Haloperidol Plasma A Therapeutic
Theodore
Van
Putten,
Levels and Clinical Response: Window Relationship
M.D., Stephen R. Marder, M.D., and Russell E. Poland, Ph.D.
Jim Mintz,
Ph.D.,
Objective: The purpose ofthe study was to assess the relationship between plasma haloperidol and clinical response. Method: Sixty-nine newly admitted drug-free schizophrenic men were randomly assigned to receive haloperidol, 5, 1 0, or 20 mg daily for 4 weeks, and clinical response
was
sensitive between ent
measured
at the
end
of the
and specific radioimmunoassay. clinical response and plasma
optimum
between
5 and
12 ng/ml.
the 5-1 2 ng/ml range, all patients When plasma levels of nonresponders ng/ml phoric.
(as in routine practice), With the 20-mg dose,
In this
sample
a plasma (Am
of newly
J
range
Psychiatry
1992;
When
improved within
they, halfthe
admitted
haloperidol
fixed-dose
Results: haloperidol
schizophrenic
of5-12
UCLA;
Dec. 1991.
27,
1 990;
From
the
and the Harbor-UCLA
revision VA,
received
Aug.
Brentwood
Medical
5, 1991;
Division,
Center,
Los
tion
500
and
Ross
Batdessarini,
M.D.,
Los Angeles.
Award Ph.D.,
for extensive
above
12 ng/ml
men,
optimal
clinical
by a
relationship with an apparwere
lowered
to
no patient deteriorated. were raised above
12
they became more dys1 2 ng/ml. Conclusions: response
occurred
with
MH-00534
for statistical editorial
too high-60 versus 120 mg (7), 0.4 mg/kg (6), and 30 mg ( 1 1 )-to have plasma levels in the subtherapeutic range. In other words, of the 1 0 fixed-dose studies with halopenidol, four, in retrospect, were designed propemly, and these four found a therapeutic window. In the four studies (2-5) that suggested a therapeutic range, there were only 26 patients that defined the proposed toxic range, and 1 7 of these were in Potkin et al.’s investigation (5) of Chinese schizophrenic patients, in which the curvilinear fit explained only 8% of the vanance of clinical improvement. Further, these studies failed to discuss the clinical state of patients with high, supraoptimal halopenidol levels. Do such patients appear overmedicated, and, more importantly, do they improve when plasma levels are lowered? This study addresses these issues. The matter is important, for if a therapeutic window relationship exists, it could have significant implications for clinical practice. In particulan, the dose would be reduced in some nonnesponding patients.
accepted Angeles;
Address
reprint requests to Dr. Van Putten, VA Medical Center, I 1301 Witshire Blvd., Bldg. 210C, Los Angeles, CA 90073. Supported by the VA Medical Center, Psychiatry Division, and by
NIMH Research Scientist Development Poland). The authors thank Wilfrid J. Dixon,
levels
to varying degrees and this therapeutic window
assayed
149:500-505)
detect a relationship between plasma halopenilevels and clinical response, patients need to be treated with a fixed dose of halopenidol. Preferably, patients are randomly assigned to a low, moderate, and high dose so that both ends of the therapeutic window (if such exists) are represented. Variable-dose studies tend to produce artifactual therapeutic windows (1). Four fixed-dose studies with halopenidol suggest a therapeutic window relationship between plasma halopenidol and clinical response. The suggested optimal drug concentrations are 6.5-1 6 ng/ml (2), 3-1 1 ng/ml (3), 4-1 1 ng/ml (4), and 4-22 ng/ml (5). Six fixed-dose studies did not find a therapeutic window relationship, but they were so designed that detection of a therapeutic window relationship (if such exists) was unlikely. Two of the studies (6, 7) used primarily poor neuroleptic responders; three used such low doses-S versus 10 mg (8), 6 mg (9), and 0.2 mg/kg (10)-that detection of an upper toxic limit was unlikely; and three used doses
27,
plasma
was
a curvilinear treatment,
ng/ml.
dol
Received
Haloperidol
on balance, deteriorated in that patients had plasma levels above
T
Sept.
period.
