• Open Access

Pharmacokinetics and pharmacodynamics of protein-unbound docetaxel in cancer patients

Authors


To whom correspondence should be addressed. E-mail: hminami@east.ncc.go.jp

Abstract

α1-Acid glycoprotein, a plasma protein that binds docetaxel, is a significant determinant of the clearance and activity of docetaxel, but its serum levels in cancer patients are variable. This emphasizes the importance of investigating the pharmacokinetics of unbound drug rather than total drug in the plasma. In the present study, the pharmacokinetics and pharmacodynamics of unbound docetaxel were investigated in cancer patients. Docetaxel was infused over a 1-h period in 69 patients. The concentration of unbound docetaxel was measured in the plasma ultrafiltrate at the end of infusion and the unbound fraction (fu) was calculated. The pharmacokinetics of total docetaxel in the plasma was investigated. The area under the concentration–time curve (AUC) of unbound docetaxel was calculated by multiplying fu by the AUC of total docetaxel. The peak concentration at the end of infusion (Cmax) and AUC of total and unbound drug were compared between patients who did or did not experience grade 4 neutropenia. The median of fu was 4.0%, ranging from 1.2 to 22.6% (5–95% percentile; 1.4–10.5%). Grade 4 neutropenia was observed in 24 patients. Although Cmax and AUC of total drug were not different in patients with or without grade 4 neutropenia, patients who experienced grade 4 neutropenia had significantly greater Cmax (92.3 vs 63.3 ng/mL, P = 0.01) and AUC (0.137 vs 0.104 µg × h/mL, P = 0.05) of unbound docetaxel. In a logistic regression analysis, the unbound Cmax and α1-acid glycoprotein were determinants of grade 4 neutropenia. Pharmacokinetics of unbound drug rather than total drug is a better predictor of neutropenia for docetaxel. (Cancer Sci 2006; 97: 235–241)

Drugs are distributed to various tissues of the body via blood flow, and must bind to their receptors to show pharmacological activity. Many drugs in plasma bind to proteins, but theoretically only unbound drugs can be distributed to tissues and into cells. In this sense, unbound drug concentrations in plasma are pharmacologically important when pharmacokinetics is investigated.(1,2) Protein binding also has a great influence on drug distribution and elimination.(3–5) For drugs that are bound highly to plasma proteins, small variations in the fraction of unbound drug in plasma (fu) cause great changes in concentrations of unbound drugs, and therefore profound variations in pharmacological activities. Therefore, it is theoretically important to investigate interpatient variations in the fu of drugs. However, concentrations of unbound drug have seldom been measured when pharmacokinetics of anticancer agents are investigated in patients.(2)

Docetaxel is an anticancer agent that is active against many cancers including breast, non-small cell lung, ovarian, head and neck, esophageal, gastric and prostate cancers.(6–18) A previous in vitro study showed extensive binding of docetaxel to plasma proteins, including α1-acid glycoprotein, albumin and lipoproteins, with high binding affinity for α1-acid glycoprotein.(19) Moreover, there was a large interindividual variability of α1-acid glycoprotein levels in cancer patients, suggesting that this moiety may be the main determinant of variability in docetaxel binding in plasma.(20–22) In a small clinical study, fu was reported to range from 4 to 10% in the plasma of patients treated with docetaxel, and α1-acid glycoprotein was found to be correlated with fu and clearance of total docetaxel.(21) In a recent study, the wide variation of fu was confirmed in 55 patients ranging from 1.2 to 8.6%, and a stronger correlation between the area under the concentration–time curve (AUC) and neutropenia was observed for unbound rather than total docetaxel.(23)

