THERAPEUTIC APPROACHES
`Pharmacologic Agents for the Treatment of Acute
`Bipolar Mania
`
`Susan L. McElroy and Paul E. Keck, Jr.
`
`The knowledge base regarding the medical treatment of
`acute bipolar mania is rapidly expanding. Information
`about agents with established antimanic properties is
`increasing, and more agents with putative antimanic
`properties are being identified. We first review the con-
`trolled studies supporting the efficacy of the established
`antimanic agents lithium, valproate, and carbamazepine
`and standard antipsychotics. We then review available
`research on two important classes of drugs that are
`emerging as potential treatments for acute mania: the
`novel antipsychotics, which include clozapine, olanzapine,
`quetiapine, risperidone, and ziprasidone, and the new
`antiepileptics, which include gabapentin, lamotrigine, ox-
`carbazepine, tiagabine, topiramate, and zonisamide. We
`conclude that although controlled data are accumulating
`to support the efficacy of several atypical antipsychotics in
`the treatment of acute bipolar mania, particularly olanza-
`pine, ziprasidone, and risperidone, the novel antiepileptics
`need more extensive study before it can be concluded that
`any of them possess specific antimanic properties. We also
`conclude that as the medical options for acute bipolar
`mania expand, treatment guidelines must remain both
`evidence based and flexible, so that they represent cutting
`edge medical science yet can be tailored to the specific
`needs of individual patients. Biol Psychiatry 2000;48:
`539–557 © 2000 Society of Biological Psychiatry
`
`Key Words: Bipolar mania, mood stabilizers, atypical
`antipsychotics, novel antiepileptics
`
`Pharmacologic Treatments for Acute Mania
`
`The knowledge base regarding the medical treatment of
`
`acute bipolar mania is rapidly expanding. Information
`about agents with established antimanic efficacy is in-
`creasing, and more agents with putative antimanic prop-
`erties are being identified.
`In this article we first review research supporting the
`efficacy of the established antimanic agents lithium, val-
`proate, and carbamazepine and standard antipsychotics in
`the short-term treatment of acute bipolar mania. Because
`
`From the Biological Psychiatry Program, Department of Psychiatry, University of
`Cincinnati College of Medicine, Cincinnati, Ohio.
`Address reprint requests to Susan L. McElroy, M.D., University of Cincinnati
`College of Medicine, Biological Psychiatry Program (ML559), 231 Bethesda
`Avenue, Cincinnati OH 45267.
`Received February 4, 2000; revised June 7, 2000; accepted June 7, 2000.
`
`© 2000 Society of Biological Psychiatry
`
`of their reputations as having established antimanic effi-
`cacy, we limit our review of these agents to double-blind,
`controlled monotherapy and placebo-controlled add-on or
`dual therapy studies. We then review available research on
`two important classes of drugs that are emerging as
`potential treatments for acute mania: the novel antipsy-
`chotics, which include clozapine, olanzapine, quetiapine,
`risperidone, and ziprasidone, and the new antiepileptics,
`which include gabapentin,
`lamotrigine, oxcarbazepine,
`tiagabine, topiramate, and zonisamide. We conclude that
`as the medical options for acute bipolar mania expand,
`treatment guidelines must remain both evidence based and
`flexible, so that
`they represent cutting edge medical
`science yet can be tailored to the specific needs of
`individual patients.
`
`Established Antimanic Agents
`Lithium
`Lithium was the first drug approved by the United States
`Food and Drug Administration (FDA; in 1970) for the
`treatment of “manic episodes of manic-depressive illness”
`(Goodwin and Jamison 1990). Five controlled studies have
`demonstrated that lithium is superior to a placebo for the
`treatment of acute mania (Bowden et al 1994; Goodwin et
`al 1969; Maggs 1963; Schou et al 1954; Stokes et al 1971;
`Table 1). Summarized below, several methodological
`limitations should be considered in interpreting these
`studies. First, only one study (conducted after lithium was
`granted its approval by the FDA for the acute treatment of
`mania) employed a parallel design (Bowden et al 1994);
`the four earlier studies were crossover trials of varying
`duration. Crossover studies are vulnerable to carryover
`and period effects, potential contamination of blindness,
`and abrupt treatment discontinuation effects (which may
`artificially lower placebo response rates via rebound re-
`currence of symptoms; Calabrese and Rapport 1999; Keck
`et al 2000b; Stallone et al 1974). Second, two studies
`utilized nonrandom assignment to lithium or a placebo
`(Goodwin et al 1969; Stokes et al 1971). Third, the four
`earlier studies essentially performed completer analyses;
`last observation carried forward (LOCF) analyses were not
`conducted. Completer analyses, which only evaluate pa-
`tients who receive a treatment for a specified duration of
`
`0006-3223/00/$20.00
`PII S0006-3223(00)00961-6
`
`Exhibit 2136
`Slayback v. Sumitomo
`IPR2020-01053
`
`1
`
`

`

`540
`
`BIOL PSYCHIATRY
`2000;48:539–557
`
`S.L. McElroy and P.E. Keck, Jr.
