The effect of body weight changes on total plasma clozapine concentrations determined by applying a statistical model to the data from a double-blind trial

Abstract

Purpose/Background

Some therapeutic drug monitoring (TDM) studies suggest that increased weight is associated with small increases in clozapine concentrations. The goal of this study is to reanalyze a US double-blind study using a sophisticated statistical model to test whether weight gains from baseline or increases in percentage of body fat from baseline, computed from a published equation, are associated with increased total plasma clozapine concentrations after controlling for the effects of smoking and sex.

Methods/Procedures

Using data from a multi-dosage randomized double-blind US clozapine trial previously published, a random intercept linear model of steady-state total plasma clozapine concentrations was fitted to 424 concentrations from 47 patients.

Findings/Results

After adjusting for sex and smoking, 1) a 1-Kg gain in body weight during clozapine treatment was significantly associated with a 1.4% increase in total plasma clozapine concentrations [95% confidence interval, CI, (0.55, 2.3)], and 2) a 1-point increase in percentage of body fat during clozapine treatment was significantly associated with a 5.4% increase in total clozapine concentration (2.5, 8.3) in females, and 1.4% (−1.1, 4.0) in males.

Implications/Conclusions

As hypothesized, weight increases during clozapine treatment, which probably reflect increases in fat tissue, were associated with increases in total plasma concentrations. Pending further replication in other samples, it appears likely that clozapine may deposit in body fat and that this may decrease clozapine clearance. This change may be small in most patients but may be clinically relevant in females with major gains in body fat.

Introduction

There is general agreement that clozapine therapeutic drug monitoring (TDM) is the best established TDM in psychiatry.¹ Using data from a multi-dosage double-blind randomized clinical trial (RCT) completed in the United Sates (US),² a statistical random-effects linear model for personalizing clozapine dosing was developed.^3–6 In this inpatient study³ under highly controlled conditions (see Box 1^3,7–15), clozapine concentration was significantly associated with daily clozapine dosage, but sex and smoking status influenced clozapine concentrations as well.⁴ At comparable dosages, males have lower concentrations than females, and smokers have lower concentrations than non-smokers.⁴ This is compatible with our pharmacological knowledge that smoking is an inducer of CYP1A2, probably the most important clozapine metabolic pathway. The reasons that females metabolize clozapine more poorly are not completely established, but female sexual hormones are clinically relevant CYP1A2 inhibitors.¹⁴

Using the model in Italian samples,¹⁵ we confirmed that the inducer phenobarbital significantly decreased total clozapine concentrations on average by 28%, while the inhibitor fluvoxamine increased them by 263%. We also found that valproate appears to be a mild clozapine inducer in smokers but may be a mild clozapine inhibitor in non-smokers.¹⁵ In Swiss samples,¹⁰ the model indicated that valproate may be a mild clozapine inducer in some patients, even after controlling for smoking effects.

In 1990, in a cross-sectional study of 148 inpatients taking clozapine, Haring et al.¹⁶ found that trough steady-state plasma clozapine concentrations were influenced by weight after controlling for dosing, sex and smoking. Not all studies have studied or replicated the effects of weight on clozapine clearance, but a set of studies using a United Kingdom and Ireland TDM database with thousands of samples consistently found that increased weight is significantly associated with higher plasma clozapine concentrations.^17–19 Rostami-Hodjegan et al.¹⁷ described that total plasma clozapine concentrations increased 5% for every increase of 10 kg of body weight. The exactitude of this estimation may be contaminated by the weaknesses of that study including: 1) lack of clarity regarding how many of the clozapine concentrations were steady-state concentrations, or how many were trough concentrations, 2) lack of separation of the effects of different individuals and repeated samples from the same individuals, and 3) contamination by the effects of co-medications. Assuming that increased weight is associated with decreased clozapine clearance would imply that the effect of obesity is similar to the effect of an inhibitor of clozapine metabolism in our model, but the weight effects are expected to be rather modest per each Kg of weight increase. In a review of clozapine half-life studies, Fang et al.²⁰ explain that the results are compatible with clozapine having a third deep compartment, which means it deposits in fat tissue. Assuming this is correct, the fat tissue would work as a buffer for the serum concentrations, and an increase in fat tissue would be associated with lower clozapine clearance. A series of studies by Procyshyn et al.^21–24 are also compatible with the hypothesis that clozapine has affinity for fat tissue.