In the clinical development of docetaxel, a population pharmacokinetic investigation was prospectively implemented, and α1-acid glycoprotein was found to be a significant covariate of docetaxel clearance.(24) Binding of docetaxel to α1-acid glycoprotein might contribute to the correlation between α1-acid glycoprotein and clearance of total docetaxel by lowering concentrations of unbound docetaxel available for metabolism. In contrast, α1-acid glycoprotein is an acute-phase protein, and it was suggested that cancer patients with elevated acute-phase proteins, including α1-acid glycoprotein and C-reactive protein, have reduced activity of CYP3A4, an enzyme that metabolizes docetaxel.(25,26) This might also explain the correlation between α1-acid glycoprotein and drug clearance. Furthermore, in patients with non-small cell lung cancer treated with docetaxel, higher α1-acid glycoprotein levels were associated with a lower risk of grade 4 neutropenia and febrile neutropenia, and with a shorter time to disease progression.(20) These findings might be explained by high levels of α1-acid glycoprotein causing decreasing concentrations of unbound docetaxel by binding the drug in the plasma and thereby reducing pharmacological activity. Although α1-acid glycoprotein levels were negatively correlated with the fu of docetaxel in plasma,(21,23) the extent of the correlation was not so high that α1-acid glycoprotein could be used instead of measuring the fu of docetaxel. Therefore, it is important to assay concentrations of unbound docetaxel when investigating its pharmacokinetics and pharmacodynamics. In order to evaluate the clinical relevance of unbound drug levels in plasma during chemotherapy with docetaxel, here we investigated the pharmacokinetics and pharmacodynamics of unbound drug compared to total drug in cancer patients treated with docetaxel as a single agent.

Patients and Methods

Patient selection

Sixty-nine patients were enrolled into this pharmacological study of docetaxel. Eligibility criteria included histologically or cytologically confirmed solid cancers against which docetaxel is active, age ≥20 years, Eastern Cooperative Oncology Group performance status 0–3, at least 3 weeks since the last chemotherapy (6 weeks for mitomycin and nitrosoureas), and adequate hematological values (white blood cells ≥3000/µL, platelet count ≥75 000/µL). Exclusion criteria were active infection, severe heart disease, uncontrolled hypertension or diabetes mellitus, pregnant/nursing women, or seropositive for human immunodeficiency virus, hepatitis C virus, hepatitis B surface antigen or syphilis.

The study protocol did not exclude patients taking drugs known to bind α1-acid glycoprotein. It was approved by the Institutional Review Board of the National Cancer Center, Japan, and all patients gave written informed consent.

Treatment and follow up

Docetaxel was infused intravenously over 1 h every 3 weeks. Most patients received the approved dose in Japan of 60 mg/m2, but attending physicians were allowed to reduce the dose depending on liver function, performance status or the extent of prior chemotherapy.

Physical examination and toxicity assessment included complete blood cell counts with differential counts as well as platelet counts, and blood chemistry. These were performed before treatment and repeated at least weekly during the first course. Data on toxicity during the first course were used for pharmacodynamic analysis in the present study.

Pharmacokinetic analysis

Blood sampling for pharmacokinetic analysis was carried out before and 30 min into an infusion, at the end of the docetaxel infusion, and 0.17, 1, 5, 10 and 24 h after the end of infusion. Heparinized blood was centrifuged immediately, and an aliquot of plasma at the end of infusion was ultrafiltered using UFC3GC membranes (Japan Millipore, Tokyo, Japan) for the measurement of protein-unbound docetaxel concentrations. Plasma and ultrafiltrate samples were frozen at −80°C until analysis.

The concentration of docetaxel in plasma (total docetaxel) and ultrafiltrate (unbound docetaxel) was determined by using a previously reported high-performance liquid chromatography method.(27) Because of concerns about the sensitivity of the assay system, we measured concentrations of unbound docetaxel only at the end of infusion. Pharmacokinetic parameters for individuals were calculated by Bayesian estimation after population pharmacokinetic parameters were estimated in the entire population. These calculations were carried out using the NONMEM program (version V, level 1.1; Globo Max, Ellicot City, MD, USA). A three-compartment open model with zero-order administration and first-order elimination (ADVAN 7 and TRANS 1) was used to describe the plasma concentration–time course for docetaxel in the entire population. Assuming a log-normal distribution for interindividual variability in pharmacokinetic parameters, the interindividual variability was modeled as (e.g. for clearance):

CLj = ĈL exp(ηjCL),

where CLj and ĈL are the estimated values in an individual j and the population mean for clearance, respectively, and ηjCL is the individual random perturbation from the population mean. Intrapatient residual variability was also described by a log-normal distribution model. Similarly, interindividual variability was modeled for intercompartment rate constants between the central compartment and the second peripheral compartment (k13 and k31). The AUC was calculated as dose/clearance in each patient, and the total volume of distribution at steady state (Vss) was computed as:

Vss = V1(1 + k12/k21 + k13/k31),

where V1 was distribution volume of the central compartment, and k12 and k21 were intercompartment rate constants between the central compartment and the first peripheral compartment.

