A Double-Blind
Comparison in the Treatment
Thomas
W. Freeman, Michael
of Valproate and Lithium of Acute Mania
M.D., Jeffrey L. Clothier, M.D., Peggy Pazzaglia, D. Lesem, M.D., and Alan C. Swann, M.D.
M.D.,
Objective: This study was carried out to compare the efficacy oflithium carbonate with that of valproate in acute mania and to determine whether pretreatment clinical characteristics, such as the presence ofa mixed affective state, might predict a differential response to the two drugs. Method: Twenty-seven patients meeting DSM-III-R criteria for acute manic episodes underwent
a 3-week,
randomized,
double-blind,
parallel-groups
trial
of treatment
with
lit!,-
ium carbonate or valproate. Symptom severity was measured by using the Schedule for Affective Disorders and Schizophrenia, change version (SADS-C), the Global Assessment Scale (GAS), and the Brief Psychiatric Rating Scale (BPRS). Drug effects were compared by using repeated measures analysis ofvariance (ANOVA). Results: At the end ofthe study, nine of 14 patients treated with valproate and 12 of 13 patients treated with lithium had responded favorably, as measured by changes in the SADS-C mania, BPRS, and GAS scores. Elevated pretreatment SADS-C depression scores were associated with good response to valproate. ANO VA revealed a significant interaction between drug and mixed affective state with respect to treatment response. Conclusions: Lithium and vaiproate were both effective in improving manic symptoms, and lithium was slightly more efficacious overall. Unlike the case with lithium, favorable response to valproate was associated with high pretreatment depression scores. Therefore, treatment with valproate alone may be particularly effective in manic patients with mixed affective states. (AmJ
Psychiatry
1992;
149:108-111)
L
ithium
lack
episodes of mania. Despite the efficacy of ill effects of lithium, a substantial
salts
are
the
standard
treatment
for
acute
and relative number of
patients with manic disorder do not respond to treatment with lithium alone. The best-defined groups of patients who do not respond well to lithium are those with mixed affective states (1-3) and with four or more cydes per year (4). Of these groups, patients with mixed presentations are the more common, representing 30%-S0% of manic episodes (1). However, prospective treatment trials provide essentially no information about effective regimens for the mixed state. Emnich et al. (5, 6) and Post et al. (7) suggested that
Received July 12, 1990; revisions received Jan. 31 and May 29, 1991; acceptedJuly 24, 1991. From the Department ofPsychiatny and Behavioral Sciences, Harris County Psychiatric Center, University of Texas Medical School, Houston. Address reprint requests to Dr. Swann, Department of Psychiatry and Behavioral Sciences, Harris County Psychiatric Center, University of Texas Medical School, P.O. Box 20708, Houston, TX 77225. The authors thank the nursing staff of the Clinical Research Unit of the Harris County Psychiatric Center; Anne Roache, Phanm.D., the research pharmacist; and Kim Stoner, the project coordinator. Copyright © 1992 American Psychiatric Association.
i08
certain cluding
tive
drugs previously canbamazepine
in manic
episodes.
used as anticonvulsants, and valproate, may
inbe effec-
These
be espe-
drugs
would
cially useful if they were effective in patients who would not be expected to respond well to lithium. Data from retrospective studies suggest that valproatc may be effective in patients with rapid cycling (8) and that open treatment with valpmoate was effective in mixed states (9). As a preliminary investigation of differential predictors of response to lithium compared with valpnoate, we have carried out a double-blind, parallel-groups trial of lithium versus valproate in patients admitted to our clinical research unit with acute manic episodes.
METHOD Twenty-seven patients, 21 women and six men, were admitted to our clinical research unit. The patients were consecutive admissions meeting DSM-III-R criteria for a manic episode. After giving informed consent, paticnts entered a 4-7-day placebo washout period during which the DSM-III-R diagnosis of manic episode was
Am J Psychiatry
149:1,
January
1992
FREEMAN,
confirmed
independently
chiatnists
(J.L.C.,
by
two
board-certified
psy-
M.D.L.,
on A.C.S.) using a scmistmuctuned interview and a DSM-III-R checklist. Baseline clinical matings and medical and laboratory evaluations, including physical and neurological examinations, CBC, liven and thyroid function tests, electrolytes, EEG, and ECG (if the patient was over SO years old), were completed during this period. Patients with abnormal EEG, liver function, thyroid function, or hematologic findings were excluded from the study, as were patients with positive urine drug screens and patients with focal neurological abnormalities. Patients were also excluded if they had DSM-III-R diagnoses of drug or alcohol dependence or drug on alcohol abuse within 12 months before admission on before the onset of major affective episodes. Eight subjects assigned to valpmoate and seven subjects assigned to lithium had histories of drug on alcohol abuse at some point during the course of their bipolar disorder. Other clinical characteristics of the two groups were also comparable. Time since the first episode averaged 12.3±6.2 years (mange=0-22) for valproate patients and 8.8±5.9 years (range=0-20) for lithium patients; each group included one patient having a first manic episode. Twelve patients in the valpnoatc group had been treated with lithium; seven had failed to respond at least once. In the lithium group, 10 patients had previously been treated with lithium, and six had at least one treatment
failure.