The authors found during fixed-dose
(to Dr. consuttaassistance.
METHOD Subjects The subjects were 82 newly at least 2 weeks but usually phnenic men. They had given
Am
J
admitted, drug-free (for several months) schizoinformed consent, and
Psychiatry
1 49:4,
April
1992
VAN
each patient suffered from a schizophrenic disorder (DSM-III-R) without significant physical disease or drug or alcohol addiction. Patients with a history of nonresponse to neuroleptic drugs were excluded, as were those who had a history of severe extrapymamidal side effects with high-potency neuroleptics. Those with a history of poor compliance with medication on treatment also were excluded. Patients were observed in a drug-free state (except for chloral hydrate) for at least 3 days, and none started treatment with halopenidol as long as there was an indication of spontaneous improvement. Characteristics of the 69 patients who completed the study are detailed in table 1. The patients’ ages averaged in the early 30s, and the patients averaged four previous hospitalizations. They had all served in the armed forces, had worked for a mean of 4.5 years at some time, were judged markedly ill on a nurses’ mating scale, and had scored at least moderate on conceptual disorganization, unusual thought content, on hallucinatory behavior on the Brief Psychiatric Rating Scale (BPRS) ( 12). In fact, their mean baseline BPRS Schizophrenia factor score was 13 (normal=3, maximum=21), indicating that they were quite psychotic at baseline. Very excited or menacing patients were not included. Procedure The 82 patients were assigned to receive halopenidol, 5, 10, on 20 mg daily for 4 weeks. Patients were assigned by cohort. The first 38 eligible patients were given the 10-mg dose, the next 22 eligible patients were given the S-mg dose, and the next 22 eligible patients were given the 20-mg daily dose. Assignment to dose was independent of the patient’s presenting clinical symptoms. Of the 82 patients, halopenidol plasma 1evels were available for 69 (of the 13 subjects lost from the study, four eloped, four had protocol violations, three refused blood drawing, and all blood samples were lost for two patients). In case of nonresponse, the physician could increase (or decrease) the dose for another 4 weeks according to clinical judgment. Clinical response was measured at baseline, weekly for the first 4 weeks, and at week 8 after the flexible-dose period. Clinical ratings were made by clinicians who were blind to plasma levels. Clinical status was assessed weekly in semistructured interviews with the first author (T.V.P.), using the BPRS and the Clinical Global Impression (CGI) (13). The rater was not blind to dose. Extrapyramidal side effects were rated with the Involuntary Movement and Extrapyramidal Side Effects Scale (unpublished manuscript; copies available on request from Dr. Van Putten). On this scale akathisia is scored on a 7-point scale and akinesia is rated for decreases in facial expression, gestunes, spontaneous movement, and speed of movement. Halopemidol was assayed by a sensitive and specific radioimmunoassay (14, 15) capable of detecting 0.5 ng/ml of halopenidol with a within-sample variance (standard deviation/mean) of less than 10%. All sam-
Am
J
Psychiatry
I 49:4,
April
1992
TABLE
1. Characteristics
PUTTEN,
of 69 Schizophrenic
Item Age (years) Previous
admissions
Age at onset symptoms
MARDER,
of psychotic (years)
MINTZ,
Men
Mean
SD
34.2
9.3
4.3
4.5
23.6
4.9
Race Black Caucasian Hispanic Living status Alone Family
Board and care Longest continuing period employment (years) Nurses’ mental Baseline
rating of degree of illness at baselinea BPRS score
TotalLy Schizophrenia a4Moderately bPerfect score=1 CPerfect score=3.
factor’ ill, 5=markedty 9.
ET AL.