Concentrations of protein-unbound docetaxel were measured in ultrafiltrates of plasma at the end of infusion, and fu was calculated as unbound concentrations divided by total concentrations in plasma at the end of infusion in each patient. The AUC of unbound docetaxel was calculated by multiplying fu and the AUC of total docetaxel, assuming that fu at the end of infusion was not changed after the infusion.

Pharmacodynamic analysis

In the pharmacodynamic analysis, patient characteristics as well as pharmacokinetic parameters were compared between patients who did or did not experience grade 4 neutropenia (<500/µL). Patient characteristics evaluated for possible association with grade 4 neutropenia were age, sex, performance status (0–1 vs 2–3), type of cancer (breast cancer vs other cancers), the number of prior chemotherapy regimens (0–1 vs 2 or more), albumin, total bilirubin, alanine aminotransferase (ALT), and α1-acid glycoprotein levels in serum. Pharmacokinetic parameters, including dose per body surface area, the peak concentration at the end of infusion (Cmax) and AUC of total and unbound docetaxel, were also compared between patients with or without grade 4 neutropenia.

Statistical analysis

To identify factors associated with grade 4 neutropenia, continuous variables were compared between patients with or without grade 4 neutropenia using the Mann–Whitney U-test, and differences in the distribution of dichotomized variables were evaluated with the χ2-test or Fisher's exact test, where appropriate. All reported P-values were two-tailed. To confirm variables significantly associated with grade 4 neutropenia, multivariate logistic regression analyses were carried out.

Results

The characteristics of the 69 patients treated in the present study are listed in Table 1. Fifty-two (75%) had breast cancer, so that female patients predominated. Seventy-five percent of patients had good performance status of 0 or 1 and 70% were minimally treated before entering this study, with the number of prior treatment regimens being 0 or 1. Although one patient with nephrotic syndrome had a low level of serum albumin (1.3 g/dL), other patients had serum albumin levels of 2.6 g/dL or greater. Fifteen patients had ALT levels exceeding the upper limit of the normal range. However, most patients had normal serum bilirubin levels and only three patients had minimally elevated serum total bilirubin levels, ranging from 1.6 to 1.9 mg/dL. As with other reports on cancer patients, serum levels of α1-acid glycoprotein varied greatly between individuals.

Table 1. Patient characteristics
CharacteristicValue
  1. ALT, alanine aminotransferase; ULN, upper limit of normal.

Age (years)
 Median 56
 Range 21–73
Sex (n)
 Female 57
 Male 12
Performance status (n)
 0 12
 1 40
 2 12
 3  5
Cancers (n)
 Breast cancer 52
 Non-small cell lung cancer  8
 Head and neck cancer  5
 Others  4
Regimens of prior chemotherapy (n)
 0 13
 1 35
 2 12
 ≥3  9
Albumin (g/dL)
 Median  3.5
 Range  1.3–4.5
ALT (IU/L)
 Normal 54
 ≤1.5 × ULN  5
 >1.5–≤2.5 × ULN  5
 >2.5 × ULN  5
Total bilirubin (mg/dL)
 Median  0.6
 Range  0.2–1.9
α1-Acid glycoprotein (mg/dL)
 Median111
 Range 51–259

Table 2 summarizes doses and pharmacokinetic parameters of docetaxel. Most patients received 60 mg/m2 of docetaxel, but some received reduced doses because of poor performance status, liver dysfunction or extensive prior treatments. A large interpatient variability of fu was observed, ranging from 1.2 to 22.6% (Fig. 1). The 5th, 25th, 50th, 75th and 95th percentiles of fu were 1.4, 3.1, 4.0, 6.1 and 10.5%, respectively. When we compared other drugs that were administered to patients with a fu less than the 25th percentile or greater than the 75th percentile, no differences were observed. One patient had a fu of 22.6% and seemed to be an outlier with regard to protein binding. This patient had elevated aspartate aminotransferase (AST) (310 IU/L) and ALT (140 IU/L) values due to extensive liver metastasis from breast cancer, but the total bilirubin level was normal. Although the dose of docetaxel was reduced to 30 mg/m2 because of liver dysfunction, grade 4 neutropenia was observed in this patient. The AUC of total docetaxel in this patient was higher than the 90th percentile of all patients, and the calculated AUC of unbound docetaxel was the highest among all patients.