No patients
had
even received
Patients were assigned randomly lithium or valproate in a double-blind iurn group, consisting of 10 women
ceived
lithium
citrate
as the
drate
and
5.0±5.6
mg of lorazepam.
but dropped out after included two patients
starting
at 0.5
sponden
and
signed
one
to lithium
end
of the
week. Medicine was given three times a day, diluted by the research pharmacist to a total volume of 30 ml with
each
diluent. drug
week,
levels
the results
were
were
monitored
sent
on
days
to the research
5-7
of
phan-
macist, and the dose was adjusted by a nonblinded, nontncating physician. The Schedule for Affective Disorders and Schizophrenia, change version (SADS-C) (10), the Global Assessment Scale (GAS) (11), and the Brief Psychiatric Rating Scale (BPRS) (12) were administened at the time of dose adjustments. Rescue medication, including chloral hydrate and
Am J Psychiatry
149:1,
January
1992
in
nonnesponden)
and
and one nonnespond-
responders
three
These (one re-
(two
patients
as-
en). Patients dropped out because of deterioration, recovery, on desire to leave the study. No patients had to leave the study because of adverse drug effects. Statistical comparisons used Student’s two-tailed t test on analysis of variance (ANOVA), as described in the Results section. Categorical response and nonresponse were defined essentially as described by Secunda et al. (1). An improvement in the SADS-C mania factor of at least 50% was defined as a favorable outcome. Patients whose pretreatment SADS-C depression scones
were
greater
than
nia. There were drug treatment
30 were
defined
four patients group.
with
as having mixed
mixed
mania
ma-
in each
RESULTS
age or sex.
at the
study ranged from 0.8 to 1 .4 rnmol/liter. The valproate group, consisting of 1 1 women and three men, was started on a fixed-dose regimen of valproate as the elixir. Valproate patients received 1500 mg/day during the first week, 2250 mg/day during the second week, and 3000 mg/day during the third week, unless their symptoms had remitted on they experienced dose-limiting side effects. Valpnoate levels were 85±1 7 jig/ml (range=67-1 1 8) at the end of the first week, 98±27 jig/ml (range=71-147) at the end of the second week, and 98±17 jig/ml (range=65-126) at the end ofthe third
flavored Plasma
patients
8-20 days of treatment. assigned to valpnoatc
provement
levels
Four
each group required no rescue medication. Five patients out of the 27 whose outcomes were evaluated did not complete the full 21-day protocol
Comparison of the lithium groups revealed no significant BPRS, GAS, or SADS-C factor There were also no differences
Lithium
ET AL.
lonazepam, was allowed for extreme behavioral problems not responding to nonphammacological interventions. The amount of rescue medication needed did not differ between the treatment groups: the patients assigned to valpnoatc received 2.9±5.9 g of chloral hydrate and 5.7±5.9 mg of lomazepam and the patients assigned to lithium received 4.4±5.4 g of chlonal hy-
per day. The dose of lithium was increased until either a maximum dose of 1 800 mg/day, level of 1.5 mmol/liter, dose-limiting side effects, or clinical immeq/kg
occurred.
PAZZAGLIA,
valproate. to treatment with fashion. The lithand three men, mc-
elixir,
CLOTHIER,
Nine
patients
of the valproate had favorable
ment. In neither responders with
patients outcomes
and
valpmoate treatment differences in initial scores (Student’s t test). between the groups in and after
12 of the lithium 3 weeks of treat-
group did responders differ from respect to drug doses or plasma
nondrug
levels or in pretreatment SADS-C mania scones. Table 1 shows initial and final SADS-C mania scores for all patients treated with lithium and with valproate and for responders to the two drugs. Pretreatment mania scones were nonsignificantly higher in patients as-
signed
to valproate.
