient 138010-147-604 relapsed during the Extension Phase, according to the relapse torm, but discontinued from the Extension Phase because of " other known cause" according to the end-of-study form.
During the Extension Phase, the primary "other known cause" (study closed by sponsor because prespecified number of relapses had been attained) For the double-blind Maintenance Phase of the study, the patients were similarly distributed between the placebo and aripiprazole groups. Since a greater percentage of aripiprazole-treated patients (50%) than placebo-treated patients (34%) completed the Maintenance Phase, there were more patients in the aripiprazole group in the Extension Phase than in the placebo group. The distribution of all randomized patients within each of the patient samples is presented by treatment group in Table 3.1.6.2. There were 35 patients excluded from the Samples because of a labeling error. After these 35 patients had been randomized to the Maintenance Phase, it was discovered that blinded supplies for this phase were labeled in error with product information. Table 3.1.6.3 presents the patient identification number of these patients.   In the Randomized Sample, the demographic characteristics of the treatment groups were similar with the exception of gender: more men were randomized to the aripiprazole group (38%) than to the placebo group (28%), and conversely, fewer women were randomized to the aripiprazole group (62%) than to the placebo group (72%). Demographic characteristics of the Enrolled the Randomized Sample is presented by treatment group in Table 3.1.6.4.

(2)
The psychiatric history of bipolar disorder of patients in the Randomized Samples is presented in Tables 3.1.6.5.   The primary efficacy endpoint was the time from randomization to relapse during the Maintenance Phase (as defined by discontinuation due to lack of efficacy). Patients were discontinued from the study because oflack of efficacy if they were hospitalized and/or required an addition to or increase in their allowed psychotropic medications, other than study medication, for manic or depressive symptoms.
As shown in Table 3.1.7.1 and Figure 3.1.7.1, patients in the placebo group relapsed sooner than patients in the aripiprazole group, as evidenced by the log-rank P-value 0.020. Moreover, the probability of not experiencing relapse by Week 26 was 49% for placebo-treated patients and 72% for aripiprazole-treated patients.  Key secondary efficacy measures were the time to manic relapse and the time to depressive relapse during the Maintenance Phase. For these analyses, a hierarchical testing procedure was used. If aripiprazole was significant versus placebo in the primary efficacy analysis, then testing of the key secondary endpoints proceeded sequentially. First, time to manic relapse was tested and if this was significant, then time to depressive relapse was tested.
The results, as displayed in  The reviewer validated the sponsor's analysis according to the protocol. The log-rank test for the primary analysis gives p-value .0199. The log-rank for the key secondary analysis gives p-value .008 for the time to a manic relapse, where there were 19 relapse in placebo and 6 relapse in aripiprazole groups, respectively; and .6838 for the time to a depressive relapse, where there were 11 relapse in placebo and 9 relapse in aripiprazole groups, respectively.
There is one subject 0009300533 who was randomized to Aripiprazole but actual treatment was placebo. Ifusing randomization code, the subject was in Aripiprazole group. The log-rank test gives p-value .039l.

All-Cause Analysis
There are 36 relapsed in placebo and 19 relapsed in aripiprazole, and 28 completers in placebo and 39 completers in aripiprazole groups, respectively.  If subjects in the above table are treated as relapsed, the log-rank test gives p-value .0640 using actual treatment. The log-rank test gives p-value .0991 using randomization code.

Protocol Deviation
There were 35 patients excluded from the Samples because of a labeling error.  The impact due to this exclusion is difficult to evaluate.  Using Cox regression, p-values for country and interaction by treatment are not significant. There are 50 centers in the study. Centers in Mexico are largest. The next largest has 8 subjects so analysis based on center is not performed.  The log-rank test gives p-value .1043 after removing center 093 due to its relatively lower relapse rate. Baseline measures are balanced between two groups for center 093. Center 093 was suggested to DSI to have an inspection after filing meeting. If the inspection result indicates that the quality of operation in center 093 is poor, one needs to be very cautions when making final decision since we don't have much experience in Mexico.

Analysis by Country
There is one subject 00093 00533 who was randomized to Aripiprazole but actual treatment was placebo. Ifusing randomization code, the subject was in Aripiprazole group.   Since majority subjects are white, no separate analysis on race is performed.

Other Special/Subgroup Populations
There is no other subgroup analysis performed.

Statistical Issues and Collective Evidence
The primary analysis is log-rank test which gives p-value .0199 where there were 36 out of83 (43%) relapsed in placebo, and 19 out of 77 (25%) relapsed in aripiprazole groups, respectively.
One issue is whether the study is robust because center 093 in Mexico, where there were 7 in placebo and 6 in aripiprazole groups, respectively, had 5 (71%) relapsed in placebo and 0 (0%) relapsed in aripiprazole groups, respectively. The primary analysis is not significant after removing this center.