Table 2. Pharmacokinetic parameters and neutropenia
ParameterValue
  1. AUC, area under the concentration–time curve; Cmax, peak concentration at the end of infusion.

Dose (n)
 60 mg/m2  47
 50 mg/m2   8
 ≥40–<50 mg/m2   9
 ≥20–<40 mg/m2   5
Dose (mg)
 Median  84
 Range  29–114
Clearance (L/h)
 Median  29.4
 Range   6.9–47.6
Volume of distribution at steady state (L)
 Median 250.0
 Range 115.1–821.6
Unbound fraction (%)
 Median   4.0
 Range   1.2–22.6
AUC (µg × h/mL)
 Median   2.68
 Range   1.35–12.20
Unbound AUC (µg × h/mL)
 Median   0.113
 Range   0.033–1.12
Cmax (ng/mL)
 Median1588
 Range 576–3535
Unbound Cmax (ng/mL)
 Median  70.9
 Range  16.9–349.8
Pretreatment neutrophil counts (per µL)
 Median3770
 Range1890–13 130
Neutrophil counts at nadir (per µL)
 Median 720
 Range  40–4070
Change in neutrophil counts (%)
 Median  80.1
 Range  18.3–98.8
Nadir neutrophil counts <500/µL (n)
 Yes  24
 No  45
Figure 1.

Distribution of the unbound fraction of docetaxel.

The primary toxicity of chemotherapy with docetaxel was neutropenia, and 24 patients (35%) experienced grade 4 neutropenia in this study (Table 2). When the characteristics of patients who did or did not develop grade 4 neutropenia were compared, the distribution of sex and cancer types were significantly different, and pretreatment serum levels of α1-acid glycoprotein in patients with grade 4 neutropenia were significantly lower than in those without (Table 3). Among pharmacokinetic parameters, the AUC and Cmax of unbound docetaxel were significantly higher in patients with grade 4 neutropenia, whereas differences in AUC and Cmax of total docetaxel were not significantly different. The fu of docetaxel in patients with grade 4 neutropenia was also higher than in those without.

Table 3. Characteristics and pharmacokinetic parameters of docetaxel in patients with or without neutropenia (neutrophils <500/µL)
NeutropeniaNeutrophils <500/µLNeutrophils ≥500/µLP
  1. AUC, area under the concentration–time curve; ALT, alanine aminotransferase; Cmax, peak concentration at the end of infusion.

Patients (n)  24  45 
Sex (n)
 Female  24  33 0.006
 Male   0  12 
Performance status (n)
 0–1  18  34 0.96
 2–3   6  11 
Cancer type (n)
 Breast cancer  24  30<0.01
 Other cancers   0  15 
Prior chemotherapy regimens (n)
 0–1  15  33 0.35
 ≥2   9  12 
Age (years)
 Median  56  57 0.90
 Range  32–71  21–73 
Albumin (g/dL)
 Median   3.5   3.6 0.72
 Range   2.6–4.5   1.3–4.5 
Total bilirubin (mg/dL)
 Median   0.7   0.5 0.29
 Range   0.2–1.8   0.2–1.9 
ALT (IU/L)
 Median  23  23 0.82
 Range   9–140   7–235 
α1-Acid glycoprotein (mg/dL)
 Median  87 123 0.02
 Range  51–259  61–241 
Pretreatment neutrophil counts (per µL)
 Median32304050 0.02
 Range1890–131301920–11320 
Dose per body surface area (mg/m2)
 Median  60  60 0.64
 Range  30–60  20–60 
Total AUC (µg × h/mL)
 Median   2.73   2.49 0.55
 Range   1.49–5.99   1.35–12.17 
Unbound AUC (µg × h/mL)
 Median   0.137   0.104 0.05
 Range   0.0415–1.12   0.0332–0.868 
Total Cmax (ng/mL)
 Median16251588 0.29
 Range1197–2892 579–3535 
Unbound Cmax (ng/mL)
 Median  92.3  63.3 0.01
 Range   2.59–349.8  16.9–317.3 

To confirm that these variables were significantly associated with grade 4 neutropenia, multivariate logistic regression analyses were carried out. The Cmax of unbound docetaxel and α1-acid glycoprotein were selected among the variables listed in Table 3, whereas neither the AUC nor Cmax of total docetaxel was selected (Table 4). The probability of grade 4 neutropenia was predicted by these two variables using the logistic regression model (Fig. 2).