`
`Table 1. Double-Blind, Placebo-Controlled Studies of Lithium Monotherapy in Acute Bipolar Mania
`
`Study
`Schou et al 1954
`
`Design
`Random, crossovera
`
`Maggs 1963
`
`Goodwin et al 1969
`
`Random, crossover,
`ABA vs. BAB
`Nonrandom, crossover
`
`Stokes et al 1971
`
`Nonrandom, crossover
`
`Bowden et al 1994
`
`Random, parallel-group,
`VPA comparison
`
`Overall monotherapy
`responsee
`
`N
`38b
`
`28
`
`12
`
`28
`
`Li 35,
`PBO 73,
`VPA 68
`85
`
`Duration (days)
`“Usually” 14 for Li
`and PBO
`
`14 for Li and PBO
`
`ND
`
`7–10 for Li and
`PBO
`
`21
`
`Outcome
`14 (37%) positive effect,c
`18 (47%) possible effect,
`6 (16%) negative effect
`Li superior to PBO for 18 patients who
`completed entire 6-week study
`9 (75%) response to Li,c
`3 (25%) worse with Li
`42 (75%) 56 response to Li,d
`17 (41%) 42 response to PBOd
`17 (49%) 35 response to Li (p ⫽ .025),
`18 (25%) response to PBO,
`33 (48% response to VPA (p ⫽ .004)
`58 (68%) response to Li
`
`Li, lithium; PBO, placebo; ND, not documented; VPA, valproate.
`aLithium sometimes administered as “open treatment for a certain period.”
`bIncludes 30 patients with “typical” and 8 patients with “atypical” (schizoaffective) manic–depressive illness.
`cWorsening with PBO substitution part of definition of response to Li.
`dRefers to number of treated periods of mania.
`eIncludes those studies in which Li response rate is quantifiable (Bowden et al 1994; Goodwin et al 1969; Schou et al 1954).
`
`time, may be biased toward showing efficacy, as opposed
`to LOCF analyses, which evaluate all patients who receive
`a treatment for any duration of time. Fourth, the diagnostic
`criteria used to define bipolar disorder in the early lithium
`studies were not necessarily comparable to those of
`DSM-III-R (American Psychiatric Association 1987) or
`DSM-IV (American Psychiatric Association 1994).
`In the first placebo-controlled, crossover study (Schou
`et al 1954) a definite response based on global impression
`of improvement was reported in 12 (40%) and a probable
`response in 15 (50%) of 30 patients with typical bipolar
`disorder. Response was less robust in eight patients with
`atypical features (which implied a schizoaffective diagno-
`sis), with two (25%) displaying a probable response. In the
`second crossover trial (Maggs 1963), which was the first
`study to use formal rating scales (i.e., the Wittenborn
`Scale) and to analyze data statistically, 28 inpatients with
`mania were randomized to three consecutive 14-day peri-
`ods of lithium–rest–placebo or placebo–rest–lithium. Nine
`patients did not complete their 6-week cycles of treatment,
`and results were based on the 18 patients who completed
`their trials. In these 18 patients, lithium was superior to a
`placebo during the second week of treatment on the
`Wittenborn Scale measures of “manic states” and “schizo-
`phrenic excitement.”
`In the first United States study (Goodwin et al 1969) the
`longitudinal efficacy of lithium was compared with a
`placebo in 12 patients with mania; eight (67%) displayed
`a complete response and one (8%) a partial response. A
`complete response was defined as complete remission of
`all manic symptoms within 2 weeks of starting lithium and
`return of symptoms during placebo periods; a partial
`response was defined as a decrease in mean mania ratings
`
`of at least three points within 2 weeks of starting lithium,
`but not complete remission of symptoms, and an increase
`in symptoms during placebo periods. In the fourth study
`(Stokes et al 1971) 38 inpatients with “typical manic
`depressive illness” were evaluated in a crossover design
`with alternating 7- to 10-day periods on lithium or a
`placebo. Although 7- to 10-day trial periods may have
`limited the patients’ ability to display a more robust
`lithium response, the equally brief placebo periods may
`have been confounded by residual lithium effects. Despite
`these caveats, mania ratings decreased in 75% of lithium
`treatment periods, as compared with 41% of placebo
`treatment periods.