N
%
37
53
30 2
44 3
35 31
51 45
3
4
of 4.5
3.7
4.9
1.1
49.6
9.7
13.1 ill, 6=severely
3.2 ill.
pies from each subject were analyzed together in the same batch. The maximum intra-assay and interassay coefficient of variation was 12%, as determined by multiple serum pool replicates analyzed repeatedly in all assays. Blood was drawn weekly in the morning, 10 to 12 hours after the last dose. Data
Analysis
Halopenidol plasma levels were averaged during fixeddose treatment, and clinical change was measuned by the change from baseline on the BPRS Psychosis cluster (Thought Disturbance and Hostile Suspiciousness clusters). Mean halopenidol plasma levels had an estimated reliability (intraclass correlation) of 0.83. These data were fit to a theoretical model that proposed the presence of two sigmoidal plasma level-effect functions, one representing positive treatment response (i.e., increasing BPRS Psychosis improvement scores) and the other representing negative effects (i.e., decreasing BPRS Psychosis improvement scores), as suggested by Teichem and Baldessarini (15). In this two-component model, the predicted BPRS change score is obtained by subtracting the negative component from the positive component. Each of the two hypothesized components is defined by its slope, minimum and maximum values. The full six-parameter model was initially estimated by using nonlinear regression (16). Very high correlations among the parameters, extremely large coefficients of variation, and poor convergence suggested that a simpier model might be better in this data set. A four-parametem model with equal slopes appeared to be optimal, yielding the same pseudo-R2 (0.46) as the full model. In the final model, coefficients of variation ranged from -0.28 to 0.34, and correlations between parameters ranged from -0.86 to 0.54.
501
HALOPERIDOL
FIGURE
PLASMA
1. Two Sigmoidal
Haloperidol -
LEVELS
Functions
Concentrations
of Outcome
in 69 Schizophrenic
component
Positive
in Relation to Plasma TABLE 2. Scores of 69 Schizophrenic Factor After Haloperidol Treatment
Men
Men on the BPRS Ps ychosis BPRS
Negathie component
Factor
.
Haloperidot Plasma Range (ng/ml) (I)
0
r
-2 2 12
8
PLASMA
HALOPERIDOL
16 (ng/ml)
FIGURE 2. Curvilinear Relationship Between Plasma Haloperidol Concentration and Change in BPRS Psychosis Factor for 69 Schizophrenic Men 20
0
U)
0
16
;
12
a.
00
U)
a.
z4 Ui
(0 z I
o-4 0
4
8
PLASMA
12
HALOPERIDOL
16
20
24
(ng/ml)
RESULTS Figure 1 shows the two sigmoidal dose-effect curves. Negative responses start at about 8 ng/ml and plateau at about 15 ng/ml. Positive treatment response plateaus at about 1 0 ng/ml. If we subtract the negative from the positive response curve, the resultant curve appears to fit the data rather well (r=0.46, df=66, p=O.OO1). Figure 2 shows this two-component model supemimposed on the scatterplot of change in the BPRS Psychosis factor versus mean plasma halopenidol. By inspection, it appears reasonable to divide the data into the following nanogram/milliliter manges: less than 2 (ineffective), 2-S (threshold), 5-12 (optimal), and more than 12 (suboptimal or toxic). Mean improvement on the BPRS Psychosis factor was significantly greater by t test in the optimal than in the suboptimal range (p=0.094) or the threshold range (p=O.OO4) (table 2). Mean improvement in the optimal range was roughly twice that in the threshold and suboptimal ranges. The standardized effect sizes were both large (0.95 and 0.90, respectively). Thirteen patients had mean plasma halopenidol levels above 12 ng/ml at the end of the fixed-dose period. Three insisted on leaving the hospital before a dose ad-
502
SD
N
12
Ineffective
(12)
I 0
Mean
Threshold
Optimal
U)
Score
-10
U)
I
Psychosis
Level