Thus, the goal of this study was to reanalyze our data from the US RCT using our statistical model to test whether increased weight gain from baseline was associated with increased total plasma clozapine concentration after controlling for the effects of smoking and sex, as well as for inter- and intra-individual variability. An extension of that goal was to replicate the analyses using estimations of changes in percentage of body fat computed with the Gomez-Ambrosi equation in place of weight changes.²⁵

Methods

The details of the US clozapine study are provided in Supplemental Box 1. The initial sample was 50 schizophrenia patients but only 47 provided enough TDM data. These 47 subjects had a mean (SD) age of 45 (10) years (range, 28–62 years). Forty-seven (22/47) percent were males, 85% (40/47) were Caucasian, 15% (7/47) African-American, and 83% (39/47) were smokers. Mean (±SD) baseline weights were 79.8 Kg (±16.3) for males and 68.6 Kg (±10.5) for females. The mean weight gains to the last available weight were 1.8 Kg (±6.3) for males and 3.0 Kg (±6.5) for females.

Percentage of body fat was computed by using the Gomez-Ambrosi’s equation.²⁵ Mean baseline percentages of body fat were 23.4% (±6.0) for males and 37.1% (±5.6) for females. Mean increases in percentage of body fat to the last available weight measure were 0.93% (±2.9) for males and 1.5% (±3.0) for females.

The 47 patients provided 470 measures of trough steady-state plasma clozapine concentrations (in ng/mL). After excluding observations for which no corresponding weight measure was available, a total of 424 concentrations were used in the current analyses. Each patient provided 1 to 18 measures of clozapine concentrations for these analyses. A random intercept linear model of steady-state total plasma clozapine concentrations was fitted to the 424 concentrations. The dependent variables were sex, smoking, weight change from baseline and the log of dosage (in mg/day) (Table 1). An additional model was fitted that included change in percentage of body fat from baseline in place of weight change as the dependent variable (Table 2). Residual analyses suggested that the models fitted well. The Stata software was used in statistical analyses (StatCorp LP, College Station, TX)

TABLE 1.

Random intercept linear model of log of total clozapine concentrations (ng/mL) for N=47 subjects who provided a total of 424 total clozapine concentrations, using weight change as independent variable.

Variable	Fixed regression coefficient Ba (95% CI)		P-value	Effect sizeb (95% CI)		Correction factorc (95% CI)
Male sexd	−0.730	(−1.01, −0.449)	<0.001	−51.8%	(−63.6, −36.2)	2.07	(1.57, 2.75)
Smokinge	−0.443	(−0.814, −0.0708)	0.02	−35.8%	(−55.7, −6.8)	1.56	(1.07, 2.26)
Weight change (Kg)f	0.0139	(0.00548, 0.0223)	0.001	1.4%	(0.55, 2.3)	0.986	(0.978, 0.995)
Log (dose)	1.20	(1.16, 1.24)	<0.001	--	--	--	--

CI: confidence interval.

^aFixed regression coefficients are represented with the letter B.

^bEffect size was computed as E=(exp(B)−1) × 100.

^cThe clozapine dose-correction factors were calculated with the formula (E/100 + 1)^{− 1}.

^dThe dichotomous variable was defined as 1 if the patient was male, 0 if female.

^eThe dichotomous variable was defined as 1 if the patient was a smoker, 0 otherwise.

^fWeight change from baseline.

TABLE 2.

Random intercept linear model of log of total clozapine concentrations (ng/mL) for N=47 subjects who provided a total of 424 total clozapine concentrations, using change in percentage of body fat as independent variable.

Variable	Fixed regression coefficient Ba (95% CI)		P-value	Effect sizeb (95% CI)		Correction factorc (95% CI)
Male sexd	−0.728	(−1.01, −0.445)	<0.001	−51.7%	(−63.6, −35.9)	2.07	(1.56, 2.74)
Smokinge	−0.444	(−0.818, −0.0705)	0.02	−35.9%	(−55.9, −6.8)	1.56	(1.07, 2.27)
Change in fat %f	0.0312	(0.0128, 0.0497)	0.001	3.2%	(1.3, 5.1)	0.969	(0.952, 0.987)
Log (dose)	1.20	(1.16, 1.24)	<0.001	--	--	--	--

^aFixed regression coefficients are represented with the letter B.

^bEffect size was computed as E=(exp(B)−1) × 100.

^cThe clozapine dose-correction factors were calculated by using the formula (E/100 + 1)^{− 1}.