`In the only randomized, double-blind, placebo-con-
`trolled, parallel-design trial of lithium published to date in
`acute bipolar mania (Bowden et al 1994) lithium was used
`as an active control substance in a study designed primar-
`ily to assess the antimanic efficacy of valproate. In this
`study, 17 (49%) of 35 patients receiving lithium displayed
`at least 50% improvement on the Mania Rating scale
`(MRS) of the Schedule for Affective Disorders and
`Schizophrenia (SADS-C) at 3 weeks, as compared with 18
`(25%) of 73 patients receiving a placebo and 33 (48%) of
`68 patients receiving valproate. Regarding onset of re-
`sponse, both lithium and valproate first separated from the
`placebo on the MRS on day 10 of treatment. On day 8 of
`treatment, the mean lithium and valproate serum concen-
`trations were 1.0 mmol/L and 77 mg/mL, respectively.
`In summary, these studies showed that lithium is superior
`in efficacy to a placebo in acute bipolar mania, usually
`requiring a 1- to 3-week trial at therapeutic levels to exert
`significant antimanic effects. The pooled response rate from
`the three placebo-controlled studies in which patient response
`
`2
`
`

`

`Pharmacologic Agents and Acute Bipolar Mania
`
`BIOL PSYCHIATRY
`2000;48:539–557
`
`541
`
`Table 2. Controlled Studies of Lithium and Standard Antipsychotics in Acute Bipolar Mania
`
`Study
`Johnson et al 1968
`
`Design
`Random, parallel-group
`
`Platman 1970
`
`Random, parallel-group
`
`Spring et al 1970
`
`Johnson et al 1971
`
`Random, parallel-group,
`crossover of nonresponders
`Random, parallel-group
`
`Prien et al 1972
`
`Random, parallel-group
`
`Shopsin et al 1975
`
`Random, parallel-group
`
`Takahashi et al 1975
`
`Random, parallel-group
`
`Garfinkel et al 1980
`
`Random, parallel-group
`
`Segal et al 1998
`
`Random, parallel-group, HAL
`comparison
`
`N
`
`Li 18
`CPZ 11
`Li 13
`CPZ 10
`Li 7
`CPZ 5
`Li 13
`CPZ 8
`
`Mildly active 130
`Highly active 125
`
`Li 10
`CPZ 10
`HAL 10
`Li 37
`CPZ 34
`Li ⫹ PBO 7
`HAL ⫹ PBO 7
`Li ⫹ HAL 7
`Li 15
`HAL 15
`
`Duration
`(days)
`21–28
`
`21
`
`21
`
`21
`
`21
`
`21
`
`35
`
`21
`
`28
`
`Outcome
`14 (78%) response to Li,
`4 (36%) response to CPZ
`Li superior to CPZ after 3 weeks (ns)
`
`6 (86%) rsponse to Li,
`3 (60%) response to CPZ (ns)
`5 CPZ and 6 Li completers showed significant
`and equal improvement on BPRS and CGI;
`Li superior to HAL on “major component”
`of TRAM; overall, ns
`Li ⫽ CPZ in mildly active group at weeks 1, 2,
`and 3; CPZ superior to Li in highly active
`group at weeks 1 and 2, equivalent to Li at
`week 3
`7 (70%) response to Li,
`1 (10%) response to CPZ,
`2 (20%) response to HAL
`25 (68%) response to Li,
`17 (50%) response to CPZ (p ⫽ .05)
`HAL ⫹ PBO ⫽ HAL ⫹ Li; both superior to Li
`⫹ PBO in improving global clinical ratings
`on days 8, 15, and 22
`Li ⫽ HAL in decreasing manic symptoms
`
`Li, lithium; CPZ, chlorpromazine; BPRS, Brief Psychiatric Rating Scale; CGI, Clinical Global Impression; HAL, haloperidol; TRAM, Treatment Response Assessment
`Method; PBO, placebo.