justment was possible (all had become more paranoid [mean BPRS Paranoia score=-3.S, SD=2.S], and two had become extremely dysphonic). One patient with a mean plasma halopenidol level of 15.6 ng/ml insisted that he required more halopenidol, and his plasma level eventually reached 90 ng/ml (see Discussion). Another patient with a plasma halopenidol level of 14.7 ng/ml was mildly retarded (BPRS Withdrawal-Retardation score=-2) but refused dose reduction. When the plasma halopemidol levels of the remaining eight patients were reduced to less than 12 ng/ml (range=4.0-10.8 ng/ml, mean=7.8), they experienced milder side effects (in particular, a subjective sense of sedation or objective akinesia) and became less psychotic (N=S), less dysphonic (N=3), and less retarded (N=8). No patient deteniorated. Two patients had developed peculiar delusions of bodily destruction, which vanished as the plasma level was lowered. Table 3 summarizes these changes. In eight relative nonnesponders (defined as patients with a global improvement rating of minimally improved or less) with plasma levels in the 2-12 ng/ml range, the plasma level was raised above 12 ng/ml (mange=12.9-24 ng/ml, mean=17.8) by increasing mean daily dose from 9.2 mg (SD=2.0) to 23.3 (SD=7.S). Six of eight patients became worse on the global improvement ratings, and, on balance, these six became more dysphonic. Table 4 summarizes the changes. Figure 2 suggests a threshold limit of 2 ng/ml. If so, patients with plasma levels lower than 2 ng/ml at the end of the fixed-dose period should improve as the plasma level is raised to 2-12 ng/ml. Patients with plasma levels of 1.8, 1.5, 1.6, 1.1, 1.1, and 1.4 ng/ml all improved (with the CGI as the criterion) as their plasma levels nose above 2.0 ng/ml (the plasma levels at which improvement occurred were 2.9, 2.6, 2.1, 2.2, 2.0, and 6.7 ng/ml, respectively). One man with a plasma level of 1 .8 ng/ml improved remarkably.
DISCUSSION A curvilinear relationship was found between plasma halopenidol and clinical response when patients were given fixed doses of halopenidol. The approximate upper and lower limits of this proposed therapeutic window were 12 and S ng/ml, respectively, with 2 ng/ml as a threshold level. When plasma halopenidol levels in relative nonmespondens were pushed beyond 12 ng/ml
Am
J
Psychiatry
1 49:4,
April
1992
VAN
TABLE
3. Cha nge in BPRS Scores of Eight Schizophrenic
Men Whe n Plasma Haloperidol Change
Hatoperidot Level (ng/mt) Weeks
Mean
1-4
15.0
2.1
5-8
7.8
2.1
mean paired
more improvement; t test, two-tailed,
aHigher
change
bt276
df=7,
ct193
df=7, df=7,
dt534
scores p
Theodore
Van
Putten,
Levels and Clinical Response: Window Relationship
M.D., Stephen R. Marder, M.D., and Russell E. Poland, Ph.D.
Jim Mintz,
Ph.D.,
Objective: The purpose ofthe study was to assess the relationship between plasma haloperidol and clinical response. Method: Sixty-nine newly admitted drug-free schizophrenic men were randomly assigned to receive haloperidol, 5, 1 0, or 20 mg daily for 4 weeks, and clinical response
was
sensitive between ent
measured
at the
end
of the
and specific radioimmunoassay. clinical response and plasma
optimum
between
5 and
12 ng/ml.
the 5-1 2 ng/ml range, all patients When plasma levels of nonresponders ng/ml phoric.
(as in routine practice), With the 20-mg dose,
In this
sample
a plasma (Am
of newly
J
range
Psychiatry
1992;
When
improved within
they, halfthe
admitted
haloperidol
fixed-dose
Results: haloperidol
schizophrenic
of5-12
UCLA;
Dec. 1991.
27,
1 990;
From
the
and the Harbor-UCLA
revision VA,
received
Aug.
Brentwood
Medical
5, 1991;
Division,
Center,
Los
tion
500
and
Ross
Batdessarini,
M.D.,
Los Angeles.
Award Ph.D.,
for extensive
above
12 ng/ml
men,
optimal
clinical
by a
relationship with an apparwere
lowered
to
no patient deteriorated. were raised above
12
they became more dys1 2 ng/ml. Conclusions: response
occurred
with
MH-00534
for statistical editorial
too high-60 versus 120 mg (7), 0.4 mg/kg (6), and 30 mg ( 1 1 )-to have plasma levels in the subtherapeutic range. In other words, of the 1 0 fixed-dose studies with halopenidol, four, in retrospect, were designed propemly, and these four found a therapeutic window. In the four studies (2-5) that suggested a therapeutic range, there were only 26 patients that defined the proposed toxic range, and 1 7 of these were in Potkin et al.’s investigation (5) of Chinese schizophrenic patients, in which the curvilinear fit explained only 8% of the vanance of clinical improvement. Further, these studies failed to discuss the clinical state of patients with high, supraoptimal halopenidol levels. Do such patients appear overmedicated, and, more importantly, do they improve when plasma levels are lowered? This study addresses these issues. The matter is important, for if a therapeutic window relationship exists, it could have significant implications for clinical practice. In particulan, the dose would be reduced in some nonnesponding patients.