^dThe dichotomous variable was defined as 1 if the patient was male, 0 if female.

^eThe dichotomous variable was defined as 1 if the patient was a smoker, 0 otherwise.

^fChange from baseline in percentage of body fat.

Results

Model with weight gain measured in Kg

After adjusting for sex and smoking, a 1-Kg gain in body weight during clozapine treatment was significantly associated with a 1.4% increase in total plasma clozapine concentrations [95% CI, (0.55, 2.3)] (Table 1).

Table 1 shows clozapine dose-correction factors. The correction factors are not recommendations for clinicians since most of them are too small to be relevant; they allow comparing the effects of investigated variables with those of inducers and inhibitors from prior studies. The correction factor for a 1-Kg increase in body weight was 0.986. To illustrate, if the clozapine therapeutic dose for a particular patient is 450 mg/day but the patient gains 1 Kg during clozapine treatment, then the patient’s dose must be reduced to 450*0.986=444 mg/day. The correction factor for body weight gains of 3 Kg is 0.959 (=0.986³) and in the same patient the dose should be reduced to 450*0.959=432 mg/day. The correction factor for bodyweight gains of 5 Kg is 0.932 (=0.986⁵); in the same patient the dose should be reduced to 450*0.932=419 mg/day. The correction factor for bodyweight gains of 10 Kg is 0.868 (=0.986¹⁰); in the same patient the dose should be reduced to 450*0.868=391 mg/day. In summary, the effects of increased weight on plasma clozapine levels were significant as we hypothesized, but small.

Model with weight gain measured in percentage of body fat

It was very interesting to observe that a 1-point increase in percentage of body fat during clozapine treatment was significantly associated with a 3.2% increase in total clozapine concentrations [95% CI, (1.3, 5.1)] (Table 2). In an additional analysis, we added to the model an interaction term between sex and change in percentage of body fat from baseline. The interaction was significant (p=0.046). Thus, the effect size of changes in percentage of body fat on total clozapine concentrations differed significantly between females and males. In females, an increase of one unit in percentage of body fat yielded a 5.4% increase in total clozapine concentrations [95% CI, (2.5, 8.3); correction factor, 0.949]. In contrast, in males the increase in total clozapine concentrations associated with a one-unit increase in percentage of body fat was significantly lower, just 1.4% [95% CI, (−1.1, 4.0); correction factor 0.986].

Table 3 shows clozapine dose-correction factors and doses after an initial therapeutic dose of 450 mg/day by particular gains in body fat percentage. Other things being equal, females require significantly larger reductions in clozapine dose to compensate for augmentations in body fat tissue. In a patient taking 450 mg/day (Table 3), increases of 5% of body fat were associated with clinically relevant changes in total plasma clozapine concentrations, which would recommend decreasing the dosage to 419 mg/day if the patient is a male, or to 346 mg/day if female. Most of our data suggest that some of our correction factors are probably not relevant for clinical purposes. However, it is possible that these correction factors may be relevant in extreme cases. This is why Table 3 provides correction factors after increases of 10% in body fat. These large increases may happen in the real world in clinical environments not as well controlled as our research inpatient unit.

TABLE 3.

Clozapine dose-correction factors and doses after increases in body fat percentage for a hypothetical patient under an initial therapeutic dose of 450 mg/day.

	Male		Female
Body fat percentage increase	Correction factora	Dose (mg/day) b	Correction factorc	Dose (mg/day)b
0%	--	450	--	450
1%	0.986	444	0.949	427
3%	0.959	431	0.855	385
5%	0.932	419	0.770	346
10%	0.868	391	0.592	267

^aThe dose-correction factor for males after an increase of n units in body fat percentage was computed as (0.986)ⁿ.

^bDose was computed as 450 multiplied by correction factor.

^cThe dose-correction factor for females after an increase of n units in body fat percentage was computed as (0.949)ⁿ.

Discussion

As hypothesized, weight changes during clozapine treatment were associated with decreased clozapine clearance. The sophisticated statistical model, the relatively large number of repeated measures and the controlled experimental design helped us to detect relatively small effects of weight changes that may not be detected in less controlled studies. In the Swiss TDM study, which included mainly cross-sectional data with few repeated measures, baseline weight was not significant in the clozapine model.¹⁰ In the Italian sample, weights were not available.¹⁵

Increasing total clozapine concentrations is a sign of decreasing clozapine clearance and equivalent to the action of an inhibitor of clozapine metabolism. In our Italian sample,¹⁵ fluvoxamine increased total clozapine concentrations by an average of 263%, which is very important from the clinical point of view. On the other hand, the average increases by fluoxetine and paroxetine were relatively small, 42% and 32% respectively, and not likely to be clinically relevant in average patients. The effect size of weight changes on total clozapine concentrations that we have found in this re-analysis suggests that it may be relevant on rare occasions. Due to the small effect size of weight on clozapine clearance, cross-sectional TDM studies^17–19 using baseline weights or weight at the time of TDM may need hundreds or even thousands of samples to detect significant weight effects.