`
`rate to lithium monotherapy was quantifiable revealed that 58
`(68%) of 85 acutely manic patients experienced at least
`partial
`improvement with lithium (Bowden et al 1994;
`Goodwin et al 1969; Schou et al 1954; Table 1). Further
`analysis of the Bowden et al (1994) study showed that a
`history of previous lithium response and pure mania, or of
`mania with predominantly elevated or elated mood and
`without depressive symptoms (Swann et al 1997), were
`associated with favorable response to lithium. In those studies
`in which response of psychotic symptoms was assessed,
`lithium also produced significant
`improvement
`in these
`symptoms (Bowden et al 1994; Goodwin et al 1969; Maggs
`1963; Stokes et al 1971); however, psychotic symptoms in
`the absence of manic symptoms (Schou et al 1954), depres-
`sive symptoms during mania (Swann et al 1997), and a
`greater number (approximately 10 or more) of prior mood
`episodes (Swann et al 1999) were associated with poor
`antimanic response to lithium.
`Lithium has also been compared with standard antipsy-
`chotic agents in nine controlled trials in the treatment of
`acute bipolar mania (Garfinkel et al 1980; Johnson et al
`1968, 1971; Platman 1970; Prien et al 1972; Segal et al
`1998; Shopsin et al 1975; Spring et al 1970; Takahashi et
`al 1975; Table 2). Interpretation of the results of virtually
`all of these studies is limited because of the inclusion of
`manic patients with schizoaffective disorder, lack of pla-
`
`cebo comparison groups, lack of standardized rating scales
`for mania, lack of performance of LOCF analyses, and/or
`the possibility of occurrence of a Type II error (the failure
`to find a significant difference between treatments because
`of a small sample size; Table 2). Nonetheless, of these
`nine studies, three involving 58 patients found lithium
`comparable to chlorpromazine (Johnson et al 1971; Spring
`et al 1970) or haloperidol (Segal et al 1998) over periods
`of 1 to 4 weeks; four studies involving 160 patients found
`lithium superior to chlorpromazine (Johnson et al 1968;
`Platman 1970; Shopsin et al 1975; Takahashi et al 1975)
`and/or haloperidol (Shopsin et al 1975) over periods of 1
`to 5 weeks; and one study (Garfinkel et al 1980) involving
`21 patients found haloperidol plus a placebo and haloper-
`idol plus lithium superior to lithium plus a placebo (and
`equivalent to one another) after 1 and 2 weeks.
`In the ninth study,
`the largest and most rigorous
`comparison of lithium and an antipsychotic conducted in
`acute bipolar mania to date, Prien et al (1972) evaluated
`the response of 225 manic inpatients to lithium versus
`chlorpromazine according to degree of psychomotor agi-
`tation by dividing patients into “highly active” (N ⫽ 125)
`or “mildly active” (N ⫽ 130) groups. The dosage of
`lithium ranged from 500 to 4000 mg/day, with a mean of
`1800 mg/day; the median lithium level was 1.4 mEq/L for
`the highly active group and 1.2 mEq/L for the mildly
`
`3
`
`

`

`542
`
`BIOL PSYCHIATRY
`2000;48:539–557
`
`S.L. McElroy and P.E. Keck, Jr.
`
`Table 3. Double-Blind, Controlled Studies of Valproate in Acute Bipolar Mania
`
`Study
`Placebo controlled
`Emrich et al 1981
`
`Brennan et al 1984
`
`Pope et al 1991
`
`Bowden et al 1994
`
`Mu¨ller-Oerlinghausen et al 2000
`
`Lithium controlled
`Freeman et al 1992
`
`Overall monotherapy response
`
`Design
`
`Random crossover, ABA
`
`Random crossover, ABA
`
`Random, parallel-group,
`Li comparison
`Random, parallel-group,
`Li comparison
`
`Random, parallel-group,
`add-on to antipsychotic
`
`Random, parallel-group
`
`ABA, placebo/valproate/placebo; VPA, valproate; PBO, placebo; Li, lithium.
`
`active group. Chlorpromazine doses ranged from 200 to
`3000 mg/day, with a mean of 1000 mg/day. In the mildly
`active group, LOCF analysis showed that both medica-
`tions produced significant and comparable improvement
`on the Brief Psychiatric Rating Scale (BPRS), the Inpa-
`tient Multidimensional Psychiatric Scale, and the Psy-
`chotic Inpatient Profile; however, side effects were more
`frequent and severe among the chlorpromazine-treated
`patients. By contrast, in the highly active group LOCF
`analysis showed that chlorpromazine produced more sig-
`nificant reductions in measures of agitation, excitement,
`grandiosity, hostility, and psychotic disorganization than
`did lithium during the first week of treatment. In addition,
`dropouts in the lithium-treated group were higher (38%)
`than in the chlorpromazine-treated group (8%). By 3
`weeks of treatment both drugs were significantly and
`comparably effective. The authors concluded that chlor-
`promazine was superior to lithium in the initial treatment
`of highly active patients, but that the two drugs were
`equally effective in mildly active patients. Of relevance
`when interpreting other lithium–antipsychotic comparator
`trials, a completer analysis of the highly active group
`showed no differences between the lithium- and chlor-
`promazine-treated patients.