accepted Angeles;
Address
reprint requests to Dr. Van Putten, VA Medical Center, I 1301 Witshire Blvd., Bldg. 210C, Los Angeles, CA 90073. Supported by the VA Medical Center, Psychiatry Division, and by
NIMH Research Scientist Development Poland). The authors thank Wilfrid J. Dixon,
levels
to varying degrees and this therapeutic window
assayed
149:500-505)
detect a relationship between plasma halopenilevels and clinical response, patients need to be treated with a fixed dose of halopenidol. Preferably, patients are randomly assigned to a low, moderate, and high dose so that both ends of the therapeutic window (if such exists) are represented. Variable-dose studies tend to produce artifactual therapeutic windows (1). Four fixed-dose studies with halopenidol suggest a therapeutic window relationship between plasma halopenidol and clinical response. The suggested optimal drug concentrations are 6.5-1 6 ng/ml (2), 3-1 1 ng/ml (3), 4-1 1 ng/ml (4), and 4-22 ng/ml (5). Six fixed-dose studies did not find a therapeutic window relationship, but they were so designed that detection of a therapeutic window relationship (if such exists) was unlikely. Two of the studies (6, 7) used primarily poor neuroleptic responders; three used such low doses-S versus 10 mg (8), 6 mg (9), and 0.2 mg/kg (10)-that detection of an upper toxic limit was unlikely; and three used doses
27,
plasma
was
a curvilinear treatment,
ng/ml.
dol
Received
Haloperidol
on balance, deteriorated in that patients had plasma levels above
T
Sept.
period.
The authors found during fixed-dose
(to Dr. consuttaassistance.
METHOD Subjects The subjects were 82 newly at least 2 weeks but usually phnenic men. They had given
Am
J
admitted, drug-free (for several months) schizoinformed consent, and
Psychiatry
1 49:4,
April
1992
VAN
each patient suffered from a schizophrenic disorder (DSM-III-R) without significant physical disease or drug or alcohol addiction. Patients with a history of nonresponse to neuroleptic drugs were excluded, as were those who had a history of severe extrapymamidal side effects with high-potency neuroleptics. Those with a history of poor compliance with medication on treatment also were excluded. Patients were observed in a drug-free state (except for chloral hydrate) for at least 3 days, and none started treatment with halopenidol as long as there was an indication of spontaneous improvement. Characteristics of the 69 patients who completed the study are detailed in table 1. The patients’ ages averaged in the early 30s, and the patients averaged four previous hospitalizations. They had all served in the armed forces, had worked for a mean of 4.5 years at some time, were judged markedly ill on a nurses’ mating scale, and had scored at least moderate on conceptual disorganization, unusual thought content, on hallucinatory behavior on the Brief Psychiatric Rating Scale (BPRS) ( 12). In fact, their mean baseline BPRS Schizophrenia factor score was 13 (normal=3, maximum=21), indicating that they were quite psychotic at baseline. Very excited or menacing patients were not included. Procedure The 82 patients were assigned to receive halopenidol, 5, 10, on 20 mg daily for 4 weeks. Patients were assigned by cohort. The first 38 eligible patients were given the 10-mg dose, the next 22 eligible patients were given the S-mg dose, and the next 22 eligible patients were given the 20-mg daily dose. Assignment to dose was independent of the patient’s presenting clinical symptoms. Of the 82 patients, halopenidol plasma 1evels were available for 69 (of the 13 subjects lost from the study, four eloped, four had protocol violations, three refused blood drawing, and all blood samples were lost for two patients). In case of nonresponse, the physician could increase (or decrease) the dose for another 4 weeks according to clinical