Table 4. Logistic regression model for grade 4 neutropenia
VariableRegression coefficient (SE)P
  1. Cmax, peak concentration at the end of infusion.

Cmax of unbound docetaxel  9.1 (4.9) × 10−30.06
α1-Acid glycoprotein−14.8 (7.3) × 10−30.04
Figure 2.

Probability of grade 4 neutropenia predicted by a logistic regression model with the concentration of unbound docetaxel at the end of infusion and the concentration of α1-acid glycoprotein.

When relationships between serum concentrations of α1-acid glycoprotein and pharmacokinetic and hematological variables were investigated, α1-acid glycoprotein was found to be negatively correlated with the clearance of total docetaxel (r2 = 0.08, P = 0.003) and positively correlated with neutrophil counts before treatment (r2 = 0.21, P = 0.0002).

Discussion

In this pharmacodynamic analysis of chemotherapy with docetaxel as a single agent, it was found that greater Cmax and AUC of unbound, but not total, docetaxel in plasma were associated with greater risk of grade 4 neutropenia. These findings support the concept that primarily unbound drugs are pharmacologically active because they can distribute into tissues or cells to bind their receptors. When interpatient variability of the unbound fraction is minimal, patients with higher concentrations of total drug have higher concentrations of unbound drug than patients with lower concentrations of total drug. In this sense, pharmacokinetic parameters of total drug can be used as surrogate measures for pharmacokinetic parameters of protein-unbound drug in pharmacodynamic analysis, provided that interpatient variability of the unbound fraction is minimal. However, when interpatient variability of the unbound fraction is large, pharmacokinetic parameters of total drug are no longer surrogates for those of unbound drug, and investigation of pharmacokinetics of unbound drug becomes important.(2) In the present study, the unbound fraction of docetaxel varied widely among patients, ranging from 1.2 to 22.6%. The 21-fold interpatient difference in the Cmax of unbound drug was clearly greater than the six-fold difference in Cmax of total drug. The association of neutropenia with concentrations of unbound drug, but not with total drug, and greater interpatient variability of exposure to unbound drug than total drug suggest that pharmacokinetic investigation of unbound drug is clinically relevant for docetaxel.

The importance of unbound drug concentrations in pharmacodynamic analysis was confirmed by the multivariate logistic regression model in which the Cmax of unbound docetaxel and α1-acid glycoprotein were selected as covariates (Table 4). In previous reports on pharmacokinetic studies of docetaxel, total drug concentrations were measured and AUC was correlated with neutropenia.(20,22,28–30) However, in contrast to the previous reports, the Cmax of unbound docetaxel was selected in the multivariate analysis for predicting grade 4 neutropenia in the present study, but the AUC of unbound docetaxel was not selected probably because it was correlated to the Cmax of unbound docetaxel (r2 = 0.23, P < 0.0001). In the present study, the AUC of unbound docetaxel was calculated by multiplying the AUC of total docetaxel and the unbound fraction at the end of infusion, assuming that the unbound fraction was not changed after the infusion. If this assumption was erroneous, it is possible that we failed to detect relationships between the AUC of unbound docetaxel and neutropenia.

In a small study of pharmacokinetics of docetaxel in combination chemotherapy with methotrexate, Loos et al. investigated the time course of unbound fractions of docetaxel in 23 patients using equilibrium dialysis, and they suggested that polysorbate 80, a surfactant that is contained in the clinical formulation of docetaxel, might increase the unbound fraction of the drug by 13%.(21) The unbound fraction was slightly increased from 5.49% in pretreatment samples to 6.37% at the end of infusion when plasma concentration of polysorbate 80 was high, and decreased to approximately 5% 5 h afterwards. This was confirmed by another study in which the unbound fraction was increased from 4.60% at pretreatment to 5.68% at the end of infusion. These findings suggested that unbound AUC calculated by multiplying the unbound fraction at the end of infusion and AUC of total docetaxel in this study might overestimate the AUC of unbound docetaxel. Nonetheless, in the report by Loos et al. the unbound fraction calculated from concentrations at the end of infusion (6.37%) was similar to the unbound fraction calculated using the AUC of total and unbound drug (6.00%).(21) Consistent with these findings, in another study, the unbound fraction at the end of infusion (5.68%) was very similar to that calculated based on the AUC (5.46%).(23) Therefore, we believe that a slight change in the unbound fraction after infusion would not have a significant impact on the results of pharmacodynamic analyses in our study, although overestimation of unbound AUC was possible.