`In summary, these data suggest that lithium is compa-
`rable and possibly superior to antipsychotics in the short
`term (i.e., 3- to 6-week treatment of acute bipolar mania).
`They also suggest that lithium exerts antipsychotic effects
`in mania; however, these data also indicate that antipsy-
`
`Duration
`(days)
`
`Variable
`
`14
`
`21
`
`21
`
`21
`
`21
`
`N
`
`5
`
`8
`
`VPA 17
`PBO 19
`VPA 68
`PBO 73
`Li 35
`
`VPA 69
`PBO 67
`
`VPA 14
`Li 13
`VPA 112
`PBO 92
`Li 48
`
`Outcome
`
`4 (80%) marked response,
`1 (20%) no response
`6 (75%) marked response,
`2 (25%) no response
`9 (53%) response to VPA,
`2 (10%) response to PBO
`33 (48%) response to VPA
`(p ⫽ .004),
`18 (25%) response to PBO,
`17 (49%) response to Li
`(p ⫽ .025)
`48 (70%) response to VPA,
`31 (46%) response to PBO
`(p ⫽ .005)
`
`9 (63%) response to VPA,
`12 (92%) response to Li (ns)
`61 (54%) response to VPA,
`20 (22%) response to PBO,
`29 (60%) response to Li
`
`chotics may have a more rapid onset of action in mania
`and, therefore, may be more effective initially (i.e., within
`the first week), especially in severely manic or highly
`agitated patients.
`It is important to note that the response rates in the
`above studies were to lithium monotherapy, and that these
`rates would be expected to be more robust with the use of
`adjunctive antimanic agents. Although there are controlled
`add-on trials in which other potential antimanic agents are
`added to lithium, we were unable to locate any such trials
`in which lithium was added to another antimanic agent.
`Nonetheless, numerous open reports suggest the antimanic
`effects of lithium may be augmented by other mood
`stabilizers, standard antipsychotics, and atypical antipsy-
`chotics (Freeman and Stoll 1998).
`
`Valproate
`Five controlled trials have shown valproate to be effica-
`cious as monotherapy for the short-term treatment of acute
`bipolar mania (Bowden et al 1994; Brennan et al 1984;
`Emrich et al 1981; Freeman et al 1992; Pope et al 1991;
`Table 3). These studies include comparisons of valproate
`and a placebo in crossover trials without concomitant
`psychotropics (Brennan et al 1984; Emrich et al 1981),
`valproate and a placebo in a parallel-group trial in lithium-
`refractory or intolerant patients (Pope et al 1991), val-
`proate and lithium in a parallel-group trial (Freeman et al
`1992), and valproate and a placebo and lithium in a
`
`4
`
`

`

`Pharmacologic Agents and Acute Bipolar Mania
`
`BIOL PSYCHIATRY
`2000;48:539–557
`
`543
`
`parallel-group trial (Bowden et al 1994). The last three
`studies (Bowden et al 1994; Freeman et al 1992; Pope et
`al 1991), which enrolled the largest patient samples,
`allowed as-needed lorazepam at low dosages during the
`initial week of 3-week trials. Two of these trials (Bowden
`et al 1994; Pope et al 1991) led to valproate being the
`second drug approved by the FDA for the treatment of the
`manic episodes associated with bipolar disorder.
`In the first double-blind, placebo-controlled, parallel-
`group study (Pope et al 1991) 36 inpatients with DSM-
`III-R bipolar disorder, manic phase, who were either
`lithium refractory or lithium intolerant, were randomly
`assigned to valproate (N ⫽ 17) or to a placebo (N ⫽ 19)
`for 7 to 21 days. Compared with placebo-treated patients,
`valproate-treated patients displayed statistically significant
`improvement on all three measures used to assess re-
`sponse:
`the Young Mania Rating Scale (YMRS),
`the
`BPRS, and the Global Assessment of Functioning Scale
`(GAF). Of the 17 patients receiving valproate, nine (53%)
`displayed a 50% or greater reduction on the YMRS,
`compared with two (10%) of the 19 patients receiving a
`placebo. Patients receiving valproate required significantly
`less lorazepam, and there was no statistically significant
`difference in the frequency of side effects between the two
`groups. Further,
`in responders the onset of antimanic
`response to divalproex was prompt, with significant im-
`provement occurring within the first week of treatment
`despite use of a gradual titration schedule (the beginning
`valproate dose was 750 mg/day).