Neither
pretreatment
scores
nor
scores after treatment differed significantly between lithiumand valproate-tncated patients. Repeated measures ANOVA, however, showed a significant drug group effect, and posttreatrncnt mania scores were higher in patients given valpnoate. Table 1 also summarizes the effects of lithium and valpnoate on BPRS scores and on global impairment as measured by the GAS. As with the SADS-C mania facton, the effects of treatment appeared highly significant. In addition, there was a trend toward a significant group effect for BPRS scores. Table 1 also shows the SADS-C depression scores in responders and nonrespondens to valproate before and after treatment. Pre-
109
COMPARISON
TABLE
Scores
OF VALPROATE
AND
LITHIUM
1. SADS-C
Mania, BPRS, GAS, and SADS-C Depression With Acute Mania Before and After Treatment or Lithium
of Patients
With Valproate
lithium and with mixed
valpnoate differ in their and nonmixed mania.
effects
on patients
Score Before Treatment Test and Group
Mean
SADS-C maniaa All patients given valpnoate (N=14) Patients who responded to valproate (N=9) All patients given lithium (N=13) Patients who responded to lithium(N=12) BPRS totalc All patients given valproate (N=14) All patients given lithium (N=13) GAS” All patients given vahpnoate (N=14) All patients given lithium (N=13) SADS-C depression All patients given vahproate (N=14) Patients who responded to valproate (N=9) Patients who did not respond to valproate (N=S) All patients given lithium (N=13) aRepeated
measures
ANOVA:
DISCUSSION
After Treatment SD
52.9
12.3
54.2
9.8
43.4
20.3
44.7
20.6
Mean
SD
20.9
271b
12.9
9.2
102b
8.8
9.0
8.2
28.4
9.4
10.6
9.1
24.1
9.8
4.3
5.7
30.9
14.9
61.9
20.8
27.6
12.6
71.2
14.6
18.9
15.9
9.1
8.4
25.6e
15.8
9.0
8.9
70’
6.3
9.4
7.5
5.7 df=1,
8.1 25, p=
23.6 group effect,
9.5 F=7.13,
0.013; response effect, F=45.8, df=1, 15, pe10; interaction, F= 0.75, df=1, 25, n.s. bSignificant difference in posttreatment score between all patients given valproate and all patients given lithium (t=2.12, df=2S, p< 0.05). cRepeated measures ANOVA: group effect, F=3.2, df=1, 25, p=O.O8; response effect, F=76.9, df=1, 25, p
Comparison in the Treatment
Thomas
W. Freeman, Michael
of Valproate and Lithium of Acute Mania
M.D., Jeffrey L. Clothier, M.D., Peggy Pazzaglia, D. Lesem, M.D., and Alan C. Swann, M.D.
M.D.,
Objective: This study was carried out to compare the efficacy oflithium carbonate with that of valproate in acute mania and to determine whether pretreatment clinical characteristics, such as the presence ofa mixed affective state, might predict a differential response to the two drugs. Method: Twenty-seven patients meeting DSM-III-R criteria for acute manic episodes underwent
a 3-week,
randomized,
double-blind,
parallel-groups
trial
of treatment
with
lit!,-
ium carbonate or valproate. Symptom severity was measured by using the Schedule for Affective Disorders and Schizophrenia, change version (SADS-C), the Global Assessment Scale (GAS), and the Brief Psychiatric Rating Scale (BPRS). Drug effects were compared by using repeated measures analysis ofvariance (ANOVA). Results: At the end ofthe study, nine of 14 patients treated with valproate and 12 of 13 patients treated with lithium had responded favorably, as measured by changes in the SADS-C mania, BPRS, and GAS scores. Elevated pretreatment SADS-C depression scores were associated with good response to valproate. ANO VA revealed a significant interaction between drug and mixed affective state with respect to treatment response. Conclusions: Lithium and vaiproate were both effective in improving manic symptoms, and lithium was slightly more efficacious overall. Unlike the case with lithium, favorable response to valproate was associated with high pretreatment depression scores. Therefore, treatment with valproate alone may be particularly effective in manic patients with mixed affective states. (AmJ
Psychiatry
1992;
149:108-111)
L
ithium
lack
episodes of mania. Despite the efficacy of ill effects of lithium, a substantial
salts
are
the
standard
treatment
for
acute
and relative number of
patients with manic disorder do not respond to treatment with lithium alone. The best-defined groups of patients who do not respond well to lithium are those with mixed affective states (1-3) and with four or more cydes per year (4). Of these groups, patients with mixed presentations are the more common, representing 30%-S0% of manic episodes (1). However, prospective treatment trials provide essentially no information about effective regimens for the mixed state. Emnich et al. (5, 6) and Post et al. (7) suggested that
Received July 12, 1990; revisions received Jan. 31 and May 29, 1991; acceptedJuly 24, 1991. From the Department ofPsychiatny and Behavioral Sciences, Harris County Psychiatric Center, University of Texas Medical School, Houston. Address reprint requests to Dr. Swann, Department of Psychiatry and Behavioral Sciences, Harris County Psychiatric Center, University of Texas Medical School, P.O. Box 20708, Houston, TX 77225. The authors thank the nursing staff of the Clinical Research Unit of the Harris County Psychiatric Center; Anne Roache, Phanm.D., the research pharmacist; and Kim Stoner, the project coordinator. Copyright © 1992 American Psychiatric Association.