We observed a large interpatient variation in the unbound fraction of docetaxel, ranging from 3.1 to 10.5% in 90% of patients. In the report by Loos et al. where equilibrium dialysis was used, the unbound fraction of docetaxel in 23 patients also varied greatly, ranging from 3.9 to 10.3% with a median of 6.18%.(21) The large interpatient variability of the unbound fraction of docetaxel observed in the two studies indicates that measuring unbound concentrations is important in pharmacokinetic studies of this drug. The median value of the unbound fraction was 4.0% in our study and seemed slightly lower than that reported by Loos et al.(21) This small difference in the unbound fraction in the two studies may be explained by differences in concentrations of coinjected polysorbate 80 because lower doses of docetaxel were used in the present study (20–60 vs 75–85 mg/m2). It may also be related to differences in patient characteristics or different methods for the measurement of the unbound fraction.

In population pharmacokinetic and pharmacodynamic analysis of docetaxel using total drug concentrations, high serum levels of α1-acid glycoprotein were significantly associated with low clearance of total docetaxel as well as mild neutropenia and diminished antitumor efficacy.(20,24)α1-Acid glycoprotein is a major binding protein of docetaxel, and it is estimated that 29% of the docetaxel in plasma is bound to α1-acid glycoprotein whereas the fraction bound by albumin is 20%.(19) This may be considered to account for the correlation between serum levels of α1-acid glycoprotein and clearance of the drug.(21,24) However, patients with acute-phase reactions were suggested to have reduced activity of CYP3A4,(25,26) which may also explain the association between high serum levels of α1-acid glycoprotein and low clearance of docetaxel. With regard to the mechanism of the association between α1-acid glycoprotein levels and the toxicity or the antitumor activity in the population pharmacodynamic analysis,(20,31) high α1-acid glycoprotein levels might be associated with low concentrations of unbound docetaxel and, therefore, with milder toxicity but decreased antitumor activity. In this sense, the level of α1-acid glycoprotein might be a surrogate for unbound drug concentrations. However, in addition to the concentration of unbound docetaxel, α1-acid glycoprotein was also a determinant of toxicity in the logistic regression model in our study, which suggested that the level of α1-acid glycoprotein determines not only the pharmacokinetics but also the pharmacodynamics.

α1-Acid glycoprotein is an acute-phase protein and its expression is induced by cytokines such as tumor necrosis factor (TNF)α, which also induces various other cytokines including granulocyte-stimulating factors.(32–34) A positive correlation between α1-acid glycoprotein and pretreatment neutrophil counts was observed in our study. High levels of α1-acid glycoprotein might be a surrogate marker for the production of neutrophils, and therefore associated with resistance to neutropenia. However, we did not measure concentrations of TNFα or granulocyte colony-stimulating factor in this study to support the hypothesis. Further studies are necessary to elucidate the mechanism accounting for the observation that α1-acid glycoprotein was a significant predictive factor of neutropenia in addition to the concentration of unbound docetaxel.

In the multivariate logistic regression analysis, not only the Cmax of unbound docetaxel but also the serum concentration of α1-acid glycoprotein was selected as a determinant of grade 4 neutropenia. By measuring these concentrations, it might be possible to predict the probability of toxicity, but future studies will be necessary to confirm the validity of such strategies.

In conclusion, the pharmacokinetics of unbound docetaxel better accounted for the pharmacodynamics of the drug than did total docetaxel. Measuring concentrations of pharmacologically active unbound drug may be important in pharmacokinetic studies of docetaxel.

Acknowledgments

This study was supported in part by Grants-in-Aid from the Ministry of Health and Welfare.

Ancillary