`In the second double-blind, parallel-group, controlled
`study (Freeman et al 1992) 27 patients with DSM-III-R
`bipolar disorder, manic episodes were randomized to
`valproate or lithium. Both drugs produced significant and
`comparable improvement as measured by the MRS of the
`SADS-C, the BPRS, and the GAF. Twelve (92%) of 13
`patients assigned to the lithium group were rated as
`responders, compared with nine (64%) of 14 patients
`assigned to the valproate group. Although the response
`rate to lithium exceeded that to valproate in this study, the
`difference was not statistically significant (p ⫽ .20 by
`Fisher exact test, two-tailed). Unlike the case with lithium,
`favorable response to valproate was associated with high
`pretreatment depression scores.
`In the second double-blind, placebo-controlled, parallel-
`group study (Bowden et al 1994) 179 inpatients meeting
`Research Diagnostic Criteria for manic disorder were
`randomized to valproate (N ⫽ 68), lithium (N ⫽ 35), or a
`placebo (N ⫽ 73) for up to 3 weeks. Both valproate- and
`lithium-treated patients had statistically significantly
`greater improvement on the primary measure—the MRS
`of the SADS-C—than placebo-treated patients by day 10
`of the study, beginning with an initial valproate dose of
`750 mg/day and using a gradual titration schedule. The
`
`proportions of patients improving at least 50% on the
`MRS were comparable for valproate (48%) and lithium
`(49%) and superior to a placebo (25%). All patients with
`rapid cycling (N ⫽ 8) were randomly assigned to dival-
`proex; four (50%) displayed at least 50% improvement on
`the MRS, which was comparable to the overall response
`rate of the divalproex-treated group. This response rate,
`though limited by the small number of patients, is notable
`because rapid cycling is associated with poor lithium
`response (Dunner and Fieve 1974). In addition, analysis of
`response according to several definitions of depressive
`mania based on the SADS-C depression subscale measure
`showed that the presence of even mild depressive symp-
`toms was associated with a poor antimanic response to
`lithium, but had no significant effect on valproate response
`(Swann et al 1997). (There was a trend, however, toward
`more improvement with valproate with the narrowest
`definition of depressive mania.) Finally, significantly
`more lithium-treated patients dropped out of this study due
`to side effects than did patients receiving valproate or a
`placebo.
`One study has compared valproate monotherapy with a
`standard antipsychotic in the treatment of acute bipolar
`mania. In that study, 36 inpatients with bipolar I disorder,
`manic or mixed phase with psychotic features by DSM-
`III-R criteria, were randomized to receive either valproate
`(20 mg/kg/day) or haloperidol (0.2 mg/kg/day) in single
`(rater)–blind fashion for 6 days (McElroy et al 1996).
`There was no placebo group. Lorazepam up to 4 mg/day
`was the only other permitted psychotropic for the man-
`agement of agitation. Valproate and haloperidol were
`equally effective in acutely reducing manic and psychotic
`symptoms as assessed by the YMRS and the Scale for
`Assessment of Positive Symptoms,
`respectively. Ten
`(48%) of 21 patients receiving valproate and five (33%) of
`15 patients receiving haloperidol were classified as re-
`sponders. The greatest rate of improvement for both drug
`regimens occurred over the first 3 days of treatment.
`Adverse effects were infrequent and minor for both drugs,
`except for extrapyramidal side effects, which were signif-
`icantly more common with haloperidol.
`In summary, pooled response rates to valproate from the
`three parallel-design, double-blind, controlled, parallel-
`design monotherapy studies (Bowden et al 1994; Freeman
`et al 1992; Pope et al 1991) revealed significant improve-
`ment (i.e., at least a partial response or a 50% or greater
`reduction in manic symptoms) in 54% of patients, as well
`as efficacy superior to that of a placebo (Bowden et al
`1994; Pope et al 1991) and efficacy equivalent to that of
`lithium (Bowden et al 1994; Freeman et al 1992). These
`studies further suggest that valproate may have a broad
`spectrum of efficacy in acute mania, with effectiveness in
`mania with and without psychotic features, with and
`
`5
`
`

`

`544
`
`BIOL PSYCHIATRY
`2000;48:539–557
`
`S.L. McElroy and P.E. Keck, Jr.