Conclusions and Recommendations
The conclusion is that the primary analysis for the time from randomization to relapse during the maintenance phase is significant comparing aripiprazole and placebo in evaluating subjects with Bipolar I Disorder but one should consider whether the quality of operations in center 093 is high, which was suggested to be inspected by DSI, when making final decision. .. o,~h"riP~rittiJl'I or typewriter versions (only) of this report: I~ additional space is required for any narrative answer (i.e., one thlilldo~$ ' ncit;requirea "Yes" or "No" response), please attach an additional page referencing the question number.     ; ·· · · ' · · ; : : . : : : "' t . : ; : : , "

Sponsors must submit the Information
2.5 Does the patent claim only a metabolite of the active ingredient pending in the NDA or supplement? (Complete the information in section 4 below if the patent claims a pending method of using the pending drug product to administer the metabolite.) 2.61 Does the patent claim only an intermediate?

2.7·
If the patent referenced in 2.1 is a product-llY'process patent, is the product claimed in the patent novel? (An answer is required only if the patent is a product-by-process patent.)

Drug Product (ComposltlonIFonnulatlon)
3. 28 If the answer to 4.2 is Use: (Submit indication or method of use information as identified specifically in the approved labeling,) "V es," identify with specificity the use with reference to the proposed labeling for the drug product.

No Relevant Patents
For this pending NDA, amendment, or supplement, there are no relevant patents that claim the drug substance (active ingredient), drug product (formulation or composition) or method(s) of use, for which the applicant is seeking approval and with respect to which a claim of patent infringement could reasonably be asserted if a person not licensed by the owner of the patent engaged in 0 Yes the manufacture, use, or sale of the drug product.
FORM FDA 3542a (7103) Page 2 PSCMcdiaAlU (101) (1) If the answer to 2 (b) is "yes," do you personally know of any reason to disagree with the applicant I s conclusion? If not applicable, answer NO.
If yes, explain: (2) If the answer to 2 (b) is "no," are you aware of published studies not conducted or sponsored by the applicant or other publicly available data that could independently demonstrate the safety and effectiveness of this drug product?
If yes, explain: (c) If the answers to (b) (1) and (b) (2) were both "no," identify the clinical investigations submitted in the application that are essential to the approval: CN138010 (Investigation 1: note only on e study was required for this i ndicat i o n) ----------------------Studies comparing two products with the same ingredient (s) are considered to be bioavailability studies for the purpose of this section.
3. In addition to being essential , investigations must be "new" to support exclusivity. The agency interprets "new clinical investigation" to mean an investigation that 1) has not been relied on by the agency to demonstrate the effectiveness of a previously approved drug for any indication and 2) does not duplicate the results of another investigation that was relied on by the agency to demonstrate the effectiveness of a previously approved drug product, i . e . , does not redemonstrate something the agency considers to have been demonstrated in an already approved application. a) For each investigation identified as "essential to the approval," has the investigation been relied on by the agency to demonstrate the effectiveness of a previously approved drug product?
(If the investigation was relied on only to support the safety of a previously approved drug, answer "no.") c) If the answers to 3(a) and 3(b) are no, identify each "new" investigation in the application or supplement that is essential to the approval (i.e., the investigations listed in #2(c), less any that are not "new"): CN138010 (Investigation 1)