i08
certain cluding
tive
drugs previously canbamazepine
in manic
episodes.
used as anticonvulsants, and valproate, may
inbe effec-
These
be espe-
drugs
would
cially useful if they were effective in patients who would not be expected to respond well to lithium. Data from retrospective studies suggest that valproatc may be effective in patients with rapid cycling (8) and that open treatment with valpmoate was effective in mixed states (9). As a preliminary investigation of differential predictors of response to lithium compared with valpnoate, we have carried out a double-blind, parallel-groups trial of lithium versus valproate in patients admitted to our clinical research unit with acute manic episodes.
METHOD Twenty-seven patients, 21 women and six men, were admitted to our clinical research unit. The patients were consecutive admissions meeting DSM-III-R criteria for a manic episode. After giving informed consent, paticnts entered a 4-7-day placebo washout period during which the DSM-III-R diagnosis of manic episode was
Am J Psychiatry
149:1,
January
1992
FREEMAN,
confirmed
independently
chiatnists
(J.L.C.,
by
two
board-certified
psy-
M.D.L.,
on A.C.S.) using a scmistmuctuned interview and a DSM-III-R checklist. Baseline clinical matings and medical and laboratory evaluations, including physical and neurological examinations, CBC, liven and thyroid function tests, electrolytes, EEG, and ECG (if the patient was over SO years old), were completed during this period. Patients with abnormal EEG, liver function, thyroid function, or hematologic findings were excluded from the study, as were patients with positive urine drug screens and patients with focal neurological abnormalities. Patients were also excluded if they had DSM-III-R diagnoses of drug or alcohol dependence or drug on alcohol abuse within 12 months before admission on before the onset of major affective episodes. Eight subjects assigned to valpmoate and seven subjects assigned to lithium had histories of drug on alcohol abuse at some point during the course of their bipolar disorder. Other clinical characteristics of the two groups were also comparable. Time since the first episode averaged 12.3±6.2 years (mange=0-22) for valproate patients and 8.8±5.9 years (range=0-20) for lithium patients; each group included one patient having a first manic episode. Twelve patients in the valpnoatc group had been treated with lithium; seven had failed to respond at least once. In the lithium group, 10 patients had previously been treated with lithium, and six had at least one treatment
failure.
No patients
had
even received
Patients were assigned randomly lithium or valproate in a double-blind iurn group, consisting of 10 women
ceived
lithium
citrate
as the
drate
and
5.0±5.6
mg of lorazepam.
but dropped out after included two patients
starting
at 0.5
sponden
and
signed
one
to lithium
end
of the
week. Medicine was given three times a day, diluted by the research pharmacist to a total volume of 30 ml with
each
diluent. drug
week,
levels
the results
were
were
monitored
sent
on
days
to the research
5-7
of
phan-
macist, and the dose was adjusted by a nonblinded, nontncating physician. The Schedule for Affective Disorders and Schizophrenia, change version (SADS-C) (10), the Global Assessment Scale (GAS) (11), and the Brief Psychiatric Rating Scale (BPRS) (12) were administened at the time of dose adjustments. Rescue medication, including chloral hydrate and
Am J Psychiatry
149:1,
January
1992
in
nonnesponden)
and
and one nonnespond-
responders
three
These (one re-
(two
patients
as-
en). Patients dropped out because of deterioration, recovery, on desire to leave the study. No patients had to leave the study because of adverse drug effects. Statistical comparisons used Student’s two-tailed t test on analysis of variance (ANOVA), as described in the Results section. Categorical response and nonresponse were defined essentially as described by Secunda et al. (1). An improvement in the SADS-C mania factor of at least 50% was defined as a favorable outcome. Patients whose pretreatment SADS-C depression scones
were
greater
than
nia. There were drug treatment
30 were
defined
four patients group.
with
as having mixed
mixed
mania
ma-
in each
RESULTS
age or sex.