`
`without depressive features, with and without rapid cy-
`cling, and with and without numerous prior mood episodes
`(Bowden et al 1994; McElroy et al 1991). Indeed, the
`Bowden et al (1994) study suggests that valproate may be
`more effective than lithium for acute mania with depres-
`sive features (Swann et al 1997) and acute mania associ-
`ated with many (10 or more) prior mood episodes (Swann
`et al 1999).
`As with the controlled studies of lithium for bipolar
`acute mania, the response rates in the above studies were
`attributable to valproate monotherapy and might be ex-
`pected to be more robust with the use of adjunctive agents.
`Indeed, valproate has recently been compared with a
`placebo as add-on therapy to standard antipsychotics in the
`treatment of acute bipolar mania. In a multicenter, paral-
`lel-group, double-blind, 3-week study from Europe, 136
`hospitalized patients with acute mania by ICD-10 criteria
`receiving open-label standard antipsychotics (haloperidol
`and/or perazine) were randomized to receive add-on ther-
`apy with valproate (20 mg/kg/day) or a placebo (Mu¨ller-
`Oerlinghausen et al 2000). The primary outcome measure
`was the mean dose of antipsychotic received for the
`21-day treatment period, converted into haloperidol equiv-
`alents. The mean antipsychotic dose declined continuously
`in the valproate group, whereas only slight dose variations
`were observed in the placebo group; the difference in
`antipsychotic dose was statistically significant for study
`weeks 2 and 3 (p ⫽ .0007). The proportion of responders
`(response was defined as 50% or greater improvement on
`the YMRS) was higher for the group receiving the
`combination (70%) than for the group receiving antipsy-
`chotics alone (46%; p ⫽ .005). The authors concluded that
`the combination of valproate and an antipsychotic was
`superior to an antipsychotic alone in treating acute mania.
`Many open reports indicate that valproate can be suc-
`cessfully combined with other typical antipsychotics, other
`mood stabilizers, and atypical antipsychotics in the treat-
`ment of acute manic, mixed, and rapid-cycling states
`(Freeman and Stoll 1998). In addition, valproate has been
`administered via the oral loading strategy of 20 to 30
`mg/kg/day (Keck et al 2000a; McElroy et al 1996) as well
`as intravenously (1200 or 1800 mg/day; Grunze et al
`1999a) to acutely manic patients with rapid onset of
`response (within 1 to 3 days) and minimal side effects.
`
`Carbamazepine
`At least 14 double-blind, controlled studies published to
`date have shown carbamazepine to be effective in acute
`mania (Keck et al 1992); however, only five of these
`studies were not confounded by the simultaneous coad-
`ministration of carbamazepine with lithium and/or stan-
`dard antipsychotics (Table 4). One of these studies was
`
`placebo controlled, two compared carbamazepine with
`lithium,
`and
`two
`compared
`carbamazepine with
`chlorpromazine.
`In the placebo-controlled study (N ⫽ 19), which utilized
`a crossover (BABA) design, 63% of patients receiving
`carbamazepine (from 11 to 56 days; mean dose 1242
`mg/day; mean plasma level 10.4 ⫾ 2.2 mg/mL) displayed
`significant improvement on nursing staff global ratings of
`mania on the Bunney–Hamburg scale (Ballenger and Post
`1978; Post et al 1984, 1987). Eight of nine responders who
`received placebo discontinuation trials displayed a “re-
`lapse in manic or psychotic symptomatology.” Factors
`significantly associated with favorable antimanic response
`to carbamazepine were greater severity of mania and
`presence of rapid cycling; greater dysphoria during mania
`and family history negative for mood disorder tended to be
`associated with carbamazepine response.
`In the first study comparing carbamazepine with lithium
`(Lerer et al 1987) 34 inpatients with bipolar disorder,
`manic phase, by DSM-III criteria were randomized to
`either drug for up to 4 weeks. Twenty-eight patients (14 on
`each drug) completed the 4-week study period and were
`included in the data analysis. Although the overall re-
`sponse to treatment was not significantly different be-
`tween the two groups, a more consistent level of improve-
`ment was seen in the lithium-treated group, as compared
`with a minority of robust responders in the carbamaz-
`epine-treated group. Specifically, both groups displayed
`significant and comparable improvement on the BPRS and
`the Beigel–Murphy Manic State Rating Scale, and trends
`toward superior improvement with lithium on both scales
`were not significant; however, the Clinical Global Impres-
`sion (CGI) change scores for the lithium-treated group
`showed statistically significant improvement, as compared
`with those for the carbamazepine group. Specifically, only
`four (29%) of 14 patients receiving carbamazepine were
`evaluated as having a good response, compared with 11
`(79%) of 14 patients receiving lithium.
