Comparability of whole-blood and plasma clozapine and norclozapine concentrations

Measurement of plasma clozapine and N-desmethylclozapine (norclozapine) concentrations can help to assess patient compliance, optimize dosage, and minimize the risk of dose-related toxicity [1, 2]. The UK Clozaril Patient Monitoring Service (CPMS; Novartis) support a plasma clozapine and norclozapine assay service via the Medical Toxicology Unit (MTU). Normally EDTA blood (2–5 ml) is sent by first-class post and plasma is separated on arrival at the MTU. However, whole (i.e. unseparated) partially or completely haemolysed blood may be supplied for analysis: (i) if a patient has died and blood has been obtained postmortem, or residues of stored whole-blood samples obtained before death are to be analysed; and (ii) if blood has been frozen or otherwise allowed to haemolyse before plasma has been separated. In addition, plasma clozapine and norclozapine data from a given patient are sometimes used to help in the interpretation of postmortem whole-blood results. Hence it is important to be clear as to the comparability of whole-blood and plasma concentrations.

Anticoagulated whole blood from a healthy volunteer studied in vitro gave plasma : whole-blood clozapine and norclozapine ratios that were similar across the range 0.1–1.5 mg l⁻¹. The slopes of the regression lines suggested that whole-blood clozapine will be some 10% lower than plasma clozapine concentrations, but that whole-blood norclozapine concentrations will be some 30% higher [3]. This may be a consequence of the higher plasma protein-binding of clozapine (5.5% unbound) compared with that of norclozapine (9.7% unbound) [4], making more norclozapine available for partition into cells. Guitton et al.[5] have reported mean washed red blood cell (RBC) : plasma ratios in blood from 10 patients (clozapine dose 400–800 mg day⁻¹) of 0.29 ± 0.14 for clozapine and 1.63 ± 0.45 for norclozapine (confidence interval not stated), again suggesting relative accumulation of norclozapine in red cells. In the pig, norclozapine accumulates in myocardium, liver, striated muscle, and kidney to concentrations some 2–3 fold higher than those of clozapine [3].

In order to extend the plasma : whole-blood distribution work [3], paired whole-blood and plasma samples from clozapine-treated patients were analysed. The protocol reflected the situation that occurs when residues of stored whole-blood samples obtained for routine haematology are analysed retrospectively to obtain information on antemortem blood clozapine and norclozapine. EDTA blood (5–10 ml) was collected from patients prescribed clozapine. Approximately equal portions of the sample were sent: (i) by first-class post to the MTU where plasma was separated on arrival and was assayed as if it was a routine clinical sample [6]; and (ii) by courier (at ambient temperature) to the CPMS for routine haematology, including haematocrit measurement. The residue of the portion of the sample sent for haematology was stored at 4 °C prior to transport in batches of 5–10 samples (at ambient temperature) to the MTU for whole-blood clozapine and norclozapine assay. Whole-blood samples were frozen on arrival at the MTU and roller-mixed (10 min) after thawing. Demographic details and other basic parameters including clozapine dose were recorded. The West London Mental Health Trust Ethics Committee approved the study. Blood was collected during routine haematological monitoring and each patient gave informed consent to the donation of an extra 2–5 ml of blood per venepuncture.

Plasma and whole-blood clozapine and norclozapine concentrations were measured in duplicate by HPLC-UV (240 nm) [7]. Assay calibration was by duplicate analysis of solutions containing clozapine and norclozapine (0.10, 0.20, 0.50, 0.80, 1.00, and 1.50 mg l⁻¹) in analyte-free calf serum. Intra-assay precision (accuracy) [RSD percentage (%)](nominal analyte concentrations 0.15, 0.40, and 1.20 mg l⁻¹ in analyte-free human serum, n = 6 each concentration) was: clozapine 6.6 (100), 5.7 (103), 6.5 (96); norclozapine 3.8 (100), 5.8 (100), 3.3 (98), respectively. Corresponding inter-assay precision (n = 63 over 3 months) was: clozapine 15.2 (93), 10.7 (98), 10.6 (96); norclozapine 12.2 (100), 7.1 (103), and 5.6 (99), respectively.

In order to investigate the possibility of matrix effects upon analyte recovery, analyte-free human whole blood [600 ml + 10 ml 0.9% (w/v) sodium chloride containing 2000 U heparin, haematocrit 0.50] was obtained from a volunteer donor under medical supervision. Standard clozapine and norclozapine solutions (0.10, 0.20, 0.50, 0.80, 1.00, and 1.50 mg l⁻¹) were prepared in 20 ml portions of this blood. After thorough roller-mixing (15 min), the solutions were allowed to stand (16 h, 4 °C). Subsequently, and after thorough mixing, portions of these solutions (2 ml) were stored at −20 °C (4 h) before clozapine and norclozapine assay. Similar results were obtained upon analysis of the haemolysed human whole-blood and the calf serum calibrators. The plots of nominal analyte concentration vs peak height ratio to the internal standard gave (serum) clozapine y = 1.35x −0.02, r² = 0.9998; norclozapine y = 0.86x − 0.00, r² = 0.9998 and (blood) clozapine y = 1.40x − 0.10, r² = 0.9991; norclozapine y = 0.83x − 0.01, r² = 0.9993.