judgment. Clinical response was measured at baseline, weekly for the first 4 weeks, and at week 8 after the flexible-dose period. Clinical ratings were made by clinicians who were blind to plasma levels. Clinical status was assessed weekly in semistructured interviews with the first author (T.V.P.), using the BPRS and the Clinical Global Impression (CGI) (13). The rater was not blind to dose. Extrapyramidal side effects were rated with the Involuntary Movement and Extrapyramidal Side Effects Scale (unpublished manuscript; copies available on request from Dr. Van Putten). On this scale akathisia is scored on a 7-point scale and akinesia is rated for decreases in facial expression, gestunes, spontaneous movement, and speed of movement. Halopemidol was assayed by a sensitive and specific radioimmunoassay (14, 15) capable of detecting 0.5 ng/ml of halopenidol with a within-sample variance (standard deviation/mean) of less than 10%. All sam-
Am
J
Psychiatry
I 49:4,
April
1992
TABLE
1. Characteristics
PUTTEN,
of 69 Schizophrenic
Item Age (years) Previous
admissions
Age at onset symptoms
MARDER,
of psychotic (years)
MINTZ,
Men
Mean
SD
34.2
9.3
4.3
4.5
23.6
4.9
Race Black Caucasian Hispanic Living status Alone Family
Board and care Longest continuing period employment (years) Nurses’ mental Baseline
rating of degree of illness at baselinea BPRS score
TotalLy Schizophrenia a4Moderately bPerfect score=1 CPerfect score=3.
factor’ ill, 5=markedty 9.
ET AL.
N
%
37
53
30 2
44 3
35 31
51 45
3
4
of 4.5
3.7
4.9
1.1
49.6
9.7
13.1 ill, 6=severely
3.2 ill.
pies from each subject were analyzed together in the same batch. The maximum intra-assay and interassay coefficient of variation was 12%, as determined by multiple serum pool replicates analyzed repeatedly in all assays. Blood was drawn weekly in the morning, 10 to 12 hours after the last dose. Data
Analysis
Halopenidol plasma levels were averaged during fixeddose treatment, and clinical change was measuned by the change from baseline on the BPRS Psychosis cluster (Thought Disturbance and Hostile Suspiciousness clusters). Mean halopenidol plasma levels had an estimated reliability (intraclass correlation) of 0.83. These data were fit to a theoretical model that proposed the presence of two sigmoidal plasma level-effect functions, one representing positive treatment response (i.e., increasing BPRS Psychosis improvement scores) and the other representing negative effects (i.e., decreasing BPRS Psychosis improvement scores), as suggested by Teichem and Baldessarini (15). In this two-component model, the predicted BPRS change score is obtained by subtracting the negative component from the positive component. Each of the two hypothesized components is defined by its slope, minimum and maximum values. The full six-parameter model was initially estimated by using nonlinear regression (16). Very high correlations among the parameters, extremely large coefficients of variation, and poor convergence suggested that a simpier model might be better in this data set. A four-parametem model with equal slopes appeared to be optimal, yielding the same pseudo-R2 (0.46) as the full model. In the final model, coefficients of variation ranged from -0.28 to 0.34, and correlations between parameters ranged from -0.86 to 0.54.
501
HALOPERIDOL
FIGURE
PLASMA
1. Two Sigmoidal
Haloperidol -
LEVELS
Functions
Concentrations
of Outcome
in 69 Schizophrenic
component
Positive
in Relation to Plasma TABLE 2. Scores of 69 Schizophrenic Factor After Haloperidol Treatment
Men
Men on the BPRS Ps ychosis BPRS
Negathie component
Factor
.
Haloperidot Plasma Range (ng/ml) (I)
0
r
-2 2 12
8
PLASMA
HALOPERIDOL
16 (ng/ml)
FIGURE 2. Curvilinear Relationship Between Plasma Haloperidol Concentration and Change in BPRS Psychosis Factor for 69 Schizophrenic Men 20
0
U)
0
16
;
12
a.