`In the second lithium comparison study (Small et al
`1991) two thirds of 52 hospitalized patients with treat-
`ment-refractory mania randomly assigned to lithium or
`carbamazepine had dropped out by 8 weeks of treatment
`because of lack of efficacy or refusal to continue. Of the
`48 patients who remained in the study for at least 3 weeks,
`33% of 24 carbamazepine-treated patients were rated as
`improved (defined as at least partial remission of symp-
`toms), as were 33% of 24 lithium-treated patients. Double-
`blind assessments revealed no statistically significant fac-
`tors associated with response to either drug.
`In the first carbamazepine–chlorpromazine comparison
`study (Okuma et al 1979) 60 acutely manic patients were
`randomized to receive carbamazepine (N ⫽ 32) or chlor-
`promazine (N ⫽ 28) in a 6-week trial. The two drugs were
`
`6
`
`

`

`Pharmacologic Agents and Acute Bipolar Mania
`
`BIOL PSYCHIATRY
`2000;48:539–557
`
`545
`
`Table 4. Double-Blind, Controlled Studies of Carbamazepine in Acute Bipolar Mania
`
`Study
`Placebo controlled
`Post et al 1984, 1987
`
`Klein et al 1984
`
`Mu¨ller and Stoll 1984
`
`Desai et al 1987
`
`Mo¨ller et al 1989
`
`Lithium controlled
`Lerer et al 1987
`
`Design
`
`Random, crossover
`
`Random, parallel-group,
`add-on to HAL
`Random, parallel-group,
`add-on to HAL
`Random, parallel-group,
`add-on to Li
`
`Na
`
`19
`
`CBZ 14
`PBO 13
`CBZ 6
`PBO 0
`CBZ 5
`
`Random, parallel-group,
`add-on to HAL, LEV
`
`CBZ 11
`PBO 9
`
`Random, parallel-group
`
`Duration
`(days)
`
`11–56
`
`21
`
`28
`
`21
`
`28
`
`21–56
`
`42
`
`21
`
`Outcome
`
`12 (63%) response to CBZ,
`8 (89%) of 9 relapse on PBO substitution
`10 (71%) response to CBZ ⫹ HAL,
`7 (53%) response to PBO ⫹ HAL
`CBZ ⫹ HAL improvement superior to PBO ⫹
`HAL improvement
`CBZ ⫹ Li superior to CBZ ⫹ PBO on BRMS
`(p ⬍ .05) and CGI (p ⬍ .05), but ⫽ on
`BPRS
`CBZ ⫹ HAL ⫽ PBO ⫹ HAL in reducing
`manic symptoms, CBZ ⫹ HAL group needed
`less LEV
`
`4 (29%) of 14 response to CBZ,b
`11 (79%) of 14 response to Li (p ⬍ .05)
`8 (33%) of 24 response CBZ,c
`8 (33%) of 24 response Li (ns)
`
`17 (65%) of 26 response to CBZ,b
`12 (52%) of 23 response to CPZ at week 3 (ns),
`20 (71%) of 28 response to CBZ,
`14 (56%) of 25 response to CPZ at week 6 (ns)
`10 (67%) of 15 response to CBZ,b
`13 (76%) of 17 response to CPZ (ns)
`51 (52%) of 98 response to CBZ,b
`19 (50%) of 38 response to Li,
`25 (63%) of 40 response to CPZ (ns)
`
`CBZ 15
`Li 19
`52
`
`CBZ 32
`CPZ 28
`
`CBZ 18
`CPZ 19
`CBZ
`Li
`CPZ
`
`Small et al 1991
`
`Random, parallel-group
`
`Antipsychotic controlled
`Okuma et al 1979
`
`Random, parallel-group
`
`Grossi et al 1984
`
`Random, parallel-group
`
`Overall monotherapy response
`
`CBZ, carbamazepine; PBO, placebo; HAL, haloperidol; Li, lithium; BRMS, Bech–Rafaelson; CGI, Clinical Global Impression; BPRS, Brief Psychiatric Rating Scale;
`LEV, levo-nepromazine; CPZ, chlorpromazine.
`aIndicates number of patients randomized to controlled treatment.
`bOutcome evaluated only in completers.
`cOutcome evaluated in patients who completed at least 3 weeks of treatment.
`
`equally eff