Paired plasma and whole-blood samples were obtained from 48 patients. There were 34 males, 14 females; median age 38 (range 24–61) years; 25 Caucasians, 14 Afro-Caribbeans, 7 Asians, 2 mixed race; 16 nonsmokers, 32 smokers; 1 patient used an illicit drug (cannabis) and 1 patient recorded ingestion of more than 5 units of alcohol weekly. The median clozapine dose was 400 (range 62.5–1025) mg day⁻¹. Concomitant medication was as follows: antidepressants (4); mood stabilizers (13); anticholinergics (8); combination therapy with these and/or other drugs (5). Plots of whole-blood vs plasma clozapine and norclozapine are shown in Figure 1. With both analytes there were three individual results that could be outliers, identified in different specimens for each analyte. Regression analysis of the data excluding outliers showed good agreement with the results obtained in the in vitro human blood study (clozapine: y = 1.13x + 0.01; r² = 0.998, n = 7, norclozapine y = 0.68x + 0.03; r² = 0.997, n = 8) [3]. Using the equation:

$$\begin{array}{l}\{\text{whole}-\text{blood concentration}-[\text{plasma concentration}\\ \phantom{\text{ha}}\times (1-\text{haematocrit})]\}÷\text{haematocrit}\end{array}$$

to calculate RBC, our data (outliers excluded) gave comparable median (10th, 90th percentile) RBC : plasma ratios [clozapine 0.31 (−0.33, 0.77), norclozapine 1.30 (0.48, 2.36); n = 45 in each case] to those reported by Guitton et al.[5].

Figure 1.

Plots of whole-blood and plasma clozapine and norclozapine concentrations in patient samples (n = 48; data points omitted in the calculation of the regression equations).

Reanalysis of whole-blood samples (n = 18) after storage for 8–12 weeks (4 °C) showed no difference in mean ± SD clozapine concentration (0.38 ± 0.28 vs 0.35 ± 0.29 mg l⁻¹; t = 1.53, NS), but a significant decrease in mean norclozapine (0.42 ± 0.36 vs 0.25 ± 0.25 mg l⁻¹; t = 4.36, P < 0.001). This unexpected finding could be attributable to reaction of norclozapine with red cell protein. Although formal stability studies have not been performed, both clozapine and norclozapine appear stable in plasma or serum at 4 °C for several weeks. Loss of norclozapine from the whole-blood samples on storage is a possible explanation for the outliers observed in the whole-blood vs plasma concentration plot (Figure 1). However, there is no obvious explanation for the outliers in the clozapine plot (details of the patients are given in Table 1).

Table 1.

Details of three patients with apparently outlying whole-blood : plasma clozapine ratios (no patient consumed alcohol or used illicit drugs).

Age (years)	Sex	Ethnicity	Smoker	Other drugs	Clozapine dose (mg day−1)	Plasma clozapine (mg l−1)	Plasma nor- clozapine (mg l−1)	Blood clozapine (mg l−1)	Blood nor- clozapine (mg l−1)	Haematocrit
36	F	Caucasian	No	Fluoxetine	275	1.56	0.41	0.60	0.43	0.42
35	F	Caucasian	Yes	Paroxetine, sodium valproate	650	1.85	0.54	0.72	0.77	0.43
41	M	Afro-Caribbean	Yes	Nil	1025	1.73	1.20	0.95	1.36	0.44

As with many basic and other drugs, clozapine may bind principally to α₁-acid glycoprotein (AAG) [8, 9]. In humans the plasma AAG concentration is normally much lower than the albumin concentration. Mean serum AAG concentration in adults is 0.78 g l⁻¹ (95% range, log-normal distribution 0.48, 1.26 g l⁻¹) [10], but its plasma concentration can increase rapidly in response to systemic tissue injury, inflammation, or infection. Increased plasma AAG concentrations provide a possible explanation for the findings of abnormally high total (free and protein-bound) plasma clozapine concentrations in some patients in the absence of clinical features of toxicity [3, 11].

It is normally assumed that plasma and serum concentrations of relatively highly protein-bound analytes such as clozapine will be comparable, because the concentrations of binding proteins are thought to be similar in these fluids. On the other hand, it has been reported that mean serum clozapine and norclozapine concentrations are some 10 and 16%, respectively, lower than in plasma [12]. However, it was not clear how the assay used was calibrated or what quality control procedures were in place. The method cited [13] used either plasma or serum to prepare calibration standards and the basic drug loxapine was used as the internal standard (Kaladjian et al.[12] used carbamazepine as internal standard). Secondly, clozapine N-oxide can undergo base-catalysed reduction to clozapine [14], and thus use of sodium hydroxide to adjust the extraction pH may have lead to overestimation of clozapine concentrations in some samples, although this would not have affected norclozapine concentrations. Perhaps more importantly, the method was based on solid-phase extraction and it is known that different matrices are likely to affect analyte recovery to differing extents using this method.

In conclusion, the results suggest that whole-blood clozapine concentrations in patient samples will be some 10% lower than plasma clozapine concentrations, but that whole-blood norclozapine concentrations will be some 30% higher. However, individual samples may vary from this norm depending on haematocrit and other possible factors such as the concentration of binding proteins in plasma. As it will not be possible to measure haematocrit in some of the circumstances in which whole-blood samples have been analysed, the comparison between whole blood and plasma presented here can be used for guidance. In any event, the differences observed between plasma and whole blood are in general relatively minor when compared with the large differences in clozapine and norclozapine concentrations that may be observed, depending on the site in the body from which blood is obtained postmortem, the time elapsed between death and sample collection, and other possible factors [3]. However, our results suggest that whole-blood samples stored at 4 °C should be analysed as soon as feasible after collection to minimize the risk of loss of norclozapine. Although the contribution of norclozapine to the efficacy and toxicity of clozapine in vivo is unknown, assessment of the blood or plasma clozapine : norclozapine ratio in individual patients can sometimes provide a valuable guide as to whether an acute overdose has been ingested [3].