The significant effect of weight gain on plasma clozapine concentrations is a first hint that increased fat during clozapine treatment may contribute to decreases in clozapine clearance. It is interesting that an estimation of body fat using a previously-developed mathematical formula²⁵ allowed detecting such strong effects in our female sample. Although we do not have any external validation that the estimations of body fat are correct, the results appear to have very high face validity. The clozapine deposit in fat tissue may explain another clinically relevant finding. Although it has not been well-studied, case reports have suggested that after death serum clozapine concentrations increase due to postmortem distribution.^26,27 It is reasonable to propose that after death some of the clozapine deposited in higher concentrations in fat tissue may be slowly released from a deep peripheral compartment, and distributed in the blood where lower concentrations were present during life.

The US study is the best-controlled clozapine TDM sample that we have analyzed, but this study was not free of limitations (Supplementary Box 1). Beyond its limitations, the great number of repeated measures helped us to properly separate sex and smoking effects, which are frequently confounded since smoking tends to be associated with male sex in most schizophrenia samples.²⁸ The attempt to use a mathematical formula to estimate percentage of body fat and introduce the estimates in a complex mathematical model may appear to be a limitation of our study, but the results appear compelling and are consistent with other clozapine research that indicates that clozapine has affinity for lipids.^21–23 Our statistical model appears to have worked out surprisingly well in showing that it is likely that clozapine clearance may decrease with increased body fat. In summary, pending further replication in other samples, it appears to be likely that clozapine may deposit in body fat which may decrease clozapine clearance. We have the impression that in most patients the decreased clearance associated with weight gain may be of interest to pharmacologists but of little clinical relevance for clinicians. However, it is possible that, in females who had major increases in body fat, clozapine dosages may need to be cut after several months, due to the decrease in clozapine clearance. As individuals with extremely heavy weight are becoming more and more frequent in Western societies,²⁹ it is possible that the effects of weight gain on clozapine clearance may not be small in some extreme subjects with particularly large weight increases.

Our study was conducted in an extremely well-controlled inpatient environment (Supplemental Box 1). This may have contributed to a low variability of plasma clozapine concentrations. For the total clozapine concentrations, the mean coefficient of variation (CV) was 20% for the 100 mg/day dose, 16% for the 300 mg/day dose and 15% for the 600 mg/day dose.⁷ Studies of outpatients have reported much higher CV values including >50%³⁰ and >30%,³¹ and a third study³² proposed that high clozapine CVs may be a sign of non-compliance. Therefore, it is possible that the small effect sizes of weight and fat percentage that we have detected may not be detectable in less controlled studies using naturalistic TDM samples with limited repeated samples. The studies of the effects of weight on TDM from other second-generation antipsychotics^33–35 or antidepressants³⁶ have been mainly negative, but we cannot rule out that they were negative because they could not detect small effects due to their limitations, including the use of outpatients and/or lack of multiple repeated samples in the same individual.

It will be important to explore the association between body fat and clozapine clearance in other races, particularly in East Asians. In China, clozapine is used in 25–60% of schizophrenia patients³⁷ and is also used in treatment-resistant bipolar disorder or even bipolar disorder.³⁸ Therefore, Chinese patients are definitively the largest group of patients taking clozapine in the world. On the other hand, Chinese,⁴⁰ Japanese⁴¹ and Koreans⁴² appear to have different distributions of body fat than Caucasians. Moreover, East Asians may metabolize clozapine more poorly than Caucasians.¹⁴

To conclude, we want to stress the potential of complex mathematical models combining genetic, environmental and personal factors in the context of TDM to better personalize dosing. Despite its complex name (statistical random-effects linear model) and its complex mathematics, our clozapine model has already helped us to propose two new pharmacological hypotheses: the influence of valproate co-treatment on clozapine metabolism in the Italian¹⁵ and Swiss studies,¹⁰ and of percentage of body fat in this US study.