4.
To be eligible for exclusivity, a new investigation that is D Yes: Please proceed to Section A.
~No: Please check all that apply: _./_Partial Waiver _./_Deferred __ Completed NOTE: More than one may apply Please proceed to Section B, Section C, and/or Section D and complete as necessary.  No problem Susan, it's perfectly understandable.
We tend to see point-by-point responses acknowledging each of the questions in an action letter, even where an issue has been settled or subsumed into another point, so we were concerned that we didn't see anything regarding these two points. We weren't sure how to interpret that. (If we'd been here yesterday we'd have been in touch then.) All looks well at this point for a Complete Class 1 response, with a two month date. You will get a formal letter from us later today, confirming this. > >We note that the January 3 response for S-005 makes no mention of the >two pending Phase 4 commitments which we consider will be met for S-005 >if they are met for S-002. We need some kind of information on the status of these two commitments, in order for your response . to be /considered complete. There is no mention of either at all, and we are >thus unable to determine BMS' intentions with respect to S-005 from >what we have in hand.
> >As the deadline for our making this decision is COB today, can you >please update us via secure email as soon as possible. Hi Teresa: Please see our responses to your questions below: I) All of the AEs for CN13S-037 are contained in ISSQADR3.xpt and identified by the variable AE]ROT when it takes the value 'CN13S-037'. These records may be selected by using the ROWS, ROW SELECTION, SELECT WHERE, selecting variable AE_PROT in the variable window, selecting 'equals' for the comparator and typing 'CN13S-037' in the value window. If you need to find an AE that occurred in CN13S-037 from a specific patient and have their uniq_id from CN13S-037 you can use the variable UNIQ_ID2 to select the AEs reported by the patient in CN13S-037. AEs that occurred during CN13S-037 are summarized in the CSR for CN13S-037.
2) We are not certain about your question regarding 'additional information' on the 35 patients who were unblinded. We did send an e-mail dated April 16, 2004 in which we responded to clinical bullets #1 & #2 in the 74-day letter (formal submission of this information was also sent on July 22). In that response we described how to find demographic, efficacy, and safety data for the 35 patients in the CN13S-01O Individual SAS Data Sets. In addition, we mentioned that the safety data for these patients is also contained in the Composite Safety Data Set "Merged" Files Structured for Viewing in SAS~JMP, but added that since these files were originally designed for review of the overall safety of aripiprazole, there was no way to specifi~ally select out the 35 patients. However, the data for the 35 patients are included in the overall safety databases and are flagged as such.
The following tables in the Clinical Study Report provide separate safety summaries and listings during the Maintenance Phase for the 35 patients: I I 2) Perhaps I am mistaken, however, as I recall , you submitted additional information regarding the 35 patients who received unblinded medication. Also, as I recall, the safety data for these patients were evaluated separately. Could you please provide the submission date and number or the location of that additional information on the 35 patients and also the location of the safety data for those patients.
Thanks fo r you r assistance in these matters . Kind regards , Teresa P.S. I am out of the office for the rest of the day.
This message was sent from Bristol-Myers Squibb, Co. across the Internet in encrypted format and was successfully decrypted, unless otherwise noted. Bristol-Myers Squibb This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. • Clinical reviewer noted that the e-submission is disorganized and therefore difficult to review. 35 patients appear to have been excluded from analysis. No explanation was found. This is a potential filing issue. • Statistical reviewer comments that primary efficacy data appear to have been placed in appendices that were not included as part of the submission, but must be requested by FDA if desired. This is a defmite RTF issue. • CMC review was completed prior to the filing meeting. Categorical exclusion can be granted.
Approval recommendation for CMC.
f Biopharm has no significant review responsibility. Interaction studies were adequately addressed in the context of 8-002, the acute mania supplement. These meeting minutes serve as documentation that no OCPB review will be required for this supplement. • PharmTox has two studies to examine which have not been previously submitted but are not expected to have significant impact on labeling. • A foreign DSI inspection (Mexico) will be required; a consult will be prepared.

REGULATORY CONCLUSIONS/DEFICIENCIES:
Meeting consensus: RTF (Statistics) if missing data cannot be provided and made accessible to primary statistics reviewer with sufficient time to evaluate their completeness prior to Filing Date.
The finn was contacted immediately and the filing issues explained. The finn was informed that the missing information had to be posted to the server in the EDR before the afternoon of March 26 (Friday) in order to allow the clinical and statistical reviewers a modicum of time to examine the contents before the filing decision date of March 30,2004. The application, following repair by the applicant of these RTF-level severe deficiencies, was rendered suitable for filing prior to the official filing date. It was therefore filed.
Clinical and statistics non-RTF filing review issues were nentified after receipt of the missing information. These will be communicated in the 74-Day Letter. For protocol CN138-010, 27 subjects were enrolled in the stabilization phase of open label treatment with aripiprazole 30 mglday. Reasons for discontinuation from the stabilization phase included withdrawal of consent, lost to follow up or adverse events.
Out of these 27 subjects, 8 subjects were randomized in the maintenance phase and only two subjects completed the study. Reasons for discontinuation included the following.
• Subject 104 was discontinued after week 4 visit due to compromising the blind as the package was imprinted with the name and dosage (aripiprazole 15 mg) by the manufacturer. • Subject 179 was terminated due to lack of efficacy: depression which required treatment with paroxetine and neurontin after week 4 visit. • Subject 238 was lost to follow up after week 12 visit.
• Subject 279 was listed as lack of efficacy on week 1. • Subject 349 was due to lack of efficacy based on the Y -MRS scores at week 14. This subject was enrolled in the maintenance phase without meeting the inclusion criterion that the patient must have 4 consecutive Y -MRS scores of equal or less than 10 to enter the maintenance phase. The study coordinator noticed this error and notified the sponsor. The sponsor approved the patient to continue in the study. • Subject 429 withdrew the consent after week 3 visit as it was noted that he was fearful to lose hisr:: J because ofhis participation in the clinical study.

b(6)
Two subjects participated in the extension phase. Subject 416 was terminated at week 60 of the extension phase as the investigator detennined that the subject met the criteria for relapse. Subject 288 discontinued from the study after week 44 due to worsening of involuntary movement in the right leg and tongue tremor (tardive dyskinesia).