00
U)
a.
z4 Ui
(0 z I
o-4 0
4
8
PLASMA
12
HALOPERIDOL
16
20
24
(ng/ml)
RESULTS Figure 1 shows the two sigmoidal dose-effect curves. Negative responses start at about 8 ng/ml and plateau at about 15 ng/ml. Positive treatment response plateaus at about 1 0 ng/ml. If we subtract the negative from the positive response curve, the resultant curve appears to fit the data rather well (r=0.46, df=66, p=O.OO1). Figure 2 shows this two-component model supemimposed on the scatterplot of change in the BPRS Psychosis factor versus mean plasma halopenidol. By inspection, it appears reasonable to divide the data into the following nanogram/milliliter manges: less than 2 (ineffective), 2-S (threshold), 5-12 (optimal), and more than 12 (suboptimal or toxic). Mean improvement on the BPRS Psychosis factor was significantly greater by t test in the optimal than in the suboptimal range (p=0.094) or the threshold range (p=O.OO4) (table 2). Mean improvement in the optimal range was roughly twice that in the threshold and suboptimal ranges. The standardized effect sizes were both large (0.95 and 0.90, respectively). Thirteen patients had mean plasma halopenidol levels above 12 ng/ml at the end of the fixed-dose period. Three insisted on leaving the hospital before a dose ad-
502
SD
N
12
Ineffective
(12)
I 0
Mean
Threshold
Optimal
U)
Score
-10
U)
I
Psychosis
Level
justment was possible (all had become more paranoid [mean BPRS Paranoia score=-3.S, SD=2.S], and two had become extremely dysphonic). One patient with a mean plasma halopenidol level of 15.6 ng/ml insisted that he required more halopenidol, and his plasma level eventually reached 90 ng/ml (see Discussion). Another patient with a plasma halopenidol level of 14.7 ng/ml was mildly retarded (BPRS Withdrawal-Retardation score=-2) but refused dose reduction. When the plasma halopemidol levels of the remaining eight patients were reduced to less than 12 ng/ml (range=4.0-10.8 ng/ml, mean=7.8), they experienced milder side effects (in particular, a subjective sense of sedation or objective akinesia) and became less psychotic (N=S), less dysphonic (N=3), and less retarded (N=8). No patient deteniorated. Two patients had developed peculiar delusions of bodily destruction, which vanished as the plasma level was lowered. Table 3 summarizes these changes. In eight relative nonnesponders (defined as patients with a global improvement rating of minimally improved or less) with plasma levels in the 2-12 ng/ml range, the plasma level was raised above 12 ng/ml (mange=12.9-24 ng/ml, mean=17.8) by increasing mean daily dose from 9.2 mg (SD=2.0) to 23.3 (SD=7.S). Six of eight patients became worse on the global improvement ratings, and, on balance, these six became more dysphonic. Table 4 summarizes the changes. Figure 2 suggests a threshold limit of 2 ng/ml. If so, patients with plasma levels lower than 2 ng/ml at the end of the fixed-dose period should improve as the plasma level is raised to 2-12 ng/ml. Patients with plasma levels of 1.8, 1.5, 1.6, 1.1, 1.1, and 1.4 ng/ml all improved (with the CGI as the criterion) as their plasma levels nose above 2.0 ng/ml (the plasma levels at which improvement occurred were 2.9, 2.6, 2.1, 2.2, 2.0, and 6.7 ng/ml, respectively). One man with a plasma level of 1 .8 ng/ml improved remarkably.
DISCUSSION A curvilinear relationship was found between plasma halopenidol and clinical response when patients were given fixed doses of halopenidol. The approximate upper and lower limits of this proposed therapeutic window were 12 and S ng/ml, respectively, with 2 ng/ml as a threshold level. When plasma halopenidol levels in relative nonmespondens were pushed beyond 12 ng/ml
Am
J
Psychiatry
1 49:4,
April
1992
VAN
TABLE
3. Cha nge in BPRS Scores of Eight Schizophrenic
Men Whe n Plasma Haloperidol Change
Hatoperidot Level (ng/mt) Weeks
Mean
1-4
15.0
2.1
5-8
7.8
2.1
mean paired
more improvement; t test, two-tailed,
aHigher
change
bt276
df=7,
ct193
df=7, df=7,
dt534
scores p