at the
study ranged from 0.8 to 1 .4 rnmol/liter. The valproate group, consisting of 1 1 women and three men, was started on a fixed-dose regimen of valproate as the elixir. Valproate patients received 1500 mg/day during the first week, 2250 mg/day during the second week, and 3000 mg/day during the third week, unless their symptoms had remitted on they experienced dose-limiting side effects. Valpnoate levels were 85±1 7 jig/ml (range=67-1 1 8) at the end of the first week, 98±27 jig/ml (range=71-147) at the end of the second week, and 98±17 jig/ml (range=65-126) at the end ofthe third
flavored Plasma
patients
8-20 days of treatment. assigned to valpnoatc
provement
levels
Four
each group required no rescue medication. Five patients out of the 27 whose outcomes were evaluated did not complete the full 21-day protocol
Comparison of the lithium groups revealed no significant BPRS, GAS, or SADS-C factor There were also no differences
Lithium
ET AL.
lonazepam, was allowed for extreme behavioral problems not responding to nonphammacological interventions. The amount of rescue medication needed did not differ between the treatment groups: the patients assigned to valpnoatc received 2.9±5.9 g of chloral hydrate and 5.7±5.9 mg of lomazepam and the patients assigned to lithium received 4.4±5.4 g of chlonal hy-
per day. The dose of lithium was increased until either a maximum dose of 1 800 mg/day, level of 1.5 mmol/liter, dose-limiting side effects, or clinical immeq/kg
occurred.
PAZZAGLIA,
valproate. to treatment with fashion. The lithand three men, mc-
elixir,
CLOTHIER,
Nine
patients
of the valproate had favorable
ment. In neither responders with
patients outcomes
and
valpmoate treatment differences in initial scores (Student’s t test). between the groups in and after
12 of the lithium 3 weeks of treat-
group did responders differ from respect to drug doses or plasma
nondrug
levels or in pretreatment SADS-C mania scones. Table 1 shows initial and final SADS-C mania scores for all patients treated with lithium and with valproate and for responders to the two drugs. Pretreatment mania scones were nonsignificantly higher in patients as-
signed
to valproate.
Neither
pretreatment
scores
nor
scores after treatment differed significantly between lithiumand valproate-tncated patients. Repeated measures ANOVA, however, showed a significant drug group effect, and posttreatrncnt mania scores were higher in patients given valpnoate. Table 1 also summarizes the effects of lithium and valpnoate on BPRS scores and on global impairment as measured by the GAS. As with the SADS-C mania facton, the effects of treatment appeared highly significant. In addition, there was a trend toward a significant group effect for BPRS scores. Table 1 also shows the SADS-C depression scores in responders and nonrespondens to valproate before and after treatment. Pre-
109
COMPARISON
TABLE
Scores
OF VALPROATE
AND
LITHIUM
1. SADS-C
Mania, BPRS, GAS, and SADS-C Depression With Acute Mania Before and After Treatment or Lithium
of Patients
With Valproate
lithium and with mixed
valpnoate differ in their and nonmixed mania.
effects
on patients
Score Before Treatment Test and Group
Mean
SADS-C maniaa All patients given valpnoate (N=14) Patients who responded to valproate (N=9) All patients given lithium (N=13) Patients who responded to lithium(N=12) BPRS totalc All patients given valproate (N=14) All patients given lithium (N=13) GAS” All patients given vahpnoate (N=14) All patients given lithium (N=13) SADS-C depression All patients given vahproate (N=14) Patients who responded to valproate (N=9) Patients who did not respond to valproate (N=S) All patients given lithium (N=13) aRepeated
measures
ANOVA:
DISCUSSION
After Treatment SD
52.9
12.3
54.2
9.8
43.4
20.3
44.7
20.6
Mean
SD
20.9
271b
12.9
9.2
102b
8.8
9.0
8.2
28.4
9.4
10.6
9.1
24.1
9.8
4.3
5.7
30.9
14.9
61.9
20.8
27.6
12.6
71.2
14.6
18.9
15.9
9.1
8.4
25.6e
15.8
9.0
8.9
70’
6.3
9.4
7.5
5.7 df=1,
8.1 25, p=
23.6 group effect,
9.5 F=7.13,
0.013; response effect, F=45.8, df=1, 15, pe10; interaction, F= 0.75, df=1, 25, n.s. bSignificant difference in posttreatment score between all patients given valproate and all patients given lithium (t=2.12, df=2S, p< 0.05). cRepeated measures ANOVA: group effect, F=3.2, df=1, 25, p=O.O8; response effect, F=76.9, df=1, 25, p