Limitations of inspection: N/ A
General observations/commentary: The protocol specified a positive screen of lithium, divalproex acid or drug of abuse as one of the exclusionary criteria for enrollment in maintenance phase of the study. The lithium and divalproex acid levels for two subjects (416 and 179) were not performed as required by the protocol.
For three subjects, the following source documents were not available for review and therefore, the FDA investigator was not able to verify the data. Subject 104: all source documents Subject 160: signed informed consent, screening source documents Subject 039: source documents from baseline evaluations 3 The protocol specified that the patient must have 4 consecutive Y-MRS scores of equal or less than 10 to enter the maintenance phase. Subject 349 was enrolled in the maintenance phase without meeting this inclusion criterion. The study coordinator noticed the error and notified the sponsor. The sponsor approved the patient to continue in the study. Recommendation: The review division should note above protocol deviations and record keeping deficiencies. The review division should consider any impact of these findings on study data. Otherwise, data appear acceptable.

2.
Ignatio Rosales,M.D. (site 93) What was inspected: For protocol CN138~OlO, 18 subjects were screened and enrolled in the stabilization phase of open label treatment with aripiprazole 30 mg/day. During the stabilization phase, 2 subjects (126 and 514) were reported by the clinical investigator that they withdrew their consent and subject 427 discontinued because of adverse event.
During the maintenance phase of the study, three subjects (101, 102 and 114) were discontinued due to compromising the blind as the package was imprinted with the name and dosage (aripiprazole 15 mg) by the manufacturer. Five subjects (184, 198,495,533 and 542) were discontinued for lack of efficacy. Subject 504 discontinued because of adverse event/lack of efficacy. Five subjects (154,196,501,532,535) entered in the extension phase of the study. There were 3 serious adverse events reported at this site.
An audit of all 18 subjects' records was conducted. A Fonn FDA-483 was issued at the end of inspection. Dr. Rosales responded to the FDA-483 in writing. DSI received a copy of Dr. Rosales response on September 28, 2004.
Limitations of inspection: The source documents were written in Spanish.
General observations/commentary: The protocol specified that patients who have a positive screen for lithium, divalproex acid or drugs of abuse be excluded from entering the maintenance phase of the study. The site did not perfonn the lithium and divalproex acid levels for four subjects (# 533,535,538 and 542). The site performed these protocol required tests after randomization to enrolhnent in the maintenance phase for three subjects (# 101, 102, and 114).
There were multiple instances of protocol required clinical laboratory tests (hematology, urine), prolactin levels, pregnancy tests and drug screens that were not performed. For all 18 subjects, the number of missing tests ranged from one to fifteen tests per subject. For subject #126, it was documented in the source document that the subject has poor compliance with the study medication and the family could not monitor the subject. The subject was discontinued from the study. The reason for discontinuation is reported on the CRF as withdrawal of consent.
The protocol specified that subjects enter the stabilization phase with a recent manic or mixed episode requiring hospitalization that began no more than 3 months before entry into the stabilization phase. During the hospitalization, the subject # 198 was treated with medication for acute mania. The source document for this subject reports no prior hospitalization.
Subject #427 refused hospitalization for the last manic episode that occurred more than three months before entry in the study. There was no documentation that the sponsor was notified and was granted permission for the subject to enter the study. The source documents report that this subject was experiencing adverse events, no mention of depressive symptoms. However, the subject was discontinued from the study for relapse with a MADRS score of 17. The CRF listed the reason for discontinuation as Lack of Efficacy while the comment section in the CRF reported that the subject did not want to continue in trial because of the adverse events.
Other observations: Although Dr. Rosales signed the Form FDA-1572, the sponsor reported that it was not submitted to the Agency and therefore, this site was listed as a non-INn site.
The protocol consisted of stabilization phase (up to 18 weeks),maintenance phase (up to 26 weeks) and extension phase (up to 74 weeks). During the stabilization phase, subjects would receive open-label treatment with aripiprazole 30 mg/day. The dose could be decreased to 15 mg/dayat any time ifnecessary for side effects. The protocol specified that subjects would continue in the stabilization phase until a Young Mania Rating Scale (Y-MRS) score of.::;10 and a A response to the first and second request in the 74 day letter for this supplement is provided attached to this e-mail. We would like some further clarifications on the requests as indicated in the attachment. Primarily, if, after reviewing the response, you are interested in JMP datasets for the secondary variables, it would be very helpful to know which specific variables would be of interest.
Please note that the request for all appendices has been fulfilled (submitted on March 25, 2004). Please let us know if you are unable to locate them. e are in the process of compiling the information for the remaining items requested and will be providing those in a separate correspondence.
As I will be on vacation next week, please contact my colleague, Dr. Chuck Wolleben, at the above e-mail address, or by phone at 203-677-3834 for further communications on this matter or for any other requests related to this application.